chapter
stringlengths
1.97k
1.53M
path
stringlengths
47
241
Interest in the thermodynamic properties of binary liquid mixtures extended over most of the 19th Century. Interest was concerned with the vapour pressures of the components of a given binary mixture. Experimental work led Raoult to propose a law of partial vapour pressures which characterises a liquid mixture whose thermodynamic properties are ideal. If liquid 1 is a component of a binary liquid mixture, components liquid 1 and liquid 2, the mixture is defined as having ideal thermodynamic properties if the partial vapour pressure at equilibrium $\mathrm{p}_{1}(\text{mix})$ is related to the mole fraction composition using equation (a). $) p (mix) x p ( * 1 1 1 = ⋅ l (a) Here $\mathrm{p}_{1}^{*}(\ell)$ is the vapour pressure of pure liquid 1 at the same temperature. A similar equation describes the vapour pressure of liquid 2, $\mathrm{p}_{2}(\text{mix})$ in the mixture. Raoult’s Law as given in equation (a) forms the basis for examining the properties of real binary liquid mixtures. Binary liquid mixtures [1-13] are interesting in their own right. Nevertheless chemists also use binary liquid mixtures as solvents for chemical equilibria and for media in which to carry out chemical reactions between solutes. In fact an enormous amount of our understanding of the mechanisms of inorganic and organic reactions is based on kinetic data describing the rates of chemical reactions in 80/20 ethanol and water mixtures. Because so much experimental information in the literature concerns the properties of solutes in binary aqueous mixtures we concentrate our attention on these systems. Indeed the properties of such solvent systems cannot be ignored when considering the properties of solutes in these mixtures. In examining the properties of binary aqueous mixtures we adopt a convention in which liquid water is chemical substance 1 and the non-aqueous liquid component is chemical substance 2. For the most part we describe the composition of a given liquid mixture (at defined temperature and pressure) using the mole fraction scale. Then if $\mathrm{n}_{1}$ and $\mathrm{n}_{2}$ are the amounts of chemical substances 1 and 2, the mole fractions are defined by equation (b) [12,13]. \[\mathrm{x}_{1}=\mathrm{n}_{1} /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right) \quad \mathrm{x}_{2}=\mathrm{n}_{2} /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ The chemical potentials of each component in a given liquid mixture, mole fraction composition $\mathrm{x}_{2}\left(=1-\mathrm{x}_{1}\right)$ are compared with the chemical potential of the pure liquid chemical potential at the same temperature and pressure, $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$. The starting point is a description of the equilibrium at temperature $\mathrm{T}$ between pure liquid and its saturated vapour in a closed system; i.e. a two phase system. [In terms of the Gibbs phase rule the number of components = 1; the number of phases = 2. Hence the number of degrees of freedom = 1. Then at a specified temperature the vapour pressure is defined.] The equilibrium is described in terms of the equality of chemical potentials of substance 1 in the two phases; equation (c). $\mu_{1}\left(\mathrm{~g} ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)=\mu_{1}\left(\ell ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)$ Thus $\mu_{1}\left(\ell ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)$ is the molar Gibbs energy of the pure liquid 1, otherwise the chemical potential. However our interest concerns the properties of binary liquid mixtures. Using the Gibbs phase rule, number of phases = 2; number of components = 2; hence number of degrees of freedom = 2. Then for a defined temperature and mole fraction composition the vapour pressure $\mathrm{p}(\text{mix})$ is fixed. Equation (d) describes the equilibrium between liquid and vapour phases with reference to liquid substance 1. $\mu_{1}\left(\mathrm{~g} ; \mathrm{p}_{1} ; \mathrm{T}\right)=\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1} ; \mathrm{p} ; \mathrm{T}\right)$ Equations (c) and (d) offer a basis for comparing the chemical potentials of chemical substance 1 in the two phases, liquid and vapour. $\mu_{1}\left(\mathrm{~g} ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)-\mu_{1}\left(\mathrm{~g} ; \mathrm{p}_{1} ; \mathrm{T}\right)=\mu_{1}\left(\ell ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)-\mu_{1}\left(\mathrm{mix} ; \mathrm{x}_{1} ; \mathrm{p} ; \mathrm{T}\right)$ Analysis of vapour pressure data is not straightforward because account has to be taken of the fact that the properties of real gases are not those of an ideal gas. However here we assume that the properties of chemical substance in the gas phase are ideal. Consequently we use the following equation to provide an equation for the r.h.s. of equation (e) in terms of two vapour pressures. $\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1} ; \mathrm{p} ; \mathrm{T}\right)=\mu_{1}\left(\ell ; \mathrm{p}_{1}^{*} ; \mathrm{T}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{p}_{1} / \mathrm{p}_{1}^{*}\right)$ Equation (a) is combined with equation (f) to yield the following equation for chemical substance 1 as component of the binary liquid mixture. $\mu_{1}\left(\operatorname{mix} ; \mathrm{id} ; \mathrm{x}_{1}\right)=\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)$ Similarly for chemical substance 2 in the liquid mixture, $\mu_{2}\left(\operatorname{mix} ; \mathrm{id} ; \mathrm{x}_{2}\right)=\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)$ Hence for an ideal binary liquid mixture, formed by mixing $\mathrm{n}_{1}$ and $\mathrm{n}_{2}$ moles respectively of chemical substance 1 and 2, the Gibbs energy of the mixture is given by equation (i). \begin{aligned} &\mathrm{G}(\operatorname{mix} ; \mathrm{id})= \ &\quad \mathrm{n}_{1} \,\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right]+\mathrm{n}_{2} \,\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)\right] \end{aligned} Bearing in mind that Gibbs energies cannot be measured for either a pure liquid or a liquid mixture, it is useful to rephrase equation (i) in terms of the change in Gibbs energy that accompanies mixing to form an ideal binary liquid mixture. We envisage a situation where before mixing the molar Gibbs energy of the system defined as $\mathrm{G}(\text{no}-\text{mix})$ is given by equation (j). $\mathrm{G}(\mathrm{no}-\operatorname{mix})=\mathrm{n}_{1} \, \mu_{1}^{*}(\ell)+\mathrm{n}_{2} \, \mu_{2}^{*}(\ell)$ The change in Gibbs energy on forming the ideal binary liquid mixture $\Delta_{\text{mix}} \mathrm{G}$ is given by the equation (k). $\Delta_{\text {mix }} \mathrm{G}(\mathrm{id})=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{n}_{1} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{n}_{2} \, \ln \left(\mathrm{x}_{2}\right)\right]$ We re-express $\Delta_{\text{mix}} \mathrm{G}$ in terms of the Gibbs energy of mixing forming one mole of the ideal binary liquid mixture. $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}=\Delta_{\text {mix }} \mathrm{G} /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ Hence, $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{2}\right)\right]$ As required, $\operatorname{limit}\left(\mathrm{x}_{1} \rightarrow 0\right) \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})=0$ $\operatorname{limit}\left(x_{2} \rightarrow 0\right) \Delta_{\text {mix }} G_{m}(\text { id })=0$ The dependence of $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{R} \, \mathrm{T}$ on mole fraction composition is defined by equation (m). For a mixture where $x_{1}= x_{2}=0.5$, $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{R} \, \mathrm{T}=2.0 \times 0.5 \times \ln (0.5)=-0.693$ In fact the molar Gibbs energy of mixing for an ideal binary liquid mixture is negative across the whole composition range. Equation (m) is the starting point of most equations used in the analysis of the properties of binary liquid mixtures. In fact most of the chemical literature concerned with liquid mixtures describes the properties of aqueous mixtures, at ambient pressure and $298.15 \mathrm{~K}$. Nevertheless an extremely important subject concerns the properties of liquid mixtures at high pressures [14,15]. Footnotes [1] K. N. Marsh, Pure Appl. Chem., 1983, 55, 467. [2] K. N. Marsh, Annu. Rep. Prog. Chem., Sect. C, Phys. Chem., 1994,91,209. [3] J. S. Rowlinson and F. L. Swinton, Liquids and Liquid Mixtures, Butterworths, 3rd. edn., London, 1982. [4] G. Scatchard, Chem.Rev., 1931,8,321;1949, 44,7. [5] E. A. Guggenheim, Liquid Mixtures, Clarendon Press, Oxford, 1952. [6] J. H. Hildebrand and R. L. Scott, Solubility of Non-Electrolytes, Reinhold, New York,3rd. edn.,1950. [7] J. H. Hildebrand, J. M. Prausnitz and R. L. Scott, Regular and Related Solutions, van Nostrand Reinhold, New York,1970. [8] A. G. Williamson, An Introduction to Non-Electrolyte Solutions, Oliver and Boyd, Edinburgh, 1967. [9] Y. Koga, J.Phys.Chem.,1996,100,5172. [10] L. S. Darken, Trans. Metallurg. Soc., A.M.E., 1967,239,80. [11] A. D. Pelton and C. W Bale, Metallurg. Trans.,!986,17A,211. [12] According to Phase Rule, P = 2 (for liquid and vapour), C = 2 then F = 2 + 2 - 2 = 2. Having defined temperature and pressure there remain no degrees of freedom - the system is completely specified. [13] Other methods of defining the composition include the following. Mass % [or w%] Mass of component $1=\mathrm{n}_{1} \, \mathrm{M}_{1}$ Mass of component $2=\mathrm{n}_{2} \, \mathrm{M}_{2}$ Then $\mathrm{w}_{1} \%=\frac{\mathrm{n}_{1} \, \mathrm{M}_{1} \, 100}{\left[\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{2} \, \mathrm{M}_{2}\right]} \quad \mathrm{w}_{2} \%=100-\mathrm{w}_{1} \%$ Volume % This definition often starts out by defining the volumes of the two liquid components used to prepare a given mixture at defined temperature and pressure. The definition does not normally refer to the volume of the actual mixture. The volume after mixing is often less than the sum of the component volumes before mixing. \begin{aligned} &\mathrm{V}_{1}^{*}(\ell)=\mathrm{n}_{1} \, \mathrm{M}_{1} / \rho_{1}^{*}(\ell) \quad \mathrm{V}_{2}^{*}(\ell)=\mathrm{n}_{2} \, \mathrm{M}_{2} / \rho_{2}^{*}(\ell) \ &\mathrm{V}_{2} \%=\frac{\mathrm{V}_{2}^{*}(\ell)}{\mathrm{V}_{1}^{*}(\ell)+\mathrm{V}_{2}^{*}(\ell)} \end{aligned} [14] G. Schneider, Pure Appl. Chem.,1983,55,479; 1976,47,277. [15] G. Schneider, Ber. Bunsenges, Phys.Chem.,1972,76,325.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.26%3A_Gibbs_Energies-_Binary_Liquid_Mixtures.txt
A given binary liquid mixture ( at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, which is close to the standard pressure) mole fraction $x_{1}\left(=1-x_{2}\right)$ can be characterised by the molar Gibbs energy of mixing, $\Delta_{\text{mix}} \mathrm{~G}_{\mathrm{m}}$ and related molar enthalpic, volumetric and entropic properties. A corresponding set of properties exists for this mixture granted that the thermodynamic properties are ideal; e.g. $\Delta_{\text {mix }} G_{m}(\mathrm{id})$. Hence we can define the corresponding excess molar property, $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}-\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}} \text { (id) }$. Interesting patterns emerge relating these properties and the corresponding partial molar properties; e.g. chemical potentials. Ideal Mixing Properties At defined $\mathrm{T}$ and $\mathrm{p}$, the molar Gibbs energy for an ideal binary liquid mixture is given by equation (a). $\mathrm{G}_{\mathrm{m}}(\mathrm{id})=\mathrm{x}_{1} \,\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right]+\mathrm{x}_{2} \,\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)\right]$ Here $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquids 1 and 2 at the same $\mathrm{T}$ and $\mathrm{p}$. If the same amounts of the two liquid had not been allowed to mix, $\mathrm{G}_{\mathrm{m}}(\text { no }-\operatorname{mix})=\mathrm{x}_{1} \, \mu_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mu_{2}^{*}(\ell)$ By definition, $\Delta_{\text {mix }} G_{m}(\mathrm{id})=x_{1} \, R \, T \, \ln \left(x_{1}\right)+x_{2} \, R \, T \, \ln \left(x_{2}\right)$ Recalling that $x_{1}+x_{2}=1$ and $dx_{1}=-d x_{2}$, $\mathrm{d} \Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)-\mathrm{R} \, \mathrm{T}+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)+\mathrm{R} \, \mathrm{T}$ We use equation (c) for $\ln \left(x_{2}\right)$ and substitute in equation (d). \begin{aligned} &\mathrm{d} \Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}= \ &-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)+\left(1 / \mathrm{x}_{2}\right) \,\left[\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})-\mathrm{x}_{1} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right] \end{aligned} \begin{aligned} &\mathrm{x}_{2} \, \mathrm{d} \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}= \ &\left.-\mathrm{x}_{2} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)+\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})-\mathrm{x}_{1} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right] \end{aligned} Hence, $\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)=\Delta_{\mathrm{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})-\mathrm{x}_{2} \, \mathrm{d} \Delta_{\mathrm{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}$ At all mole fractions, $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})$ is negative, the plot of $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})$ against $x_{1}$ being symmetric about ‘$x_{1}=0.5$’. At the extreme, where $\mathrm{d} \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}$ is zero; equation (c) shows that $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) \text { equals } [\mathrm{R} \, \mathrm{T} \, \ln (0.5)]$. At $298 \mathrm{~K}$, the latter quantity equals $– 1.72 \mathrm{~kJ mol}^{-1}$. Using equation (c), the ratio $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{T}$ is given by equation (h) $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{T}=\mathrm{R} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{2}\right)\right]$ Hence using the Gibbs-Helmholtz it follows that the molar enthalpy of mixing is zero at all mole fractions. Similarly the molar isobaric heat capacity of mixing is zero at all mole fractions. In terms of entropies, $\Delta_{\operatorname{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})=-\mathrm{R} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{2}\right)\right]$ But at all (real) mole fractions (other than $x_{1} =1 \text { and } x_{2} = 1$) $\left[x_{1} \, \ln \left(x_{1}\right)+x_{2} \, \ln \left(x_{2}\right)\right]<0$. Hence $\Delta_{\operatorname{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})>0$. In the event that mixing of two liquids at temperature $\mathrm{T}$ to produce a binary liquid mixture having ideal properties, then $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})$ is negative because $\Delta_{\operatorname{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})$ is positive, $\Delta_{\text {mix }} \mathrm{H}_{\mathrm{m}}(\mathrm{id})$ being zero. In other words we have a reference against which to examine the properties of real liquid mixtures. Excess properties The molar Gibbs energy of a real binary liquid mixture is related to the mole fraction composition using equation (j). \begin{aligned} &\mathrm{G}_{\mathrm{m}}= \ &\mathrm{x}_{1} \,\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)\right]+\mathrm{x}_{2} \,\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)\right] \end{aligned} For the corresponding mixture having ideal thermodynamic properties, \begin{aligned} &\mathrm{G}_{\mathrm{m}}(\mathrm{id})= \ &\mathrm{x}_{1} \,\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right]+\mathrm{x}_{2} \,\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)\right] \end{aligned} The excess molar Gibbs energy ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is therefore given by equation (l). $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{f}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{f}_{2}\right)\right]$ We differentiate equation (l) with respect to mole fraction, $x_{1}$. $\frac{1}{R \, T} \, \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}=\ln \left(\mathrm{f}_{1}\right)+\mathrm{x}_{1} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{d \mathrm{x}_{1}}-\ln \left(\mathrm{f}_{2}\right)+\mathrm{x}_{2} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{2}\right)}{\mathrm{dx}_{1}}$ But from the Gibbs-Duhem equation at fixed $\mathrm{T}$ and $\mathrm{p}$, $x_{1} \, \frac{d \ln \left(\mu_{1}\right)}{d x_{1}}+x_{2} \, \frac{d \ln \left(\mu_{2}\right)}{d x_{1}}=0$ Hence, $x_{1} \, \frac{d \ln \left(f_{1}\right)}{d x_{1}}+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{1}}=0$ From equation (k), $\ln \left(f_{2}\right)=\ln \left(f_{1}\right)-\frac{1}{R \, T} \, \frac{d G_{m}^{E}}{d x_{1}}$ Hence using equation (j), $\frac{\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{R} \, \mathrm{T}}=\ln \left(\mathrm{f}_{1}\right)-\frac{\mathrm{X}_{2}}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}}$ Or $\ln \left(\mathrm{f}_{1}\right)=\frac{\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{R} \, \mathrm{T}}=+\frac{\mathrm{x}_{2}}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ Equation (o) has an interesting feature. At the mole fraction composition where $\frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ is zero, $\frac{\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{R} \, \mathrm{T}}$ offers a direct measure of $\ln \left(f_{1}\right)$ at that mole fraction. Perhaps the most direct measure of the extent to which the thermodynamic properties of a given binary liquid mixture differs from that defined as ideal is afforded by the molar enthalpy of mixing which is therefore the excess molar enthalpy of mixing ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}$. Thus, $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}-\mathrm{T} \, \mathrm{S}_{\mathrm{m}}^{\mathrm{E}}$ where $\mathrm{C}_{\mathrm{pmn}}^{\mathrm{E}}=\left(\partial \mathrm{H}_{\mathrm{m}}^{\mathrm{E}} / \partial \mathrm{T}\right)_{\mathrm{p}}$ In many reports the properties of a given binary liquid mixture at $298.15 \mathrm{~K}$ and ambient pressure are summarised in a plot showing the three properties ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$, ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}$ and $\mathrm{T} \, \mathrm{S}_{\mathrm{m}}^{\mathrm{E}}$ as a function of the mole fraction composition. An enormous amount of information can be summarised in such plots. In order to understand the various patterns which emerge two liquid mixtures are often taken as models against which to compare the properties of other liquid mixtures. 1. Trichloromethane + Methanol For this mixture both ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$ and ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}$ are negative, their minima being at approx. mole fractions at 0.5. The pattern in understood in terms of strong inter-component interaction, hydrogen bonding. In these terms the mixing is ‘favourable’ and exothermic. 2. Tetrachloromethane + Methanol For this mixture the mixing is, for the most part endothermic and ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$ is positive with a maximum at mole fractions equal to 0.5. Thus the mixing is unfavourable and endothermic. This pattern points to the impact of added tetrachloromethane disrupting the intermolecular hydrogen bonding between methanol molecules. These two liquid mixtures provide a basis for the examination of the properties of binary aqueous mixtures for which there is an immense published information. In most cases ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$ is either positive or negative across the mole fraction range for a given liquid mixture although plots of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$ and $\mathrm{T} \, \mathrm{S}_{\mathrm{m}}^{\mathrm{E}}$ against mole fraction composition are often S-shaped, nevertheless operating to produce a smooth change in ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$. Plots of ${\mathrm{C}_{\mathrm{pm}}}^{\mathrm{E}}$ and excess molar volume of mixing ${\mathrm{V}_{\mathrm{m}}}^{\mathrm{E}}$ against mole fraction are often quite complicated. In a few cases the plot of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}$ against composition is S-shaped. One such system is the mixture, water + 1,1,1,3,3,3-hexafluoropropanol [1]. Explanations of such complex patterns are not straightforward. However we might for an alcohol + water mixture envisage a switch from 1. at low $x_{2}$ strong water-water interactions with weak alcohol-water interactions to 2. at high $x_{2}$ strong alcohol-water interactions. Definition of excess thermodynamic properties is not straightforward in all instances; e.g. isentropic compressibilities.[2] Footnotes [1] M. J. Blandamer, J. Burgess, A. Cooney, H. J. Cowles, I. M. Horn, K. J. Martin, K. W. Morcom and P. Warwick, J. Chem. Soc. Faraday Trans.,1990,86,2209` [2] G. Douheret, C . Moreau and A. Viallard, Fluid Phase Equilibrium, 1985,22 ,277; 289.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.27%3A_Gibbs_Energies-_Binary_Liquid_Mixtures-_General_Properties.txt
A given binary liquid mixture (at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, which is close to the standard pressure) mole fraction $x_{1} \left(= 1 - x_{2}\right)$ is characterised by the molar Gibbs energy of mixing, $\Delta_{\text{mix}}\mathrm{G}_{\mathrm{m}}$ and related molar enthalpic, volumetric and entropic properties. A corresponding set of properties for this mixture exist granted that the thermodynamic properties are ideal; e.g. $\Delta_{\text{mix}}\mathrm{G}_{\mathrm{m}}(\text{id})$. As a consequence we define the corresponding excess molar property, $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}-\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}} \text { (id) }$. Interesting patterns emerge relating these properties and the corresponding partial molar properties; e.g. chemical potentials. For chemical substance 1, (which we will conventionally take as water) the chemical potential in the liquid mixture is related to the chemical potential of the pure liquid at the same $\mathrm{T}$ and using equation (a) where $x_{1}$ is the mole fraction and $\mathrm{f}_{1}$ is the rational activity coefficient. $\mu_{1}(\operatorname{mix})=\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ where $\operatorname{limit}\left(\mathrm{x}_{1} \rightarrow 1\right) \mathrm{f}_{1}=1.0 \text { at all T and } p$ Similarly for component 2, $\mu_{2}(\operatorname{mix})=\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)$ where $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 1\right) \mathrm{f}_{2}=1.0 \text { at all } \mathrm{T} \text { and } \mathrm{p}$ Here $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquids at the same $\mathrm{T}$ and $\mathrm{p}$; $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$ are rational activity coefficients. These (rational) activity coefficients approach unity at opposite ends of the mixture composition range. For the aqueous component, as $x_{1}$ approaches 1, so $\mathrm{f}_{1}$ approaches unity (at the same $\mathrm{T}$ and $\mathrm{p}$). At the other end of the scale, as $x_{1}$ approaches zero so the chemical potential of water in the binary system approaches ‘minus infinity’. If across the whole composition range (at all $\mathrm{T}$ and $\mathrm{p}$), both $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$ are unity, the thermodynamic properties of the liquid mixture are ideal. A given liquid mixture (at fixed $\mathrm{T}$ and $\mathrm{p}$) is formed by mixing $\mathrm{n}_{1}$ moles of liquid 1 and $\mathrm{n}_{2}$ moles of liquid 2. Before mixing the total Gibbs energy of the system, defined as G(no-mix) is given by the following equation where $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquids at the same $\mathrm{T}$ and $\mathrm{p}$. Then, $\mathrm{G}(\mathrm{no}-\operatorname{mix})=\mathrm{n}_{1} \, \mu_{1}^{*}(\ell)+\mathrm{n}_{2} \, \mu_{2}^{*}(\ell)$ After mixing, the Gibbs energy of the mixture is given by equation (f). \begin{aligned} &\mathrm{G}(\operatorname{mix})= \ &\quad \mathrm{n}_{1} \,\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)\right]+\mathrm{n}_{2} \,\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)\right] \end{aligned} By definition, $\Delta_{\operatorname{mix}} \mathrm{G}=\mathrm{G}(\operatorname{mix})-\mathrm{G}(\mathrm{no}-\mathrm{mix})$ Hence the Gibbs energy of mixing , $\Delta_{\text {mix }} \mathrm{G}=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{n}_{1} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)\right]+\left[\mathrm{n}_{2} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)\right]$ We re-express $\Delta_{\text{mix}}\mathrm{G}$ in terms of the Gibbs energy of mixing for one mole of liquid mixture. Thus $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}=\Delta_{\text {mix }} \mathrm{G} /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ Hence, $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)\right]$ Or, $\Delta_{\text {mix }} G_{m}=R \, T \,\left[x_{1} \, \ln \left(x_{1}\right)+x_{1} \, \ln \left(f_{1}\right)+x_{2} \, \ln \left(x_{2}\right)+x_{2} \, \ln \left(f_{2}\right)\right]] By definition, \[\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})=\mathrm{x}_{1} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{2} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)$ Recalling that $x_{1} + x_{2} = 1$ and $dx_{1} = -dx_{2}$, $\mathrm{d} \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)-\mathrm{R} \, \mathrm{T}+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)+\mathrm{R} \, \mathrm{T}$ We use equation (l) for $\ln \left(\mathrm{x}_{2}\right)$ and substitute in equation (m). \begin{aligned} &\mathrm{d} \Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}= \ &-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)+\left(1 / \mathrm{x}_{2}\right) \,\left[\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})-\mathrm{x}_{1} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)\right] \end{aligned} or, \begin{aligned} &x_{2} \, d \Delta_{\text {mix }} G_{m}(\mathrm{id}) / d x_{2}= \ &\left.-x_{2} \, R \, T \, \ln \left(x_{1}\right)+\Delta_{\text {mix }} G_{m} \text { (id) }-x_{1} \, R \, T \, \ln \left(x_{1}\right)\right] \end{aligned} Hence, $\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)=\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})-\mathrm{x}_{2} \, \mathrm{d} \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}$ At all mole fractions, $\Delta_{\operatorname{mix}} G_{m}(\text { id })$ is negative; the plot of $\Delta_{\text {mix }} G_{m} \text { (id) }$ against $x_{1}$ is symmetric about ‘$x_{1} = 0.5$’. At the extremum, where $\mathrm{d} \Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{dx} \mathrm{x}_{2}$ is zero, equation (l) shows that $\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}(\mathrm{id})$ equals $\mathrm{R} \, \mathrm{T} \, \ln (0.5)$. We define an excess chemical potential for each of the two components of a binary liquid mixture. For liquid component 1, $\mu_{1}^{\mathrm{E}}(\mathrm{mix})=\mu_{1}(\mathrm{mix})-\mu_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{f}_{1}\right)$ Similarly for liquid component 2, $\mu_{2}^{\mathrm{E}}(\mathrm{mix})=\mu_{2}(\mathrm{mix})-\mu_{2}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{f}_{2}\right)$ Rational activity coefficients $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$ depend on mixture composition, $\mathrm{T}$ and $\mathrm{p}$. In summary ‘excess’ means excess over ideal. Excess properties provide a mutually consistent set of perspectives of a given liquid mixture. We define a reference state for binary liquid mixtures so that the thermodynamic properties of a given liquid mixture can be correlated with a common model. A convenient approach defines an excess molar Gibbs energy of mixing. $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}}-\Delta_{\operatorname{mix}} \mathrm{G}_{\mathrm{m}} \text { (id) }$ Then $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{x}_{1} \, \ln \left(\mathrm{f}_{1}\right)+\mathrm{x}_{2} \, \ln \left(\mathrm{f}_{2}\right)\right]$ Where $\operatorname{limit}\left(x_{1} \rightarrow 1\right) \mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=0$ And $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 1\right) \mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=0$ Other than the latter two conditions we cannot predict the dependence of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ on a mixture composition for m a given real mixture.[1] In general terms excess molar properties of binary aqueous mixtures are expressed in terms of the following general equation with respect to the thermodynamic variable $\mathrm{Q} (= \mathrm{~G}, \mathrm{~V}, \mathrm{~H} \text { and } \mathrm{S})$. $\mathrm{Q}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{Q}_{\mathrm{m}}(\text { mix })-\mathrm{Q}_{\mathrm{m}}(\text { mix } ; \text { ideal })$ Returning to the Gibbs energies, we differentiate equation (t) with respect to mole fraction $x_{1}$ at fixed $\mathrm{T}$ and $\mathrm{p}$. $\frac{1}{R \, T} \, \frac{d G_{m}^{E}}{d x_{1}}=\ln \left(f_{1}\right)+x_{1} \, \frac{d \ln \left(f_{1}\right)}{d x_{1}}-\ln \left(f_{2}\right)+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{1}}$ According to the Gibbs-Duhem equation, at fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{x}_{1} \, \frac{\mathrm{d} \ln \left(\mu_{1}\right)}{\mathrm{dx}_{1}}+\mathrm{x}_{2} \, \frac{\mathrm{d} \ln \left(\mu_{2}\right)}{\mathrm{dx}}=0$ Hence, $x_{1} \, \frac{d \ln \left(f_{1}\right)}{d x_{1}}+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{1}}=0$ From equation (x), $\ln \left(f_{2}\right)=\ln \left(f_{1}\right)-\frac{1}{R \, T} \, \frac{d G_{m}^{E}}{d x_{1}}$ Hence using equation (t), $\frac{\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{R} \, \mathrm{T}}=\ln \left(\mathrm{f}_{1}\right)-\frac{\mathrm{x}_{2}}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ Or $\ln \left(f_{1}\right)=\frac{G_{m}^{E}}{R \, T}+\frac{x_{2}}{R \, T} \, \frac{d G_{m}^{E}}{d x_{1}}$ Equation (zc) has an interesting feature. At the mole fraction composition where $\frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ is zero, $\frac{\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{R} \, \mathrm{T}}$ offers a direct measure of $\ln \left(\mathrm{f}_{1}\right)$ at that mole fraction. In some systems the plot of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ against composition is S-shaped so we have this information at two mole fractions. Turning to volumetric properties, the molar volume of an ideal binary liquid mixture is given by equation (zd). $\mathrm{V}_{\mathrm{m}}=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ Hence, $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \,\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]+\mathrm{x}_{2} \,\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ \begin{aligned} &\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx} \mathrm{x}_{1}}= \ &{\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]+\mathrm{x}_{1} \, \frac{\mathrm{dV}_{1}}{\mathrm{dx}_{1}}-\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]+\mathrm{x}_{2} \, \frac{\mathrm{dV}_{2}}{\mathrm{dx}_{1}}} \end{aligned} Using the Gibbs-Duhem equation, $\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}=\left[\mathrm{V}_{1}(\operatorname{mix})-\mathrm{V}_{1}^{*}(\ell)\right]-\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ or, $\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]=\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx} \mathrm{x}_{1}}+\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \, \frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx} \mathrm{x}_{1}}+\mathrm{x}_{1} \,\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]+\mathrm{x}_{2} \,\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ Hence, $\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]=\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}-\mathrm{x}_{1} \, \frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx} \mathrm{x}_{1}}$ At the composition where $\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ is zero, $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]=\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$. The analysis set out above is repeated for excess molar enthalpies and excess molar isobaric heat capacities for binary liquid mixtures. Interesting proposals have been made in which the dependence of excess thermodynamic properties on mixture composition are examined in different composition domains; e.g. the four segment model [2-9]. Footnotes [1] R. Schumann, Metallurg. Trans.,B,1985,16B,807. [2] M. I. Davis, M. C. Molina and G. Douheret, Thermochim. Acta, 1988,131,153. [3] M. I. Davis, Thermochim. Acta 1984,77,421; 1985,90, 313; 1987, 120,299; and references therein. [4] G. Douheret, A. H. Roux, M. I .Davis, M. E. Hernandez, H. Hoiland and E. Hogseth, J. Solution Chem.,1993,22,1041. [5] G. Douheret, C. Moreau and A. Viallard, Fluid Phase Equilib.,1986,26,221. [6] G. Douheret, A. Pal and M. I. Davis, J.Chem.Thermodyn., 1990,22,99. [7] H. Hoiland, O. Anowi and M. I. Davis, J. Chem. Thermodyn., 1991,23,569. [8] G. Douheret, A. Pal, H. Hoiland, O. Anowi and M. I. Davis, J. Chem. Thermodyn., 1991,23,569. [9] G. Douheret, J.C.R. Reis, M. I. Davis, I. J. Fjellanger and H. Hoiland, Phys.Chem. Chem.Phys.,2004,6,784.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.28%3A_Gibbs_Energies-_Liquid_Mixtures-_Thermodynamic_Patterns.txt
The properties of binary mixtures are complicated. As a point of reference the excess molar properties of two non-aqueous binary liquid mixtures are often discussed. The mixtures are (A) trichloromethane + propanone, and (B) tetrachloromethane + methanol. A snapshot of the thermodynamic properties of a given binary mixture (at fixed $\mathrm{T}$ and $\mathrm{p}$) is provided by combined plots of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$, ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ and $\mathrm{T} \, \mathrm{S}_{\mathrm{m}}^{\mathrm{E}}$ as a function of mixture composition [1-5]. In effect the starting point is the Gibbs energy leading to first, second, third and fourth derivatives [6]. At this stage we make some sweeping (and dangerous) generalizations. For most binary aqueous mixtures, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is a smooth function of water m mole fraction $x_{1}$, with an extremum near $x_{1} = 0.5$. Rarely for a given mixture does the sign of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ change across the mole faction range m although this feature is not unknown; e.g. water + 1,1,1,3,3,3- hexafluropropan-2-ol mixtures at $298.15 \mathrm{~K}$ [7] but contrast water + 2,2,2- trifluorethanol mixtures [8] where at $298.2 \mathrm{~K} {\mathrm{~G}_{\mathrm{m}}}^{\mathrm{E}}$ is positive across the m whole mole fraction range. However a change in sign of ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ and $\mathrm{T} \, {\mathrm{S}_{\mathrm{m}}}^{\mathrm{E}}$ and ${\mathrm{V}_{\mathrm{m}}}^{\mathrm{E}}$ with change in mole fraction composition is quite common. For mixture A, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is negative indicating that $\Delta_{\text{mix}}\mathrm{G}_{\mathrm{m}}$ is more negative than in the case of an ideal binary liquid mixture. In the case of Mixture A, the negative ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is linked with a marked exothermic mixing; ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}} < 0$. The latter is attributed to strong inter-component hydrogen bonding. For both mixtures A and B the signs of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ and ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ are the same. This feature is characteristic of binary non-aqueous liquid mixtures where in most instances, $\left|\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right|>\left|\mathrm{T} \, \mathrm{S}_{\mathrm{mm}}^{\mathrm{E}}\right|$. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ and ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ are both positive for mixture B. Here the pattern is understood in terms of disruption of methanol-methanol hydrogen bonding ( i.e. intracomponent interaction) by the second component. Again we note that a positive ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ means that the tendency for $\Delta_{\text{mix}}\mathrm{G}_{\mathrm{m}}$ to be negative (cf. ideal mixtures) is opposed. Through a series of mixtures with increasing ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$, a stage is reached where the magnitude of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is such that phase separation occurs [1]. For m many binary non-aqueous binary liquid mixtures the phase diagram for liquid miscibility has an upper critical solution temperature UCST. In other words only at high temperatures is the liquid mixture miscible in all proportions. Often binary aqueous mixtures are used as solvents for the following reason. The solubilities of salts in water(l) are high because ‘water is a polar solvent’ but the solubilities of apolar solutes are low. However the solubilities of apolar substances in organic solvents (e.g. ethanol) are high. If the chemical reaction being studied involves both polar and apolar solutes, judicious choice of the composition of a binary aqueous mixture leads to a solvent where the solubilities of both polar and apolar solutes are high. Nevertheless the task of accounting for the properties of binary aqueous mixtures is awesome. For this reason the classification introduced by Franks [9] has considerable merit. A distinction is drawn between Typically Aqueous and Typically Non-Aqueous Binary Aqueous Mixtures, based on the the thermodynamic excess functions, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$, ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ and $\mathrm{T} \, {\mathrm{S}_{\mathrm{m}}}^{\mathrm{E}}$. Davis has explored how the properties of many binary aqueous mixtures can be subdivided on the basis of the ranges of mole fraction compositions [10]. Footnotes [1] J. S. Rowlinson and F. L. Swinton, Liquids and Liquid Mixtures, Butterworths, London, 3rd. edn., 1982. [2] K. N. Marsh, Annu. Rep. Prog. Chem., Sect. C, Phys. Chem.,1984, 81 , 209-245; Pure Appl. Chem.,1983,55,467. [3] G. Scatchard. Chem. Rev.,1931,8,321. [4] G. Scatchard, Chem. Rev.,1940, 44,7. [5] With respect to compressibilities; G. Douheret, C. Moreau and A. Viallard, Fluid Phase Equilib., 1985, 22,289. [6] Y. Koga, J.Phys.Chem.,1996,100,5172; Y. Koga, K. Nishikawa and P. Westh, J. Phys. Chem.A,2004,108,3873. [7] 1. M. J. Blandamer, J. Burgess, A. Cooney, H. J. Cowles, I. M. Horn, K. J. Martin, K. W. Morcom and P. Warrick, J. Chem. Soc. Faraday Trans.,1990,86,2209. 2. A. Kivinen, J. Murto and A. Viit, Suomen Kemist.,Sect. B,1967,40,298. [8] R. Jadot and M. Fraiha, J. Chem. Eng. Data, 1988, 33,237. [9] F. Franks, in Hydrogen –Bonded Solvent Systems, ed. A. K. Covington and P. Jones, Taylor and Francis, London, 1968, pp.31-47. [10] 1. M. I. Davis, Thermochim Acta, 1984,77,421; 1985,90,313; 1987,120,299. 2. M. I. Davis, M. C. Molina and G. Douheret.1988, 131, 153. 3. G. Douheret, A. Pal and M. I. Davis, J. Chem. Thermodyn., 1990, 22, 99. 4. G. Douheret, A. Pal, H. Hoiland, O. Anowi and M. I. Davis, J. Chem. Thermodyn., 1991, 23,569. 5. G. Douheret, A. H. Roux, M. I. Davis, M. E. Hernandez, H. Hoiland and E. Hogseth, J. Solution Chem.,1993,22,1041. 6. G. Douheret, C. Moreau and A. Viallard, Fluid Phase Equilib., 1985, 22, 277, 289.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.29%3A_Gibbs_Energies-_Binary_Liquid_Mixtures-_Excess_Thermodynamic_Variab.txt
For binary liquid mixtures at fixed $\mathrm{T}$ and $\mathrm{p}$, an important task is to fit the dependence of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ on $x_{2}$ to an equation in order to calculate the derivative $\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{2}$ at required mole fractions. The Guggenheim - Scatchard [1,2] (commonly called the Redlich - Kister [3] ) equation is one such equation. This equation has the following general form. $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{2} \,\left(1-\mathrm{x}_{2}\right) \, \sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{k}} \mathrm{A}_{\mathrm{i}} \,\left(1-2 \, \mathrm{x}_{2}\right)^{\mathrm{i}-1}$ $\mathrm{A}_{\mathrm{i}}$ are coefficients obtained from a least squares analysis of the dependence of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ on $x_{2}$.The equation clearly satisfies the condition that ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is zero at $x_{2} = 0$ and at $x_{2} = 1$. In fact the first term in the $\mathrm{G} - \mathrm{S}$ equation has the following form. $X_{m}^{E}=X_{2} \,\left(1-X_{2}\right) \, A_{1}$ According to equation (b) ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ is an extremum at $x_{2} = 0.5$, the plot being symmetric about the line from ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ to ‘$x_{2} = 0.5$’. In fact for most systems the $\mathrm{A}_{1}$ term is dominant. For the derivative $\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx} \mathrm{x}_{2}$, we write equation (a) in the following general form. $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\left(\mathrm{x}_{2}-\mathrm{x}_{2}^{2}\right) \, \mathrm{Q}$ Then $\mathrm{dX}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx} \mathrm{x}_{2}=\mathrm{x}_{2} \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{dQ} / \mathrm{dx} \mathrm{x}_{2}+\left(1-2 \, \mathrm{x}_{2}\right) \, \mathrm{Q}$ where $\mathrm{dQ} / \mathrm{dx}_{2}=-2 \, \sum_{\mathrm{i}=2}^{\mathrm{i}=\mathrm{k}}(\mathrm{i}-1) \, \mathrm{A}_{\mathrm{i}} \,\left(1-2 \, \mathrm{x}_{2}\right)^{\mathrm{i}-2}$ Equation (a) fits the dependence with a set of contributing curves which all pass through points, ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}=0$ at $x_{1} = 0$ and $x_{1} =1$. The usual procedure involves fitting the recorded dependence using increasing number of terms in the series, testing the statistical significance of including a further term. Although equation (a) has been applied to many systems and although the equation is easy to incorporate into computer programs using packaged least square and graphical routines, the equation suffers from the following disadvantage. As one incorporates a further term in the series, (e.g. $\mathrm{A}_{j}$) estimates of all the previously calculated parameters (i.e. $\mathrm{A}_{2}, \mathrm{~A}_{3 \ldots} \ldots \mathrm{A}_{\mathrm{j}-1}$) change. For this reason orthogonal polynomials have been increasingly favoured especially where the appropriate computer software is available. The only slight reservation is that derivation of explicit equations for the required derivative $\mathrm{d}\mathrm{X}_{\mathrm{m}}{ }^{\mathrm{E}}$ is not straightforward. The problem becomes rather more formidable when the second and higher derivatives are required. The derivative $\mathrm{d}^{2} \mathrm{X}_{\mathrm{m}}{ }^{\mathrm{E}}$ is sometimes required by calculations concerning the properties of binary liquid mixtures. The derivative $\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx} \mathrm{x}_{1}$ and ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ are combined to yield an equation for $\ln\left(\mathrm{f}_{1}\right)$. $\ln \left(f_{1}\right)=\frac{G_{m}^{E}}{R \, T}+\frac{\left(1-x_{1}\right)}{R \, T} \, \frac{d G_{m}^{E}}{d x_{1}}$ A similar equation leads to estimates of $\ln\left(\mathrm{f}_{2}\right)$. Hence the dependences are obtained of both $\ln\left(\mathrm{f}_{1}\right)$ and $\ln\left(\mathrm{f}_{2}\right)$ on mixture composition. It is of interest to explore the case where the coefficients $\mathrm{A}_{2}, \mathrm{~A}_{3} \ldots$ in equation (a) are zero. Then $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{2} \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{A}_{1}$ and $\mathrm{dX}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx_{2 }}=\left(1-2 \, \mathrm{X}_{2}\right) \, \mathrm{A}_{1}$ With reference to the Gibbs energies, $\ln \left(\mathrm{f}_{2}\right)=(1 / \mathrm{R} \, \mathrm{T}) \,\left[\mathrm{x}_{2} \,\left(1-\mathrm{x}_{2}\right)+\left(1-\mathrm{x}_{2}\right) \,\left(1-2 \, \mathrm{x}_{2}\right)\right] \, \mathrm{A}_{1}^{\mathrm{G}}$ $\ln \left(\mathrm{f}_{2}\right)=\left(\mathrm{A}_{1}^{\mathrm{G}} / \mathrm{R} \, \mathrm{T}\right) \,\left[1-2 \, \mathrm{x}_{2}+\mathrm{x}_{2}^{2}\right]$ or, $\ln \left(\mathrm{f}_{2}\right)=\left(\mathrm{A}_{1}^{\mathrm{G}} / \mathrm{R} \, \mathrm{T}\right) \,\left[1-\mathrm{x}_{2}\right]^{2}$ In fact the equation reported by Jost et al. [4] has this form. Rather than using the Redlich-Kister equation, recently attention has been directed to the Wilson equation [5] written in equation (l) for a two-component liquid [6]. $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{R} \, \mathrm{T}=-\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)-\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{2}+\Lambda_{21} \, \mathrm{x}_{1}\right)$ Then , for example [7], $\ln \left(\mathrm{f}_{1}\right)=-\ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)+\mathrm{x}_{2} \,\left(\frac{\Lambda_{12}}{\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}}-\frac{\Lambda_{21}}{\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}}\right)$ The Wilson equation forms the basis for two further developments, described as the NRTL (non-random, two-liquid) equation [8-10] and the UNIQUAC equation [9-10]. Nevertheless Douheret et al. [11] show how an excess property must be carefully defined. Davis et al. have explored how excess molar properties for liquid mixtures can be analysed in terms of different mole fraction domains [12]. Footnotes [1] E. A. Guggenheim, Trans. Faraday Soc.,1937,33,151; equation 4.1. [2] G. Scatchard, Chem. Rev.,1949,44,7;see page 9. [3] O. Redlich and A. Kister, Ind. Eng. Chem.,1948,40,345; equation 8. [4] F. Jost, H. Leiter and M. J. Schwuger, Colloid Polymer Sci., 1988, 266, 554. [5] G. M. Wilson, J. Am. Chem. Soc.,1964,86,127. [6] See also 1. R. V. Orye and J. M. Prausnitz, Ind. Eng. Chem.,1965,57,19. 2. S. Ohe, Vapour-Liquid Equilibrium Data, Elsevier, Amsterdam, 1989. 3. C. W. Bale and A. D. Pelton, Metallurg. Trans.,1974,5,2323. 4. R. Schuhmann, Metallurg. Trans.,B,1985,16,807. 5. M. Prehal, V. Dohnal and F. Versely, Collect. Czech. Chem.Commun.,1982,47,3171. [7] From equation (l), \begin{aligned} \frac{1}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}=&-\ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)-\frac{\mathrm{x}_{1} \,\left(1-\Lambda_{12}\right)}{\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}} \ &+\ln \left(\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}\right)-\frac{\mathrm{x}_{2} \,\left(\Lambda_{21}-1\right)}{\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}} \end{aligned} Then using equation (f) with $1 − x_{1} = x_{2}$, \begin{aligned} \ln \left(\mathrm{f}_{1}\right)=&-\mathrm{x}_{1} \, \ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)-\mathrm{x}_{2} \, \ln \left(\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}\right) \ &-\mathrm{x}_{2} \, \ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)-\frac{\mathrm{x}_{1} \, \mathrm{x}_{2} \,\left(1-\Lambda_{12}\right)}{\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}} \ &+\mathrm{x}_{2} \, \ln \left(\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}\right)+\frac{\left(\mathrm{x}_{2}\right)^{2} \,\left(1-\Lambda_{21}\right)}{\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}} \end{aligned} Or, \begin{aligned} \ln \left(f_{1}\right) &=-\left(x_{1}+x_{2}\right) \, \ln \left(x_{1}+\Lambda_{12} \, x_{2}\right) \ +x_{2} \,\left[\frac{\Lambda_{12} \, x_{1}-x_{1}}{x_{1}+\Lambda_{12} \, x_{2}}-\frac{\Lambda_{21} \, x_{2}-x_{2}}{\Lambda_{21} \, x_{1}+x_{2}}\right] \end{aligned} But $\Lambda_{12} \, \mathrm{x}_{1}-\mathrm{x}_{1}=\Lambda_{12} \,\left(1-\mathrm{x}_{2}\right)-\mathrm{x}_{1}=\Lambda_{12}-\left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)$ Hence, \begin{aligned} \ln \left(\mathrm{f}_{1}\right) &=-\ln \left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right) \ &+\mathrm{x}_{2} \,\left[\frac{\Lambda_{12}-\left(\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}\right)}{\mathrm{x}_{1}+\Lambda_{12} \, \mathrm{x}_{2}}-\frac{\Lambda_{21}-\left(\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}\right)}{\Lambda_{21} \, \mathrm{x}_{1}+\mathrm{x}_{2}}\right] \end{aligned} Or, $\ln \left(f_{1}\right)=-\ln \left(x_{1}+\Lambda_{12} \, x_{2}\right)+x_{2} \,\left[\frac{\Lambda_{12}}{x_{1}+\Lambda_{12} \, x_{2}}-\frac{\Lambda_{21}}{\Lambda_{21} \, x_{1}+x_{2}}\right]$ [8] D. Abrams and J. M. Prausnitz, AIChE J.,1975,21,116. [9] R. C. Reid, J. M. Prausnitz and E. B. Poling, The Properties of Gases and Liquids, McGraw-Hill, New York, 4th edn.,1987, chapter 8. [10] J. M. Prausnitz, R. N. Lichtenthaler and E. G. de Azevedo, Molecular Thermodynamics of Fluid Phase Equilibria, Prentice –Hall, Upper Saddle River, N.J., 3rd edn.,1999,chapter 6. [11] G. Douheret, C. Moreau and A. Viallard, Fluid Phase Equilib.,1985,22,277,287. [12] 1. M. I. Davis, Thermochim. Acta, 1984,77,421; 1985,90,313;1987,120,299; 1990,157,295. 2. M. I. Davis, M. C. Molina and G. Douheret, Thermochim Acta, 1988,131,153. 3. G. Douheret, A. Pal and M. I. Davis, J.Chem.Thermodyn.,1990,22,99. 4. G. Douheret, A. H. Roux, M. I. Davis, M. E .Hernandez, H. Hoiland and E. Hogseth, J. Solution Chem.,1993,22,1041. 5. G. Douheret, M. I. Davis., J. C. R. Reis , I. J. Fjellanger, M. B. Vaage and H. Hoiland, Phys.Chem.Chem.Phys.,2002,4,6034. [13] Finally we note that the Redlich-Kister equation can be expressed in the following form. $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \,\left(1-\mathrm{x}_{1}\right) \, \sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{k}} \mathrm{B}_{\mathrm{i}} \,\left(1-2 \, \mathrm{x}_{1}\right)^{i-1}$ Then $\mathrm{A}_{\mathrm{i}}=\mathrm{B}_{\mathrm{i}} \,(-1)^{i}$ J. D. G. de Oliveira and J. C. R.Reis, Thermochim. Acta 2008,468, 119 1.10.31: Gibbs Energies- Liquid Mixtures- Ideal The molar Gibbs energy of mixing for an ideal binary liquid mixture is given by equation (a); $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})=\mathrm{R} \, \mathrm{T} \,\left[\mathrm{x}_{1} \, \ln \mathrm{x}_{1}+\mathrm{x}_{2} \, \ln \mathrm{x}_{2}\right]$ $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id}) / \mathrm{R} \, \mathrm{T}=\mathrm{x}_{1} \, \ln \mathrm{x}_{1}+\mathrm{x}_{2} \, \ln \mathrm{x}_{2}$ In fact the molar Gibbs energy of mixing for an ideal binary mixture is negative across the complete composition range. According to Gibbs - Helmholtz equation, the molar enthalpy of mixing for an ideal binary mixture is given by equation (c). $\Delta_{\operatorname{mix}} \mathrm{H}_{\mathrm{m}}(\mathrm{id})=\frac{\mathrm{d}}{\mathrm{d}\left(\mathrm{T}^{-1}\right)}\left[\frac{\Delta_{\operatorname{mix}} \mathrm{G}(\mathrm{id})}{\mathrm{T}}\right]_{\mathrm{p}}$ But mole fractions are not dependent on temperature. Hence, $\Delta_{\text {mix }} \mathrm{H}_{\mathrm{m}}(\mathrm{id})=0$ This important result offers a point of reference. At fixed pressure, the mixing of two liquids to form an ideal binary liquid mixture is athermal. Hence a recorded heat of mixing is a direct measure of the extent to which the properties of a given mixture differ from those defined as ideal. But, $\Delta_{\text {mix }} \mathrm{G}_{\mathrm{m}}(\mathrm{id})=\Delta_{\mathrm{mix}} \mathrm{H}_{\mathrm{m}}(\mathrm{id})-\mathrm{T} \, \Delta_{\text {mix }} \mathrm{S}_{\mathrm{m}}(\mathrm{id})$ For an ideal binary liquid mixture the partial molar entropies of the two liquid components are given by the following equations. $\mathrm{S}_{1}(\operatorname{mix} ; \mathrm{id})=\mathrm{S}_{1}^{*}(\ell)-\mathrm{R} \, \ln \left(\mathrm{x}_{1}\right)$ $S_{2}(\operatorname{mix} ; \mathrm{id})=\mathrm{S}_{2}^{*}(\ell)-\mathrm{R} \, \ln \left(\mathrm{x}_{2}\right)$ $\mathrm{S}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \,\left[\mathrm{S}_{1}^{*}(\ell)-\mathrm{R} \, \ln \left(\mathrm{x}_{1}\right)\right]+\mathrm{x}_{2} \,\left[\mathrm{S}_{2}^{*}(\ell)-\mathrm{R} \, \ln \left(\mathrm{x}_{2}\right)\right]$ From equation (h), $\Delta_{\operatorname{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})=-\mathrm{R} \,\left[\mathrm{x}_{1} \, \ln \mathrm{x}_{1}+\mathrm{x}_{1} \, \ln \mathrm{x}_{2}\right]$ or, $\mathrm{T} \, \Delta_{\text {mix }} S_{m}(\text { id })=-R \, T\left[x_{1} \, \ln x_{1}+x_{2} \, \ln x_{2}\right]$ But across the complete mole fraction range $\left[x_{1} \, \ln x_{1}+x_{2} \, \ln x_{2}\right] \leq 0$. Over the same range, $\mathrm{T} \, \Delta_{\operatorname{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})>0$ Thus the sign and magnitude of $\Delta_{\text{mix}}\mathrm{G}_{\mathrm{m}}(\text{id})$ and (with opposite sign) $\mathrm{T} \, \Delta_{\mathrm{mix}} \mathrm{S}_{\mathrm{m}}(\mathrm{id})$ are defined. A further consequence of equation (d) is that the corresponding isobaric heat capacity variable, $\Delta_{\mathrm{mix}} \mathrm{C}_{\mathrm{pm}}(\mathrm{id})$ is zero across the whole mole fraction range. Using equation (a), the molar volume of mixing is given by equation (l). Thus, $\Delta_{\text {mix }} V_{m}(\mathrm{id})=\frac{\partial}{\partial p}\left[\Delta_{\text {mix }} G_{m}(\mathrm{id})\right]_{p}$ Hence for a binary liquid mixture having ideal properties, across the complete mole fraction range, $\Delta_{\operatorname{mix}} \mathrm{V}_{\mathrm{m}}(\mathrm{id})=0$. The latter condition requires that the volume of a liquid mixture equals the sum of the volumes of the two liquid components used to prepare the mixture at fixed temperature and pressure. For such a mixture, $\mathrm{V}_{\text {mix }}(\mathrm{id})=\mathrm{V}_{2}^{*}(\ell)+\mathrm{x}_{1} \,\left[\mathrm{V}_{1}^{*}(\ell)-\mathrm{V}_{2}^{*}(\ell)\right]$ This simple pattern is not observed. In fact the molar volume of a real binary mixture is usually less than $\mathrm{V}_{\text{mix} (\text{id})$. With the benefit of hindsight, we distinguish between Gibbsian and non-Gibbsian on the one hand and between first and second law (thermodynamic) variables on the other hand. Variables $\mathrm{H}$, $\mathrm{V}$ and $\mathrm{C}_{\mathrm{p}}$ are Gibbsian first law variables such that the molar property of an ideal binary liquid mixture is given by the mole fraction weighted sum of the properties of the pure liquids. However Gibbsian second law properties (e.g. entropies and Gibbs energies) require combinatorial terms arising from the irreversible entropy of mixing. These simple rules do not apply in the case of molar non-Gibbsian properties (e.g. isentropic compressions and isochoric heat capacities) of ideal mixtures.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.30%3A_Gibbs_Energies-_General_Equations.txt
For many binary aqueous liquid mixtures, the pattern shown by the molar excess thermodynamic parameters is $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}>0$; $\left|\mathrm{T} \, \mathrm{S}_{\mathrm{m}}^{\mathrm{E}}\right|>\left|\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right|$. This pattern of excess molar properties defines TA mixtures. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is positive because the excess molar entropy of mixing is large in magnitude and negative in sign. In these terms mixing is dominated by the entropy change. The excess molar enthalpy of mixing is smaller in magnitude than either ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ or $\mathrm{T} \, {\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ but exothermic in water-rich mixtures. The word ‘Typically’ in the description stems from observation that this pattern in thermodynamic variables is rarely shown by non-aqueous systems. At the time the classification was proposed [1], most binary aqueous liquid mixtures seemed to follow this pattern. Among the many examples of this class of system are aqueous mixtures formed by ethanol, 2-methyl propan-2-ol and cyclic ethers including tetrahydrofuran[2]. In water-rich mixtures, a large in magnitude but negative in sign $\mathrm{T} \, {\mathrm{S}_{\mathrm{m}}}^{\mathrm{E}}$ produces a large (positive) ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$. For mixtures rich in the apolar component m endothermic mixing produces a positive ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$. The key point is that in m water-rich mixtures a positive ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ emerges from a negative $\mathrm{T} \, {\mathrm{S}_{\mathrm{m}}}^{\mathrm{E}}$. With reference to volumetric properties of these system, the partial molar volume $\mathrm{V}(\mathrm{ROH})$ for monohydric alcohols can be extrapolated to infinite dilution; i.e. $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 0\right) \mathrm{V}(\mathrm{ROH})=\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})^{\infty}$ where $x_{2}$ is the mole fraction of alcohol. The difference, $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})^{\infty}-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ is negative. In fact this pattern is observed for both TA and Typical Non-aqueous binary aqueous mixtures. Examples where this pattern is observed included aqueous mixtures formed by DMSO, $\mathrm{H}_{2}\mathrm{O}_{2}$ and $\mathrm{CH}_{3}\mathrm{CN}$. Significantly $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ is negative for TA mixtures, decreasing from $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})^{\infty}-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ with increase in mole fraction of $\mathrm{ROH}$, accompanying by a tendency to immiscibility. The initial decrease in $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ with increase in $x_{2}$ is more dramatic the more hydrophobic the non-aqueous component; for 2-methyl propan-2-ol aqueous mixtures at $298.2 \mathrm{~K}$ and ambient pressure, the minimum occurs at an alcohol mole fraction 0.04 (at $298.2 \mathrm{~K}$). Many explanations have been offered for the complicated patters shown by the dependence of $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ on mole fraction composition. In one model, the negative $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ at low mole fractions of $\mathrm{ROH}$ is accounted for in terms of a liquid clathrate in which part of the hydrophobic R-group ‘occupies’ a guest site in the water lattice. The decrease in $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ is accounted for in terms of an increasing tendency towards a clathrate structure. But with increase in $x_{2}$ there comes a point where there is insufficient water to construct a liquid clathrate water host. Hence $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ increases. An important characteristic of TA mixtures is a tendency towards and in some cases actual decrease in liquid miscibility with increase in temperature. At ambient $\mathrm{T}$ and $\mathrm{p}$, the mixture 2-methyl propan-2-ol + water is miscible (but only just!) in all molar proportions. The corresponding mixtures prepared using butan-1-ol and butan-2-ol are partially miscible. TA systems are therefore often characterised by a Lower Critical Solution Temperature LCST. In fact nearly all examples quoted in the literature of systems having an LCST involve water as one component; e.g. $\mathrm{LCST} = 322 \mathrm{~K}$ for 2-butoxyethanol + water [3]. This tendency to partial miscibility is often signalled by the properties of the completely miscible systems. Returning to the patterns shown by relative partial molar volumes, $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$, a stage is reached whereby with increase in mole fraction of the non-aqueous component, this property increases after a minimum. Other properties of the mixtures also change dramatically including a marked increase in $\left(\alpha_{a} / v^{2}\right)$ where $\alpha_{\mathrm{a}}$ is the amplitude attenuation constant and $ν$ is the frequency of the sound wave in the MHz range; e.g. $70 \mathrm{~MHz}$. Actually the pattern is complicated. Over the range of mixture mole fractions $x_{2}$ where $\left[\mathrm{V}(\mathrm{ROH} ; \mathrm{aq})-\mathrm{V}^{*}(\mathrm{ROH} ; \ell)\right]$ decreases with increase in $x_{2}$, the ratio $\left(\alpha_{a} / v^{2}\right)$ hardly changes although the speed of sound increases. At a mole fraction $x_{2}$ characteristic of the temperature and the non-aqueous component, $\left(\alpha_{a} / v^{2}\right)$ increases sharply, reaching a maximum where the mixture has a strong tendency to immiscibility. This interplay between in-phase and out-of-phase components of the complex isentropic compressibility when the mole fraction composition of the mixture is changed highlights the molecular complexity of these systems. By way of contrast the ratio $\left(\alpha_{a} / v^{2}\right)$ for DMSO + water mixtures (a TNAN system) changes gradually when the mole fraction of DMSO is changed. For TA mixtures where $\left(\alpha_{a} / v^{2}\right)$ is a maximum [4], other evidence points to the fact these mixtures are micro-heterogeneous; cf. excess molar isobaric heat capacities. Phase separation of the mixture 2-methyl propan-2-ol is observed when butane gas is dissolved in the liquid mixture. The miscibility curve shows an LCST near $282 \mathrm{~K}$ [5]. Footnotes [1] F. Franks in Hydrogen-Bonded Solvent Systems, ed. A. K. Covington and P. Jones, Taylor and Francis, London,1968, pp.31-47. [2] The following references refer to properties of TA binary liquid mixtures. 1. alcohol + water mixtures. Isobaric heat capacities. 1. H. Ogawa and S. Murakami, Thermochim. Acta, 1986, 109,145. 2. G. I. Makhatadze and P. L. Privalov, J. Solution Chem.,1989,18,927. 3. R. Arnaud, L. Avedikian and J.-P. Morel,J. Chim. Phys., 1972,45. 2. fluoroalkanol + water mixtures. 1. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ R. Jadot and M.Fralha,J. Chem.Eng. Data,1988,33,237. J. Murto and A. Kiveninen, Suomen Kemist. Ser. B, 1967,40,258. 2. Enthalpies M. Denda, H. Touhara and K. Nakanishi, J. Chem. Thermodyn., 1987, 19,539. A. Kivinen, J. Murto and A.Vhtala, Suomen Kemist.1967,40,298. 3. Volumes; J. Murto, A. Kivinen, S. Kivimaa and R. Laakso, Suomen Kemist., Ser. B, 1967, 40,250. 3. amine + water mixtures 1. ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ and miscibility; J. L. Copp and D. H. Everett, m Discuss. Faraday Soc.,1953, 15,174. 2. Et3N + water; miscibility; A. Bellemans, J.Chem.Phys.,1953, 21, 368. 4. THF + water 1. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ J. Matous, J. P. Novak, J. Sobr and J. Pick, Collect. Czech. Chem.Commun.,1972,37,2653. C. Treiner, J.-F. Bocquet and M. Chemla, J. Chim..Phys., 1973,70, 72. 2. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ for $\mathrm{THF} + \mathrm{~D}_{2}\mathrm{O}$; J. Lejcek, J. Matous, J. P. Novak and J. Pick, J. Chem. Thermodyn., 1975,7,927. 3. Vapour composition; W. Hayduk, H. Laudie and O. H. Smith, J. Chem.Eng Data,1973,18,373. 4. ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ H. Nakayama and K. Shinoda, J. Chem. Thermodyn., 1971,3,401. 5. Activity of water K. L. Pinder, J. Chem. Eng. Data, 1973,18,275. 6. Thermal expansivities O. Kiyohara, P. J. D’Arcy and G. C. Benson, Can. J. Chem., 1978,56,2803. 7. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ and ${\mathrm{V}_{\mathrm{m}}}^{\mathrm{E}}$. Signer, H. Arm and H. Daeniker, Helvetica Chimica Acta, 1969, 52, 2347. 8. 2-Methyl propan-2-ol + water In the chemical literature, 2-methyl propan-2-ol is often called t-butanol but as Prof. David J. G. Ives often pointed out, there is no organic compound, t-butane. 9. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ and ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$ Y. Koga, W. W. Y. Siu and T. Y. H. Wong, J. Phys. Chem., 1990, 94,7700. 10. Enthalpies Y. Koga, Can. J.Chem.,1986,64,206;1988,66,1187,3171. 11. Volumes A. Hvidt, R. Moss and G. Nielsen, Acta Chem. Scand.,Sect. B, 1978, B32, 274. M. Sakurai, Bull. Chem. Soc. Jpn.,1987,160,1. 12. $\mathrm{C}_{\mathrm{p}}$ and $\mathrm{V}$ B. de Visser, G. Perron, and J. E. Desnoyers, Can J. Chem.,1977,55,856. 13. X-ray scattering K. Nishikawa, Y. Kodera and T. Iijima, J. Phys. Chem., 1987, 91,3694. 14. Sound velocity H. Endo and O. Nomoto,Bull. Chem. Soc. Jpn., 1973, 46, 3004. 15. Isentropic compressibilities; J. Lara and J. E. Desnoyers, J. Solution Chem., 1981,10,465. 16. Propan-1-ol + water 17. ${\mathrm{V}_{\mathrm{m}}}^{\mathrm{E}}$ G. C. Benson and O. Kiyohara, J. Solution Chem.,1980,9,791. M. I. Davis, Thermochim. Acta, 1990,157,295. C. De Visser, G. Perron and J. E. Desnoyers, Can. J.Chem.,1977,55,856. 18. Propanone + water ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ references; M. J. Blandamer, N. J. Blundell, J. Burgess, H. J. Cowles and I. M. Horn, J. Chem. Soc. Faraday Trans.,1990,86,283. 19. Methyl vinyl ketone + water LCST = $301 \mathrm{~K}$; UCST = $356 \mathrm{~K}$; J. Vojtko and M.Cihova, J. Chem. Eng. Data, 1972,17,337. [3] F. Elizalde, J Gracia and M. Costas, J Phys.Chem.,1988,93,3565. [4] M. J Blandamer and D. Waddington, Adv. Mol. Relax.Processes,1970,2,1. [5] R. W. Cargill and D. E. MacPhee, J. Chem. Soc. Faraday Trans.1, 1989, 85, 2665; an excellent observation! [5] ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$; M. J. Blandamer, J. Burgess, A. Cooney, H. J. Cowles, I. M. Horn, K. M. Martin, K.W. Morcom and P. Warrick, J. Chem. Soc. Faraday Trans., 1990, 86, 2209.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.32%3A_Gibbs_Energies-_Liquid_Mixtures-_Typically_Aqueous_%28TA%29.txt
For this sub-group of binary aqueous liquid mixtures, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is positive. An example of such a mixture is ‘water + ethanenitrile’. The positive ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ reflects endothermic mixing across nearly all the mole fraction range. These mixtures have a tendency to be partially miscible with an Upper Critical Solution Temperature, UCST. For aqueous mixtures the composition at the UCST is often ‘water-rich’. For ethanenitrile + water, the UCST is $272 \mathrm{~K}$. The positive ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ and endothermic mixing are attributed to disruption of water-water hydrogen bonding by added mathrm{MeCN}; cf. $\mathrm{CCl}_{4} + \mathrm{~MeOH}$ [1,2]. Footnotes [1] MeCN + water. 1. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$; H. T. French, J.Chem.Thermodyn.,1987, 19,1155. 2. ${\mathrm{H}_{\mathrm{m}}}^{\mathrm{E}}$; 1. R. H. Stokes, J.Chem.Thermodyn.,1987,19,977; 2. K. W. Morcom and R. W. Smith, J. Chem. Thermodyn., 1969,1,503. 3. Volumes 1. A. J. Easteal and L. A. Woolf, J.Chem.Thermodyn.,1988,20,693,701. 2. D. A. Armitage, M. J. Blandamer, M. J. Foster, N. J. Hidden, K. W. Morcom, M. C. R. Symons and M. J. Wootten, Trans. Faraday Soc.,1968,64,1193. 4. Sound Absorption M. J. Blandamer, M. J. Foster and D. Waddington, Trans. Faraday Soc.,1970,66,1369. 5. Miscibility; M. Hurth and D. Woermann, Ber. Bunsenges. Phys. Chem., 1987, 91, 614. 6. Excess thermodynamic properties; G. Douheret, C. Moreau and A. Viallard, Fluid Phase Equilib., 1986,26,221 7. Analysis of TD properties; M. J. Blandamer, N. J. Blundell, J. Burgess, H. J. Cowles and I. M. Horn, J. Chem. Soc. Faraday Trans.,1990, 86,277. [2] Propylene Carbonate Miscibility; N. F. Catherall and A. G. Williamson, J. Chem. Eng. Data, 1971, 16, 335. 1.10.34: Gibbs Energies- Liquid Mixtures- Typically Non-Aqueous Negative TNA For this sub-group, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is negative because there is strong inter- m component interaction which also produces exothermic mixing [1-4]. Examples of aqueous mixtures which fall into this class are 1. water + DMSO and 2. water + hydrogen peroxide. Footnotes [1] DMSO + water 1. ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ 1. T. C. Chan and W. A. Van Hook. J. Solution Chem.,1976,5,107. 2. J. Kenttamaa and J. J. Lindberg, Suomen Kem. J. Sect. B,,1960,33,98. 3. S. Biswas and A. R. Gupta, Indian J. Chem.,1975, 13,1036. 4. S. Y. Lam and R. L. Benoit, Can. J.Chem.,1974,52,718. 2. volumes 1. D. J. Pruett and L. K. Felker, J. Chem. Eng. Data, 1985,30,452. 2. C. F. Lau, P. T. Wilson and D. V. Fenby, Aust. J. Chem.,19709,23,1143. 3. Thermodynamics; M. J. Blandamer, N. J. Blundell, J. Burgess,H. J. Cowles and I. M. Horn, J. Chem. Soc. Faraday Trans.,1990,86,277. 4. Sound velocities W. Kaatze, M. Brai, F.-D. Scholle and R. Pottel, J. Mol. Liq., 1990,44,197. 5. Relative permittivities; E. Tommila and A. Pajunen, Suomen Kemist., Sect. B 1968, 41,172. [2] Ethane-1,2- diol+ water; 1. J.-Y. Huot, E. Battistel, R. Lumry, G. Villeneuve, J.-C. Lavallee, A. Anusiem and C. Jolicoeur, J. Solution Chem.,1988,17,601. 2. G. Douheret, A. Pal, H. Hoiland, O. Anowi and M. Davis, J.Chem.Thermodyn.,1991,23,569. [3] 2-methoxyethanol + water; M. Page, J.-Y. Huot and C. Jolicoeur, J.Chem. Thermodyn., 1993,25,139. [4] Hydrogen peroxide +water; G. Scatchard, G. M. Kavanagh and L. B. Ticknor, J. Am. Chem.Soc.,1952,74,3715. 1.10.35: Gibbs Energies- Liquid Mixtures- Immiscibility For a given binary liquid mixture (at defined $\mathrm{T}$ and $\mathrm{p}$) characterised by a plot of excess Gibbs energy ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ against mole fraction, ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ can be positive. Indeed if $\left[\mathrm{G}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{R} \, \mathrm{T}\right]$ strongly exceeds 0.5, the mixture is partially miscible. That is to say the liquid comprises two liquid phases having different mole fraction compositions. A fascinating variety of patterns emerge in the context of partial miscibilities. 1. Some binary liquid mixtures are completely miscible but become partially miscible with increase in temperature. The corresponding miscibility curve has a minimum at a Lower Critical Solution Temperature, LCST. For example in the case of 2-butoxyethanol + water , the LCST is at $322.2 \mathrm{~K}$ where $x\left(\mathrm{H}_{2}\mathrm{O}\right)=0.942$ [1]. In fact all commonly quoted examples of this class of systems have water as one component. A fascinating example concerns propionitrile+ polystyrene mixtures. The miscibility curves indicate that the LCST occurs at negative pressures; in effect when the mixture is ‘stretched’ [2]. 2. Many binary liquid mixtures (e.g/ phenol + water has (at ambient pressure).are partially miscible, becoming completely miscible on raising the temperature. The miscibility curve has a maximum at an Upper Critical Solution Temperature, UCST. At ambient pressure a small number of liquid mixtures exhibit both UCST and LCST. In other words the miscibility plot forms a closed loop. Partial miscibility plots also show deuterium isotope effects. In the case of $\mathrm{CH}_{3}\mathrm{CN}+\mathrm{H}_{2}\mathrm{O} \left(\text{component } 2 = \mathrm{~CH}_{3}\mathrm{CN}\right)$ the UCST is $272.10 \mathrm{~K}$ at $x_{2}=0.38$ [2]. Footnotes [1] A. Imre and W.A. Van Hook, J. Polym Sci.; Part B; Polymer Physics,1994,32,2283. [2] M. J. Blandamer, M. J. Foster and D. Waddington, Trans. Faraday Soc.,1970,66,1369.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.33%3A_Gibbs_Energies-_Liquid_Mixtures-_Typically_Non-Aqueous_Positive_TNA.txt
The solubilities of chemical substance $j$ in two liquids $\ell_{1}$ and $\ell_{2}$ (at the same $\mathrm{T}$ and $\mathrm{p}$) offers a method for comparing the reference chemical potentials, using the transfer parameter $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu_{\mathrm{j}}^{0}$. A similar argument is advanced in the context of salt solutions in which comparison of the solubility of salt j in two liquids leads to the transfer parameter for the salt. However the argument does not stop there. In the case of, for example a 1:1 salt $\mathrm{M}^{+} \mathrm{X}^{-}$, the derived transfer for the salt is re-expressed as the sum of transfer parameters for the separate ions $\mathrm{M}^{+}$ and $\mathrm{X}^{-}$. Thus $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{M}^{+} \mathrm{X}^{-}\right)=\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{M}^{+}\right)+\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{X}^{-}\right)$ However granted that we can obtain an estimate of the transfer parameter for the salt, $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{M}^{+} \mathrm{X}^{-}\right)$, thermodynamics does not offer a method for calculating the corresponding ionic transfer parameters. Several extra-thermodynamic procedures yield estimated single ion thermodynamic transfer parameters. The simplest approach adopts a reference ion (e.g. $\mathrm{H}^{+}$) and reports relative transfer ionic chemical potentials. $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{H}^{+}\right)=0$ For example; $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}\left(\mathrm{C} \ell^{-}\right)=\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu^{0}(\mathrm{HC} \ell)$ Solubilities and Transfer Parameters A closed system (at fixed $\mathrm{T}$ and ambient pressure) contains a solid salt $j$ in equilibrium with salt $j$ in aqueous solution. At equilibrium, $\mu_{\mathrm{j}}^{*}(\mathrm{~s})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{Q} \, \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) \, \gamma_{\pm}^{\mathrm{cq}}(\mathrm{aq}) / \mathrm{m}^{0}\right)$ Similarly for an equilibrium system where the solvent is a binary aqueous mixture, mole fraction $x_{2}$, $\mu_{\mathrm{j}}^{*}(\mathrm{~s})=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{x}_{2}\right)+\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{Q} \, \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) \, \gamma_{\pm}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}^{0}\right)$ Then, \begin{aligned} \Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln ) &=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{x}_{2}\right)-\mu_{\mathrm{j}}^{0}(\mathrm{aq}) \ &=-\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) \, \gamma_{\pm}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) \, \gamma_{\pm}^{\mathrm{eq}}(\mathrm{aq})\right] \end{aligned} A key assumption sets the ratio of mean ionic activity coefficients to unity. In effect we assume that the solubilities do not change dramatically as $x_{2}$ is changed. Therefore, $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )=-\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})\right]$ Thus the ratio $\left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})\right]$ is effectively the ratio of solubilities of salt $j$ in the mixed aqueous solutions and aqueous solution. If the solubility of the salt increases with increase in $x_{2}$, $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )$ is negative. In other words, the salt in aqueous solutions is stabilised by adding the co-solvent. Granted that solubility data lead to an estimate for $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )$, this quantity involves contributions from both cations and anions. For a salt containing two ionic substances $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )=\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{+}^{0}(\mathrm{~s} \ln )+\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{-}^{0}(\mathrm{~s} \ln )$ The background to this type of analysis centres on classic studes into the electrical conductivities of salt solutions. For a given salt in a solvent (at fixed $\mathrm{T}$ and $\mathrm{p}$), the molar conductivity approaches a limiting value with decrease in concentration; $\operatorname{limit}\left(c_{j} \rightarrow 0\right) \Lambda_{j}=\Lambda_{j}^{0}$. The limiting molar conductivity of a salt solution ${\Lambda^{0}}_{j}$ containing a 1:1 salt can be written as the sum of limiting ionic conductivities ${\lambda_{i}}^{0}$ of anions and cations. $\Lambda_{\mathrm{j}}^{0}=\lambda_{+}^{0}+\lambda_{-}^{0}$ The transport number of an ion $\mathrm{t}_{j}$ measures the ratio $\lambda_{\mathrm{j}}^{0} / \Lambda$. Both $\mathrm{t}_{j}$ and $\Lambda$ can be measured and hence $\lambda_{\mathrm{j}}^{0}$ calculated in the limit of infinite dilution characterizes ion $j$ in a given solvent at defined $\mathrm{T}$ and $\mathrm{p}$. Discrimination between anions and cations arises from their electrical charges and hence the direction of migration of ions in an electric field. Nevertheless the task of measuring both $\mathrm{t}_{j}$ and $\Lambda$ is not trivial and some simple working hypothesis is often sought. The argument is advanced that the molar conductivities are equal in magnitude for two ions having similar size and solvation characteristics. This ‘extrathermodynamic’ assumption has been applied [1-8] to a range of ‘onium salts including 1. $\mathrm{Bu}_{4} \mathrm{N}^{+} \mathrm{~Ph}_{3} \mathrm{FB}^{-}$, 2. $\text { iso }-\mathrm{Bu}_{3} \mathrm{N}^{+} \mathrm{~H} \mathrm{~Ph}_{4} \mathrm{B}^{-}$, 3. $\mathrm{Bu}_{4} \mathrm{N}^{+} \mathrm{~Ph}_{4} \mathrm{B}^{-}$, 4. $\text { iso }-\mathrm{Am}_{3} \mathrm{BuN}^{+} \mathrm{~H}_{4} \mathrm{B}^{-}$ 5. $\text { iso }-\mathrm{Am}_{4} \mathrm{N}^{+} \text { iso }-\mathrm{Am}_{4} \mathrm{~B}^{-}$; so $\lambda^{0}$ (big cation) = $\lambda^{0}$ (big anion). This ‘big ion – big ion’ assumption is carried over to the analysis of thermodynamic properties where we lack the discrimination between cations and anions based on their mobilities in an applied electric potentials. gradient. Then for example the change in solubility of one such salt in aqueous solution on adding a cosolvent ( e.g. ethanol) can be understood in terms of equal transfer thermodynamic potentials. $(1 / 2) \, \Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big cation big anion; } \mathrm{s} \ln )=\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big cation; } \mathrm{s} \ln )=\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big anion; } \mathrm{l} \ln )$ For example having obtained $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big cation; } \mathrm{ln})$, the difference in solubilities of the corresponding salt iodide is used to obtain the transfer parameter for iodide ions in the two solvents. \begin{aligned} \Delta(\mathrm{aq} \rightarrow&\left.\mathrm{x}_{2}\right) \mu^{0}\left(\mathrm{I}^{-} ; \mathrm{s} \ln \right)=\ & \Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big cation iodide; } \mathrm{s} \ln )-\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu^{0}(\text { big cation } \mathrm{s} \ln ) \end{aligned} Considerable information is available in the chemical literature concerning ionic transfer parameters, particularly for solutes in binary aqueous mixtures at $298.2 \mathrm{~K}$ and ambient pressure.8-21 Unfortunately there is no agreed composition scale for transfer parameters. Information includes transfer parameters based on concentration, molality and mole fractions scales for the solutes. The situation is further complicated by the fact that different scales are used to express composition of liquid mixtures. Common scales include mass%, mole fraction and vol%. Conversion between these scales is a tedious. Some examples of the required equations are presented in an Appendix to this Topic. Footnotes 1. M. A. Coplan and R. M. Fuoss, J. Phys. Chem.,1964,68,1181. 2. R. L. Kay, J .Hales and G. P. Cunningham, J. Phys. Chem.,1967,67, 3925. 3. M. R. Coplan and R.M.Fuoss, J. Phys. Chem.,1964,68,1177. 4. D. F. Evans, J. Thomas, J. A. Nadas and M. A. Matesich, J. Phys. Chem.,1971,75,1714. 5. C. Treiner and R. M. Fuoss, Z. Phyik. Chemie, 1965,228,343. 6. J. E. Coetzee and G. P. Cunningham, J. Am. Chem.Soc.,1964,86,3403; 1965,87,2529. 7. B. S. Krumgal’z, Russ. J.Phys.Chem.,1972,46,858. 8. G. Petrella, A. Sacco, M. Castagnolo, M. Della Monica, and A. De Giglio, J. Solution Chem.,1977,6,13. 9. M. H. Abraham, T. Hill, H. C. Ling R. A. Schultz and R. A. C. Watt, J. Chem. Soc. Faraday Trans.,1,1984,80,489. 10. H. Talukdar and K. K. Kundu, J. Phys. Chem.,1992,96,970. 11. Y. Marcus, J. Chem. Soc. Faraday Trans.,1987, 83,858. 12. M. R. J. Dack, K. J. Bird and A. J. Parker, Aust. J. Chem.,1975,28,955. 13. P. Singh, I. D. MacLeod and A. J. Parker, J. Solution Chem.,1984,13,103. 14. Y. Marcus, J. Solution Chem.,1986,15,291. 15. E. A. Gomaa, Thermochim Acta, 1987,120,183. 16. M. Booij and G. Somsen, Electrochim Acta, 1983,28,1883. 17. C. V. Krishnan and H. L. Friedman, J. Phys. Chem.,1970,74,2356. 18. J. I. Kim and E.A.Gomaa, Bull. Soc., Chem. Bwel;g.,1981,90,391. 19. H. L. Friedman, J. Phys. Chem.,1967,71,1723. 20. A.F. Danil de Namor, T. Hill and E. Sigsted, J. Chem. Soc Faraday Trans.,I,1983,79,2713. 21. I. N. B. Mullick and K. K. Kundu, Indian J. Chem.,1984, 23A,812. Appendix Solute j in Aqueous Solution We write down two equations for the same quantity, the chemical potential of solute $j$. For the chemical potential of solute $j$ in an ideal aqueous solution at ambient pressure ( i.e. close to the standard pressure, $\mathrm{p}^{0}$), $\mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right]$ Here $\mathrm{m}_{j}$ is the molality of solute $j$; $\mathrm{m}^{0} = 1 \mathrm{~mol kg}^{-1}$, the reference molality. However we may decide to express the composition of the solution in terms of the mole fraction of solute. If the properties of the solute are ideal, the chemical potential of solute $j$, $\mu_{j}(\mathrm{aq})$ is related to the mole fraction of solute $x_{j}$. $\mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}=1\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{x}_{\mathrm{j}}\right]$ Equations (i) and (ii) describe the same property, $\mu_{j}(\mathrm{aq})$. The property $\mu_{j}^{0}\left(\mathrm{aq} ; x_{j}=1\right)$ is interesting because it describes the chemical potential of solute $j$ in aqueous solution where the mole fraction of solute is unity; it is clearly an ‘extrapolated’ property of the solute. If $\mathrm{n}_{j}$ is the amount of solute in a solution prepared using $10^{2} \mathrm{~kg}$ of water, we can combine equations (i) and (ii); $\mathrm{x}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} /\left[\left(10^{2} / \mathrm{M}_{1}\right)+\mathrm{n}_{\mathrm{j}}\right]$ where for a dilute solution $\left(10^{2} / \mathrm{M}_{1}\right)>>\mathrm{n}_{\mathrm{j}}$; $\mathrm{M}_{1}$ is the molar mass of water. $\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right]=\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}=1\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{l}} / 10^{2}\right]$ Or, $\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{m}^{0}\right)-\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}=1\right)=\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}^{0} \, \mathrm{M}_{1}\right]$ We note that $\left[\mathrm{m}^{0} \, \mathrm{M}_{1}\right]=\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{kg} \mathrm{mol}^{-1}\right]=[1]$ Solute j in a Solvent prepared as a Binary Aqueous Mixture If $\mathrm{n}_{j}$ is the amount of solute $j$ in $10^{2} \mathrm{~kg}$ of a solvent mixture, the chemical potential of solute $j$ is given by equation (iv) $\mu_{\mathrm{j}}(\operatorname{mix})=\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right]$ We note that $\left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right]=\left[\mathrm{mol} / \mathrm{kg} \, \mathrm{mol} \, \mathrm{kg}^{-1}\right]=[1]$. If the binary solvent mixture comprises $w_{2} \%$ of the non-aqueous component, for a dilute solution of solute $j$, the mole fraction of solute $x_{j}$ is given by equation (vi) where $\mathrm{M}_{2}$ is the molar mass of the cosolvent. $\mathrm{x}_{\mathrm{j}}=\frac{\mathrm{n}_{\mathrm{j}}}{\left\{\left[\left(10^{2}-\mathrm{w}_{2} \%\right] / \mathrm{M}_{1}\right\}+\left\{\mathrm{w}_{2} \% / \mathrm{M}_{2}\right\}\right.}$ Using the mole fraction scale for solute $j$, the chemical potential of solute $j$ in the mixture, composition $w_{2} \%$ is given by equation (vii). $\mu_{\mathrm{j}}(\mathrm{mix})=\mu_{\mathrm{j}}^{0}\left(\mathrm{mix} ; \mathrm{x}_{\mathrm{j}}=1\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\mathrm{n}_{\mathrm{j}}}{\left\{\left[\left(10^{2}-\mathrm{w}_{2} \%\right] / \mathrm{M}_{1}\right\}+\left\{\mathrm{w}_{2} \% / \mathrm{M}_{2}\right\}\right.}\right]$ Equations (v) and (vii) describe the same property, the chemical potential of solute $j$ in a mixed solvent system. Hence, \begin{aligned} &\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right] \ &=\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{x}_{\mathrm{j}}=1\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\mathrm{n}_{\mathrm{j}}}{\left\{\left[\left(10^{2}-\mathrm{w}_{2} \%\right] / \mathrm{M}_{1}\right\}+\left\{\mathrm{w}_{2} \% / \mathrm{M}_{2}\right\}\right.}\right] \end{aligned} Or, \begin{aligned} &\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{m}^{0}\right) \ &=\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{x}_{\mathrm{j}}=1\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{10^{2} \, \mathrm{m}^{0}}{\left\{\left[\left(10^{2}-\mathrm{w}_{2} \%\right] / \mathrm{M}_{1}\right\}+\left\{\mathrm{w}_{2} \% / \mathrm{M}_{2}\right\}\right.}\right] \end{aligned} Conversion of Scales. It is convenient at this point to comment on the difference in reference chemical potentials of solute $j$ in aqueous solutions and a solvent mixture. Thus from equation (iv). $\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{m}^{0}\right)-\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}=\mathrm{l}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}^{0} \, \mathrm{M}_{1}\right]$ And from equation (ix) \begin{aligned} \mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{m}^{0}\right)-\mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{x}_{\mathrm{j}}\right.&=1) \ &=\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{10^{2} \, \mathrm{m}^{0}}{\left\{\left[\left(10^{2}-\mathrm{w}_{2} \%\right] / \mathrm{M}_{1}\right\}+\left\{\mathrm{w}_{2} \% / \mathrm{M}_{2}\right\}\right.}\right] \end{aligned} The difference between equations (x) and (xi) yields an equation relating transfer parameters for solute $j$ on the two composition scales. $\begin{array}{r} \mu_{\mathrm{j}}^{0}\left(\operatorname{mix} ; \mathrm{m}^{0}\right)-\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{m}^{0}\right)=\mu_{\mathrm{j}}^{0}\left(\mathrm{mix} ; \mathrm{x}_{\mathrm{j}}=1\right)-\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{x}_{1}=1\right) \ -\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\left\{10^{2}-\mathrm{w}_{2} \%\right\}+\left\{\mathrm{w}_{2} \% \, \mathrm{M}_{1} / \mathrm{M}_{2}\right\}}{10^{2}}\right] \end{array}$ Hence \begin{aligned} \Delta(\mathrm{aq} \rightarrow \mathrm{mix}) & \mu_{\mathrm{j}}^{0}(\mathrm{~m}-\mathrm{scale})=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{x}-\text { scale }) \ &-\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\left\{10^{2}-\mathrm{w}_{2} \%\right\}+\left\{\mathrm{w}_{2} \% \, \mathrm{M}_{1} / \mathrm{M}_{2}\right\}}{10^{2}}\right] \end{aligned} Or, \begin{aligned} &\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~m}-\mathrm{scale})=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{x}-\text { scale }) \ &\quad-\mathrm{R} \, \mathrm{T} \, \ln \left[1-\left(\mathrm{w}_{2} \% / 10^{2}\right)+\left(\mathrm{w}_{2} \% / 10^{2}\right) \, \mathrm{M}_{\mathrm{l}} / \mathrm{M}_{2}\right] \end{aligned} Or, $\begin{array}{r} \Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~m}-\text { scale })=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{x}-\text { scale }) \ -\mathrm{R} \, \mathrm{T} \, \ln \left\{1-\left[1-\left(\mathrm{M}_{1} / \mathrm{M}_{2}\right)\right] \,\left(\mathrm{w}_{2} \% / 10^{2}\right)\right\} \end{array}$ If solute $j$ is a salt which is completely dissociated into $ν$ ions in both aqueous solution and in the mixed solvent system, \begin{aligned} \Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~m}-\mathrm{scale})=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{x}-\mathrm{scale}) \ -\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left\{1-\left[1-\left(\mathrm{M}_{1} / \mathrm{M}_{2}\right)\right] \,\left(\mathrm{w}_{2} \% / 10^{2}\right)\right\} \end{aligned} Thus for each ionic substance contributing to the transfer property for the salt, \begin{aligned} \Delta(\mathrm{aq}&\rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )=\ v_{+} \, \Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mu_{+}^{0}(\mathrm{~s} \ln )+v_{-} \, \Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{-}^{0}(\mathrm{~s} \ln ) \end{aligned} Equations (xv) and (xvi) show that the difference between the transfer chemical potentials on the x- and m- scales is independent of temperature. The difference is based on the mass of the solvent components in the mixture. Consequently the transfer enthalpies on the two scales are equal. $\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~m}-\text { scale })=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{x}-\text { scale })$ Therefore the difference in the transfer chemical potentials can be traced to differences in the transfer entropies. At constant pressure, \begin{aligned} &-\Delta(\mathrm{aq} \rightarrow \text { mix }) \mathrm{S}_{\mathrm{j}}^{0}(\mathrm{~m}-\text { scale })=\mathrm{d} \Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mu_{\mathrm{j}}^{0} / \mathrm{dT} \ &=-\Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mathrm{S}_{\mathrm{j}}^{0}(\mathrm{x}-\text { scale }) \ &+\mathrm{v} \, \mathrm{R} \, \ln \left\{\left[1-\left[1-\left(\mathrm{M}_{1} / \mathrm{M}_{2}\right)\right] \,\left(\mathrm{w}_{2} \% / 10^{2}\right)\right\}\right\} \end{aligned} A similar argument notes that the masses of the solvents forming the mixed solvents are independent of pressure (at fixed temperature) Therefore the volumes of transfer on molality and mole fraction scales are equal. In summary (at fixed $\mathrm{T}$ and \mathrm{p}\)), \begin{aligned} \Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{s} \ln ) &=\mathrm{H}_{\mathrm{j}}^{\infty}(\operatorname{mix})-\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq}) \ &=-\mathrm{T}^{2} \,\left[\partial\left\{\Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{sln} ; \mathrm{T}) / \mathrm{T}\right\} / \mathrm{dT}\right] \end{aligned} Further, for the isobaric partial molar heat capacities, \begin{aligned} \Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{s} \ln ) &=\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{mix})-\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq}) \ &=\left[\partial\left\{\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{s} \ln ; \mathrm{T})\right\} / \partial \mathrm{T}\right]_{\mathrm{p}} \end{aligned} Also $\Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{s} \ln )=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{mix})-\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$ Transfer Parameters: Molality and Concentration Scales. The procedures described above are repeated but now in a comparison of the molality and concentration scales. For a solute $j$ (at fixed $\mathrm{T}$ and $\mathrm{p}$) in a solution having ideal thermodynamic properties, the chemical potential of solute $j$ is related to concentration of solute $j$, $\mathrm{c}_{j}$ which by convention is expressed in terms of amount of solute in $1 \mathrm{~dm}^{3}$ of solution at defined $\mathrm{T}$ and $\mathrm{p}$; i.e. $\mathrm{c}_{\mathrm{j}}=\left[\mathrm{mol} \mathrm{~dm} ^ {-3} \right]$. A reference concentration $\mathrm{c}_{r}$ describes a solution where one $\mathrm{dm}^{3}$ of solution contains one mole of solute. Because the volume of a liquid depends on both temperature and pressure, these variables must be specified. Thus $\mu_{j}(a q)=\mu_{j}^{0}(c-s c a l e ; a q)+R \, T \, \ln \left[c_{j}(a q) / c_{r}\right]$ The units of both $\mathrm{c}_{j}(\mathrm{aq})$ and $\mathrm{c}_{r}$ are $\left[\mathrm{mol dm}^{-3}\right]$.Hence using equations (i) and (xxiii), \begin{aligned} &\mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}(\mathrm{~m} ; \mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}(\mathrm{aq}) / \mathrm{m}^{0}\right] \ &=\mu_{\mathrm{j}}^{0}(\mathrm{c}-\mathrm{scale} ; \mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{c}_{\mathrm{j}}(\mathrm{aq}) / \mathrm{c}_{\mathrm{r}}\right] \end{aligned} For a solution in $10^{2} \mathrm{~kg}$ of solvent, $\mathrm{m}_{\mathrm{j}}(\mathrm{aq})=\mathrm{n}_{\mathrm{j}} / 10^{2} \mathrm{~mol} \mathrm{~kg}{ }^{-1}$ For a dilute solution, density $\rho(\mathrm{aq})=\rho_{1}^{*}(\ell)$ Volume of a dilute solution with mass $10^{2} \mathrm{~kg}=10^{2} / \rho_{1}^{*}(\ell)$ Concentration, $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} \, \rho_{\mathrm{l}}^{*}(\ell) / 10^{2}$ Therefore equation (xxiv) can be written in the following form. \begin{aligned} \mu_{\mathrm{j}}^{0}(\mathrm{~m} ; \mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right] \ &=\mu_{\mathrm{j}}^{0}(\mathrm{c}-\mathrm{scale} ; \mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} \, \rho_{1}^{*}(\ell) / \mathrm{c}_{\mathrm{r}}\right] \end{aligned} For the solution in a binary aqueous mixture, \begin{aligned} \mu_{\mathrm{j}}^{0}(\mathrm{~m} ; \mathrm{mix})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} / 10^{2} \, \mathrm{m}^{0}\right] \ &=\mu_{\mathrm{j}}^{0}(\mathrm{c}-\text { scale; mix })+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{n}_{\mathrm{j}} \, \rho(\mathrm{mix}) / \mathrm{c}_{\mathrm{r}}\right] \end{aligned} Then, \begin{aligned} \Delta(\mathrm{aq} \rightarrow&\mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~m}) \ &=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{c}-\mathrm{scale})+\mathrm{R} \, \mathrm{T} \, \ln \left[\rho(\mathrm{mix}) / \rho_{\mathrm{1}}^{*}(\ell)\right] \end{aligned} In the event that solute is a salt which produces $ν$ moles of ions for each mole of salt, \begin{aligned} \Delta(\mathrm{aq} \rightarrow&\mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{~m}) \ &=\Delta(\mathrm{aq} \rightarrow \mathrm{mix}) \mu_{\mathrm{j}}^{0}(\mathrm{c}-\text { scale })+\mathrm{v} \, \mathrm{R} \, \mathrm{T} \, \ln \left[\rho(\mathrm{mix}) / \rho_{1}^{*}(\ell)\right] \end{aligned} For each ionic substances, e.g. a cation \begin{aligned} \Delta(\mathrm{aq} \rightarrow&\operatorname{mix}) \mu_{+}^{0}(\mathrm{~m}) \ &=\Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mu_{+}^{0}(\mathrm{c}-\mathrm{scale})+v \, \mathrm{R} \, \mathrm{T} \, \ln \left[\rho(\operatorname{mix}) / \rho_{1}^{*}(\ell)\right] \end{aligned} Because the densities of water and each mixture depends on temperature at fixed pressure, the transfer enthalpies on molality and concentration scales differ. Thus \begin{aligned} \Delta(\mathrm{aq} \rightarrow&\operatorname{mix}) \mathrm{H}_{+}^{\infty}(\mathrm{m}) \ &=\Delta(\mathrm{aq} \rightarrow \operatorname{mix}) \mathrm{H}_{+}^{\infty}(\mathrm{c}-\mathrm{scale})-\mathrm{v} \, \mathrm{R} \, \mathrm{T}^{2} \,\left[\partial \ln \left[\rho(\mathrm{mix}) / \rho_{1}^{*}(\ell)\right] / \partial \mathrm{T}\right]_{\mathrm{p}} \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.10%3A_Gibbs_Energies/1.10.36%3A_Gibbs_Energies-_Salt_Solutions-_Aqueous_Mixtures.txt
From the definition of enthalpy $\mathrm{H}$, an infinitesimal small change in enthalpy is related to the corresponding change in thermodynamic energy $\mathrm{dU}$ by equation (a). $\mathrm{dH}=\mathrm{dU}+\mathrm{p} \, \mathrm{dV}+\mathrm{V} \, \mathrm{dp}$ If only ‘$\mathrm{p}-\mathrm{V}$’ work is involved, $\mathrm{dU}=\mathrm{q}-\mathrm{p} \, \mathrm{dV}$ Then $\mathrm{dH}=\mathrm{q}+\mathrm{V} \, \mathrm{dp}$ But in general terms, $\mathrm{q}=\mathrm{C} \, \mathrm{dT}$ Here $\mathrm{C}$ is the heat capacity of the system, an extensive variable. Hence for a change at constant pressure, $\mathrm{dH}=\mathrm{C}_{\mathrm{p}} \, \mathrm{dT}$ Isobaric heat capacity is related to the change in enthalpy accompanying a change in temperature. $\mathrm{C}_{p}=(\partial \mathrm{H} / \partial \mathrm{T})_{\mathrm{p}}$ $\mathrm{C}_{\mathrm{p}}$ is an extensive variable; $\mathrm{C}_{\mathrm{pm}}$ is the corresponding molar property. We develop the above analysis in a slightly different way in order to make an important point. We explore the relationship between the dependence of ($\mathrm{G}/\mathrm{T}$) on temperature at 1. constant affinity and 2. constant extent of reaction. A calculus operation yields the following equation. $\left[\frac{\partial(\mathrm{G} / \mathrm{T})}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}}=\left[\frac{\partial(\mathrm{G} / \mathrm{T})}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi}-\frac{1}{\mathrm{~T}} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ But at equilibrium, $A=-\left[\frac{\partial G}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}=0$ Using the Gibbs-Helmholtz Equation, $\mathrm{H}(\mathrm{A}=0)=\mathrm{H}\left(\xi^{\mathrm{eq}}\right)$ This result is expected because enthalpy $\mathrm{H}$ is a strong state variable, a function of state which does not need a description of a pathway. This is not the case for isobaric heat capacities. Using the same calculus operation, $\left[\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}}=\left[\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi}-\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \mathrm{H}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ We cannot assume that the triple product term is zero. Hence there are two limiting isobaric heat capacities; the equilibrium isobaric heat capacity, $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$ and the ‘frozen’ isobaric heat capacity, $\mathrm{C}_{\mathrm{p}\left(\xi_{\mathrm{eq}\right)$. $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)=\mathrm{C}_{\mathrm{p}}\left(\xi^{\mathrm{eq}}\right)-\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \mathrm{H}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ In other words, the isobaric heat capacity is not a strong function of state. The property is concerned with a pathway between states. The term \left[-\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \mathrm{H}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}\right]\) is the relaxational isobaric heat capacity. $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$, the equilibrium heat capacity, signals that when heat q passes into a system, the composition - organization of the system changes in order that the Gibbs energy of the system remains at a minimum. In contrast $\mathrm{C}_{\mathrm{p}}\left(\xi_{\mathrm{eq}}\right)$, the frozen heat capacity, signals that no changes occur in the composition – organization in the system such that the Gibbs energy is displaced from the original minimum. Moreover the equilibrium isobaric heat capacity is always larger than the frozen isobaric heat capacity. Indeed we can often treat the extensive equilibrium property $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$ as a function of state. Certainly isobaric heat capacities differentiate water as a solvent from other associated liquids [1,2] such as $\mathrm{H}_{2}\mathrm{O}_{2}$, and $\mathrm{N}_{2}\mathrm{H}_{4}$ and low melting fused salts such as ethylammonium nitrate. Interestingly, among liquids, water has one of the highest heat capacitances; i.e. heat capacities per unit volume [3]. Therefore hypothermia is often life threatening for babies and old persons because in order to raise their temperature a large amount of thermal energy has to be passed into the body in order to raise their temperature. This is often difficult without damaging the skin and other body tissues--- a consequence of humans being effectively concentrated aqueous systems. The isobaric heat capacity of a solution prepared using $\mathrm{n}_{1}$ moles of solvent (water) and $\mathrm{n}_{j}$ moles of solute $j$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ is defined by equation (1). $\mathrm{C}_{\mathrm{p}}=\mathrm{C}_{\mathrm{p}}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ We assume the system is at thermodynamic equilibrium such that the affinity for spontaneous change is zero at a minimum in Gibbs energy. The isobaric heat capacity of the solution is related to the composition using equation (m). $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{pl}}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}(\mathrm{aq})$ Here $\mathrm{C}_{\mathrm{p}1}(\mathrm{aq})$ and $\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})$ are the partial molar isobaric heat capacities enthalpies of solvent and solute respectively. Alternatively $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})$ is given by equation (n) where $\mathrm{C}_{\mathrm{p} 1}^{*}(\ell)$ and $\phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ are the molar isobaric heat capacity of the pure solvent and the apparent molar isobaric heat capacity of the solute $j$ respectively. Thus $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ Footnotes [1] M. Allen, D. F. Evans and R. Lumry, J. Solution Chem.,1985,14,549. [2] M. Hadded, M. Biquard, P. Letellier and R. Schaal, Can. J. Chem.,1985,63,565. [3] Liquid Heat Capacitance $\mathrm{C}_{\mathrm{p}} / \mathrm{~J K}^{-1} \mathrm{~cm}^{-3}$ Water 4.18 Propane 1.67 Cyanomethane 2.26 Ethanol 1.92 Tetrachloroethane 1.38 See J. K. Grime, in Analytical Solution Calorimetry, ed. J. K.Grime, Wiley, New York, 1985, chapter 1. For isochoric and isobaric heat capacities of liquids see, D. Harrison and E. A. Moelwyn-Hughes, Proc. R. Soc. London, Ser.A,1957, 239, 230.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.01%3A_Heat_Capacities-_Isobaric-_Solutions.txt
Equilibrium isobaric heat capacities of solutions can be treated for most purposes as extensive variables. Thus for an aqueous solution prepared using $\mathrm{n}_{1}$ moles of solvent (water) and $\mathrm{n}_{j}$ moles of a simple solute $j$ the isobaric (equilibrium) heat capacity of the solution $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})$ can be related to the composition of the solution using equations (a) and (b) [1]. $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}(\mathrm{aq})$ where $\mathrm{C}_{\mathrm{pl}}(\mathrm{aq})=\left(\frac{\partial \mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{j})} \quad \text { and } \mathrm{C}_{\mathrm{pj}}(\mathrm{aq})=\left(\frac{\partial \mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{l})}$ Similar equations are encountered in a discussion of the partial molar enthalpies but with reference to these properties we develop a number of strategies because it is not possible to determine the enthalpy of a solution. In the present case the outlook is much more favourable because it is possible to measure isobaric heat capacities of solutions [1-3]. The fact that we can measure the temperature dependence of the equilibrium enthalpy of a solution but not the actual enthalpy is an interesting philosophical point. Nevertheless it is informative to develop the analysis starting from equations relating partial molar enthalpies and compositions of solutions. A given aqueous solution is prepared using $\mathrm{n}_{1}$ moles of solvent (water) and $\mathrm{n}_{j}$ moles of solute. The partial molar enthalpies are related to the composition of the solution by the following equations. Thus $\mathrm{H}_{1}(\mathrm{aq})=\mathrm{H}_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,(\partial \phi / \partial \mathrm{T})_{\mathrm{p}}$ and $\mathrm{H}_{\mathrm{j}}(\mathrm{aq})=\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq})-\mathrm{R} \, \mathrm{T}^{2} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right)_{\mathrm{p}}$ At all $\mathrm{T}$ and $\mathrm{p}$, $\operatorname{limit}\left(m_{j} \rightarrow 0\right) \phi=1 \text { and } \gamma_{j}=1$ By definition, $\mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})=\left(\frac{\partial \mathrm{C}_{\mathrm{p}}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{j})}=\left(\frac{\partial \mathrm{H}_{1}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left(\frac{\partial^{2} \mathrm{H}}{\partial \mathrm{n}_{1} \, \partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{n}(\mathrm{j})}$ And, $\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})=\left(\frac{\partial \mathrm{C}_{\mathrm{p}}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(1)}=\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left(\frac{\partial^{2} \mathrm{H}}{\partial \mathrm{n}_{\mathrm{j}} \, \partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{n}(1)}$ The latter two equations trace the story from the enthalpy of the solution to partial molar isobaric heat capacities. Using equation (f) in conjunction with equation (c) we obtain an equation for dependence for $\mathrm{C}_{\mathrm{p}1}(\mathrm{aq})$ on molality $\mathrm{m}_{j}$. $\mathrm{C}_{\mathrm{pl} 1}(\mathrm{aq})=\mathrm{C}_{\mathrm{pl} 1}^{*}(\ell)+2 \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,(\partial \phi / \partial \mathrm{T})_{\mathrm{p}}+\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\partial^{2} \phi / \partial \mathrm{T}^{2}\right)_{\mathrm{p}}$ Similarly using equations (d) and (g), we obtain an equation relating $\mathrm{C}_{\mathrm{p}j}(\mathrm{aq})$ and molality $\mathrm{m}_{j}$; the origin of the two minus signs is the Gibbs - Helmholtz Equation. $\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})-2 \, \mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right]_{\mathrm{p}}-\mathrm{R} \, \mathrm{T}^{2} \,\left[\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}^{2}\right]_{\mathrm{p}}$ We consider a solution prepared using $1 \mathrm{~kg}$ of water and $\mathrm{m}_{j}$ moles of solute $j$ [4]. $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl}}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}} \text { (aq) }$ If the thermodynamic properties of the solution are ideal, from the definitions of both practical osmotic coefficient $\phi$ and activity coefficient $\gamma_{j}$, the last two terms in equations (h) and (i) are zero. $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ This is an interesting equation because two experiments yield $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{id})$ and $\mathrm{C}_{\mathrm{p} 1}^{*}(\ell)$. Hence, granted the ideal conditions, we obtain an estimate of $\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$, limiting isobaric heat capacity of solute $j$ in solution. Unfortunately the assumption concerning ideal properties of a solution is often unrealistic. Nevertheless equation (k) offers a reference against which we can examine the properties of real solutions. Footnotes [1] An important technological development was the design of the Picker flow calorimeter; P. Picker, P.-A. Leduc, P. R. Philip and J. E. Desnoyers, J. Chem. Thermodyn.,1971, 3,631. [2] For details of calibration of the Picker calorimeter; D. E.White and R. H. Ward, J. Solution Chem.,1982,11,223. [3] For extension to measurement of thermal expansion coefficients; J. F. Alary, M. N. Simard, J. Dumont and C. Jolicoeur, J. Solution Chem.,1982,11,755. [4] $\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]=\left[\mathrm{kg} \, \mathrm{mol}^{-1}\right]^{-1} \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]+\left[\operatorname{mol~kg}{ }^{-1}\right] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]$ 1.12.03: Heat Capacities- Isobaric and Isochoric When heat $\mathrm{q}$ passes smoothly (reversibly) into a closed system from the surroundings, the temperature of the system increases (if there are no phase changes, e.g. liquid to vapour). The increase in temperature $\Delta \mathrm{T}$ is related to heat $\mathrm{q}$ using equation (a). $\mathrm{q}=\mathrm{C} \, \Delta \mathrm{T}$ Heat capacity $\mathrm{C}$ is an extensive property of a system whereas $\Delta \mathrm{T}$ is the change in an intensive variable. For a given amount of heat, a more dramatic increase in temperature is produced the lower is the heat capacity $\mathrm{C}$. Moreover as defined by equation (a) the heat capacity of a system is not a thermodynamic function of state because heat capacity describes a pathway accompanying a change in temperature. Hence, we define precisely the pathway taken by the system. Two important classes of heat capacities are 1. isobaric, $\mathrm{C}_{\mathrm{p}}$, and 2. isochoric, $\mathrm{CV}$. Isochoric and isobaric heat capacities are related to the isobaric expansions $\mathrm{E}_{\mathrm{p}}$ and isothermal compression $\mathrm{KT}$ using equation (b) [1]. $\mathrm{C}_{\mathrm{V}}=\mathrm{C}_{\mathrm{p}}-\mathrm{T} \,\left(\mathrm{E}_{\mathrm{p}}\right)^{2} / \mathrm{K}_{\mathrm{T}}$ Heat capacities and compressions are simply related [2]. $\mathrm{K}_{\mathrm{T}} / \mathrm{K}_{\mathrm{S}}=\mathrm{C}_{\mathrm{p}} / \mathrm{C}_{\mathrm{V}}$ Footnotes [1] According to a calculus operation, the dependences of entropy on temperature at constant volume and constant pressure are related. $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ A Maxwell equation requires that $\left(\frac{\partial S}{\partial p}\right)_{T}=-\left(\frac{\partial V}{\partial T}\right)_{p}$ Hence, $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ But the isobaric expansion, $E_{p}=\left(\frac{\partial V}{\partial T}\right)_{p}$ And the isothermal compression, $\mathrm{K}_{\mathrm{T}}=-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ From the Gibbs –Helmholtz equation, $\mathrm{C}_{\mathrm{V}}=\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ And $C_{p}=T \,\left(\frac{\partial S}{\partial T}\right)_{p}$ Then $C_{V}=C_{p}-T \,\left(E_{p}\right)^{2} / K_{T}$ The latter equation is correct under the condition of either ‘at constant affinity $\mathrm{A}$’ or ‘at constant composition’. [2] The starting point is the following equation. $\begin{array}{r} \left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=-\left(\frac{\partial \mathrm{T}}{\partial \mathrm{p}}\right)_{\mathrm{V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=-\left(\frac{\partial \mathrm{T}}{\partial \mathrm{p}}\right)_{\mathrm{V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{S}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \ \left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{S}}=-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{S}}\right)_{\mathrm{p}}=-\left(\frac{\partial \mathrm{T}}{\partial \mathrm{p}}\right)_{\mathrm{v}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{S}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{V}} \end{array}$ Then $(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{T}} /(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{S}}=(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{p}} /(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{V}}$ Hence, $\mathrm{K}_{\mathrm{T}} / \mathrm{K}_{\mathrm{S}}=\mathrm{C}_{\mathrm{p}} / \mathrm{C}_{\mathrm{V}}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.02%3A_Heat_Capacity-_Isobaric-_Partial_Molar-_Solution.txt
An aqueous solution is prepared using $\mathrm{n}_{1}$ moles of water($\ell$) and $\mathrm{n}_{j}$ moles of solute at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The system is at equilibrium where the affinity for spontaneous change is zero. Hence, $\mathrm{H}^{\mathrm{eq}}=\mathrm{H}^{\mathrm{eq}}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}, \mathrm{A}=0, \xi^{\mathrm{eq}}\right]$ Heat capacities describe a pathway. There are two limiting pathways. The system can be displaced either to a nearby state along a pathway for which $\xi$ is constant or along a pathway for which the affinity for spontaneous change is constant. The accompanying differential changes in enthalpies are unlikely to be the same. In fact they are related using a calculus procedure. $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{A}, \mathrm{p}}=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\xi, \mathrm{p}}-\left(\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right)_{\xi, \mathrm{p}} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}, \mathrm{p}} \,\left(\frac{\partial \mathrm{H}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}$ If the original state was an equilibrium state we write this equation in the following form which incorporates an equation for the dependence of affinity $\mathrm{A}$ on temperature at equilibrium. $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{A}=0, \mathrm{p}}=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\xi \mathrm{eq}, \mathrm{p}}-\frac{1}{\mathrm{~T}} \,\left[\left(\frac{\partial \mathrm{H}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}\right]^{2} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}$ Then from the definition of isobaric heat capacity [1], $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)=\mathrm{C}_{\mathrm{p}}\left(\xi^{\mathrm{eq}}\right)-\frac{1}{\mathrm{~T}} \,\left[\left(\frac{\partial \mathrm{H}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}\right]^{2} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}$ Here $(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}$, is the enthalpy of reaction. For a stable equilibrium state $(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}$, is negative. Hence, $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)>\mathrm{C}_{\mathrm{p}}\left(\xi^{\mathrm{eq}}\right)$ Here $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$ is the equilibrium heat capacity signalling that when heat q passes into the system the composition/organisation of the system changes in order that the Gibbs energy of the system remains at a minimum. In contrast $\mathrm{C}_{\mathrm{p}}\left(\xi^{\mathrm{eq}}\right)$ is the frozen capacity signalling that no changes occur in the composition/organisation in the system such that the Gibbs energy of the system is displaced from the original minimum. Moreover, equation (e) shows that the equilibrium isobaric heat capacity is always larger than the frozen isobaric heat capacity. Indeed we can often treat the extensive equilibrium property $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$ as a function of state (although it is not). Footnote [1] $\frac{1}{T} \,\left[\left(\frac{\partial \mathrm{H}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}^{\mathrm{eq}}\right]^{2} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}, \mathrm{p}}=\frac{1}{[\mathrm{~K}]} \,\left[\frac{\mathrm{J}}{\mathrm{mol}}\right]^{2} \,\left[\frac{\mathrm{mol}}{\mathrm{J} \mathrm{mol}^{-1}}\right]=\left[\mathrm{J} \mathrm{K}^{-1}\right]$ 1.12.05: Heat Capacity- Isobaric- Solutions- Excess A given solution is prepared using $1 \mathrm{~kg}$ of solvent (water) and $\mathrm{m}_{j}$ moles of solute $j$. If the thermodynamic properties of this solution are ideal, the isobaric heat capacity can be expressed as follows [1]. $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ On the other hand for a real solution the isobaric heat capacity can be expressed in terms of the apparent molar heat capacity of the solute, $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$. $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl}}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ The difference between $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ and $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ defines the relative isobaric heat capacity of the solution $\mathrm{J}$, an excess property. $\mathrm{J}(\mathrm{aq})=\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ Thermodynamics does not define the magnitude or sign of $\mathrm{J}(\mathrm{aq})$. However, from the definitions of ideal and real partial molar isobaric capacities of solvent and solute, the following condition must hold. $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{J}(\mathrm{aq})=0$ Relative quantities can also be defined for solute and solvent. $\mathrm{J}_{j}(\mathrm{aq})=\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})-\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ $\mathrm{J}_{1}(\mathrm{aq})=\mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})-\mathrm{C}_{\mathrm{p} 1}^{*}(\ell)$ Also, $\phi\left(\mathrm{J}_{\mathrm{j}}\right)=\phi\left[\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})\right]-\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ Hence, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{J}_{\mathrm{j}}(\mathrm{aq})=\mathrm{J}_{1}(\mathrm{aq})=\phi\left(\mathrm{J}_{\mathrm{j}}\right)=0$ Equation (c) defines a property $\mathrm{J}$ which is an excess isobaric heat capacity of a solution prepared using $1 \mathrm{~kg}$ of water. Thus, $\mathrm{C}_{\mathrm{p}}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{J}(\mathrm{aq})=\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ From equations (a) and (b), $\mathrm{C}_{\mathrm{p}}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{m}_{\mathrm{j}} \,\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})\right]$ From equation (g), $\mathrm{C}_{\mathrm{p}}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{J}_{\mathrm{j}}\right)$ Thus $\phi \left(\mathrm{J}_{j}\right)$ is the relative apparent molar isobaric heat capacity of the solute in a given real solution. Isobaric heat capacities of solutions and related partial molar isobaric heat capacities reflect in characteristic fashion the impact of added solutes on water water interactions Footnote [1] $\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]=\left[\mathrm{kg} \mathrm{mol}^{-1}\right]^{-1} \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]+\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.04%3A_Heat_Capacities-_Isobaric-_Equilibrium_and_Frozen.txt
Interesting cases emerge where an equilibrium isobaric heat capacity reflects a change in composition as a consequence of the system changing composition in order to hold the system at equilibrium. The development of sensitive scanning calorimeters stimulated research in this subject [1], particularly with respect to biochemical research [2]; e.g. multilamellar systems [3]. We consider the case where a solution is prepared using $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{\mathrm{x}}$ moles of solute $\mathrm{X}$. In solution at temperature $\mathrm{T}$ (and fixed pressure $\mathrm{p}$) the following chemical equilibrium is established. $\mathrm{X}(\mathrm{aq}) \Longrightarrow \mathrm{Y}(\mathrm{aq})$ At equilibirum $\mathrm{n}_{\mathrm{x}}^{0}-\xi$ $\xi$ moles or, $\mathrm{n}_{\mathrm{x}}^{0} \,\left[1-\frac{\xi}{\mathrm{n}_{\mathrm{x}}^{0}}\right]$ $\mathrm{n}_{\mathrm{x}}^{0} \,\left[\frac{\xi}{\mathrm{n}_{\mathrm{x}}^{0}}\right]$ moles By definition $\alpha=\xi / \mathrm{n}_{\mathrm{x}}^{0}$, the degree of reaction forming substance $\mathrm{Y}$ at equilibrium. Then $\mathrm{n}_{\mathrm{x}}^{\mathrm{eq}}=\mathrm{n}_{\mathrm{x}}^{0} \,(1-\alpha)$ and $\mathrm{n}_{\mathrm{y}}^{\mathrm{eq}}=\alpha \, \mathrm{n}_{\mathrm{x}}^{0}$. If $\mathrm{w}_{1}$ is the mass of solvent, water($\ell$), the equilibrium molalities are $\mathrm{m}_{\mathrm{x}}^{0} \,(1-\alpha)$ for chemical substance $\mathrm{X}$ and $\alpha \, \mathrm{m}_{\mathrm{x}}^{0}$ for chemical substance $\mathrm{Y}$. For the purposes of the arguments advanced here we assume that the thermodynamic properties of the solution are ideal. The equilibrium composition of the closed system at defined temperature and pressure is described by the equilibrium constant $\mathrm{K}^{0}$. Then, $\mathrm{K}^{0}=\alpha /(1-\alpha)$ Hence the (dimensionless and intensive) degree of reaction, $\alpha=\mathrm{K}^{0} /\left(1+\mathrm{K}^{0}\right)$ Because $\mathrm{K}^{0}$ is dependent on temperature then so is the degree of reaction. The extent to which an increase in temperature favours or disfavours formation of more $\mathrm{Y}(\mathrm{aq})$ depends on the sign of the enthalpy of reaction, $\Delta_{r}\mathrm{H}^{0}$. Thus [4,5], $\frac{\mathrm{d} \alpha}{\mathrm{dT}}=\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}} \,\left[\frac{\mathrm{d} \ln \mathrm{K}^{0}}{\mathrm{dT}}\right]$ The analysis at this point is considerably simplified if we assume that the limiting enthalpy of reaction, $\Delta_{r}\mathrm{H}^{\infty}(\mathrm{aq})$ for the chemical reaction is independent of temperature (at pressure $\mathrm{p}$). Hence using the van’t Hoff equation, $\frac{\mathrm{d} \alpha}{\mathrm{dT}}=\frac{1}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)} \, \Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$ Thus the shift in the composition of the solution depends on the sign of the limiting enthalpy of reaction. If $\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})<0$, an increase in temperature favours an increase in the amount of $\mathrm{X}(\mathrm{aq})$ at the expense of $\mathrm{Y}(\mathrm{aq})$. At temperature $\mathrm{T}$, the enthalpy of the solution is given by equation (e) where $\mathrm{H}_{1}^{*}(\ell)$ is the molar enthalpy of the solvent. $\mathrm{H}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{n}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{x}}^{0} \,(1-\alpha) \, \mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\mathrm{n}_{\mathrm{x}}^{0} \, \alpha \, \mathrm{H}_{\mathrm{y}}^{\infty}(\mathrm{aq})$ or, $\mathrm{H}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{n}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{x}}^{0} \, \mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\alpha \, \mathrm{n}_{\mathrm{x}}^{0} \,\left[\mathrm{H}_{\mathrm{y}}^{\infty}(\mathrm{aq})-\mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})\right]$ or, $\mathrm{H}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{n}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{x}}^{0} \, \mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\alpha \, \mathrm{n}_{\mathrm{x}}^{0} \, \Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$ We assume that $\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$ is independent of temperature together with the amount $\mathrm{n}_{1}$. Then, $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)=\left(\frac{\partial \mathrm{H}(\mathrm{aq}: \mathrm{A}=0)}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Hence, $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)=\left\{\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{x}}^{0} \, \mathrm{C}_{\mathrm{px}}^{\infty}(\mathrm{aq})\right\}+\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq}) \, \mathrm{n}_{\mathrm{x}}^{0}(\mathrm{~d} \alpha / \mathrm{dT})$ The terms in the { } brackets are not (formally) dependent on temperature and constitute a frozen contribution to $\mathrm{C}_{\mathrm{p}}(\mathrm{aq}, \mathrm{A}=0)$, $\mathrm{C}_{\mathrm{p}}(\mathrm{aq}: \xi)$. Then equations (d) and (i) yield an equation for $\mathrm{C}_{\mathrm{p}}(\mathrm{aq}: \mathrm{A}=0)$ in terms of $\left[\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})\right]^{2}$. $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{C}_{\mathrm{p}}(\xi: \mathrm{aq})+\left[\frac{\left[\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})\right]^{2}}{\mathrm{RT}^{2}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}\right] \, \mathrm{n}_{\mathrm{x}}^{0}$ in terms of one mole of chemical substance $\mathrm{X}$ (i.e. $\mathrm{n}_{\mathrm{x}}^{0}=1 \mathrm{~mol}$) [6], $\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{A}=0)=\mathrm{C}_{\mathrm{p}}(\xi ; \mathrm{aq})+\left[\frac{\left[\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})\right]^{2}}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}\right]$ According to equation (k) a large equilibrium heat capacity is favoured by a high $\mathrm{C}_{\mathrm{p}}(\xi ; \mathrm{aq})$ and a large enthalpy of reaction. The term $\left[\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})\right]^{2}$ ensures that irrespective of whether the reaction (as written) is exothermic or endothermic, $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)-\mathrm{C}_{\mathrm{p}}(\xi)$ is positive. The dependence of $\left[\mathrm{C}_{\mathrm{p}}(\mathrm{aq}: \mathrm{A}=0)-\mathrm{C}_{\mathrm{p}}(\xi: \mathrm{aq})\right]$ on temperature forms a bell-shaped plot covering the range of temperatures when all added substance X is completely in the form of $\mathrm{X}$ or of $\mathrm{Y}$. The maximum in the bell occurs near the temperature at which $\mathrm{K}^{0}$ is unity [7]. If $\mathrm{K}^{0}$ is unity, at this temperature $\Delta_{r}\mathrm{G}^{0}$ is zero. In other words, at this temperature the reference chemical potentials of $\mathrm{X}$ and $\mathrm{Y}$, $\mu^{0}(\mathrm{X})$ and $\mu^{0}(\mathrm{Y})$ respectively are equal. Clearly therefore the temperature at the maximum in $\mathrm{C}_{p}(\mathrm{~A}=0)-\mathrm{C}_{p}(\xi)$ is characteristic of the two solutes [8]. Equation (k) forms the basis of the technique of differential scanning calorimetry (DSC) as applied to the investigation of the thermal stability of biologically important macromolecules [2,9,10]. In the text book case, a plot of isobaric heat capacity against temperature forms a bell-shaped curve, the maximum corresponding to temperature at which the equilibrium constant for an equilibrium having the simple form discussed above is unity. The area under the curve yields the enthalpy change characterising the transition between the two forms $\mathrm{X}$ and $\mathrm{Y}$ of a single substance. The possibility exists that the temperature dependences of$\mu^{0}(\mathrm{X})$ and $\mu^{0}(\mathrm{Y})$ are such that the two plots intersect at two temperatures producing two maxima in the plot of $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)-\mathrm{C}_{\mathrm{p}}(\xi)$ against temperature. The patterns recorded by DSC scans for a $\mathrm{X} \leftrightarrows \mathrm{Y}$ system can be understood in terms of the separate dependences of $\left(\mu_{\mathrm{X}}^{0} / \mathrm{T}\right)$ and $\left(\mu_{\mathrm{Y}}^{0} / \mathrm{T}\right)$ on temperature, where $\mu^{0}(\mathrm{X})$ and $\mu^{0}(\mathrm{Y})$ are the standard chemical potentials of substances $\mathrm{X}$ and $\mathrm{Y}$. The maximum in the recorded heat capacity occurs where the plots of $\left(\mu_{\mathrm{X}}^{0} / \mathrm{T}\right)$ and $\left(\mu_{\mathrm{Y}}^{0} / \mathrm{T}\right)$ against temperature cross [11,12]. If these curves have a more complicated shape there is the possibility that they will cross at two temperatures. In fact this observation raises the possibility of identifying hot and cold denaturation of proteins using DSC. Similar extrema in isobaric heat capacities are recorded for gel-to-liquid transitions in vesicles [13,14]. In more complex systems, the overall DSC scan can indicate the presence of domains in a macromolecule which undergo structural changes when the temperature is raised [15,16]. Analysis of extrema in heat capacities becomes somewhat more complicated when two or more equilibria are coupled [17,18]. Footnotes [1] V. V. Plotnikov, J. M. Brandts, L.-N. Lin and J. F. Brandts, Anal. Biochem., 1997, 250,237. [2] J. M. Sturtevant, Ann. Rev. Phys.Chem.,1987,38,463. [3] S. Mabrey and J. M. Sturtevant, Proc. Natl. Acad., Sci. USA, 1976, 73, 3862. [4] From equation (b) \begin{aligned} &\frac{\mathrm{d} \alpha}{\mathrm{dT}}=\frac{\mathrm{d}}{\mathrm{dT}}\left[\mathrm{K}^{0} \,\left(1+\mathrm{K}^{0}\right)^{-1}\right]=\left[\frac{1}{1+\mathrm{K}^{0}}-\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}}\right] \, \frac{\mathrm{dK}}{\mathrm{dT}} \ &=\frac{1}{\left(1+\mathrm{K}^{0}\right)^{2}} \, \frac{\mathrm{dK}^{0}}{\mathrm{dT}}=\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}} \, \frac{1}{\mathrm{~K}^{0}} \, \frac{\mathrm{dK}^{0}}{\mathrm{dT}}=\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}} \,\left[\frac{\mathrm{d} \ln \mathrm{K}^{0}}{\mathrm{dT}}\right] \end{aligned} [5] $\frac{\mathrm{d} \alpha}{\mathrm{dT}}=\frac{1}{\left[\mathrm{~J} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right]} \, \frac{1}{[\mathrm{~K}]^{2}} \, \frac{[1]}{[1]} \,\left[\mathrm{J} \mathrm{mol}^{-1}\right]=\frac{1}{[\mathrm{~K}]}$ [6] $\frac{\left[\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})\right]^{2}}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}=\frac{\left[\mathrm{J} \mathrm{mol}^{-1}\right]^{2}}{\left.\left[\mathrm{~J} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \mathrm{K}\right]^{2}} \, \frac{[1]}{[1]^{2}}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]$ [7] With $\mathrm{h}=\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$, the second term in equation (k) can be written as follows $\mathrm{y}=\frac{\mathrm{h}^{2}}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}}$. The pattern formed by the dependence of $\mathrm{y}$ on temperature is given by $\frac{\mathrm{dy}}{\mathrm{dT}}=\frac{\mathrm{h}^{2}}{\mathrm{R} \, \mathrm{T}^{2}} \,\left[\frac{1}{\left(1+\mathrm{K}^{0}\right)^{2}}-\frac{2 \, \mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{3}}\right] \frac{\mathrm{dK}^{0}}{\mathrm{dT}}-\frac{2 \, \mathrm{h}^{2} \, \mathrm{K}^{0}}{\mathrm{R} \, \mathrm{T}^{3} \,\left(1+\mathrm{K}^{0}\right)^{2}}$ Or, $\frac{\mathrm{dy}}{\mathrm{dT}}=\frac{\mathrm{h}^{2} \, \mathrm{K}^{0}}{\mathrm{R} \, \mathrm{T}^{2} \,\left(1+\mathrm{K}^{0}\right)^{2}} \,\left[\frac{1-\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)} \, \frac{\mathrm{d} \ln \left(\mathrm{K}^{0}\right)}{\mathrm{dT}}-\frac{2}{\mathrm{~T}}\right]$ Since $\mathrm{d} \ln \left(\mathrm{K}^{0}\right) / \mathrm{dT}=\mathrm{h} / \mathrm{R} \, \mathrm{T}^{2}$, then $\frac{\mathrm{dy}}{\mathrm{dT}}=\frac{\mathrm{h}^{2} \, \mathrm{K}^{0}}{\mathrm{R} \, \mathrm{T}^{3} \,\left(1+\mathrm{K}^{0}\right)^{2}} \,\left[\frac{1-\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)} \, \frac{\mathrm{h}}{\mathrm{R} \, \mathrm{T}}-2\right]$ Hence the condition for an extremum in $\mathrm{y}$ as a function of $\mathrm{T}$ is $\frac{1-\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)} \, \frac{\mathrm{h}}{\mathrm{R} \, \mathrm{T}}-2=0$ Or $\frac{1-\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)}=\frac{2 \, \mathrm{R} \, \mathrm{T}}{\mathrm{h}}$ Then $\mathrm{K}^{0}=\frac{1-(2 \, \mathrm{R} \, \mathrm{T} / \mathrm{h})}{1+(2 \, \mathrm{R} \, \mathrm{T} / \mathrm{h})}$ By definition $h=\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$. Therefore if the magnitude of $\Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})$ is much larger than $2 \, \mathrm{~R} \, \mathrm{T}$, the top of the bell shaped curve is reached at a temperature where $\mathrm{K}^{0}$ is unity. In the general case, at approximately this temperature $\mathrm{y}$ is a maximum. [8] M. J. Blandamer, J. Burgess and J. M. W. Scott, Ann. Rep. Prog. Chem., Sect. C, Phys. Chem.,1985,82,77. [9] J. M. Sturtevant, Thermodynamic Data for Biochemistry and Biotechnlogy, ed. H. J. Hinz, Springer-Verlag, Berlin,1974, pp. 349- 376. [10] Th. Ackermann, Angew. Chem. Int. Ed. Engl.,1989, 28, 981. [11] 1. M. J. Blandamer, B. Briggs, J. Burgess and P. M. Cullis, J. Chem. Soc. Faraday Trans., 1990, 86, 1437. 2. V. Edge, N. M. Allewell and J. M. Sturtevant, Biochemistry.,1985,24,5899. 3. P. L. Privalov, Yu. V. Griko, S.Yu. Venyaminov and V. P. Kutyshenko, J. Mol. Biol.,1986,190,487. 4. W. M. Jackson and J. M. Brandts, Biochem.istry,1970,9,2294. [12] 1. F. Franks and T. Wakabayashi, Z. Phys. Chem. Neue Folge, 1987, 155, 171. 2. F. Franks and R. M. Hateley, Cryo-Letters, 1985,6,171;1986,7,226. 3. F. Franks, R. H. M. Hateley and H. L. Friedman, Biophys. Chem., 1988, 31, 307. [13] M. J. Blandamer, B. Briggs, M. D. Butt, M. Waters, P. M. Cullis, J. B. F. N. Engberts, D. Hoekstra and R. K. Mohanty, Langmuir, 1994, 10, 3488. [14] M. J. Blandamer, B. Briggs, P. M. Cullis, J. B. F. N. Engberts and D. Hoekstra, J. Chem. Soc. Faraday Trans., 1994, 90, 1905. [15] C. O. Pabo, R. T. Sauer, J. M. Sturtevant and M. Ptashne, Proc. Natl. Acad. .Sci., USA, 1979, 76, 1608. [16] M. J. Blandamer, B. Briggs, P. M. Cullis, A. P. Jackson, A. Maxwell and R. J. Reece, Biochemistry, 1994, 33, 7510. [17] M. J. Blandamer, J. Burgess and J. M. W. Scott, J. Chem. Soc. Faraday Trans.1, 1984, 80, 2881. [18] G. J. Mains, J. W. Larson and L. G. Hepler, J.Phys.Chem.,1985,88,1257.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.06%3A_Heat_Capacities-_Isobaric-_Dependence_on_Temperature.txt
The volume of a closed system at thermodynamic equilibrium containing two chemical substances is defined by equation (a). $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}\right]$ At temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the chemical composition / organisation $\xi^{\mathrm{eq}}$ corresponds to the state where the affinity for spontaneous change is zero. Isobaric Expansions The system is displaced by a change in temperature to a neighbouring state where the affinity for spontaneous change is also zero; the organisation/composition changes to $\xi^{\mathrm{eq}}(\mathrm{new})$. $\mathrm{V}(\text { new })=\mathrm{V}\left[\mathrm{T}(\text { new }), \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}(\text { new })\right]$ The differential dependence on temperature of the volume defined in equation. (a) is the equilibrium isobaric thermal expansion, $\mathrm{E}_{\mathrm{p}}(\mathrm{A}=0)$ [1]. $\mathrm{E}_{\mathrm{p}}(\mathrm{A}=0)=\left(\frac{\partial \mathrm{V}}{\partial T}\right)_{\mathrm{p}, \mathrm{A}=0}$ The chemical composition/organisation changes to hold the affinity for spontaneous change at zero. Indeed the perturbation in the form of a change in temperature might have to be extremely slow so that the change in organisation/chemical composition keeps in step with the change in temperature. The isobaric expansion $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})$ for an aqueous solution containing solute $j$ is related to the partial molar isobaric expansions of solute and solvent; equation (d). $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{pl}}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ Alternatively using the concept of an apparent molar property, we define an (equilibrium) apparent molar isobaric expansion for solute $j$, $\phi\left(E_{p j}\right)$. $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{pl} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right) / \partial \mathrm{T}\right]_{\mathrm{p}}$ Isentropic Expansions Generally little interest has been shown in either partial molar or apparent molar isentropic expansions of solutes. Complications are encountered in understanding isentropic expansions without the redeeming feature of practical accessibility via an analogue of the Newton -Laplace equation. The isentropic expansions $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ is defined by equation (g). $\mathrm{E}_{\mathrm{S}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{S}(\mathrm{aq})}$ The constraint on the partial derivative refers to the entropy of the solution $\mathrm{S}(\mathrm{aq})$. As we change the amount of solute nj for fixed temperature and fixed pressure and amount of solvent $\mathrm{n}_{1}$, so both $\mathrm{V}(\mathrm{aq})$ and $\mathrm{S}(\mathrm{aq})$ change yielding a new isentropic thermal expansion, $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ at a new entropy $\mathrm{S}(\mathrm{aq})$. For a series of solutions having different molalities of solute, comparison of $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ is not straightforward because ${\mathrm{S}(\mathrm{aq})$ is itself a function of solution composition. Further comparison cannot be readily drawn with the isentropic thermal expansion of the solvent, $\mathrm{E}_{\mathrm{SI}}^{*}(\ell)$; equation (h). $\mathrm{E}_{\mathrm{S} 1}^{*}(\ell)=\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right) \quad \text { at constant } \quad \mathrm{S}_{1}^{*}(\ell)$ $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ is a non-Gibbsian property. Consequently familiar thermodynamic relationships involving partial molar properties are not valid in the case of partial molar isentropic thermal expansions which are non-Lewisian properties. $\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{s}(\mathrm{aq})}$ is a semi-partial property. Footnote [1] For a system at equilibrium where $\mathrm{A} = 0$, $\frac{\partial^{2} G}{\partial T \, \partial p}=\frac{\partial^{2} G}{\partial p \, \partial T}$ Therefore, $\mathrm{E}_{\mathrm{p}}(\mathrm{A}=0)=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0}=-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}=0}$ 1.12.08: Expansions- Solutions- Isobaric- Partial and Apparent Molar The volume of a given aqueous solution containing $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ moles of solute $j$ is related to the composition by equation (a). $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ $\mathrm{V}_{1}(\mathrm{aq})$ and $\mathrm{V}_{j}(\mathrm{aq})$ are the partial molar volumes of water and solute $j$ respectively. The (equilibrium) isobaric thermal expansion of the solution (at fixed pressure) $\mathrm{E}_{\mathrm{p}}$ characterises the differential dependence of $\mathrm{V}(\mathrm{aq})$ on temperature. $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=[\partial \mathrm{V}(\mathrm{aq}) / \partial \mathrm{T}]_{\mathrm{p}, \mathrm{A}=0}$ $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})$ is an extensive property of the solution [1]. Two partial molar isobaric thermal expansions are defined, characteristic of solute and solvent [2]. $\mathrm{E}_{\mathrm{p} 1}(\mathrm{aq})=\left(\partial \mathrm{V}_{1}(\mathrm{aq}) / \partial \mathrm{T}\right)_{\mathrm{p}}$ $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\left(\partial \mathrm{V}_{\mathrm{j}}(\mathrm{aq}) / \partial \mathrm{T}\right)_{\mathrm{p}}$ From equation (a), $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{p} 1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ In the treatment of volumetric properties of solutions we define an apparent molar volume of the solute, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$. By analogy we rewrite equation (e) in a form which defines the apparent molar isobaric expansion of the solute, $\phi\left(\mathrm{E}_{\mathrm{j}}\right)$. Thus, $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ Here [3], $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ For the pure solvent, $\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)=\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Footnotes [1] $\mathrm{E}_{\mathrm{p}}$ is an extensive property; the larger the volume $\mathrm{V}$ the larger the change in volume for a given increase in temperature. [2] $E_{p}=\left[m^{3} K^{-1}\right] \quad E_{p 1}=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \quad E_{p j}=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right]$ [3] $\phi\left(\mathrm{E}_{\mathrm{j}}\right)=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right]$ 1.12.09: Expansions- Apparent Molar Isobaric- Composition Dependence For many aqueous solutions at ambient temperature and pressure the dependence of apparent molar isobaric expansions for solute $j$ $\phi\left(E_{p j}\right)$ on molality $\mathrm{m}_{j}$ is accounted for using an equation having the following general form. [The reason for choosing the molality scale is that $\mathrm{m}_{j}$ is independent of $\mathrm{T}$ and $\mathrm{p}$ whereas concentration $\mathrm{c}_{j}$ is not. $\phi\left(E_{p j}\right)=a_{1}+a_{2} \,\left(m_{j} / m^{0}\right)+a_{3} \,\left(m_{j} / m^{0}\right)^{2} \ldots . .$ At low solute molalities the linear term is dominant. Granted therefore that equation (a) accounts for the observed pattern, we need to explore the analysis a little further. There are advantages in linking $\phi\left(E_{p j}\right)$ and the partial molar property $\mathrm{E}_{p j}(\mathrm{aq})$. For an aqueous solution prepared using $1 \mathrm{~kg}$ of water and $\mathrm{m}_{j}$ moles of solute $j$ at fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence, $\mathrm{E}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ $\left(\frac{\partial \mathrm{E}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)=\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{E}_{\mathrm{pj}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)+\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ But $\left(\frac{\partial \mathrm{E}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right.}{\partial \mathrm{m}_{\mathrm{j}}}\right)=\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{E}_{\mathrm{pj}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)+\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ Hence the partial molar isobaric expansions for solute $j$ can be calculated using the apparent molar isobaric expansions and its dependence on molality. Further if equation (a) accounts for the dependence of $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ on $\mathrm{m}_{j}$, then $\mathrm{E}_{\mathrm{pj}}=\mathrm{a}_{1}+2 \, \mathrm{a}_{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+3 \, \mathrm{a}_{3} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2} \ldots \ldots$ Therefore, using equations (a) and (g), $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ In the next stage of the analysis we develop an argument starting with an equation for the chemical potential of solute $j$ in solution. $\mu_{j}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)+\int_{\mathrm{p}^{0}}^{p} \mathrm{~V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T}) \, \mathrm{dp}$ Then with $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})=\left(\frac{\partial \mu_{\mathrm{j}}(\mathrm{aq})}{\partial \mathrm{p}}\right)_{\mathrm{T}}$, $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}$ With $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\left(\frac{\partial \mathrm{V}_{\mathrm{j}}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{p}}$, $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\mathrm{R} \,\left(\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}+\mathrm{R} \, \mathrm{T} \,\left(\frac{\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{p} \, \partial \mathrm{T}}\right)$ In the case of dilute solutions we might assert that $\ln \left(\gamma_{\mathrm{j}}\right)$ is a linear function of molality $\mathrm{m}_{j}$. Thus[1], $\ln \left(\gamma_{\mathrm{j}}\right)=\mathrm{S}_{\gamma} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ By definition [2], $\mathrm{S}_{\mathrm{V}}=\left(\frac{\partial \mathrm{S}_{\gamma}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ and[3] $\mathrm{S}_{\mathrm{Ep}}=\mathrm{S}_{\mathrm{V}}+\mathrm{T} \,\left(\frac{\partial \mathrm{S}_{\mathrm{V}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Then[4] $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{S}_{\mathrm{Ep}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Thus we identify the basis of the parameter a2 in equation (a). Footnotes [1] $\ln \left(\gamma_{\mathrm{j}}\right)=[1] \,\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{mol} \mathrm{kg}{ }^{-1}\right]^{-1}$ [2] $\mathrm{S}_{\mathrm{V}}=[1] /\left[\mathrm{N} \mathrm{m}^{-2}\right]=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}$ [3] $\mathrm{S}_{\mathrm{Ep}}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}+[\mathrm{K}] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \,\left[\mathrm{K}^{-1}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}\right.$ [4] \begin{aligned} \mathrm{E}_{\mathrm{pj}}(\mathrm{aq}) &=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \,\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{mol} \mathrm{kg}^{-1}\right]^{-1} \ &=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\left[\mathrm{N} \mathrm{m} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \ &=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.07%3A_Expansions-_Isobaric_and_Isentropic.txt
The volume of an aqueous solution $\mathrm{V}(\mathrm{aq})$ is related to the amounts of solvent and solute through the molar volume of water $\mathrm{V}_{1}^{*}(\ell)$ and the apparent molar volume of solute $\phi \left(\mathrm{V}_{j}\right) at the same temperature and pressure; equation (a). $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ The isobaric temperature dependence of the apparent molar volume of solute \(\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ yields the apparent molar (isobaric) expansion of solute $j$, $\phi\left(\mathrm{E}_{j}\right)$. $\phi\left(E_{p j}\right)=\left(\frac{\partial \phi\left(V_{j}\right)}{\partial T}\right)_{p}$ Equation (a) (as in most treatments of volumetric properties) is the starting equation for the development of equations which relate apparent molar isobaric expansions of a solute $j$ to the measured isobaric expansibilities of solvent and solution. The following four equivalent equations are frequently quoted [1-8]. A method is also available for direct determination of $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ from density data determined as functions of $\mathrm{T}$ and $\mathrm{m}_{j}$ [9]. Molality Scale [1-3] $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{pl}}^{*}(\ell)\right]+\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ $\begin{gathered} \phi\left(E_{p j}\right)=\left[m_{j} \, \rho(a q) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{p}(a q) \, \rho_{1}^{*}(\ell)-\alpha_{p 1}^{*}(\ell) \, \rho(a q)\right] \ +\alpha_{p}(a q) \, M_{j} \,[\rho(a q)]^{-1} \end{gathered}$ Concentration Scale [4 - 7] $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\frac{1}{\mathrm{c}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq}) \, \rho_{1}^{*}(\ell)-\rho(\mathrm{aq}) \, \alpha_{\mathrm{pl}}^{*}(\ell)\right]+\alpha_{\mathrm{p} 1}^{*}(\ell) \, \mathrm{M}_{\mathrm{j}} / \rho_{1}^{*}(\ell)$ $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{pl}}^{*}(\ell)\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{p} 1}^{*}(\ell)$ The four equations (c) - (f) are thermodynamically correct, no assumptions being made in their derivation. The partial molar isobaric expansion $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ is obtained using equation (g) [8]. $E_{p j}(a q)=\phi\left(E_{p j}\right)+m_{j} \,\left(\frac{\partial \phi\left(E_{p j}\right)}{\partial m_{j}}\right)_{p}$ Footnotes [1] From equation (a) with respect to the dependence of $\mathrm{V}(\mathrm{aq}$ on temperature at constant $\mathrm{p}$ and at “$\mathrm{A} = 0$”. $(\partial \mathrm{V}(\mathrm{aq}) / \partial \mathrm{T})_{\mathrm{p}}=\mathrm{n}_{1} \,\left(\partial \mathrm{V}_{1}^{*}(\ell) / \partial \mathrm{T}\right)_{\mathrm{p}}+\mathrm{n}_{\mathrm{j}} \,\left(\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right) / \partial \mathrm{T}\right)_{\mathrm{p}}$ Using equation (b), $(\partial \mathrm{V}(\mathrm{aq}) / \partial \mathrm{T})_{\mathrm{p}}=\mathrm{n}_{1} \,\left(\partial \mathrm{V}_{1}^{*}(\ell) / \partial \mathrm{T}\right)_{\mathrm{p}}+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ Hence, $\left(\frac{1}{V(a q)}\right) \,\left(\frac{\partial \mathrm{V}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\mathrm{n}_{1} \,\left(\frac{\mathrm{V}_{1}^{*}(\ell)}{\mathrm{V}_{1}^{*}(\ell)}\right) \, \frac{1}{\mathrm{~V}(\mathrm{aq})} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{V}(\mathrm{aq})}\right) \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ Thus, $\alpha_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \,\left(\frac{\mathrm{V}_{1}^{*}(\ell)}{\mathrm{V}(\mathrm{aq})}\right) \, \alpha_{\mathrm{p} 1}^{*}(\ell)+\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{V}(\mathrm{aq})} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ or, $\mathrm{V}(\mathrm{aq}) \, \alpha_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ We again use equation (a) for $\mathrm{V}(\mathrm{aq}$, $\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \, \alpha_{\mathrm{p}}(\mathrm{aq})-\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{pl}}^{*}(\ell)$ But, $\mathrm{V}_{1}^{*}(\ell)=\mathrm{M}_{1} / \rho_{1}^{*}(\ell)$ where $\mathrm{M}_{1}$ is the molar mass of the solvent water. $\phi\left(E_{p j}\right)=\frac{n_{1} \, M_{1}}{n_{j} \, \rho_{1}^{*}(\ell)} \, \alpha_{p}(\mathrm{aq})-\frac{n_{1} \, M_{1}}{n_{j} \, \rho_{1}^{*}(\ell)} \, \alpha_{1}^{*}(\ell)+\alpha_{p}(\mathrm{aq}) \, \phi\left(V_{j}\right)$ But $\mathrm{m}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{\mathrm{l}} \, \mathrm{M}_{\mathrm{l}}$. $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{p} 1}^{*}(\ell)\right]+\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence we obtain equation (c). [2] With reference to equation (c), \begin{aligned} &{\left[\frac{1}{\mathrm{~m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{p} 1}^{*}(\ell)\right]=\left[\frac{\mathrm{kg}}{\mathrm{mol}}\right] \,\left[\frac{\mathrm{m}^{3}}{\mathrm{~kg}}\right] \,\left[\mathrm{K}^{-1}\right]=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right]} \ &\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)=\left[\mathrm{K}^{-1}\right] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \end{aligned} [3] From $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{p} 1}^{*}(\ell)\right]+\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ and, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\frac{\mathrm{V}(\mathrm{aq})-\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)}{\mathrm{n}_{\mathrm{j}}}$ $\phi\left(E_{p j}\right)=\left[m_{j} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{p}(a q)-\alpha_{p l}^{*}(\ell)\right]+\alpha_{p}(a q) \,\left[\frac{\mathrm{V}(\mathrm{aq})-\mathrm{n}_{1} \, V_{1}^{*}(\ell)}{\mathrm{n}_{\mathrm{j}}}\right]$ $\phi\left(E_{p j}\right)=\left[m_{j} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{p}(\mathrm{aq})-\alpha_{p 1}^{*}(\ell)\right]+\alpha_{p}(\mathrm{aq}) \,\left[\frac{1}{\mathrm{c}_{j}}-\frac{\mathrm{n}_{1} \, \mathrm{M}_{1}}{\rho_{1}^{*}(\ell) \, \mathrm{n}_{\mathrm{j}}}\right]$ But $\frac{1}{c_{j}}=\frac{M_{j}}{\rho(a q)}+\frac{1}{m_{j} \, \rho(a q)}$ \begin{aligned} &\phi\left(E_{\mathrm{pj}}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{pl}}^{*}(\ell)\right]+\left[\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\mathrm{m}_{\mathrm{j}} \, \rho(\mathrm{aq})}+\frac{\alpha_{\mathrm{p}}(\mathrm{aq}) \, \mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}-\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\rho_{1}^{*}(\ell) \, \mathrm{m}_{\mathrm{j}}}\right] \ &\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}-\frac{\alpha_{\mathrm{pl}}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}+\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\mathrm{m}_{\mathrm{j}} \, \rho(\mathrm{aq})}+\frac{\alpha_{\mathrm{p}}(\mathrm{aq}) \, \mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}-\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\rho_{1}^{*}(\ell) \, \mathrm{m}_{\mathrm{j}}} \end{aligned} Hence we obtain equation (d). $\begin{gathered} \phi\left(E_{p j}\right)=\left[m_{j} \, \rho(a q) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{p}(a q) \, \rho_{1}^{*}(\ell)-\alpha_{p 1}^{*}(\ell) \, \rho(a q)\right] \ +\alpha_{p}(a q) \, M_{j} \,[\rho(a q)]^{-1} \end{gathered}$ [4] From $(\partial \mathrm{V}(\mathrm{aq}) / \partial \mathrm{T})_{\mathrm{p}}=\mathrm{n}_{1} \,\left(\partial \mathrm{V}_{1}^{*}(\ell) / \partial \mathrm{T}\right)_{\mathrm{p}}+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pij}}\right)$ $\alpha_{\mathrm{p}}(\mathrm{aq}) \, \mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{p} j \mathrm{j}}\right)$ Or, $\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\mathrm{V}(\mathrm{aq}) \, \alpha_{\mathrm{p}}(\mathrm{aq})-\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)$ But, $\rho(\mathrm{aq})=\left(\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right) / \mathrm{V}(\mathrm{aq})$ or, $\mathrm{n}_{1}=\left(\mathrm{V}(\mathrm{aq}) \, \rho(\mathrm{aq})-\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right) / \mathrm{M}_{1}$ \begin{aligned} &n_{j} \, \phi\left(E_{p j}\right)=\left[V(a q) \, \alpha_{p}(a q)\right]-\left[V(a q) \, \rho(a q)-n_{j} \, M_{j}\right] \, V_{1}^{*}(\ell) \, \alpha_{p l}^{*}(\ell) / M_{1} \ &\phi\left(E_{p j}\right)=\left[\frac{V(a q) \, \alpha_{p}(a q)}{n_{j}}\right]-\left[\frac{V(a q) \, \rho(a q) \, V_{1}^{*}(\ell) \, \alpha_{1}^{*}(\ell)}{n_{j} \, M_{1}}\right]+\left[\frac{V_{1}^{*}(\ell) \, \alpha_{p 1}^{*}(\ell) \, M_{j}}{M_{1}}\right] \end{aligned} But concentration $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}(\mathrm{aq})$ and $\rho_{1}^{*}=\mathrm{M}_{1} / \mathrm{V}_{1}^{*}(\ell)$. $\phi\left(\mathrm{E}_{\mathrm{j}}\right)=\left[\frac{\alpha_{\mathrm{p}}(\mathrm{aq})}{\mathrm{c}_{\mathrm{j}}}\right]-\left[\frac{\rho(\mathrm{aq}) \, \alpha_{\mathrm{pl}}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}\right]+\left[\frac{\alpha_{\mathrm{p} 1}^{*}(\ell) \, \mathrm{M}_{\mathrm{j}}}{\rho_{1}^{*}(\ell)}\right]$ Hence we obtain equation (e). $\phi\left(E_{p j}\right)=\left[\frac{1}{c_{j} \, \rho_{1}^{*}(\ell)}\right] \,\left[\alpha_{p}(\mathrm{aq}) \, \rho_{1}^{*}(\ell)-\rho(a q) \, \alpha_{p 1}^{*}(\ell)\right]+\alpha_{p 1}^{*}(\ell) \, M_{j} / \rho_{1}^{*}(\ell)$ [5] With reference to equation (e), \begin{aligned} &{\left[\frac{1}{\mathrm{c}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq}) \, \rho_{1}^{*}(\ell)-\rho(\mathrm{aq}) \, \alpha_{\mathrm{pl} 1}^{*}(\ell)\right]=\left[\frac{\mathrm{m}^{3}}{\mathrm{~mol}}\right] \,\left[\frac{\mathrm{m}^{3}}{\mathrm{~kg}}\right] \,\left[\mathrm{K}^{-1}\right] \,\left[\mathrm{kg} \mathrm{m}^{-3}\right]} \ &=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \end{aligned} [6] The volume of a solution, $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{1} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Concentration $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}(\mathrm{aq})$ or, $\mathrm{c}_{\mathrm{j}}=\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}$ But molality $\mathrm{m}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{\mathrm{l}} \, \mathrm{M}_{\mathrm{l}}$ $\mathrm{c}_{\mathrm{j}}=\frac{\mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}}{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}$ or, $\frac{1}{\mathrm{c}_{\mathrm{j}}}=\frac{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}}+\frac{\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\mathrm{n}_{1} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}}$ or, $\frac{1}{c_{j}}=\frac{1}{m_{j} \, \rho_{1}^{*}(\ell)}+\phi\left(V_{j}\right)$ or, $\frac{1}{\mathrm{~m}_{\mathrm{j}}}=\frac{\rho_{1}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}}}-\rho_{1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ From equation (c). $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\frac{1}{\rho_{1}^{*}(\ell)}\right] \,\left[\frac{\rho_{1}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}}}-\rho_{1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{p} 1}^{*}(\ell)\right]+\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Or, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\frac{1}{\mathrm{c}_{\mathrm{j}}}\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})\right.&\left.-\alpha_{\mathrm{pl}}^{*}(\ell)\right]-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{p}}(\mathrm{aq})+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{pl}}^{*}(\ell) \ &+\alpha_{\mathrm{p}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} We obtain equation (f) $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{pl}}^{*}(\ell)\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{pl}}^{*}(\ell)$ [7] With reference to equation (f) $\left[\frac{1}{\mathrm{c}_{\mathrm{j}}}\right] \,\left[\alpha_{\mathrm{p}}(\mathrm{aq})-\alpha_{\mathrm{pl}}^{*}(\ell)\right]=\left[\frac{\mathrm{m}^{3}}{\mathrm{~mol}}\right] \,\left[\mathrm{K}^{-1}\right]=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right]$ [8] From $\mathrm{E}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{p} j}\right)$ Then, $\left(\frac{\partial E_{p}}{\partial n_{j}}\right)_{T, p, n_{j}}=n_{j} \,\left[\frac{\partial \phi\left(E_{p j}\right)}{\partial n_{j}}\right]+\phi\left(E_{p j}\right)$ Or, $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{m}_{\mathrm{j}} \,\left[\frac{\partial \phi\left(\mathrm{E}_{\mathrm{pj}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right]+\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ [9] M. J. Blandamer and H. Hoiland, Phys.Chem.Chem.Phys.,1999,1,1873. This method starts out with the measured dependence of the density $\rho(\mathrm{aq})$ on temperature and molality at fixed pressure about density $\rho_{1}^{*}(\ell, \theta)$, at temperature $\theta$ at same pressure. For example the data might be fitted to an equation having the following form yielding the b-coefficients. \begin{aligned} &\rho\left(\mathrm{m}_{\mathrm{j}}, \mathrm{T}\right)=\rho_{1}^{*}(\ell, \theta)+\mathrm{b}_{2} \,(\mathrm{T}-\theta) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta+\mathrm{b}_{3} \,(\mathrm{T}-\theta) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2} / \theta \ &+\mathrm{b}_{4} \,(\mathrm{T}-\theta)^{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta^{2}+\mathrm{b}_{5} \,(\mathrm{T}-\theta)^{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2} / \theta^{2} \ &\left(\frac{\partial \rho\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T}\right)}{\partial \mathrm{T}}\right)_{\mathrm{m}_{\mathrm{j}}}=\mathrm{b}_{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta+\mathrm{b}_{3} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2} / \theta \end{aligned} $+2 \, b_{4} \,(\mathrm{T}-\theta) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta^{2}+2 \, \mathrm{b}_{5} \,(\mathrm{T}-\theta) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta^{2}$ and, \begin{aligned} &\left(\frac{\partial \rho\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T}\right)}{\partial\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)}\right)_{\mathrm{T}}=\mathrm{b}_{2} \,(\mathrm{T}-\theta) / \theta+2 \, \mathrm{b}_{3} \,(\mathrm{T}-\theta) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta \ &+\mathrm{b}_{4} \,(\mathrm{T}-\theta)^{2} / \theta^{2}+2 \, \mathrm{b}_{5} \,(\mathrm{T}-\theta)^{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) / \theta^{2} \end{aligned} The density $\rho(\mathrm{aq})$ of an aqueous solution molality $\mathrm{m}_{j}$ prepared using $1 \mathrm{~kg}$ of water is given by the following equation. $\rho(\mathrm{aq})=\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] / \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] / \rho(\mathrm{aq})$ Also, $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ $\left(\frac{\partial \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}}=\frac{\mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}-\frac{1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}}{[\rho(\mathrm{aq})]^{2}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}}$ But, $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})=\left(\frac{\partial \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}}$ And, $E_{p j}(a q)=\left(\frac{\partial V_{j}(a q)}{\partial T}\right)_{p, m_{j}}$ $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\frac{\partial}{\partial \mathrm{T}} \,\left\{\frac{\mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}-\left(\frac{1}{\rho(\mathrm{aq})}\right)^{2} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right]\right\}$ \begin{aligned} &\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=-\frac{\mathrm{M}_{\mathrm{j}}}{(\rho(\mathrm{aq}))^{2}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{m}_{\mathrm{j}}}+\frac{2}{\left(\rho(\mathrm{aq})^{3}\right)} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{m}_{\mathrm{j}}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] \ &-\frac{1}{(\rho(\mathrm{aq}))^{2}} \, \frac{\partial}{\partial \mathrm{T}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] \end{aligned} Using equation (k) in conjunction with equations (a) - (c), partial molar isobaric expansion $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$ is calculated from the density and its dependence on both temperature and molality of solute. In another development $\mathrm{E}_{\mathrm{pj}}$ is related to $\alpha_{\mathrm{p}}$ and its dependence on molality of solute. By definition, $\alpha_{p}(\mathrm{aq})=-\frac{1}{V(\mathrm{aq})} \,\left(\frac{\partial \mathrm{V}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{m}(\mathrm{j})}$ Or, $\alpha_{p}(a q)=-\frac{1}{\rho(a q)} \,\left(\frac{\partial \rho(a q)}{\partial T}\right)_{p, m(j)}$ At temperature $\mathrm{T}$ and molality $\mathrm{m}_{j}$, $\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{m}(\mathrm{j})}=-\alpha_{\mathrm{p}}(\mathrm{aq}) \, \rho(\mathrm{aq})$ Using equation (n) \begin{aligned} &E_{p j}(a q)=\frac{M_{j} \, \alpha_{p}}{\rho(a q)}-\frac{2}{(\rho(a q))^{2}} \, \alpha_{p} \,\left(\frac{\partial \rho(a q)}{\partial m_{j}}\right) \,\left[1+M_{j} \, m_{j}\right] \ &-\frac{1}{(\rho(a q))^{2}} \, \frac{\partial}{\partial m_{j}} \,\left(\frac{\partial \rho}{\partial T}\right) \,\left[1+M_{j} \, m_{j}\right] \end{aligned} But from equation (n) $\frac{\partial}{\partial m_{j}} \,\left(\frac{\partial \rho}{\partial T}\right)=-\alpha_{p} \, \frac{\partial \rho(\mathrm{aq})}{\partial m_{j}}-\rho(\mathrm{aq}) \,\left(\frac{\partial \alpha_{p}}{\partial m_{j}}\right)$ Therefore, \begin{aligned} &\mathrm{E}_{\mathrm{pj}}=\frac{\mathrm{M}_{\mathrm{j}} \, \alpha_{\mathrm{p}}}{\rho(\mathrm{aq})}-\frac{2}{(\rho(\mathrm{aq}))^{2}} \, \alpha_{\mathrm{p}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] \ &+\frac{1}{(\rho(\mathrm{aq}))^{2}} \, \alpha_{\mathrm{p}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right]+\frac{1}{\rho(\mathrm{aq})} \,\left(\frac{\partial \alpha_{\mathrm{p}}}{\partial \mathrm{m}_{\mathrm{j}}}\right) \,\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right] \end{aligned} or, (with reordering of terms) $\mathrm{E}_{\mathrm{p} j}=-\frac{\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right]}{(\rho(\mathrm{aq}))^{2}} \, \alpha_{\mathrm{p}} \,\left(\frac{\partial \rho(\mathrm{aq})}{\partial \mathrm{m}_{\mathrm{j}}}\right)+\frac{1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}}{\rho(\mathrm{aq})} \,\left(\frac{\partial \alpha_{\mathrm{p}}}{\partial \mathrm{m}_{\mathrm{j}}}\right)+\frac{\mathrm{M}_{\mathrm{j}} \, \alpha_{\mathrm{p}}}{\rho(\mathrm{aq})}$ Using equation (g) for $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ $\mathrm{E}_{\mathrm{pj}}=\mathrm{V}_{\mathrm{j}}(\mathrm{aq}) \, \alpha_{\mathrm{p}}+\frac{\left[1+\mathrm{M}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right]}{\rho(\mathrm{aq})} \,\left(\frac{\partial \alpha_{\mathrm{p}}}{\partial \mathrm{m}_{\mathrm{j}}}\right)$ The partial molar isobaric expansion $\mathrm{E}_{\mathrm{pj}}$ is calculated from isobaric expansibility and its dependence on molality of solute.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.10%3A_Expansions-_Solutions_Apparent_Molar_Isobaric_Expansions-_Determination.txt
For many aqueous solutions at ambient pressure and temperature, the dependence of $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ on molality of a neutral solute $j$, $\mathrm{m}_{j}$ is accounted for by an equation having the following general form. [The reason for choosing the molality scale is again the fact that $\mathrm{m}_{j}$ is independent of $\mathrm{T}$ and $\mathrm{p}$ but concentration $\mathrm{c}_{j}$ is not.] $\phi\left(E_{p j}\right)=a_{1}+a_{2} \,\left(m_{j} / m^{0}\right)+a_{3} \,\left(m_{j} / m^{0}\right)^{2}+a_{4} \,\left(m_{j} / m^{0}\right)^{3}+\ldots$ At low molalities, the linear term is dominant. Granted therefore that equation(a) accounts for the observed pattern, we need a quantitative description which accounts for this pattern. There are advantages in linking directly the apparent property $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ and the partial molar property $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})$. For an aqueous solution at fixed temperature and pressure, $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{E}_{\mathrm{pj}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)+\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ Hence the partial molar isobaric expansion of solute $j$ can be calculated from the apparent molar isobaric expansion and its dependence on molality, $\mathrm{m}_{j}$. Hence if equation (a) satisfactorily describes the observed dependence of $\phi\left(E_{p j}\right)$ on $\mathrm{m}_{j}$, $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{a}_{1}+2 \, \mathrm{a}_{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+3 \, \mathrm{a}_{3} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}+\ldots$ Therefore, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi \mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})=\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}$ Consequently the parameter $\mathrm{a}_{1}$ in equations (a) and (b) is the limiting partial molar isobaric expansion of solute $j$. For dilute solutions, equation (c) takes the following simple form. $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{S}_{\mathrm{Ep}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ In these terms we can identify the basis of the parameter $\mathrm{a}_{2}$ in equations (a) and (c). Desrosiers et al [1] used a quadratic (cf. equation (a)) to express the dependence of $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ at $298 \mathrm{~K}$ on molality of urea in aqueous solutions; $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=0.07 \mathrm{~cm}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}$, coefficient $\mathrm{a}_{2}$ being positive and coefficient $\mathrm{a}_{3}$ being negative. The majority of published information concerns the dependence on temperature of $\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}$. A survey [2] based on a dilatometric study of 15 non-electrolytes in aqueous solution indicates that $\left[\mathrm{d} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}\right]$ is less than $\left[\mathrm{dV}_{\mathrm{j}}^{*}(\ell) / \mathrm{dT}\right]$ for the pure liquid substance $j$; the second derivative $\left[\mathrm{d}^{2} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}^{2}\right]$ is positive. However for hydrophilic solutes $\left[\mathrm{d} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}\right]$ is larger than $\left[\mathrm{dV}_{\mathrm{j}}^{*}(\ell) / \mathrm{dT}\right]$ and $\left[\mathrm{d}^{2} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}^{2}\right]$ is negative [3]. A similar pattern is observed for sucrose and urea for which $\left[\mathrm{d}^{2} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}^{2}\right]$ is negative. Indeed Hepler [4] classified solutes in aqueous solutions as either structure-breaking (negative) or structure forming (positive) on the basis of the sign for $\left[\mathrm{d}^{2} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}^{2}\right]$. The dependence of $\phi\left(V_{j}\right)^{\infty}$ on temperature for both glycine and alanine in $\mathrm{NaCl}(\mathrm{aq})$ is small [5], For monosacchrides(aq) $\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ is positive. Footnotes [1] N. Desrosiers, G. Perron, J. G. Mathieson, B. E. Conway and J. E. Desnoyers, J Solution Chem.,1974,3,789. [2] J. I. Neal and D. A. I. Goring, J. Phys. Chem.,1970,74,658. [3] J. Sengster, T.-T. Ling and F. Lenzi, J Solution Chem,1976,5,575. [4] L.G. Hepler, Can J.Chem.,1969,47,4613. [5] B. S. Lark, K.Balat and S. Singh, Indian J Chem., Sect A,25,534. [6] S. Paljk, K. Balat and S. Singh, J. Chem. Eng Data, 1990,35.41. 1.12.12: Expansions- Isobaric- Salt Solutions- Apparent Molar In general terms the dependence of apparent molar isobaric expansions for salt $j$ on the composition of a given solution can be described using the following empirical equation. $\phi\left(E_{p j}\right)=a_{1}+a_{2} \,\left(m_{j} / m^{0}\right)^{1 / 2}+a_{3} \,\left(m_{j} / m^{0}\right)$ The presence of a term in $\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ is not unexpected in the case of salt solutions. Moreover for dilute solutions the term in $\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ is dominant. Hence $\left[\partial \phi\left(E_{p j}\right) / \partial m_{j}\right]=(1 / 2) \, a_{2} \,\left(m_{j} \, m^{0}\right)^{-1 / 2}+a_{3} \,\left(1 / m^{0}\right)$ Then, $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{a}_{1}+(3 / 2) \, \mathrm{a}_{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}+2 \, \mathrm{a}_{3} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Therefore parameter $\mathrm{a}_{1}$ is the limiting partial molar and apparent molar isobaric expansion of solute $j$ in solution. An explanation of the term in $\left(m_{j} / m^{0}\right)^{1 / 2}$ based on the Debye-Huckel Limiting Law (DHLL). In general terms the chemical potential of salt $j$ in aqueous solution at fixed $\mathrm{T}$ and $\mathrm{p}$ is related to molality $\mathrm{m}_{j}$ using equation (d). $\mu_{j}(a q)=\mu_{j}^{0}(a q)+v \, R \, T \, \ln \left(Q \, m_{j} \, \gamma_{\pm} / m^{0}\right)+\int_{p^{0}}^{p} V_{j}^{\infty}(a q) \, d p$ $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})+\mathrm{v} \, \mathrm{R} \, \mathrm{T} \,\left(\partial \ln \gamma_{\pm} / \partial \mathrm{p}\right)_{\mathrm{T}}$ Therefore $\left.\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\mathrm{v} \, \mathrm{R} \,\left\{\left[\mathrm{d} \ln \left(\gamma_{\pm}\right) / \mathrm{dp}\right)\right]_{\mathrm{T}}+\mathrm{T} \,\left[\mathrm{d}^{2} \ln \left(\gamma_{\pm}\right) / \mathrm{dp} \, \mathrm{dT}\right]\right\}$ According to the DHLL, $\ln \left(\gamma_{\pm}\right)=-S_{\gamma} \,\left(m_{j} / m^{0}\right)^{1 / 2}$ By definition $\mathrm{S}_{\mathrm{V}}=\left\lfloor\partial \mathrm{S}_{\gamma} / \partial \mathrm{p}\right\rfloor_{\mathrm{T}}$ Then [1], $\mathrm{T} \,\left[\partial \ln \left(\gamma_{\pm}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}=-\mathrm{T} \, \mathrm{S}_{\mathrm{v}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ Hence we write [2] $\mathrm{E}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})-\mathrm{v}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{S}_{\mathrm{Ep}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ where [3], $\mathrm{S}_{\mathrm{Ep}}=\mathrm{S}_{\mathrm{V}}+\mathrm{T} \,\left[\partial \mathrm{S}_{\mathrm{V}} / \partial \mathrm{T}\right)_{\mathrm{p}}$ Therefore a linear dependence of $\mathrm{E}_{\mathrm{p j}}(\mathrm{aq})$ on $\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ for dilute solutions is predicted by the DHLL. Hence for dilute solutions $\phi\left(E_{p j}\right)=\phi\left(E_{p j}\right)^{\infty}-(2 / 3) \, v \, R \, S_{E p} \,\left(m_{j} / m^{0}\right)^{1 / 2}$ For tetra-alkylammonium iodides(aq) $\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \partial \mathrm{T}\right]_{\mathrm{p}}$ is positive, the magnitude increasing on going from $\mathrm{Me}_{4}\mathrm{N}^{+}$ to $\mathrm{Bu}_{4}\mathrm{N}^{+}$ [4,5,]. Apparent molar isobaric expansions for divalent metal chlorides(aq) lead to estimates of ionic molar isobaric expansions based on $\mathrm{E}_{p}^{\infty}\left(\mathrm{Cl}^{-} ; \mathrm{aq}\right)$ set at $+0.046 \mathrm{~cm}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}$ [6,7]. The ionic estimates show a linear dependence on $\left(\mathrm{r}_{\mathrm{j}}\right)^{-1}$, a pattern predicted by the Born equation. $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}$ for $\mathrm{NaBPh}_{4}$ decreases gradually over the range 0 to 60 Celsius [8]. $\phi\left(E_{p j}\right)^{\infty}$ for $\mathrm{NaF}(\mathrm{aq})$, $\mathrm{Na}_{2}\mathrm{SO}_{4}(\mathrm{aq})$ and $\mathrm{KCl}(\mathrm{aq})$ is positive [9]. Footnotes [1] \begin{aligned} &\mathrm{S}_{\gamma}=[1] \quad \mathrm{S}_{\mathrm{V}}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \quad \mathrm{~T} \, \mathrm{S}_{\mathrm{V}}=[\mathrm{K}] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \ &\left\{\left[\partial \ln \left(\gamma_{\pm}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}+\mathrm{T} \,\left[\partial^{2} \ln \left(\gamma_{\pm}\right) / \partial \mathrm{p} \, \partial \mathrm{T}\right]\right\} \ &=\left\{\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}+[\mathrm{K}] \,[1] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}[\mathrm{~K}]^{-1}\right\}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \end{aligned} [2] \begin{aligned} &\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]-[1] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \ &=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]-\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right] \end{aligned} [3] $\mathrm{S}_{\mathrm{Ep}}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}+[\mathrm{K}] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \,\left[\mathrm{K}^{-1}\right]=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}$ [4] R. Gopal and M. A. Siddiqi, J.Phys.Chem.,1968,72,1814. [5] F. Franks and H. T. Smith, Trans. Faraday Soc.,1967,63,2586. [6] F. J. Millero and W. Drost –Hansen, J. Phys.Chem.,1968,72,1758. [7] F. J. Millero, J. Phys. Chem., 1968, 72, 4589. [8] F. J. Millero, J. Chem. Eng. Data, 1970,15,562. [9] F. J. Millero and J. H. Knox, J. Chem. Eng. Data, 1973,18,407.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.11%3A_Expansions-_Isobaric-_Apparent_Molar-_Neutral_Solutes.txt
The isobaric (equilibrium) expansion of a liquid, volume $\mathrm{V}$, is defined by equation (a). $\mathrm{E}_{\mathrm{p}}=\left(\frac{\partial V}{\partial T}\right)_{p}$ Both $\mathrm{E}_{\mathrm{p}}$ and $\mathrm{V}$ are extensive properties of a mixture. Therefore it is convenient to refer to the molar property, $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix})$. Thus $\mathrm{E}_{\mathrm{pm}}(\mathrm{mix})=\left(\frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ At fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ for a binary liquid mixture is related to the partial molar volumes of the two components. $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})=\mathrm{x}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{x}_{2} \, \mathrm{V}_{2}(\mathrm{mix})$ From equation (b) $\mathrm{E}_{\mathrm{pm}}(\mathrm{mix})=\mathrm{x}_{1} \,\left(\frac{\partial \mathrm{V}_{1}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{x}_{2} \,\left(\frac{\partial \mathrm{V}_{2}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ For a binary mixture having molar volume $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ and density $\rho(\mathrm{mix})$, $\rho(\operatorname{mix})=\left(\mathrm{x}_{1} \, \mathrm{M}_{1}+\mathrm{x}_{2} \, \mathrm{M}_{2}\right) / \mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ Here $\mathrm{M}_{1}$ and $\mathrm{M}_{2}$ are the molar masses of liquids 1 and 2 respectively. $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})=\left(\mathrm{x}_{1} \, \mathrm{M}_{1}+\mathrm{x}_{2} \, \mathrm{M}_{2}\right) / \rho(\mathrm{mix})$ Hence, \begin{aligned} {\left[\partial \mathrm{V}_{\mathrm{m}}(\operatorname{mix}) / \partial \mathrm{T}\right]_{\mathrm{p}} } &=\ &-\left[\left(\mathrm{x}_{1} \, \mathrm{M}_{1}+\mathrm{x}_{2} \, \mathrm{M}_{2}\right) / \rho(\operatorname{mix})\right] \,[\partial \ln \{\rho(\operatorname{mix})\} / \partial \mathrm{T}]_{\mathrm{p}} \end{aligned} $\mathrm{E}_{\mathrm{pm}}(\mathrm{mix})$ is obtained for a given mixture from the isobaric dependence of density on temperature. There is merit in considering equations for $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})$ of a binary mixture having ideal thermodynamic properties and hence for the related excess molar expansion $\mathrm{E}_{\mathrm{pm}}^{\mathrm{E}}$. With, $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{E}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{2}^{*}(\ell)$ $\mathrm{E}_{\mathrm{pm}}^{\mathrm{E}}=\mathrm{E}_{\mathrm{pm}}(\mathrm{mix})-\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{id})$ $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})$ is the mole fraction weighted sum of the isobaric expansions of the pure liquid components at the same $\mathrm{T}$ and $\mathrm{p}$. The isobaric expansibility of an ideal binary liquid mixture $\alpha_{p}(\operatorname{mix} ; \mathrm{id})$ is given by equation (j). $\alpha_{\mathrm{p}}(\operatorname{mix} ; \mathrm{id})=\frac{\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)}$ Or, $\alpha_{\mathrm{p}}(\operatorname{mix} ; \text { id })=\frac{\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)}{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)}+\frac{\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)}$ Hence, $\alpha_{\mathrm{p}}(\mathrm{mix} ; \mathrm{id})=\frac{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)}{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)}+\frac{\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell) \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)}$ Hence, expansibility $\alpha_{p}(\operatorname{mix} ; 1 \mathrm{~d})$ can be expressed in terms of the volume fractions of the corresponding ideal binary liquid mixture. $\alpha_{p}(\operatorname{mix} ; \text { id })=\phi_{1}(\operatorname{mix} ; \text { id }) \, \alpha_{p 1}^{*}(\ell)+\phi_{2}(\operatorname{mix} ; \text { id }) \, \alpha_{p 2}^{*}(\ell)$ The excess (equilibrium) isobaric expansivity $\alpha_{p}^{E}(\operatorname{mix})$ is given by mix equation (n) [1]. $\alpha_{\mathrm{p}}^{\mathrm{E}}(\mathrm{mix})=\frac{1}{\mathrm{~V}_{\mathrm{m}}(\mathrm{mix})} \,\left[\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\mathrm{V}_{\mathrm{m}}^{\mathrm{E}} \, \alpha_{\mathrm{p}}(\mathrm{mix} ; \mathrm{id})\right]$ From another standpoint the thermal expansion of a binary liquid mixture is analysed in terms of the differential dependence of rational activity coefficients on temperature and pressure. For liquid component 1 at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, $\mu_{1}(\operatorname{mix})=\mu_{1}^{0}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)+\int_{\mathrm{p}}^{\mathrm{p}} \mathrm{V}_{1}^{*}(\ell) \, \mathrm{dp}$ Then $\mathrm{V}_{1}(\mathrm{mix})=\mathrm{V}_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}$ At temperature $\mathrm{T}$, $\mathrm{E}_{\mathrm{p}_{1}}(\operatorname{mix})=\mathrm{E}_{\mathrm{p} 1}(\operatorname{mix} ; \mathrm{id})+\mathrm{R} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}+\mathrm{R} \, \mathrm{T} \,\left[\frac{\partial}{\partial \mathrm{T}}\left(\frac{\partial \ln \left(\mathrm{f}_{1}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right]_{\mathrm{p}}$ \begin{aligned} &\mathrm{E}_{\mathrm{p} 2}(\operatorname{mix})= \ &\quad \mathrm{E}_{\mathrm{p} 2}(\mathrm{mix} ; \mathrm{id})+\mathrm{R} \,\left[\partial \ln \left(\mathrm{f}_{2}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}+\mathrm{R} \, \mathrm{T} \,\left[\frac{\partial}{\partial \mathrm{T}}\left(\frac{\partial \ln \left(\mathrm{f}_{2}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right]_{\mathrm{p}} \end{aligned} Two equations follow for the excess partial molar isobaric expansions of the components of the mixture. $\mathrm{E}_{\mathrm{p} 1}^{\mathrm{E}}(\mathrm{mix})=\mathrm{R} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}+\mathrm{R} \, \mathrm{T} \,\left[\frac{\partial}{\partial \mathrm{T}}\left(\frac{\partial \ln \left(\mathrm{f}_{1}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right]_{\mathrm{p}}$ $E_{\mathrm{p} 2}^{\mathrm{E}}(\mathrm{mix})=\mathrm{R} \,\left[\partial \ln \left(\mathrm{f}_{2}\right) / \partial \mathrm{p}_{\mathrm{T}}+\mathrm{R} \, \mathrm{T} \,\left[\frac{\partial}{\partial \mathrm{T}}\left(\frac{\partial \ln \left(\mathrm{f}_{2}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right]_{\mathrm{p}}\right.$ Therefore for the mixture, $\mathrm{E}_{\mathrm{pm}}^{\mathrm{E}}(\mathrm{mix})=\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{\mathrm{E}}(\mathrm{mix})+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{\mathrm{E}}(\mathrm{mix})$ Footnotes [1] For a binary liquid mixture at defined $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})=\mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})+\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ $\alpha_{p}(\operatorname{mix})=\frac{1}{V_{m}(\operatorname{mix})} \, \frac{\partial}{\partial T}\left[V_{m}(\text { mix } ; 1 \mathrm{~d})+V_{m}^{E}\right]$ Or, $\alpha_{p}(\operatorname{mix})=\frac{1}{V_{m}(\operatorname{mix})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})}{\partial \mathrm{T}}+\frac{1}{\mathrm{~V}_{\mathrm{m}}(\mathrm{mix})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}$ But, $\alpha_{p}(\operatorname{mix} ; \mathrm{id})=\frac{1}{\mathrm{~V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})}{\partial \mathrm{T}}$ By definition, \begin{aligned} &\alpha_{p}^{E}=\alpha_{p}(\operatorname{mix})-\alpha_{p}(\operatorname{mix} ; \text { id })\ &\alpha_{p}^{E}(\operatorname{mix})=\left[\frac{1}{V_{m}(\operatorname{mix})}-\frac{1}{V_{m}(\operatorname{mix} ; i \mathrm{~d})}\right] \, \frac{\partial V_{m}(\operatorname{mix} ; \mathrm{id})}{\partial \mathrm{T}}+\frac{1}{\mathrm{~V}_{\mathrm{m}}(\mathrm{mix})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\ &\alpha_{p}^{\mathrm{E}}(\operatorname{mix})=\left[\frac{\mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})-\mathrm{V}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{V}_{\mathrm{m}}(\mathrm{mix}) \, \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})}\right] \, \frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})}{\partial \mathrm{T}}+\frac{1}{\mathrm{~V}_{\mathrm{m}}(\operatorname{mix})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\ &\alpha_{\mathrm{p}}^{\mathrm{E}}(\operatorname{mix})=-\left[\frac{\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{V}_{\mathrm{m}}(\mathrm{mix})}\right] \, \alpha_{\mathrm{p}}(\mathrm{mix} ; \mathrm{id})+\frac{1}{\mathrm{~V}_{\mathrm{m}}(\mathrm{mix})} \, \frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}} \end{aligned} Hence, $\alpha_{p}^{E}(\operatorname{mix})=\frac{1}{V_{m}(\operatorname{mix})} \,\left[\left(\frac{\partial V_{m}^{E}}{\partial T}\right)_{p}-V_{m}^{E} \, \alpha_{p}(\operatorname{mix} ; \text { id })\right]$ 1.12.14: Expansibilities- Isobaric- Binary Liquid Mixtures A given binary liquid mixture is prepared using liquids 1 and 2 at defined $\mathrm{T}$ and $\mathrm{p}$. The molar volume of this mixture is given by equation (a). In the event that thermodynamic properties of the mixture are ideal, the molar volume is given by equation (a). $\mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ At fixed pressure, $\left(\frac{\partial \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\mathrm{x}_{1} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{x}_{2} \,\left(\frac{\partial \mathrm{V}_{2}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ \begin{aligned} &\frac{\mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})}{\mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})} \,\left(\frac{\partial \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})}{\partial \mathrm{T}}\right)_{\mathrm{p}}= \ &\mathrm{x}_{1} \, \frac{\mathrm{V}_{1}^{*}(\ell)}{\mathrm{V}_{1}^{*}(\ell)} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{x}_{2} \, \frac{\mathrm{V}_{2}^{*}(\ell)}{\mathrm{V}_{2}^{*}(\ell)} \,\left(\frac{\partial \mathrm{V}_{2}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}} \end{aligned} Hence, $\mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id}) \, \alpha_{\mathrm{p}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell) \, \alpha_{\mathrm{p} 2}^{*}(\ell)$ But $\phi_{1}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell) / \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})$ And, $\phi_{2}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell) / \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})$ Hence $\alpha_{p}(\operatorname{mix} ; \text { id })=\phi_{1}(\operatorname{mix} ; \text { id }) \, \alpha_{p 1}^{*}(\ell)+\phi_{2}(\operatorname{mix} ; \text { id }) \, \alpha_{p 2}^{*}(\ell)$ For a real binary liquid mixture, $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)+\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}(\operatorname{mix})$ At fixed pressure, $\left(\frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\mathrm{x}_{1} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{x}_{2} \,\left(\frac{\partial \mathrm{V}_{2}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Or, \begin{aligned} &\frac{\mathrm{V}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{V}_{\mathrm{m}}(\mathrm{mix})} \,\left(\frac{\partial \mathrm{V}_{\mathrm{m}}(\mathrm{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}= \ &\mathrm{x}_{1} \, \frac{\mathrm{V}_{1}^{*}(\ell)}{\mathrm{V}_{1}^{*}(\ell)} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{x}_{2} \, \frac{\mathrm{V}_{2}^{*}(\ell)}{\mathrm{V}_{2}^{*}(\ell)} \,\left(\frac{\partial \mathrm{V}_{2}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \end{aligned} \begin{aligned} &\mathrm{V}_{\mathrm{m}}(\mathrm{mix}) \, \alpha_{\mathrm{p}}(\mathrm{mix})= \ &\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \alpha_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell) \, \alpha_{\mathrm{p} 2}^{*}(\ell)+\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \end{aligned} Or, \begin{aligned} &\alpha_{p}(\operatorname{mix})= \ &\frac{1}{V_{\mathrm{m}}(\operatorname{mix})} \,\left[x_{1} \, V_{1}^{*}(\ell) \, \alpha_{p 1}^{*}(\ell)+x_{2} \, V_{2}^{*}(\ell) \, \alpha_{p 2}^{*}(\ell)+\left(\frac{\partial V_{m}^{E}}{\partial T}\right)_{p}\right] \end{aligned} We may also define an excess property using equation (k) but it is important to note that $\alpha_{\mathrm{p}}^{\mathrm{E}}$ is not a simple second derivative of the excess molar Gibbs energy, $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}$. $\alpha_{p}^{E}(\operatorname{mix})=\alpha_{p}(\operatorname{mix})-\alpha_{p}(\operatorname{mix} ; \text { id })$ We start out using an alternative expression for $\alpha_{p}(\operatorname{mix})$. $\alpha_{p}(\operatorname{mix})=\frac{1}{V_{m}(\operatorname{mix})} \,\left[V_{m}(\operatorname{mix} ; \mathrm{id}) \, \alpha_{p}(\operatorname{mix} ; \mathrm{id})+\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{E}}{\partial T}\right)_{\mathrm{p}}\right]$ \begin{aligned} &\alpha_{p}^{E}(\operatorname{mix})= \ &\frac{1}{V_{m}(\operatorname{mix})} \,\left[V_{m}(\operatorname{mix} ; \text { id }) \, \alpha_{p}(\operatorname{mix} ; \mathrm{id})+\left(\frac{\partial V_{m}^{E}}{\partial T}\right)_{p}\right]-\alpha_{p}(\operatorname{mix} ; \text { id }) \end{aligned} \begin{aligned} &\alpha_{\mathrm{p}}^{\mathrm{E}}(\operatorname{mix})= \ &\frac{1}{\mathrm{~V}_{\mathrm{m}}(\operatorname{mix})} \,\left[\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left[\mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})-\mathrm{V}_{\mathrm{m}}(\operatorname{mix})\right] \, \alpha_{\mathrm{p}}(\operatorname{mix} ; \mathrm{id}]\right. \end{aligned} Hence, [1] $\alpha_{p}^{E}(\operatorname{mix})=\frac{1}{V_{m}(\operatorname{mix})} \,\left[\left(\frac{\partial V_{m}^{E}}{\partial T}\right)_{p}-V_{m}^{E}(\operatorname{mix}) \, \alpha_{p}(\text { mix } ; \text { id }]\right.$ Footnotes [1] \begin{aligned} &{\alpha_{\mathrm{p}}^{\mathrm{E}}(\operatorname{mix})=\left[\mathrm{K}^{-1}\right]} \ &{\left[\left(\frac{\partial \mathrm{V}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\mathrm{V}_{\mathrm{m}}^{\mathrm{E}} \, \alpha_{\mathrm{p}}(\mathrm{mix} ; \mathrm{id})\right]=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]}{[\mathrm{K}]}-\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{K}^{-1}\right]=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]} \ &\frac{1}{V_{m}(m i x)} \,\left[\left(\frac{\partial V_{m}^{E}}{\partial T}\right)_{p}-V_{m}^{E} \, \alpha_{p}(\text { mix } ; i d)\right] \ &\quad \quad =\frac{1}{\left[\mathrm{~m}^{3} \mathrm{~mol}^{-1}\right]} \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \quad \mathrm{~K}^{-1}\right]=\left[\mathrm{K}^{-1}\right] \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.13%3A_Expansions-_Isobaric-_Binary_Liquid_Mixtures.txt
The Gibbs energy of a closed system at thermodynamic equilibrium containing two chemical substances is defined by equation (a) where the molecular composition/organisation is signalled by $\xi^{\mathrm{eq}}$. $\mathrm{G}=\mathrm{G}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}\right]$ $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}\right]$ $\mathrm{S}=\mathrm{S}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}\right]$ A common feature is the use of the two intensive variables, temperature and pressure, in the definition of extensive properties $\mathrm{G}$, $\mathrm{V}$ and $\mathrm{S}$. The properties $\mathrm{G}$, $\mathrm{V}$ and $\mathrm{S}$ are Gibbsian. The system is perturbed by an increase in temperature along a path such that the affinity for spontaneous change remains zero and the entropy remains equal to that defined by equation (c). In principle we plot volume $\mathrm{V}$ as a function of temperature at constant $\mathrm{n}_{1}$, $\mathrm{n}_{2}$, at '$\mathrm{A}=0$' and at a constant entropy equal to that defined by equation (c). The gradient of the plot at the point where the volume is defined by equation (b) yields the equilibrium isentropic expansion, $\mathrm{E}_{\mathrm{S}}(\mathrm{A}=0)$ [1]; equation (d); isentropic = adiabatic and at equilibrium. $\mathrm{E}_{\mathrm{S}}(\mathrm{A}=0)=(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{S} . \mathrm{A}=0}$ $\mathrm{E}_{\mathrm{S}}(\mathrm{A}=0)$ characterises the system defined by the Gibbsian set of independent variables $\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{~A}=0, \xi^{\mathrm{eq}}\right]$. As we change the amount of solute $\mathrm{n}_{j}$ for a fixed temperature, pressure and amount of solvent $\mathrm{n}_{1}$, so both $\mathrm{V}(\mathrm{aq})$ and $\mathrm{S}(\mathrm{aq})$ change yielding a new isentropic thermal expansion $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ at a new entropy $\mathrm{S}(\mathrm{aq})$. For a series of solutions having different molalities, comparison of $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ is not straightforward because entropy $\mathrm{S}(\mathrm{aq})$ is itself a function of solution composition. Further comparison cannot be readily drawn with the isentropic thermal expansion of the pure solvent $\mathrm{E}_{\mathrm{Sl}}^{*}(\ell)$ equation (e). $\mathrm{E}_{\mathrm{S} 1}^{*}(\ell ; \mathrm{A}=0)=\left(\partial \mathrm{V}_{1}^{*}(\ell) / \partial \mathrm{T}\right)_{\mathrm{A}=0} \text { at constant } \mathrm{S}_{1}^{*}(\ell)$ $\mathrm{E}_{\mathrm{S}}(\mathrm{A}=0 ; \mathrm{aq})$ is a non-Gibbsian property [2]. Consequently, familiar thermodynamic relationships involving partial molar properties are not valid in the case of partial molar isentropic (thermal) expansions which are non-Lewisian properties [2,3]. $\left[\partial \mathrm{V}_{\mathrm{j}}(\mathrm{aq}) / \partial \mathrm{T}\right]$ for solute-$j$ in aqueous solution at constant $\mathrm{S}(\mathrm{aq})$ is a semi-partial molar property [4]. For an aqueous solution having entropy $\mathrm{S}(\mathrm{aq})$, two partial molar isentropic expansions are defined for the solvent and solute. At $\mathrm{S}(\mathrm{aq})$ characterised by $\mathrm{T}$, $\mathrm{p}$, $\mathrm{n}_{1}$ and $\mathrm{n}_{j}$, $\mathrm{E}_{\mathrm{Sl}}(\mathrm{aq} ; \mathrm{def})=\left[\partial \mathrm{E}_{\mathrm{S}}(\mathrm{aq}) / \partial \mathrm{n}_{1}\right] \text { at fixed } \mathrm{T}, \mathrm{p} and \mathrm{n}_{j}$ and $\mathrm{E}_{\mathrm{Sj}}(\mathrm{aq} ; \mathrm{def})=\left[\partial \mathrm{E}_{\mathrm{S}}(\mathrm{aq}) / \partial \mathrm{n}_{\mathrm{j}}\right] \text { at fixed } \mathrm{T}, \mathrm{p} \text { and } \mathrm{n}_{1}$ So that, $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{Sl}}(\mathrm{aq} ; \operatorname{def})+\mathrm{n}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{Sj}}(\mathrm{aq} ; \text { def })$ Equation (h) relates $\mathrm{E}_{\mathrm{S}}(\mathrm{aq})$ to the partial molar intensive isentropic properties of both solvent and solute. A similar problem is encountered in defining an apparent molar isentropic expansion for solute-$j$, $\phi\left(\mathrm{E}_{\mathrm{Sj} \mathrm{j}}\right)$. We might assert that $\phi\left(\mathrm{E}_{\mathrm{Sj} \mathrm{j}}\right)$ is defined by the isentropic differential dependence $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ on temperature. Alternatively, we use an equation by analogy to those used to relate, for example, $\mathrm{V}(\mathrm{aq})$ to $V_{1}^{*}(\ell)$ and $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$. Equation (i) relates $\mathrm{V}(\mathrm{aq})$ to the apparent molar volume of solute j, φ(Vj). $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Differentiation of equation (i) at constant entropy again raises the problem that the molar entropy $\mathrm{S}(\mathrm{aq})$ does not equal the molar entropy of the pure solvent, $\mathrm{S}_{1}^{*}(\ell)$. However, by analogy with the definition of $\phi\left(\mathrm{K}_{\mathrm{Sj}} ; \mathrm{def}\right)$ we define a quantity $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \text { def }\right)$ using equation (j). $\mathrm{E}_{\mathrm{s}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{E}_{\mathrm{Sl}}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)$ $\mathrm{E}_{\mathrm{Sl}}^{*}(\ell)$ is the molar intensive property of the solvent. The isentropic expansion of the solution at entropy $\mathrm{S}(\mathrm{aq})$ is linked with that of the pure solvent at entropy $\mathrm{S}_{1}^{*}(\ell)$. Further [5] $\phi\left(E_{\mathrm{s} j} ; \operatorname{def}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ $\phi\left(E_{\mathrm{sj}} ; \text { def }\right)=\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right] \,\left(\mathrm{c}_{\mathrm{j}}\right)^{-1}+\alpha_{\mathrm{s} 1}^{*}(1) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Footnotes [1] From $\left(\frac{\partial^{2} \mathrm{U}}{\partial \mathrm{S} \, \partial \mathrm{V}}\right)=\left(\frac{\partial^{2} \mathrm{U}}{\partial \mathrm{V} \, \partial \mathrm{S}}\right), \quad\left(\frac{\partial \mathrm{T}}{\partial \mathrm{V}}\right)_{\mathrm{s}}=-\left(\frac{\partial \mathrm{p}}{\partial \mathrm{S}}\right)_{\mathrm{V}}$ We invert the latter equation. Hence, $E_{S}=\left(\frac{\partial V}{\partial T}\right)_{s}=-\left(\frac{\partial S}{\partial p}\right)_{V}$ The isentropic dependence of volume on temperature equals (with reversed sign) the isochoric dependence of entropy on pressure. [2] J. C. R. Reis, M. J. Blandamer, M. I. Davis and G. Douheret., Phys. Chem. Chem. Phys.,2001,3,1465. [3] J. C. R. Reis, J. Chem. Soc. Faraday Trans. 2,1982,78,1595. [4] M. J. Blandamer, M. I. Davis, G. Douheret and J. C. R. Reis., Chem. Soc. Rev., 2001,30,8. [5] From \begin{aligned} &\phi\left(E_{\mathrm{S} j} ; \text { def }\right)=\frac{E_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{S} 1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}}} \ &\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \text { def }\right)=\frac{\mathrm{E}_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)}{\mathrm{m}_{\mathrm{j}}}-\frac{\mathrm{E}_{\mathrm{S} 1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}} \ &\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)=\frac{\alpha_{\mathrm{s}}(\mathrm{aq}) \, \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)}{\mathrm{m}_{\mathrm{j}}}-\frac{\alpha_{\mathrm{S} 1}^{*}(\ell) \, \mathrm{V}_{1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}} \end{aligned} But $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1} / \mathrm{kg}=1\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Then $\phi\left(\mathrm{E}_{\mathrm{s} j} ; \mathrm{def}\right)=\frac{\alpha_{\mathrm{s}}(\mathrm{aq}) \,\left[\left(1 / \mathrm{M}_{1}\right) \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]}{\mathrm{m}_{\mathrm{j}}}-\frac{\alpha_{\mathrm{s} 1}^{*}(\ell) \, \mathrm{V}_{1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}}$ Or, $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \text { def }\right)=\frac{\mathrm{V}_{1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{sl}}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence, $\phi\left(\mathrm{E}_{\mathrm{s} \mathrm{j}} ; \operatorname{def}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ But $\frac{1}{\mathrm{~m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}=\frac{1}{\mathrm{c}_{\mathrm{j}}}-\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Then \phi\left(E_{s j} ; \operatorname{def}\right)=\left[\frac{1}{c_{j}}-\phi\left(V_{j}\right)\right] \,\left[\alpha_{s}(\mathrm{aq})-\alpha_{\mathrm{s}_{1}}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\) Or, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right] \ &-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s}}(\mathrm{aq})+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\ell)+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Hence, \begin{aligned} &\phi\left(\mathrm{E}_{\mathrm{sj}} ; \mathrm{def}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\ell) \ \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \text { def }\right)=\left[\mathrm{m}^{3} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.15%3A_Expansions-_Isentropic-_Solutions.txt
In the context of the properties of aqueous solutions the concept of apparent molar properties is important with respect to the analysis of experimental results; e.g. apparent molar volume for solute $j$ $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ calculated from the densities of a given solution and solvent at fixed $\mathrm{T}$ and $\mathrm{p}$. Similarly apparent molar isobaric expansions $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$ characterise the dependence of \phi\left(\mathrm{V}_{j}\right) on temperature at fixed pressure. Nevertheless problems emerge when we turn attention to comparable isentropic properties. The way ahead involves definition of apparent molar isentropic expansions $\phi\left(\mathrm{E}_{\mathrm{sj}} ; \text { def }\right)$ and apparent molar isentropic compressions $\phi\left(\mathrm{K}_{\mathrm{sj}} ; \text { def }\right)$. These two properties are related [1]; equation(a). \begin{aligned} &\phi\left(\mathrm{E}_{\mathrm{sj}} ; \mathrm{def}\right)= \ &-\frac{\alpha_{\mathrm{S} 1}^{*}(\ell)}{\alpha_{\mathrm{p}}(\mathrm{aq})} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)+\frac{\alpha_{\mathrm{s}}(\mathrm{aq})}{\kappa_{\mathrm{s}}(\mathrm{aq})} \, \phi\left(\mathrm{K}_{\mathrm{s}}\right)+\frac{\alpha_{\mathrm{s}}(\mathrm{aq}) \, \kappa_{\mathrm{S} 1}^{*}(\ell)}{\kappa_{\mathrm{s}}(\mathrm{aq}) \, \sigma(\mathrm{aq})} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) \ &\quad+\left[\alpha_{\mathrm{p} 1}^{*}(\ell) \,\left(1+\frac{\alpha_{\mathrm{pl} 1}^{*}(\ell)}{\alpha_{\mathrm{p}}(\mathrm{aq})}\right)-\frac{\alpha_{\mathrm{s}}(\mathrm{aq}) \, \kappa_{\mathrm{S} 1}^{*}(\ell)}{\kappa_{\mathrm{s}}(\mathrm{aq})} \,\left(1+\frac{\sigma_{1}^{*}(\ell)}{\sigma(\mathrm{aq})}\right)\right] \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Equation (b) relates the corresponding properties at infinite dilution [1]. $\frac{\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)^{\infty}}{\alpha_{\mathrm{S} 1}^{*}(\ell)}=-\frac{\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}}{\alpha_{\mathrm{p} 1}^{*}(\ell)}+\frac{\phi\left(\mathrm{K}_{\mathrm{Sj}} ; \mathrm{def}\right)^{\infty}}{\kappa_{\mathrm{S} 1}^{*}(\ell)}+\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}}{\sigma_{1}^{*}(\ell)}$ ‘Semi’ apparent molar isentropic expansions and compressions are related using equation (c) $\frac{1}{\alpha_{\mathrm{S}}(\mathrm{aq})} \,\left[\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right]_{\mathrm{S}(\mathrm{aq})}=-\frac{1}{\kappa_{\mathrm{S}}(\mathrm{aq})} \,\left[\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right]_{\mathrm{S}(\mathrm{aq})}$ Equations (a) , (b) and (c) illustrate the power of thermodynamics in drawing together and relating the several properties of a solution. Footnotes [1] In the following we simplify the algebra by omitting the descriptors (aq) and ($\ell$). The starting point is the following equation. $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}} \, \sigma} \,\left(\kappa_{\mathrm{s}} \, \sigma-\kappa_{\mathrm{s}}^{*} \, \sigma^{*}\right)-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \,\left(\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right)$ The latter equation is effectively an identity. From equation (a), $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\alpha_{\mathrm{s}}-\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}} \, \sigma} \, \kappa_{\mathrm{s}}^{*} \, \sigma^{*}-\alpha_{\mathrm{s}}^{*}+\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \, \alpha_{\mathrm{p}}^{*}$ From equation (b), $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}+\frac{\kappa_{\mathrm{s}} \, \sigma}{\mathrm{T} \, \alpha_{\mathrm{p}} \, \kappa_{\mathrm{s}} \, \sigma} \, \kappa_{\mathrm{s}}^{*} \, \sigma^{*}-\frac{\kappa_{\mathrm{S}}^{*} \, \sigma^{*}}{\alpha_{\mathrm{p}} \, \mathrm{T}}$ or, $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}$ But as an identity, $\kappa_{\mathrm{S}} \, \sigma-\kappa_{\mathrm{S}}^{*} \, \sigma^{*}=\sigma \,\left(\kappa_{\mathrm{S}}-\kappa_{\mathrm{S}}^{*}\right)+\kappa_{\mathrm{S}}^{*} \,\left(\sigma-\sigma^{*}\right)$ From equations (a) and (c). $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \,\left(\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \,\left(\sigma-\sigma^{*}\right)-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}}\left(\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right)$ But, \begin{aligned} &\phi\left(\mathrm{E}_{\mathrm{s} j} ; \text { def }\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s}}^{*} \ &\phi\left(\mathrm{E}_{\mathrm{S}_{\mathrm{j}}} ; \operatorname{def}\right)=\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \, \frac{1}{\mathrm{c}_{\mathrm{j}}} \,\left(\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \, \frac{1}{\mathrm{c}_{\mathrm{j}}} \,\left(\sigma-\sigma^{*}\right) \ &-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \, \frac{1}{\mathrm{c}_{\mathrm{j}}} \,\left(\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right)+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} But for the isobaric heat capacities $\phi\left(\mathrm{C}_{\mathrm{pj}}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\sigma-\sigma^{*}\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \sigma^{*}$ Also, $\begin{array}{r} \phi\left(\mathrm{K}_{\mathrm{s} j} ; \text { def }\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left[\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \kappa_{\mathrm{s}}^{*} \ \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \text { def }\right)=\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{S}}} \,\left[\phi\left(\mathrm{K}_{\mathrm{Sj}} ; \text { def }\right)-\kappa_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \end{array}$ Hence, \begin{aligned} &+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \,\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\sigma^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \ &-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \,\left[\phi\left(\mathrm{E}_{\mathrm{pj}}\right)-\alpha_{\mathrm{p}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} With a little reorganisation, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)=-& \frac{\alpha_{\mathrm{S}}^{*}}{\alpha_{\mathrm{p}}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)+\frac{\alpha_{\mathrm{S}}}{\kappa_{\mathrm{S}}} \, \phi\left(\mathrm{K}_{\mathrm{Sj}} ; \operatorname{def}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{S}} \, \sigma} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) \ &+\left[\alpha_{\mathrm{s}}^{*} \,\left(1+\frac{\alpha_{\mathrm{p}}^{*}}{\alpha_{\mathrm{p}}}\right)-\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{S}}}\left(1+\frac{\sigma^{*}}{\sigma}\right)\right] \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Hence in the limit of infinite dilution, $\frac{\phi\left(E_{\mathrm{S} j} ; \mathrm{def}\right)^{\infty}}{\alpha_{\mathrm{s}}^{*}}=-\frac{\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}}{\alpha_{\mathrm{p}}^{*}}+\frac{\phi\left(\mathrm{K}_{\mathrm{s} j} ; \mathrm{def}\right)^{\infty}}{\kappa_{\mathrm{s}}^{*}}+\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}}{\sigma^{*}}$ [2] \begin{aligned} &{\frac{\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)^{\infty}}{\alpha_{\mathrm{s}}^{*}}=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]}{\left[\mathrm{K}^{-1}\right]}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]} \ &\frac{\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}}{\alpha_{\mathrm{p}}^{*}}=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]}{\left[\mathrm{K}^{-1}\right]}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \ &\frac{\phi\left(\mathrm{K}_{\mathrm{Sj}} ; \mathrm{def}\right)^{\infty}}{\kappa_{\mathrm{S}}^{*}}=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} / \mathrm{N} \mathrm{m}^{-2}\right]}{\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \ &\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}}{\sigma^{*}}=\frac{\left[\mathrm{J} \mathrm{mol}^{-1}\right]}{\left[\mathrm{J} \mathrm{m}^{-3}\right]}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \end{aligned} 1.12.17: Expansions- Isentropic- Solutions- Apparent and Partial Molar A given solution is prepared using $\mathrm{n}_{1}$ moles of solvent (water) and $\mathrm{n}_{j}$ moles of solute $j$. The volume of the system is defined by equation (a). $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ We consider the case where the closed system is at equilibrium and hence where the affinity for spontaneous change is zero. The entropy of the system (at equilibrium) is defined by the same set of independent variables. Thus $\mathrm{S}=\mathrm{S}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ The system is perturbed at constant pressure by a change in temperature. The path followed by the system is such that the affinity for spontaneous change remains at zero (i.e. at equilibrium) and that the entropy of the system $\mathrm{S}(\mathrm{aq})$ remains constant at that given by equation (b). The equilibrium isentropic expansion of the system is defined by equation (c). $\mathrm{E}_{\mathrm{s}}(\mathrm{A}=0)=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{S}(\mathrm{aq}), \mathrm{A}=0}$ $\mathrm{E}_{\mathrm{S}}(\mathrm{A}=0)$ is an extensive property of the system. Nevertheless it is convenient to consider an intensive property. For example, $\mathrm{E}_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{A}=0 ; \mathrm{w}_{1}=1 \mathrm{~kg} ; \mathrm{m}_{\mathrm{j}}\right)$ is the equilibrium isentropic expansion of a solution molality $\mathrm{m}_{j}$ prepared using $1 \mathrm{~kg}$ of water($\ell$). For a system comprising pure solvent at defined $\mathrm{T}$ and $\mathrm{p}$ we define a molar (equilibrium) isentropic expansion, $\mathrm{E}_{\mathrm{S}}^{*}(\ell)$; equation (d). $\mathrm{E}_{\mathrm{S} 1}^{*}(\ell ; \mathrm{A}=0)=\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{s}_{1}^{*}(\ell), \mathrm{A}=0}$ The volume of a solution, molality $\mathrm{m}_{j}$, prepared using $1 \mathrm{~kg}$ of water($\ell$) is related to the composition using either equations (e) or (f). $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ A key problem emerges. We note that the conditions on the partial differential in equation (c) relate to the entropy of the aqueous solution. The latter condition is not the same as that invoked in equation (d) which refers to the molar entropy of the pure solvent. We could of course differentiate equation (e) with respect to temperature at fixed entropy S(aq). However we would encounter a term $\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{s}(\mathrm{aq})}$. This is a complicated derivative where we might have hoped for a term $\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{s}_{1}^{*}(\ell)}$. The way forward is to accept the problem and define a property, by analogy with the corresponding isobaric property, a property $\phi\left(\mathrm{E}_{\mathrm{S} j} ; \text { def }\right)$ which has the appearance of proper thermodynamic apparent property. Then, $\mathrm{E}_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{A}=0 ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{Sl}}^{*}(\ell ; \mathrm{A}=0)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)$ There is a subtle problem with respect to equation (f) which can be differentiated with respect to $\mathrm{T}$ at constant $\mathrm{S}(\mathrm{aq})$ as defined by equation (b). Then $\mathrm{E}_{\mathrm{s}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \,\left(\frac{\partial \mathrm{V}_{1}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{S}(\mathrm{aq})}+\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \mathrm{V}_{\mathrm{j}}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{S}(\mathrm{aq})}$ Partial molar isentropic expansions $\mathrm{E}_{\mathrm{S}1}(\mathrm{aq})$ and $\mathrm{E}_{\mathrm{S}j}(\mathrm{aq})$ are defined by the following equations. $\mathrm{E}_{\mathrm{Sl}}(\mathrm{aq})=\left(\frac{\partial \mathrm{E}_{\mathrm{s}}(\mathrm{aq})}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{j})}$ $\mathrm{E}_{\mathrm{Sj}}(\mathrm{aq})=\left(\frac{\partial \mathrm{E}_{\mathrm{j}}(\mathrm{aq})}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(1)}$ But $\mathrm{E}_{\mathrm{S}1}$ and $\mathrm{E}_{\mathrm{S}j}$ are non-Lewisian partial molar properties. Hence $\mathrm{E}_{\mathrm{Sl}}(\mathrm{aq}) \neq\left(\frac{\partial \mathrm{V}_{1}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{S}(\mathrm{aq})}$ $\mathrm{E}_{\mathrm{S} j}(\mathrm{aq})=\left(\frac{\partial \mathrm{V}_{\mathrm{j}}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{s}(\mathrm{aq})}$ Then, $\mathrm{E}_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{S} 1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{Sj}}(\mathrm{aq})$ In practical terms equation (n) follows from equation (g), $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)=\left(1 / \mathrm{m}_{\mathrm{j}}\right) \,\left[\mathrm{E}_{\mathrm{S}}\left(\mathrm{aq} ; \mathrm{A}=0 ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{S} 1}^{*}(\ell ; \mathrm{A}=0)\right]$ Two practical equations follow from equation (n) allowing $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)$ to be calculated from the isentropic expansibilities of solutions and solvent, both volume intensive variables [1]. $\phi\left(\mathrm{E}_{\mathrm{s} j} ; \operatorname{def}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{sl}}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ $\phi\left(E_{\mathrm{Sj}} ; \text { def }\right)=\left[c_{j}\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\alpha_{\mathrm{S} 1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ where $\alpha_{\mathrm{s}}(\mathrm{aq})=\frac{1}{\mathrm{~V}(\mathrm{aq})} \,\left(\frac{\partial \mathrm{V}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{s}(\mathrm{aq})}$ $\alpha_{\mathrm{S} 1}^{*}(\ell)=\frac{1}{\mathrm{~V}_{1}^{*}(\ell)} \,\left(\frac{\partial \mathrm{V}_{1}^{*}(\ell)}{\partial \mathrm{T}}\right)_{\mathrm{s}_{1}^{*}(\ell)}$ Footnotes [1] From equation (n), $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)=\left(1 / \mathrm{m}_{\mathrm{j}}\right) \,\left[\mathrm{E}_{\mathrm{s}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{sl}}^{*}(\ell)\right]$ We use equation (m) for a solution prepared using $1 \mathrm{~kg}$ of water. $\mathrm{E}_{\mathrm{s}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{S} 1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{Sj}}(\mathrm{aq})$ Then $\phi\left(\mathrm{E}_{\mathrm{S} j} ; \mathrm{def}\right)=\left(1 / \mathrm{m}_{\mathrm{j}}\right) \,\left[\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{S} 1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{E}_{\mathrm{Sj}}(\mathrm{aq})-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{E}_{\mathrm{Sl} 1}^{*}(\ell)\right]$ Or, $\phi\left(\mathrm{E}_{\mathrm{s} j} ; \operatorname{def}\right)=\mathrm{E}_{\mathrm{S} j}(\mathrm{aq})+\left(1 / \mathrm{M}_{1}\right) \,\left(1 / \mathrm{m}_{\mathrm{j}}\right) \,\left[\mathrm{E}_{\mathrm{Sl}}(\mathrm{aq})-\mathrm{E}_{\mathrm{s} 1}^{*}(\ell)\right]$ Hence using equation (m), $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)=\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \mathrm{E}_{\mathrm{s}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \,\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \mathrm{E}_{\mathrm{Sl}}^{*}(\ell)$ Using equations (q) and (r), $\phi\left(\mathrm{E}_{\mathrm{s}} ; \mathrm{def}\right)=\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s}}(\mathrm{aq}) \, \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \,\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\ell) \, \mathrm{V}_{1}^{*}(\ell)$ Or, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{s} j} ; \text { def }\right)=&\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s}}(\mathrm{aq}) \,\left[\left(1 / \mathrm{M}_{1}\right) \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \ &-\left(1 / \mathrm{M}_{1}\right) \,\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\mathrm{l}) \, \mathrm{V}_{1}^{*}(\ell) \end{aligned} Or $\phi\left(E_{\mathrm{Sj}} ; \operatorname{def}\right)=\left(\mathrm{V}_{1}^{*}(\ell) / \mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}\right) \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Or $\phi\left(E_{\mathrm{s} j} ; \operatorname{def}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Also $\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1}=\left[1 / \mathrm{c}_{\mathrm{j}}\right]-\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Then, $\phi\left(\mathrm{E}_{\mathrm{sj}} ; \mathrm{def}\right)=\left[\left(1 / \mathrm{c}_{\mathrm{j}}\right)-\phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{sl}}^{*}(\ell)\right]+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Or, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)=\left(1 / \mathrm{c}_{\mathrm{j}}\right) \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{s} 1}^{*}(\ell)\right] \ &-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s}}(\mathrm{aq})+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\ell)+\alpha_{\mathrm{s}}(\mathrm{aq}) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Hence, $\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)=\left(1 / \mathrm{c}_{\mathrm{j}}\right) \,\left[\alpha_{\mathrm{s}}(\mathrm{aq})-\alpha_{\mathrm{S} 1}^{*}(\ell)\right]+\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \alpha_{\mathrm{s} 1}^{*}(\ell)$ For further details see---- J.C.R.Reis, G. Douheret, M.I.Davis, I.J.Fjellanger and H.Hoiland, Phys. Chem. Chem. Phys., 2008,10, 561.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.16%3A_Expansions-_Solutions-_Apparent_Molar_Isentropic_and_Isobaric.txt
The starting point is the following calculus operation. $\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ Also $\left(\frac{\partial \phi\left(V_{\mathrm{j}}\right)}{\partial T}\right)_{\mathrm{s}}=\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{T}} \,\left(\frac{\partial \mathrm{T}}{\partial \mathrm{V}}\right)_{\mathrm{p}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ Or, $\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{T}} \, \frac{\mathrm{V}}{\mathrm{V}} \,\left(\frac{\partial \mathrm{T}}{\partial \mathrm{V}}\right)_{\mathrm{p}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ Hence, $\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial T}\right)_{\mathrm{s}}=\frac{\sigma}{\mathrm{T} \, \alpha_{\mathrm{p}}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ But $\frac{\sigma}{\mathrm{T}}=\frac{\left[\alpha_{p}\right]^{2}}{\delta}$ Then $\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\frac{\alpha_{\mathrm{p}}}{\delta} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ Also, $\alpha_{\mathrm{s}}=-\kappa_{\mathrm{s}} \, \sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ Then $\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}}=-\frac{\alpha_{\mathrm{p}}}{\delta}$ Hence $\frac{\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right) / \partial \mathrm{T}\right]_{\mathrm{s}}}{\alpha_{\mathrm{s}}}=-\frac{\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{s}}}{\kappa_{\mathrm{s}}}$ 1.12.19: Expansions- Solutions- Isentropic Dependence of Partial Molar Volume on We switch the condition on a derivative expressing the dependence of partial molar volume on temperature. $\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}}$ Then $\left(\frac{\partial V_{j}}{\partial T}\right)_{\mathrm{s}}=E_{p j}-\left(\frac{\partial V_{j}}{\partial p}\right)_{T} \, \frac{(\partial S / \partial T)_{p}}{(\partial S / \partial p)_{T}}$ Or, $\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{S}}=\mathrm{E}_{\mathrm{pj}}+\frac{(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{p}}}{(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}} \, \mathrm{K}_{\mathrm{T}_{\mathrm{j}}}$ But $\frac{(\partial S / \partial T)_{p}}{(\partial S / \partial p)_{T}}=-\frac{C_{p}}{T \,(\partial V / \partial T)_{p}}=-\frac{C_{p}}{T \, V \, \alpha_{p}}$ Or, $\frac{(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{p}}}{(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}}=-\frac{\sigma}{\mathrm{T} \, \alpha_{\mathrm{p}}}$ Hence, $\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\mathrm{E}_{\mathrm{pj}}-\frac{\sigma}{\mathrm{T} \, \alpha_{\mathrm{p}}} \, \mathrm{K}_{\mathrm{T}_{\mathrm{j}}}$ 1.12.20: Expansions- Isentropic- Liquid Mixtures A given binary liquid mixture has mole fraction $x_{1}\left[=1-x_{2}\right]$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The system is at equilibrium at a minimum in Gibbs energy where the affinity for spontaneous change is zero. The molar volume and molar entropy of the mixtures are given by equations (a) and (b). $\mathrm{V}_{\mathrm{m}}=\mathrm{V}_{\mathrm{m}}\left[\mathrm{T}, \mathrm{p}, \mathrm{x}_{1}\right]$ $\mathrm{S}_{\mathrm{m}}=\mathrm{S}_{\mathrm{m}}\left[\mathrm{T}, \mathrm{p}, \mathrm{x}_{1}\right]$ These two equations describe the properties of the system in the $\mathrm{T}-\mathrm{p}$-composition domain; i.e. a Gibbsian description. We consider two dependences of the volume on temperature under the constraint that the affinity for spontaneous change remains at zero; i.e equilibrium expansions.The isobaric expansion is defined by equation (c). $\mathrm{E}_{\mathrm{p}}(\operatorname{mix})=\left(\frac{\partial \mathrm{V}(\operatorname{mix})}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ The isentropic expansion is defined by equation (d) $\mathrm{E}_{\mathrm{S}}(\operatorname{mix})=\left(\frac{\partial \mathrm{V}(\operatorname{mix})}{\partial \mathrm{T}}\right)_{\mathrm{S}}$ In the latter case the system tracks a path with increase in temperature where the affinity for spontaneous change remains at zero and the entropy remains the same at that defined by equation (b). [$\mathrm{NB} \mathrm{~E}_{p}(\operatorname{mix})$ and $E_{S}(\operatorname{mix})$ as defined by equations (c) and (d) are extensive properties.] The two expansions are related through the (equilibrium) isobaric heat capacity $\mathrm{C}_{\mathrm{p}} (\operatorname{mix})$and the (equilibrium) isothermal compression $\mathrm{K}_{\mathrm{T}}(\operatorname{mix})$ [1]. Thus $\mathrm{E}_{\mathrm{S}}(\operatorname{mix})=\mathrm{E}_{\mathrm{p}}(\operatorname{mix})-\frac{\mathrm{C}_{\mathrm{p}}(\operatorname{mix}) \, \mathrm{K}_{\mathrm{T}}(\operatorname{mix})}{\mathrm{T} \, \mathrm{E}_{\mathrm{p}}(\operatorname{mix})}$ In the context of the property $\mathrm{E}_{\mathrm{p}(\operatorname{mix})$, the entropy of the system changes with an increase in temperature at constant pressure. But by definition the entropy does not change for an isentropic expansion, $\mathrm{E}_{\mathrm{S}(\operatorname{mix})$. For a binary liquid mixture having ideal thermodynamic properties, $E_{S}(\operatorname{mix} ; i d)=E_{p}(\operatorname{mix} ; i d)-\frac{C_{p}(\operatorname{mix} ; i d) \, K_{T}(\operatorname{mix} ; i d)}{T \, E_{p}(\operatorname{mix} ; i d)}$ In this comparison we note that $\mathrm{E}_{\mathrm{p}(\operatorname{mix})$ and $\mathrm{E}_{\mathrm{p}}(\operatorname{mix} ; \mathrm{id})$ refer to the same pressure but the entropies referred to in $\mathrm{E}_{\mathrm{S}}(\operatorname{mix})$ and $\mathrm{E}_{\mathrm{S}}(\operatorname{mix} ; \mathrm{id})$ are not the same. The same contrast arises when we set out the two equations describing expansions of the pure liquids. $\mathrm{E}_{\mathrm{S} 1}^{*}(\ell)=\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)-\frac{\mathrm{C}_{\mathrm{p} 1}^{*}(\ell) \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell)}{\mathrm{T} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)}$ $\mathrm{E}_{\mathrm{S} 2}^{*}(\ell)=\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)-\frac{\mathrm{C}_{\mathrm{p} 2}^{*}(\ell) \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell)}{\mathrm{T} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}$ The subject is complicated by the galaxy of entropies implied by the phrase ‘at constant entropy’. Footnote [1] Using a calculus operation, $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{S}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{S}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ We note two Maxwell equations. From $\mathrm{U}=\mathrm{U}[\mathrm{S}, \mathrm{V}], \quad \partial^{2} \mathrm{U} / \partial \mathrm{S} \, \partial \mathrm{V}=\partial^{2} \mathrm{U} / \partial \mathrm{V} \, \partial \mathrm{S}$ Then $\left(\frac{\partial \mathrm{T}}{\partial \mathrm{V}}\right)_{\mathrm{s}}=-\left(\frac{\partial \mathrm{p}}{\partial \mathrm{S}}\right)_{\mathrm{V}}$ We invert the latter equation. Hence \begin{aligned} \mathrm{E}_{\mathrm{S}}=&\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{V}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{s}} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{p}} \ &=-\mathrm{K}_{\mathrm{S}} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=-\mathrm{K}_{\mathrm{s}} \, \mathrm{C}_{\mathrm{p}} / \mathrm{T} \, \mathrm{E}_{\mathrm{p}} \end{aligned} Similarly $\partial^{2} \mathrm{G} / \partial \mathrm{T} \, \partial \mathrm{p}=\partial^{2} \mathrm{G} / \partial \mathrm{p} \, \partial \mathrm{T}$ Then, $E_{p}=\left(\frac{\partial V}{\partial T}\right)_{p}=-\left(\frac{\partial S}{\partial p}\right)_{T}$ Also at equilibrium, $\mathrm{S}=-\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ But $\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$. Then $\mathrm{H}=\mathrm{G}-\mathrm{T} \,\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ $\frac{\partial \mathrm{H}}{\partial \mathrm{T}}=\frac{\partial \mathrm{G}}{\partial \mathrm{T}}-\mathrm{T} \,\left(\frac{\partial^{2} \mathrm{G}}{\partial \mathrm{T}^{2}}\right)-\frac{\partial \mathrm{G}}{\partial \mathrm{T}}$ Further, $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\mathrm{C}_{\mathrm{p}}=-\mathrm{T} \,\left(\frac{\partial^{2} \mathrm{G}}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}=\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Based on equation (a), $\mathrm{E}_{\mathrm{S}}=\mathrm{E}_{\mathrm{p}}-\mathrm{C}_{\mathrm{p}} \, \mathrm{K}_{\mathrm{T}} / \mathrm{T} \, \mathrm{E}_{\mathrm{p}}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.18%3A_Expansions-_Solutions-_Isentropic_Dependence_of_Apparent_Molar_Volume_o.txt
The starting point is equation (a). $\mathrm{E}_{\mathrm{p}}=\mathrm{E}_{\mathrm{S}}+\frac{\mathrm{K}_{\mathrm{T}} \, \mathrm{C}_{\mathrm{p}}}{\mathrm{T} \, \mathrm{E}_{\mathrm{p}}}$ We differentiate this equation with respect to the amount of solute $\mathrm{n}_{j}$ at fixed $\mathrm{T}$, $\mathrm{p}$ and amount of solvent $\mathrm{n}_{1}$. $\mathrm{E}_{\mathrm{pj}}=\mathrm{E}_{\mathrm{Sj}}+\frac{1}{\mathrm{~T}} \,\left[\frac{\mathrm{K}_{\mathrm{T}}}{\mathrm{E}_{\mathrm{p}}} \, \mathrm{C}_{\mathrm{pj}}+\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{E}_{\mathrm{p}}} \, \mathrm{K}_{\mathrm{Tj}}-\frac{\mathrm{K}_{\mathrm{T}} \, \mathrm{C}_{\mathrm{p}}}{\left(\mathrm{E}_{\mathrm{p}}\right)^{2}} \, \mathrm{E}_{\mathrm{pj}}\right]$ Or, $\mathrm{E}_{\mathrm{pj}}=\mathrm{E}_{\mathrm{Sj}}+\frac{1}{\mathrm{~T}} \, \frac{\mathrm{K}_{\mathrm{T}} \, \mathrm{C}_{\mathrm{p}}}{\mathrm{E}_{\mathrm{p}}} \,\left[\frac{\mathrm{C}_{\mathrm{p} j}}{\mathrm{C}_{\mathrm{p}}}+\frac{\mathrm{K}_{\mathrm{T} \mathrm{J}}}{\mathrm{K}_{\mathrm{T}}}-\frac{\mathrm{E}_{\mathrm{p} j}}{\mathrm{E}_{\mathrm{p}}}\right]$ We convert to volume intensive variables. $\mathrm{E}_{\mathrm{pj}}=\mathrm{E}_{\mathrm{S} \mathrm{j}}+\frac{1}{\mathrm{~T}} \, \frac{\kappa_{\mathrm{T}} \, \mathrm{C}_{\mathrm{p}}}{\alpha_{\mathrm{p}} \, \mathrm{V}} \,\left[\frac{\mathrm{V} \, \mathrm{C}_{\mathrm{pj}}}{\mathrm{C}_{\mathrm{p}}}+\frac{\mathrm{V} \, \mathrm{K}_{\mathrm{T} j}}{\mathrm{~K}_{\mathrm{T}}}-\frac{\mathrm{V} \, \mathrm{E}_{\mathrm{p} j}}{\mathrm{E}_{\mathrm{p}}}\right]$ Or, $E_{p j}=E_{S j}+\frac{1}{T} \, \frac{K_{T} \, \sigma}{\alpha_{p}} \,\left[\frac{C_{p j}}{\sigma}+\frac{K_{T_{j}}}{K_{T}}-\frac{E_{p j}}{\alpha_{p}}\right]$ But $\varepsilon=K_{\mathrm{T}} \, \sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ Then, $\frac{\mathrm{E}_{\mathrm{pj}}-\mathrm{E}_{\mathrm{S} j}}{\varepsilon}=-\frac{\mathrm{E}_{\mathrm{pj}}}{\alpha_{\mathrm{p}}}+\frac{\mathrm{K}_{\mathrm{T} j}}{\kappa_{\mathrm{T}}}+\frac{\mathrm{C}_{\mathrm{pj}}}{\sigma}$ Hence for an aqueous solution in the limit of infinite dilution, $\frac{\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})-\mathrm{E}_{\mathrm{sj}}^{\infty}(\mathrm{aq})}{\varepsilon_{1}^{*}(\ell)}=-\frac{\mathrm{E}_{\mathrm{pj}}^{\infty}(\mathrm{aq})}{\alpha_{\mathrm{p} 1}^{*}(\ell)}+\frac{\mathrm{K}_{\mathrm{T} j}^{\infty}(\mathrm{aq})}{\kappa_{\mathrm{T} 1}^{*}(\ell)}+\frac{\mathrm{C}_{\mathrm{pj}}^{\infty}}{\sigma_{1}^{*}(\ell)}$ We start with the equation, $\mathrm{E}_{\mathrm{s}}=-\frac{\mathrm{K}_{\mathrm{s}} \, \mathrm{C}_{\mathrm{p}}}{\mathrm{T} \, \mathrm{E}_{\mathrm{p}}}$ The latter equation is differentiated with respect to the amount of solute $\mathrm{n}_{j}$ in a solution at fixed $\mathrm{T}$, fixed $\mathrm{p}$ and fixed amount of solvent, $\mathrm{n}_{1}$. \begin{aligned} \left(\frac{\partial E_{S}}{\partial n_{j}}\right)_{T, p, n(1)}=-\frac{C_{p}}{T \, E_{p}} \,\left(\frac{\partial K_{s}}{\partial n_{j}}\right)_{T, p, n(1)}-\frac{K_{s}}{T \, E_{p}} \,\left(\frac{\partial C_{p}}{\partial n_{j}}\right)_{T, p, n(1)} \ &+\frac{K_{s} \, C_{p}}{T \,\left(E_{p}\right)^{2}} \,\left(\frac{\partial E_{p}}{\partial n_{j}}\right)_{T, p, n(1)} \end{aligned} Or, $E_{S_{j}}=-\frac{C_{p}}{T \, E_{p}} \, \frac{K_{s}}{K_{s}} \, K_{s_{j}}-\frac{K_{s}}{T \, E_{p}} \, \frac{C_{p}}{C_{p}} \, C_{p j}+\frac{K_{s} \, C_{p}}{T \,\left(E_{p}\right)^{2}} \, E_{p j}$ We rewrite the latter equation in terms of volume intensive variables. $\mathrm{E}_{\mathrm{Sj}}=-\frac{1}{\mathrm{~T}} \, \frac{\sigma}{\alpha_{\mathrm{p}}} \, \frac{\kappa_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \, \mathrm{K}_{\mathrm{Sj}}-\frac{1}{\mathrm{~T}} \, \frac{\kappa_{\mathrm{S}}}{\alpha_{\mathrm{p}}} \, \frac{\sigma}{\sigma} \, \mathrm{C}_{\mathrm{pj}}+\frac{\kappa_{\mathrm{s}} \, \sigma}{\mathrm{T} \,\left(\alpha_{\mathrm{p}}\right)^{2}} \, \mathrm{E}_{\mathrm{pj}}$ But $\alpha_{\mathrm{s}}=-\frac{\kappa_{\mathrm{s}} \, \sigma}{\mathrm{T} \, \alpha_{\mathrm{p}}}$ Then $E_{\mathrm{S}_{\mathrm{j}}}=\alpha_{\mathrm{s}} \, \frac{\mathrm{K}_{\mathrm{s} \mathrm{j}}}{\kappa_{\mathrm{s}}}+\alpha_{\mathrm{s}} \, \frac{\mathrm{C}_{\mathrm{p} j}}{\sigma}-\alpha_{\mathrm{s}} \, \frac{\mathrm{E}_{\mathrm{pj}}}{\alpha_{\mathrm{p}}}$ Therefore (with a change of order) $\frac{E_{\mathrm{s} j}}{\alpha_{p}}=-\frac{E_{p j}}{\alpha_{p}}+\frac{K_{S j}}{K_{s}}+\frac{C_{p j}}{\sigma}$ 1.12.22: Expansions and Compressions- Solutions- Isentropic Dependence of Volume The starting point is the calculus operation for a double differential. $\frac{\partial^{2} U}{\partial S \, \partial V}=\frac{\partial^{2} U}{\partial V \, \partial S}$ Then, $\left(\frac{\partial T}{\partial V}\right)_{s}=-\left(\frac{\partial p}{\partial S}\right)_{v}$ Or, $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{V}}$ But, $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{s}} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{p}}$ Also we note that $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Then, $\left(\frac{\partial V}{\partial T}\right)_{s}=\left(\frac{\partial V}{\partial p}\right)_{s} \,\left(\frac{\partial S}{\partial T}\right)_{p} \,\left(\frac{\partial T}{\partial V}\right)_{p}$ However from the Gibbs - Helmholtz Equation, $\left(\frac{\partial S}{\partial T}\right)_{p}=\frac{C_{p}}{T}$ Then $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{s}} \,\left(\frac{\partial \mathrm{T}}{\partial \mathrm{V}}\right)_{\mathrm{p}} \, \frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{T}}$ Or, $\mathrm{E}_{\mathrm{s}}=-\frac{\mathrm{K}_{\mathrm{s}} \, \mathrm{C}_{\mathrm{p}}}{\mathrm{T} \, \mathrm{E}_{\mathrm{p}}}$ We divide both sides of equation (f) by volume $\mathrm{V}$. Hence $\alpha_{\mathrm{s}}=-\kappa_{\mathrm{s}} \, \sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ 1.12.23: Expansions- The Difference For a solution, $\varepsilon=\alpha_{p}-\alpha_{s}=\kappa_{T} \, \sigma / T \, \alpha_{p}$ In order to simplify the algebra, we omit (aq) and ($\ell$) when describing the properties of an aqueous solution and the pure liquid respectively. Superscript '*' identifies the pure solvent. $\varepsilon^{*}=\alpha_{\mathrm{p}}^{*}-\alpha_{\mathrm{S}}^{*}=\kappa_{\mathrm{T}}^{*} \, \sigma^{*} / \mathrm{T} \, \alpha_{\mathrm{p}}^{*}$ Hence, $\varepsilon-\varepsilon^{*}=\frac{\varepsilon}{\kappa_{\mathrm{T}} \, \sigma} \,\left[\kappa_{\mathrm{T}} \, \sigma-\kappa_{\mathrm{T}}^{*} \, \sigma^{*}\right]-\frac{\varepsilon^{*}}{\alpha_{\mathrm{p}}} \,\left[\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right]$ The latter equation is effectively an identity. According to equation (c) $\varepsilon-\varepsilon=\varepsilon-\frac{\varepsilon}{\kappa_{\mathrm{T}} \, \sigma} \, \kappa_{\mathrm{T}}^{*} \, \sigma^{*}-\varepsilon+\frac{\varepsilon}{\alpha_{\mathrm{p}}} \, \alpha_{\mathrm{p}}^{*}$ We use equations (a) and (b) in the second and fourth terms on the right hand side of the latter equation. $\varepsilon-\varepsilon^{*}=\varepsilon-\frac{\varepsilon}{\kappa_{\mathrm{T}} \, \sigma} \, \varepsilon^{*} \, \mathrm{T} \, \alpha_{\mathrm{p}}^{*}-\varepsilon^{*}+\frac{\varepsilon^{*} \, \alpha_{\mathrm{p}}^{*} \, \mathrm{T} \, \varepsilon}{\kappa_{\mathrm{T}} \, \sigma}$ Or $\varepsilon-\varepsilon^{*}=\varepsilon-\varepsilon^{*}$ Further, as an identity, $\kappa_{\mathrm{T}} \, \sigma-\kappa_{\mathrm{T}}^{*} \, \sigma^{*}=\sigma \,\left(\kappa_{\mathrm{T}}-\kappa_{\mathrm{T}}^{*}\right)+\kappa_{\mathrm{T}}^{*} \,\left(\sigma-\sigma^{*}\right)$ From equation (c), $\varepsilon-\varepsilon^{*}=\frac{\varepsilon}{\kappa_{\mathrm{T}}} \,\left[\kappa_{\mathrm{T}}-\kappa_{\mathrm{T}}^{*}\right]+\frac{\varepsilon \, \kappa_{\mathrm{T}}^{*}}{\kappa_{\mathrm{T}} \, \sigma} \,\left(\sigma-\sigma^{*}\right)-\frac{\varepsilon^{*}}{\alpha_{\mathrm{p}}} \,\left[\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right]$ But $\phi\left(\mathrm{E}_{\mathrm{sj}} ; \mathrm{def}\right)=\frac{\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}}{\mathrm{c}_{\mathrm{j}}}+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ The analogue for $\phi\left(E_{p j}\right)$ is the following equation. $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)=\frac{\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}}{\mathrm{c}_{\mathrm{j}}}+\alpha_{\mathrm{p}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence $\phi\left(E_{p j}\right)-\phi\left(E_{S_{j}} ; \operatorname{def}\right)=\frac{\varepsilon-\varepsilon^{*}}{c_{j}}+\varepsilon^{*} \, \phi\left(V_{j}\right)$ From equation (g), dividing by $\mathrm{c}_{j}$, $\begin{gathered} \frac{\varepsilon-\varepsilon}{c_{j}}=\frac{\varepsilon}{\kappa_{T}} \, \frac{1}{c_{j}} \,\left[\kappa_{T}-\kappa_{T}^{*}\right]+\frac{\varepsilon \, \kappa_{T}^{*}}{\kappa_{T} \, \sigma} \, \frac{1}{c_{j}} \,\left[\sigma-\sigma^{*}\right] \ -\frac{\varepsilon^{*}}{\alpha_{p}} \, \frac{1}{c_{j}} \,\left[\alpha_{p}-\alpha_{p}^{*}\right] \end{gathered}$ But from equation (i) $\frac{\varepsilon-\varepsilon^{*}}{c_{j}}=\phi\left(E_{p j}\right)-\phi\left(E_{S j} ; \operatorname{def}\right)-\varepsilon^{*} \, \phi\left(V_{j}\right)$ Equations having similar form for $\left(\kappa_{\mathrm{T}}-\kappa_{\mathrm{T}}^{*}\right),\left(\sigma-\sigma^{*}\right)$ and $\left(\alpha_{p}-\alpha_{p}^{*}\right)$ are readily generated. Hence \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{pj}}\right)-\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right) &=\frac{\varepsilon}{\kappa_{\mathrm{T}}} \,\left[\phi\left(\mathrm{K}_{\mathrm{Tj}}\right)-\kappa_{\mathrm{T}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\frac{\varepsilon \, \kappa_{\mathrm{T}}^{*}}{\kappa_{\mathrm{T}} \, \sigma} \,\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\sigma^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \ &-\frac{\varepsilon^{*}}{\alpha_{\mathrm{p}}} \,\left[\phi\left(\mathrm{E}_{\mathrm{pj}}\right)-\alpha_{\mathrm{p}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\varepsilon^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Therefore \begin{aligned} \phi\left(E_{\mathrm{pj}}\right)-\phi\left(E_{\mathrm{Sj}} ; \operatorname{def}\right) &=-\frac{\varepsilon}{\alpha_{\mathrm{p}}} \, \phi\left(\mathrm{E}_{\mathrm{pj}}\right)+\frac{\varepsilon}{\kappa_{\mathrm{T}}} \, \phi\left(\mathrm{K}_{\mathrm{Tj}}\right)+\frac{\varepsilon \, \kappa_{\mathrm{T}}^{*}}{\kappa_{\mathrm{T}} \, \sigma} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) \ &+\left[\varepsilon * \,\left(1+\frac{\alpha_{\mathrm{p}}^{*}}{\alpha_{\mathrm{p}}}\right)-\frac{\varepsilon \, \kappa_{\mathrm{T}}^{*}}{\kappa_{\mathrm{T}}} \,\left(1+\frac{\sigma^{*}}{\sigma}\right)\right] \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \quad(\mathrm{m}) \end{aligned} In the limit of infinite dilution, $\frac{\phi\left(E_{\mathrm{pj}}\right)^{\infty}-\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \mathrm{def}\right)^{\infty}}{\varepsilon_{1}^{*}(\ell)}=-\frac{\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}}{\alpha_{\mathrm{p} 1}^{*}(\ell)}+\frac{\phi\left(\mathrm{K}_{\mathrm{T}}\right)^{\infty}}{\kappa_{\mathrm{Tl}}^{*}(\ell)}+\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}}{\sigma_{1}^{*}(\ell)}$ 1.12.24: Expansions- Equations We simplify the algebra by omitting the descriptors (aq) and ($\ell$) in the following equations. The starting point is the following equation. $\alpha_{S}-\alpha_{S}^{*}=\frac{\alpha_{S}}{\kappa_{S} \, \sigma} \,\left(\kappa_{S} \, \sigma-\kappa_{S}^{*} \, \sigma^{*}\right)-\frac{\alpha_{S}^{*}}{\alpha_{p}} \,\left(\alpha_{p}-\alpha_{p}^{*}\right)$ The latter equation is effectively an identity. Thus from equation (a) $\alpha_{S}-\alpha_{S}^{*}=\alpha_{S}-\frac{\alpha_{S}}{\kappa_{S} \, \sigma} \, \kappa_{S}^{*} \, \sigma^{*}-\alpha_{S}^{*}+\frac{\alpha_{S}^{*}}{\alpha_{p}} \, \alpha_{p}^{*}$ But $\alpha_{\mathrm{s}}=-\kappa_{\mathrm{s}} \, \sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ and $\alpha_{\mathrm{p}}^{*} / \alpha_{\mathrm{s}}^{*}=-\kappa_{\mathrm{s}}^{*} \, \sigma^{*} / \mathrm{T}$ Then from (b), $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}+\frac{\kappa_{\mathrm{s}} \, \sigma}{\mathrm{T} \, \alpha_{\mathrm{p}} \, \kappa_{\mathrm{s}} \, \sigma} \, \kappa_{\mathrm{s}}^{*} \, \sigma^{*}-\frac{\kappa_{\mathrm{s}}^{*} \, \sigma^{*}}{\alpha_{\mathrm{p}} \, \mathrm{T}}$ or $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}$ But as an identity, $\kappa_{\mathrm{S}} \, \sigma-\kappa_{\mathrm{S}}^{*} \, \sigma^{*}=\sigma \,\left(\kappa_{\mathrm{S}}-\kappa_{\mathrm{S}}^{*}\right)+\kappa_{\mathrm{S}}^{*} \,\left(\sigma-\sigma^{*}\right)$ Then from equations (a) and (c), $\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}=\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \,\left(\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \,\left(\sigma-\sigma^{*}\right)-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \,\left(\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right)$ But, $\phi\left(E_{\mathrm{S}} ; \operatorname{def}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\alpha_{\mathrm{s}}-\alpha_{\mathrm{s}}^{*}\right)+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence \begin{aligned} \phi\left(E_{\mathrm{s} j} ; \operatorname{def}\right)=& \frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \,\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \,\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\sigma-\sigma^{*}\right) \ &-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \,\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\alpha_{\mathrm{p}}-\alpha_{\mathrm{p}}^{*}\right)+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} For isobaric heat capacities, $\phi\left(\mathrm{C}_{\mathrm{pj}}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\sigma-\sigma^{*}\right)+\sigma^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Also $\phi\left(K_{\mathrm{S}_{j}} ; \operatorname{def}\right)=\left[\mathrm{c}_{\mathrm{j}}\right]^{-1} \,\left(\kappa_{\mathrm{s}}-\kappa_{\mathrm{s}}^{*}\right)+\kappa_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Hence \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{s} j} ; \operatorname{def}\right)=& \frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \,\left[\phi\left(\mathrm{K}_{\mathrm{s} j} ; \operatorname{def}\right)-\kappa_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \,\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\sigma^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \ &-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \,\left[\phi\left(\mathrm{E}_{\mathrm{pj}}\right)-\alpha_{\mathrm{p}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\alpha_{\mathrm{s}}^{*} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Then with a little reorganisation, \begin{aligned} \phi\left(\mathrm{E}_{\mathrm{s} j} ; \operatorname{def}\right)=&-\frac{\alpha_{\mathrm{s}}^{*}}{\alpha_{\mathrm{p}}} \, \phi\left(\mathrm{E}_{\mathrm{p} j}\right)+\frac{\alpha_{\mathrm{s}}}{\kappa_{\mathrm{s}}} \, \phi\left(\mathrm{K}_{\mathrm{s} j} ; \operatorname{def}\right)+\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}} \, \sigma} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) \ &+\left[\alpha_{\mathrm{s}}^{*} \,\left(1+\frac{\alpha_{\mathrm{p}}^{*}}{\alpha_{\mathrm{p}}}\right)-\frac{\alpha_{\mathrm{s}} \, \kappa_{\mathrm{s}}^{*}}{\kappa_{\mathrm{s}}} \,\left(1+\frac{\sigma^{*}}{\sigma}\right)\right] \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \end{aligned} Hence, in the limit of infinite dilution, $\frac{\phi\left(\mathrm{E}_{\mathrm{Sj}} ; \operatorname{def}\right)^{\infty}}{\alpha_{\mathrm{S} 1}^{*}(\ell)}=-\frac{\phi\left(\mathrm{E}_{\mathrm{pj}}\right)^{\infty}}{\alpha_{\mathrm{p} 1}^{*}(\ell)}+\frac{\phi\left(\mathrm{K}_{\mathrm{s} j} ; \mathrm{def}\right)^{\infty}}{\kappa_{\mathrm{s} 1}^{*}(\ell)}+\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}}{\sigma_{1}^{*}(\ell)}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.12%3A_Expansions/1.12.21%3A_Expansions-_Solutions-_Partial_Molar_Isobaric_and_Isentropic.txt
The Gibbs energy $\mathrm{G}$ of a given closed system is characterised by the independent variables temperature $\mathrm{T}$, pressure $\mathrm{p}$ and composition $\xi$. $\mathrm{G}=\mathrm{G}[\mathrm{T}, \mathrm{p}, \xi] \label{a}$ In the state defined by Equation \ref{a} the affinity for spontaneous change is $\mathrm{A}$. Starting with the system in the state defined by equation (a) it is possible to change the pressure (at fixed temperature) and thereby perturb the system to neighbouring states where the affinity $\mathrm{A}$ is the same. The differential dependence of $\mathrm{G}$ on pressure along this path is given by the partial differential $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \mathrm{A}}$. Returning to the state defined by Equation \ref{a} we envisage a perturbation by a change in pressure (at fixed temperature) along a path such that the extent of chemical reaction $\xi$ remains constant; the corresponding differential dependence of $\mathrm{G}$ is given by $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \xi}$.The two partial derivatives are related by equation (b) for a system at constant temperature. $\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{A}}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right]_{\xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{p}} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{p}} \label{b}$ The important result which emerges from this equation concerns the properties of a system at chemical equilibrium where the affinity for spontaneous change is zero, the rate of change $\mathrm{d} \xi / \mathrm{dt}$ is zero, the Gibbs energy is a minimum and, significantly, $(\partial G / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero. Hence $V=\left[\frac{\partial G}{\partial p}\right]_{T, A=0}=\left[\frac{\partial G}{\partial p}\right]_{T, \xi(e q)}$ Thus we confirm that the volume $\mathrm{V}$ of a system is a strong state variable, the dependence of $\mathrm{G}$ on pressure (at constant $\mathrm{T}$) at constant ‘$\mathrm{A}=0$’ and at constant composition, $\xi^{\mathrm{eq}}$ are identical. However if we turn our attention on to expansibilities and compressibilities we find that it is important to distinguish between two sets of properties, equilibrium and frozen. 1.13.02: Equilibrium- Isochoric and Isobaric Paramenters In a description of a given closed system we define two extensive state variables, the Gibbs energy $\mathrm{G}$ and the Helmholtz energy $\mathrm{F}$. $\mathrm{G}=\mathrm{U}+\mathrm{p} \, \mathrm{V}-\mathrm{T} \, \mathrm{S}$ $\mathrm{F}=\mathrm{U}-\mathrm{T} \, \mathrm{S}$ Hence, $\mathrm{G}=\mathrm{F}+\mathrm{p} \, \mathrm{V}$ The latter interesting equation links two practical thermodynamic potentials; 1. $\mathrm{G}$ for processes at fixed $\mathrm{T}$ and $\mathrm{p}$, 2. $\mathrm{F}$ for processes at fixed $\mathrm{T}$ and $\mathrm{V}$. The dependence of $\mathrm{G}$ on extent of reaction at constant temperature and pressure is related to the differential dependence of $\mathrm{F}$ on $\xi$ at fixed temperature and pressure. $\left(\frac{\partial \mathrm{G}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}=\left(\frac{\partial \mathrm{F}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}+\mathrm{p} \,\left(\frac{\partial \mathrm{V}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}$ At equilibrium where $\mathrm{A} = 0$, $\xi = \xi^{\mathrm{eq}$ and the Gibbs energy is a minimum [i.e. $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}=0$], $\left(\frac{\partial F}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{A}=0}=\mathrm{p} \,\left(\frac{\partial \mathrm{V}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{A}=0}$ In other words the differential dependence of the Helmholtz energy on extent of reaction at equilibrium (at constant $\mathrm{T}$ and $\mathrm{p}$) is related to the volume of reaction. We rewrite equation (c); $\mathrm{F}=\mathrm{G}-\mathrm{p} \, \mathrm{V}$ At constant temperature and volume, $\left(\frac{\partial \mathrm{F}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{V}}=\left(\frac{\partial \mathrm{G}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{V}}-\mathrm{V} \,\left(\frac{\partial \mathrm{p}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{V}}$ At equilibrium (at constant $\mathrm{T}$ and $\mathrm{V}$) where the Helmholtz energy $\mathrm{F}$ is a minimum, clearly the Gibbs energy is not at a minimum. The dependence of both $\mathrm{G}$ and $\mathrm{F}$ on temperature at equilibrium can be expressed using two Gibbs - Helmholtz equations. Thus, $\left[\frac{\partial(\Delta \mathrm{G} / \mathrm{T})}{\partial(1 / \mathrm{T})}\right]_{\mathrm{p}, \mathrm{A}=0}^{\mathrm{eq}}=\Delta \mathrm{H}^{\mathrm{eq}}$ $\left[\frac{\partial(\Delta \mathrm{F} / \mathrm{T})}{\partial(1 / \mathrm{T})}\right]_{\mathrm{V}, \mathrm{A}=0}^{\mathrm{eq}}=\Delta \mathrm{U}^{\mathrm{eq}}$ From a practical standpoint, determination of $\Delta\mathrm{H}^{\mathrm{eq}}$ is reasonably straightforward because over a range of temperatures the isobaric condition is readily satisfied. Thus we probe this differential dependence at a series of defined temperatures at fixed pressure; i.e. over the range $\mathrm{T}-\delta \mathrm{T}$ to $\mathrm{T}+\delta \mathrm{T}$ about $\mathrm{T}$ for a number of temperatures. The condition ‘at constant volume’ presents problems. In principle we change the pressure to hold $\mathrm{V}$ constant over a range of temperatures. Then we probe the differential dependence of $(\Delta \mathrm{F} / \mathrm{T})$ at a series of fixed temperatures; e.g. over the range $\mathrm{T}-\Delta \mathrm{T}$ to $\mathrm{T}-\Delta \mathrm{T}$ about a given temperature T. If the range of temperatures is large, there is a high probability that very high pressures will be required to hold the global isochoric condition. Another approach probes the dependence of $(\Delta \mathrm{F} / \mathrm{T})$ on temperature at a series of temperatures where volume $\mathrm{V}$ is held constant by changing the pressure over the range $\mathrm{T}_{\mathrm{i}}-\delta \mathrm{T}\) to \[\mathrm{T}_{\mathrm{i}}-\delta \mathrm{T}$ about $\mathrm{T}_{\mathrm{i}}$. Volume $\mathrm{V}_{\mathrm{i}}$ is constant over a small range of temperature. Here the isochoric condition is local to temperature $\mathrm{T}$; thus $\Delta\mathrm{U}$ is obtained at $\mathrm{T}_{\mathrm{i}}$ and $\mathrm{V}_{\mathrm{i}}$. Under these circumstances, comparison of derived $\Delta \mathrm{U}$ - quantities as a function of temperature is not straightforward. Interestingly the solvent water presents pairs of temperatures either side of the TMD where molar volume of water is the same at, for example, ambient pressure. It might be possible to explore this feature by assuming that the volumes of two very dilute solutions are also identical at matched pairs of temperatures.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.13%3A_Equilibrium/1.13.01%3A_Equilibrium_and_Frozen_Properties.txt
A given homogeneous liquid system comprises two chemical substances $\mathrm{i}$ and $\mathrm{j}$ at known $\mathrm{T}$ and $\mathrm{p}$. The temperature and/or pressure are changed. Consequently chemical substance $\mathrm{j}$ spontaneously separates out as a solid phase but substance $\mathrm{i}$ does not. Hence the liquid becomes richer in chemical substance $\mathrm{i}$. The starting point of the analysis is the following equation for the affinity for spontaneous transfer of substance $\mathrm{j}$ from phase II to phase I [1]. \begin{aligned} \delta\left(\frac{A_{j}}{T}\right)=& \frac{\left[\Delta_{\text {trans }} H_{j}^{0}(T, p)\right]}{T^{2}} \, \delta T \ &-\frac{\left[\Delta_{\text {trans }} V_{j}^{0}(T, p)\right]}{T} \, \delta p+R \, \delta \ln \left[\frac{x_{j}(I) \, f_{j}(I)}{x_{j}(\text { II }) \, f_{j}(I I)}\right] \end{aligned} For two equilibrium states such that $\delta\left(\mathrm{A}_{\mathrm{j}} / \mathrm{T}\right)$ is zero for the transfer of chemical substance $\mathrm{j}$ from phase II to phase I, $\mathrm{R} \, \delta \ln \left[\frac{\mathrm{x}_{\mathrm{j}}(\mathrm{II}) \, \mathrm{f}_{\mathrm{j}}(\mathrm{II})}{\mathrm{x}_{\mathrm{j}}(\mathrm{I}) \, \mathrm{f}_{\mathrm{j}}(\mathrm{I})}\right]=\frac{\left[\Delta_{\text {trans }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{T}^{2}} \, \delta \mathrm{T}-\frac{\left[\Delta_{\text {trans }} \mathrm{V}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{T}} \, \delta \mathrm{p}$ In this application, chemical substance $\mathrm{i}$ cannot exist in phase I. Then the equilibrium states are determined by substance $\mathrm{j}$. Further we consider the case where state I corresponds to pure $\mathrm{j}$ such that $x_{j}(I) \, f_{j}(I)$ is unity at reference temperature $\mathrm{T}_{\text{ref}}$ and reference pressure pref. We integrate equation (b) between these two states. \begin{aligned} &\ln \left[\mathrm{x}_{\mathrm{j}}(\mathrm{II}) \, \mathrm{f}_{\mathrm{j}}(\mathrm{II})\right]= \ &\qquad \int_{\mathrm{T}(\mathrm{ref})}^{\mathrm{T}} \frac{\left[\Delta_{\text {trans }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R} \, \mathrm{T}^{2}} \, \mathrm{dT}-\int_{\mathrm{p}(\mathrm{ref})}^{\mathrm{p}} \frac{\left[\Delta_{\text {trans }} \mathrm{V}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R} \, \mathrm{T}} \, \mathrm{dp} \end{aligned} In the event that the pressure is constant, $\ln \left[\mathrm{x}_{\mathrm{j}}(\mathrm{II}) \, \mathrm{f}_{\mathrm{j}}(\mathrm{II})\right]=\int_{\mathrm{T}(\mathrm{ref})}^{\mathrm{T}} \frac{\left[\Delta_{\text {trans }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R} \, \mathrm{T}^{2}} \, \mathrm{dT}$ Footnote [1] By definition, for the transfer of one mole of chemical substance j from phase II to phase I, $A_{j}=-\left[\mu_{j}(\mathrm{I})-\mu_{j}(\mathrm{II})\right] ; \mathrm{Or}, \mathrm{A}_{\mathrm{j}}=\mu_{\mathrm{j}}(\mathrm{II})-\mu_{\mathrm{j}}(\mathrm{I})$ 1.13.04: Equilibrium- Liquid-Solid- Schroeder - van Laar Equation A given homogeneous binary liquid system (at pressure $\mathrm{p}$) contains two chemical substances $\mathrm{i}$ and $\mathrm{j}$ at temperature $\mathrm{T}$. The liquid system is cooled and only substance $\mathrm{j}$ separates out as the pure solid substance $\mathrm{j}$. Hence, $\ln \left[\mathrm{x}_{\mathrm{j}}(\ell) \,\mathrm{f}_{\mathrm{j}}(\ell)\right]=\int_{\mathrm{T}_{\mathrm{j}}^{0}}^{\mathrm{T}} \frac{\left[\Delta_{\text {trans }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R} \,\mathrm{T}^{2}} \,\mathrm{dT} \label{a}$ Here $x_{j}(\ell)$ is the mole fraction composition of the liquid; $f_{j}(\ell)$ is the rational activity coefficient of substance $j$ in the liquid mixture at mole fraction $x_{j}(\ell)$ and temperature $\mathrm{T}$. $\mathrm{T}_{\mathrm{j}}^{0}$ is the melting point of pure $j$ substance $j$ at pressure $\mathrm{p}$; i.e., both liquid and solid phases are pure chemical substance $j$. In the event that $\Delta_{\text {trans }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})$ is independent of temperature [i.e. $\Delta_{\text {trans }} C_{p j}^{0}(T, p)$ is zero] Equation \ref{a} is integrated to yield Equation \ref{b}. $-\ln \left[\mathrm{x}_{\mathrm{j}}(\ell) \,\mathrm{f}_{\mathrm{j}}(\ell)\right]=\dfrac{\Delta_{\text {fus}} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})}{\mathrm{R}} \, \left(\dfrac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\mathrm{j}}^{0}}\right) \label{b}$ The phenomenon under consideration is fusion so that $\Delta_{\text {fus }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})$ is the enthalpy of fusion of chemical substance $j$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. In the event that the thermodynamic properties of the liquid-solid system are ideal, Equation \ref{b} simplifies to Equation \ref{c}. $-\ln \left[\mathrm{x}_{\mathrm{j}}(\ell)\right]=\frac{\Delta_{\mathrm{f}} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T}, \mathrm{p})}{\mathrm{R}} \left(\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\mathrm{j}}^{0}}\right) \label{c}$ Equation \ref{c} is the Schroeder- van Laar Equation [1]. Footnote [1] I. Prigogine and R Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Greeen, London, 1953. 1.13.05: Equilibrium- Eutectics A given homogeneous binary liquid system (at pressure $\mathrm{p}$) contains two chemical substances $\mathrm{i}$ and $\mathrm{j}$ at temperature $\mathrm{T}$. The liquid system is cooled and only substance $\mathrm{j}$ separates out as the pure solid substance $\mathrm{j}$ leaving the liquid richer in chemical substance $\mathrm{i}$. The mole fraction composition of the liquid is given by the Schroeder-van Laar equation written in the following form. $-\ln \left[\mathrm{x}_{\mathrm{j}}(\ell) \mathrm{f}_{\mathrm{j}}(\ell)\right]=\frac{\left[\Delta_{\text {fus }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\text {fus; } \mathrm{j}}^{0}}\right] \label{a}$ In many cases as the mole fraction composition of substance $\mathrm{i}$ in the liquid increases the equilibrium temperature $\mathrm{T}$ decreases until at the eutectic temperature $\mathrm{T}_{\mathrm{e}}$ and mole fraction $\left(\mathbf{X}_{\mathrm{j}}\right)_{\mathrm{e}}$ the system comprises a solid, the eutectic mixture. In the event that the thermodynamic properties of the system can be described as ideal, Equation \ref{a} simplifies to Equation \ref{b} where it is assumed that $\mathrm{f}_{\mathrm{j}}(\ell)$ is unity at all temperatures. Then $-\ln \left[\mathrm{x}_{\mathrm{j}}(\ell)\right]=\frac{\left[\Delta_{\mathrm{fus}} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\text {fus; } ; \mathrm{j}}^{0}}\right] \label{b}$ For the other component $\mathrm{i}$, a corresponding plot is obtained when on cooling the liquid mixture only pure solid $\mathrm{i}$ separates out. $-\ln \left[\mathrm{x}_{\mathrm{i}}(\ell) \mathrm{f}_{\mathrm{i}}(\ell)\right]=\frac{\left[\Delta_{\mathrm{fus}} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\text {fus }, \mathrm{i}}^{0}}\right]$ If the thermodynamic properties of the system are ideal then the analogue of equation (b) is equation (d). $-\ln \left[\mathrm{x}_{\mathrm{i}}(\ell)\right]=\frac{\left[\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\text {fuss } ; \mathrm{i}}^{0}}\right]$ The two curves described by equations (a) and (c) meet at the eutectic temperature. $\mathrm{T}_{\mathrm{e}}$. Granted that the thermodynamic properties of the system are ideal, the following two equations follow from equations (b) and (d). $-\ln \left[\mathrm{x}_{\mathrm{j}}^{\mathrm{e}}(\ell)\right]=\frac{\left[\Delta_{\mathrm{fus}} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}_{\mathrm{e}}}-\frac{1}{\mathrm{~T}_{\text {fus } ; j}^{0}}\right]$ $-\ln \left[\mathrm{x}_{\mathrm{i}}^{\mathrm{e}}(\ell)\right]=-\ln \left[1-\mathrm{x}_{\mathrm{j}}^{\mathrm{e}}\right]=\frac{\left[\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}_{\mathrm{e}}}-\frac{1}{\mathrm{~T}_{\text {fuss; } \mathrm{i}}^{0}}\right]$ In the event that $\frac{\left[\Delta_{\text {fus }} \mathrm{H}_{\mathrm{j}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}_{\mathrm{e}}}-\frac{1}{\mathrm{~T}_{\text {fuss } ; \mathrm{j}}^{0}}\right]=\frac{\left[\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T})\right]}{\mathrm{R}} \left[\frac{1}{\mathrm{~T}_{\mathrm{e}}}-\frac{1}{\mathrm{~T}_{\text {fuss } ; \mathrm{i}}^{0}}\right]$ then $x_{i}^{e}=x_{j}^{e}=0.5$. The impact of the non-ideal thermodynamic properties can be explored using equation (a) and (c) in conjunction with empirical equations relating, for example, $\mathrm{f}_{\mathrm{j}}(\ell)$ and $\mathrm{x}_{\mathrm{j}}(\ell)$; e.g. equation (g). $\ln \left[\mathrm{f}_{\mathrm{j}}(\ell)\right]=\alpha \left[1-\mathrm{x}_{\mathrm{j}}(\ell)\right]^{2}$ [1] I. Prigogine and R. Defay, Chemical Thermodynamics, tranls. D. H. Everett, Longmans Greeen, London, 1953. 1.13.06: Equilibrium- Depression of Freezing Point of a Solvent by a Solute A given homogeneous liquid system (at pressure $\mathrm{p}$) comprises solvent $\mathrm{i}$ and solute $\mathrm{j}$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. In the absence of solute $\mathrm{j}$, the freezing point of the solvent is $\mathrm{T}_{1}^{0}$. But in the presence of solute $\mathrm{j}$ the freezing point is temperature $\mathrm{T}$ where $\mathrm{T} < \mathrm{T}_{1}^{0}$. The depression of freezing point $\theta\left[=\mathrm{T}_{1}^{0}-\mathrm{T}\right]$ is recorded for a solution where the mole fraction of solvent is $\mathrm{x}_{1}(\mathrm{sln})$. If the solution is dilute, we can assume that the thermodynamic properties of the solution are ideal. From the Schroeder-van Laar equation, $-\ln \left[x_{1}(s \ln )\right]=\frac{\left[\Delta_{f} H_{1}^{0}(T)\right]}{R} \,\left[\frac{1}{T}-\frac{1}{T_{1}^{0}}\right]$ $-\ln \left[\mathrm{x}_{1}(\mathrm{~s} \ln )\right]=\frac{\Delta_{\mathrm{f}} \mathrm{H}_{1}^{0}}{\mathrm{R}} \, \frac{\theta}{\left(\mathrm{T}_{1}^{0}-\theta\right) \, \mathrm{T}_{1}^{0}}$ If $\mathrm{T}_{1}^{0}-\theta \cong \mathrm{T}_{1}^{0},-\ln \left[\mathrm{x}_{1}(\mathrm{~s} \ln )\right]=\frac{\Delta_{\mathrm{f}} \mathrm{H}_{1}^{0}}{\mathrm{R}} \, \frac{\theta}{\left(\mathrm{T}_{1}^{0}\right)^{2}}$ Or, $\ln \left[\frac{1}{x_{1}(s \ln )}\right]=\frac{\Delta_{\mathrm{f}} H_{1}^{0}}{R} \, \frac{\theta}{\left(T_{1}^{0}\right)^{2}}$ Hence [2] $\theta=\left[\frac{\mathrm{R} \,\left(\mathrm{T}_{1}^{0}\right)^{2} \, \mathrm{M}_{1}}{\Delta_{\mathrm{f}} \mathrm{H}_{1}^{0}}\right] \, \mathrm{m}_{\mathrm{j}}$ The quantity enclosed in the […] brackets is characteristic of the solvent. Footnotes [1] $\theta=\mathrm{T}_{1}^{0}-\mathrm{T}$; $\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{1}^{0}}=\frac{\mathrm{T}_{1}^{0}-\mathrm{T}}{\mathrm{T} \, \mathrm{T}_{1}^{0}}=\frac{\mathrm{T}_{1}^{0}-\mathrm{T}}{\left(\mathrm{T}_{1}^{0}-\theta\right) \, \mathrm{T}_{1}^{0}}=\frac{\theta}{\left(\mathrm{T}_{1}^{0}-\theta\right) \, \mathrm{T}_{1}^{0}}$ [2] $\frac{1}{x_{1}}=\frac{1}{1-x_{j}}=\frac{1}{1-\left[n_{j} /\left(n_{1}+n_{j}\right)\right]}=\frac{n_{1}+n_{j}}{n_{1}+n_{j}-n_{j}}$ For a solution where the molality of solute $j=m_{j}$ $\mathrm{m}_{\mathrm{j}}=\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1} \, \mathrm{M}_{1}}$ Then, $\frac{1}{\mathrm{x}_{1}}=\frac{\mathrm{n}_{1}+\mathrm{n}_{1} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}}{\mathrm{n}_{1}}$ $-\ln \left[\mathrm{x}_{1}(\mathrm{~s} \ln )\right]=-\ln \left[1+\mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right]$; $-\ln \left[x_{1}(\operatorname{sln})\right]=-\ln \left[1-x_{j}(s \ln )\right] \approx x_{j}$ $x_{j}=\frac{m_{j}}{\left(1 / M_{1}\right)+m_{j}} \approx m_{j} \, M_{1}$ [3] see I Prigogine and R Defay, Chemical Thermodynamics, trans. D. H. Everett, Longmans Green, London, 1953. 1.13.07: Equilibrium- Liquid-Solids- Hildebrand Rules A given homogeneous liquid system (at pressure $\mathrm{p}$) contains two chemical substances $\mathrm{i}$ and $\mathrm{j}$ at temperature $\mathrm{T}$. Chemical substance $\mathrm{j}$ at temperature $\mathrm{T}$ and $\mathrm{p}$ is a liquid which being in vast excess in this system is the solvent. The system is cooled and pure solid substance $\mathrm{i}$ separates out leaving the system less concentrated in the solute $\mathrm{i}$. The solution is dilute and we assume that the thermodynamic properties of the solution are ideal. Then, from the Schroeder–van Laar Equation $-\ln \left[\mathrm{x}_{\mathrm{i}}(\mathrm{s} \ln )\right]=\frac{\left[\Delta_{\mathrm{fius}} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R}} \,\left[\frac{1}{\mathrm{~T}}-\frac{1}{\mathrm{~T}_{\mathrm{i}}^{0}}\right] \label{a}$ $\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}$ is the molar enthalpy of fusion of chemical substance $\mathrm{i}$, melting point $\mathrm{T}_{\mathrm{i}}^{0}$. Mole fraction $\mathrm{x}_{\mathrm{i}}(\mathrm{s} \ln )$ is the composition of the saturated solution at temperature $\mathrm{T}$; i.e. the solubility of substance $\mathrm{i}$. From Equation \ref{a}, $\ln \left[\frac{1}{\mathrm{x}_{\mathrm{i}}(\mathrm{s} \ln )}\right]^{\mathrm{eq}}=\frac{\left[\Delta_{\mathrm{fus}} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T}, \mathrm{p})\right]}{\mathrm{R}} \,\left[\frac{\mathrm{T}_{\mathrm{i}}^{0}-\mathrm{T}}{\mathrm{T} \, \mathrm{T}_{\mathrm{i}}^{0}}\right] \label{b}$ Equation \ref{b} forms the background to several generalisations concerning solubilities; i.e. Hildebrand Rules [1]. We note that $\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T}, \mathrm{p})$ and $\mathrm{T}_{\mathrm{i}}^{0}$ characterise the solute. 1. Solubilities increase with increase in temperature. 2. For two solutes with equal $\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T}, \mathrm{p})$, the solute with lower $\mathrm{T}_{\mathrm{i}}^{0}$ will be more soluble at a common temperature $\mathrm{T}$. 3. For two solutes with the same $\mathrm{T}_{\mathrm{i}}^{0}$, the solid with lower $\Delta_{\text {fus }} \mathrm{H}_{\mathrm{i}}^{0}(\mathrm{~T}, \mathrm{p})$ will be more soluble. Footnote [1] see I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Greeen, London, 1953.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.13%3A_Equilibrium/1.13.03%3A_Equilibirium-_Solid-Liquid.txt
An enormous chemical literature describes the effects of solvents on rates of chemical reactions. C. K. Ingold [1] in his classic monograph actually uses the phrase 'solvent polarity' when commenting on the relative rates of reactions through a series of solvents of diminishing "polarity". One of the reactions discussed by Ingold concerns the solvolysis of 2-chloro-2-methyl -propane, $\left(\mathrm{CH}_{3}\right)_{3}\mathrm{CCl}$. In 1948, Winstein and Grunwald [2] used this reaction as a basis for a quantitative treatment of solvent polarities leading to the definition of solvent Y-value. The basis of their analysis can be understood using an extrathermodynamic analysis [3]. We use a description based on substituent zone $\mathrm{R}$ and reaction zone $\mathrm{X}$ for a solute molecule $\mathrm{RX}$. Here the interaction between the two zones is solvent dependent. The starting point is kinetic data describing an (assumed) unimolecular first order solvolysis of a solute $\mathrm{RX}$. The chemical reaction proceeds through a transition state $\mathrm{RX}^{neq}$. For a given solvent medium $\mathrm{M}$ at defined $\mathrm{T}$ and $\mathrm{p}$, transition state theory [9] describes the standard activation Gibbs energy as follows. $\Delta^{\neq} \mathrm{G}^{0}(\mathrm{RX} ; \mathrm{M})=\mu^{0}\left(\mathrm{RX} \mathrm{X}^{\neq} ; \mathrm{M}\right)-\mu^{0}(\mathrm{RX} ; \mathrm{M})$ The basic postulate states that the reference chemical potential of solute $\mathrm{RX}$ in solution, $\mu^{0}(\mathrm{RX} ; \mathrm{sln})$ at defined $\mathrm{T}$ and $\mathrm{p}$ is given by the sum of contributions from the substituent zone, $\mu^{0} (\mathrm{R})$ and reaction zone , $\mu^{0} (\mathrm{X})$ together with terms describing the interaction of $\mathrm{R}$ and $\mathrm{X}$ with the solvent, $\mathrm{I}(\mathrm{R}, \mathrm{M})$ and $\mathrm{I}(\mathrm{X}, \mathrm{M})$ and the effect of solvent on this interaction $\mathrm{II}(\mathrm{R}, \mathrm{X}, \mathrm{M})$. \begin{aligned} &\mu^{0}(\mathrm{RX} ; \text { in medium } \mathrm{M})= \ &\qquad \mu^{0}(\mathrm{R})+\mu^{0}(\mathrm{X})+\mathrm{I}(\mathrm{R}, \mathrm{M})+\mathrm{I}(\mathrm{X}, \mathrm{M})+\mathrm{II}(\mathrm{R}, \mathrm{X}, \mathrm{M}) \end{aligned} Thus the solvent $\mathrm{M}$ contributes to the interaction between $\mathrm{R}$ and $\mathrm{X}$. A key postulate is advanced at this stage which states that the interactions terms can be factorised. \begin{aligned} &\mu^{0}(\mathrm{X} ; \text { in medium } \mathrm{M})= \ &\mu^{0}(\mathrm{R})+\mu^{0}(\mathrm{X})+\mathrm{I}(\mathrm{R}) \, \mathrm{I}(\mathrm{M})+\mathrm{I}(\mathrm{X}) \, \mathrm{I}(\mathrm{M})+\mathrm{II}(\mathrm{R}) \, \mathrm{II}(\mathrm{X}) \, \mathrm{II}(\mathrm{M}) \end{aligned} A similar equation is set down for the transition state. \begin{aligned} &\mu^{0}\left(\mathrm{RX}^{\neq} ; \text {in medium } \mathrm{M}\right)= \ &\mu^{0}\left(\mathrm{R}^{\neq}\right)+\mu^{0}\left(\mathrm{X}^{\neq}\right)+\mathrm{I}\left(\mathrm{R}^{\neq}\right) \, \mathrm{I}(\mathrm{M})+\mathrm{I}\left(\mathrm{X}^{\neq}\right) \, \mathrm{I}(\mathrm{M})+\mathrm{II}\left(\mathrm{R}^{\neq}\right) \, \mathrm{II}\left(\mathrm{X}^{\neq}\right) \, \mathrm{II}(\mathrm{M}) \end{aligned} Equations (c) and (d) are combined with equation (a). For reaction in solvent medium $\mathrm{M}$, \begin{aligned} &\Delta^{\neq} \mathrm{G}^{0}(\mathrm{RX}, \mathrm{M})=\ &\left.\left[\mu^{0}\left(R^{\neq}\right)-\mu^{0}(R)\right]+\left[\mu^{0}\left(X^{\neq}\right)-\mu^{0} X\right)\right]\ &+\mathrm{I}(\mathrm{M}) \,\left[\mathrm{I}\left(\mathrm{R}^{\neq}\right)-\mathrm{I}(\mathrm{R})\right]+\mathrm{I}(\mathrm{M}) \,\left[\mathrm{I}\left(\mathrm{X}^{\neq}\right)-\mathrm{I}(\mathrm{X})\right]\ &+\mathrm{II}(\mathrm{M}) \,\left[\mathrm{II}\left(\mathrm{R}^{*}\right) \, \mathrm{II}\left(\mathrm{X}^{\neq}\right)-\mathrm{II}(\mathrm{R}) \, \mathrm{II}(\mathrm{X})\right] \end{aligned} A second postulate state that $\mathrm{II}(\mathrm{M})$ and $\mathrm{I}(\mathrm{M})$ are simply related; i.e. equation (f). $\mathrm{II}(\mathrm{M})=\alpha \, \mathrm{I}(\mathrm{M})$ If $\Delta_{m} \Delta^{\neq} G^{0}(\mathrm{RX})$ describes the effect of solvent $\mathrm{M}$ on $\Delta^{\neq} \mathrm{G}^{0}(\mathrm{RX})$, $\Delta_{\mathrm{m}} \Delta^{\neq} \mathrm{G}^{0}(\mathrm{RX})$ is given by the product of a (solvent operator) and a (substrate operator). By definition, $\Delta_{\mathrm{m}} \Delta^{\pm} \mathrm{G}^{0}(\mathrm{RX})=\Delta_{\mathrm{m}} \mathrm{Y} \,(\text { substrate operator })$ Originally the substrate operator was set to unity for $\left(\mathrm{CH}_{3}\right)_{3}\mathrm{CCl}$, and $\mathrm{Y}$ was set to zero for an 80:20 ethanol + water mixture [4]. The outcome was a set of $\mathrm{Y}$-values for many solvents, particularly alcohol + water mixtures at $298.5 \mathrm{~K}$ and ambient pressure Footnotes [1] C. K. Ingold, Structure and Mechanism in Organic Chemistry, G. Bell, London, 1953; see page 347. [2] E. Grunwald and S. Winstein, J. Am. Chem. Soc., 1948,70, 841; 846. [3] J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions, Wiley, New York, 1963; Dover Publications , New York,1989. [4] $20 \mathrm{~cm}^{3}$ of ethanol($\ell$) was poured from a volumetric flask containing $1 \mathrm{~dm}^{3}$ of ethanol($\ell$). The liquid in the flask was then ‘topped up’ with water($\ell$). The mixture is a good solvent for both apolar and polar solutes. Unfortunately the exact composition of the mixture is unknown. As rarely stated, the volume of water required is slightly larger than $20 \mathrm{~cm}^{3}$. Professor Ross E Robertson (University of Calgary) viewed with interest that so much information in the chemical literature describes rates of chemical reactions where the solvent is vodka.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.10:_Extrathermodynamics_-_Solvent_Effects_in_Chemical_Ki.txt
A given liquid mixture at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ ($\cong \mathrm{p}^{0}$) contains $\mathrm{i}$-liquid chemical substances. The chemical potential of liquid component $\mathrm{j}$ is given by equation (a) where $\mu_{\mathrm{j}}^{*}(\ell)$ is the chemical potential of liquid component $\mathrm{j}$ at the same $\mathrm{T}$ and $\mathrm{p}$. $\mu_{\mathrm{j}}(\operatorname{mix})=\mu_{\mathrm{j}}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{\mathrm{j}} \, \mathrm{f}_{\mathrm{j}}\right)$ Here $\operatorname{limit}\left(\mathrm{x}_{\mathrm{j}} \rightarrow 1\right) \mathrm{f}_{\mathrm{j}}=1.0$ at all $\mathrm{T}$ and $\mathrm{p}$. If the thermodynamic properties of the liquid mixture are ideal, $\mu_{j}(\operatorname{mix} ; \mathrm{id})=\mu_{\mathrm{j}}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{\mathrm{j}}\right)$ The excess chemical potential for liquid substance $\mathrm{j}$, $\mu_{\mathrm{j}}^{\mathrm{E}}(\mathrm{mix})=\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{f}_{\mathrm{j}}\right)$ DSF 1.14.3: Excess Thermodynamic Properties- Aqueous Solutions A given aqueous solution, at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ ($\cong \mathrm{p}^{0}$), contains $\mathrm{i}$-solutes, with $\mathrm{n}_{\mathrm{j}}$ moles of each solute $\mathrm{j}$, and $\mathrm{n}_{1}$ moles of water($\ell$).The Gibbs energy of the solution is given by equation (a). $\mathrm{G}(\mathrm{aq})=\mathrm{n}_{1} \, \mu_{1}(\mathrm{aq})+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}(\mathrm{aq})$ For a solution prepared using $1 \mathrm{~kg}$ of water($\ell$), in vast molar excess, $\mathrm{G}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mu_{1}(\mathrm{aq})+\sum_{\mathrm{j}=1}^{\mathrm{s}} \mathrm{m}_{\mathrm{j}} \, \mu_{\mathrm{j}}(\mathrm{aq})$ We assert that the system is at thermodynamic equilibrium. For each solute $\mathrm{j}$, $\mu_{j}(\mathrm{aq})$ is related to the molality $\mathrm{m}_{\mathrm{j}}$ and the reference chemical potential for solute $\mathrm{j}$ in a solution where $\mathrm{m}_{\mathrm{j}} = 1 \mathrm{~mol kg}^{-1}$ and the thermodynamic properties of the solute are ideal. Then, $\left\{\mathrm{m}^{0}=1 \mathrm{~mol} \mathrm{~kg}^{-1}\right\} \quad \mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)$ where $\operatorname{limit}\left(m_{j} \rightarrow 0\right) \gamma_{j}=1.0$ at all $\mathrm{T}$ and $\mathrm{p}$. For the solvent we express the properties in terms of a practical osmotic coefficient, $\phi$. $\mu_{1}(\mathrm{aq})=\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}}$ At all $\mathrm{T}$ and $\mathrm{p}$, $\operatorname{limit}\left(\sum_{j=1}^{j=i} m_{j} \rightarrow 0\right) \phi=1.0$ For the solution, \begin{aligned} \mathrm{G}\left(\mathrm{aq} ; \mathrm{w}_{1}=\right.&1 \mathrm{~kg})=\left(1 / \mathrm{M}_{1}\right) \,\left[\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}}\right] \ &+\sum_{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}} \,\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)\right] \end{aligned} If the thermodynamic properties of the solution are ideal, \begin{aligned} \mathrm{G}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=\right.&1 \mathrm{~kg})=\left(1 / \mathrm{M}_{1}\right) \,\left[\mu_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}}\right] \ &+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}} \,\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)\right] \end{aligned} By definition the solution excess Gibbs energy of the solution, $\mathrm{G}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{G}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\mathrm{G}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$ $\mathrm{G}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right) \text { is expressed in }\left[\mathrm{J} \mathrm{kg}^{-1}\right]$. Then $\mathrm{G}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \,(1-\phi) \, \sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\gamma_{\mathrm{j}}\right)$ $\mathrm{G}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right) / \mathrm{R} \, \mathrm{T}=(1-\phi) \, \sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}} \, \ln \left(\gamma_{\mathrm{j}}\right)$ For a solution containing a single solute $\mathrm{j}$, $\mathrm{G}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right) / \mathrm{R} \, \mathrm{T}=\left[1-\phi+\ln \left(\gamma_{\mathrm{j}}\right)\right] \, \mathrm{m}_{\mathrm{j}}$ If the thermodynamic properties of the solution are ideal, the chemical potential of the solute is given by equation (k). $\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{id})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Equation (c) describes the properties of solute $\mathrm{j}$ in a real solution. By definition the excess chemical potential $\mu_{\mathrm{j}}^{\mathrm{E}}(\mathrm{aq})$ is given by equation (l). $\mu_{\mathrm{j}}^{\mathrm{E}}(\mathrm{aq})=\mu_{\mathrm{j}}(\mathrm{aq})-\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{id})$ Then, $\mu_{\mathrm{j}}^{\mathrm{E}}(\mathrm{aq})=\mathrm{R} \, \mathrm{T} \, \ln \left(\gamma_{\mathrm{j}}\right)$ Often an excess chemical potential $\mu_{\mathrm{j}}^{\mathrm{E}}(\mathrm{aq})$ is written in the form $\mathrm{G}_{\mathrm{j}}^{\mathrm{E}}$. In the case of the solvent, water($\ell$) the corresponding equations for the chemical potentials in solutions having either real or ideal thermodynamic properties are given by equations (n) and (o). $\mu_{1}(\mathrm{aq} ; \mathrm{id})=\mu_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ $\mu_{1}(\mathrm{aq})=\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ $\mu_{1}^{\mathrm{E}}(\mathrm{aq})=(1-\phi) \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ Footnotes [1] For further comments see— 1. M. I. Davis and G. Douheret, Thermochim. Acta, 1991,190,267. 2. H. L. Friedman, J. Chem.Phys.,1969,32,1351.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.2:_Excess_Thermodynamic_Properties_-_Liquid_Mixtures.txt
The terms, variables and properties are synonymous. Nevertheless a given thermodynamic property of a system can be classified as either intensive or extensive. Intensive Properties. The magnitude of an intensive variable does NOT depend on the amount of chemical substance in a given closed system; e.g. density. Extensive Properties. The magnitude of an extensive variable depends on the amount of chemical substances in a closed system; e.g. volume. Let us ask – is temperature an intensive or extensive variable? Consider two conical flasks. Flask A contains $10 \mathrm{~cm}^{3}$ of water($\ell$) at $298 \mathrm{~K}$. Flask B contains $5 \mathrm{~cm}^{3}$ of water($\ell$) at $298 \mathrm{~K}$. The contents of Flask are poured into Flask B. 1. What is the volume of liquid in flask B? The answer is clearly $15 \mathrm{~cm}^{3}$. 2. What is the temperature of the liquid in flask B? Based on the answer to the previous question, we might answer $596 \mathrm{~K}$, being the sum $298 + 298$. This is clearly wrong. We have not distinguished between extensive variable variable and the intensive variable temperature. Temperature is an intensive variable as, for example, is the density of liquids. 3. A quick test to decide whether a given variable is either extensive or intensive is to ask what happens to the number value if the amount of chemical substance in a system increases by a factor of two. If the variable (e.g. volume) also increases by a factor of two, the variable is extensive. If the variable ( e.g. temperature, equilibrium constant…) remains unchanged , the variable is intensive. Otherwise the variable is neither extensive nor intensive (e.g. the inverse of volume). Footnote [1] O. Redlich ( J. Chem.Educ.,1970,42,154) presents a provocative discussion of the distinction between intensive and extensive variables. 1.14.5: Extent of Reaction For chemists, chemical reaction is the key thermodynamic process. By definition chemical reaction produces a change in composition of a closed system. The extent of chemical reaction is measured by a quantity $\mathrm{d}\xi$, where the chemical composition is described by the symbol $\xi$. An example makes the point. An aqueous solution is prepared at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ contains solute $\mathrm{X}$. The latter undergoes spontaneous chemical reaction to form chemical substance $\mathrm{Y}$. Thus $\mathrm{X}(\mathrm{aq})$ $\rightarrow$ $\mathrm{Y}(\mathrm{aq})$ At $t = 0$ $\mathrm{n}_{\mathrm{X}}^{0}$   $0 \mathrm{~mol}$ At time $t$, $\mathrm{n}_{\mathrm{X}}^{0}-\xi$   $\xi \mathrm{~mol}$ Rate of reaction $= \mathrm{d}\xi / \mathrm{dt}$ [Time is a legitimate thermodynamic property.] A key concept states that spontaneous chemical reaction is driven by the affinity for spontaneous change, $\mathrm{A}$. Then by definition equilibrium corresponds to the state where $\mathrm{A} = 0$, and $\mathrm{d}\xi / \mathrm{dt} = 0$. General Terms For a system containing $\mathrm{i}$-chemical substances, the chemical potential of chemical substance $\mathrm{j}$ is given by equation (a). $\mu_{\mathrm{j}}=\left(\frac{\partial \mathrm{G}}{\partial n_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})}$ Then, $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mu_{\mathrm{j}} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ But, $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}-\mathrm{A} \, \mathrm{d} \xi$ By comparison, $A \, d \xi=-\sum_{j=1}^{j=i} \mu_{j} \, d n_{j}$ But $\mathrm{dn}_{\mathrm{j}}=\mathrm{v}_{\mathrm{j}} \, \mathrm{d} \xi$ where $\mathrm{ν}_{\mathrm{j}}$ is positive for products and negative for reactants. Hence, $A=-\sum_{j=1}^{j=i} v_{j} \, \mu_{j}$ This remarkable equation relates the affinity for chemical reaction $A$ with the chemical potentials of the chemical substances involved in the chemical reaction. Moreover at equilibrium, $A$ is zero. Hence, $\sum_{j=1}^{j=i} v_{j} \, \mu_{j}^{e q}=0$ We have a condition describing chemical equilibrium in terms of the chemical potentials of reactants and products at equilibrium.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.4:_Extensive_and_Intensive_Variables.txt
The variable $\xi$ describes in quite general terms the molecular composition/organisation. For a closed system at fixed $\mathrm{T}$ and $\mathrm{p}$, there is a composition/organisation $\xi^{\mathrm{eq}}$ corresponding to a minimum in Gibbs energy where the affinity for spontaneous change is zero. In general terms there is an extent of reaction $\xi$ corresponding to a given affinity $\mathrm{A}$ at defined $\mathrm{T}$ and $\mathrm{p}$. In fact we can express $\xi$ as a dependent variable defined by the independent variables $\mathrm{T}$, $\mathrm{p}$, and $\mathrm{A}$. Thus $\xi=\xi[\mathrm{T}, \mathrm{p}, \mathrm{A}]$ The general differential takes the following form. $\mathrm{d} \xi=\left(\frac{\partial \xi}{\partial T}\right)_{\mathrm{p}, \mathrm{A}} \, \mathrm{dT}+\left(\frac{\partial \xi}{\partial \mathrm{p}}\right)_{\mathrm{p}, \mathrm{A}} \, \mathrm{dp}+\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}, \mathrm{p}} \, \mathrm{dA}$ The change in chemical composition occurs spontaneously. The change in composition is described in terms of the extent of chemical reaction, $\xi$. In a given aqueous solution, the chemical reaction is: $\mathrm{CH}_{3} \mathrm{COOC}_{2} \mathrm{H}_{5}+\mathrm{OH}^{-}(\mathrm{aq}) \rightarrow \mathrm{CH}_{3} \mathrm{COO}^{-}(\mathrm{aq})+\mathrm{C}_{2} \mathrm{H}_{5} \mathrm{OH}(\mathrm{aq})$ At each stage the extent of chemical reaction is represented by the symbol $\xi$ [1]. $\mathrm{CH}_{3} \mathrm{COOC}_{2} \mathrm{H}_{5}(\mathrm{aq})+\mathrm{OH}^{-}(\mathrm{aq}) \rightarrow \mathrm{CH}_{3} \mathrm{COO}^{-}(\mathrm{aq})+\mathrm{C}_{2} \mathrm{H}_{5} \mathrm{OH}(\mathrm{aq})$ At $t = 0$ (i.e. as prepared) $n(\text { ester })^{0} \quad n\left(\mathrm{OH}^{-}\right)^{0} \quad 0 \quad 0$ After extent of reaction $\xi$ (at some time later) $n(\text { ester })^{0} - \xi \quad n\left(\mathrm{OH}^{-}\right)^{0} -\xi \quad \xi \quad \xi$ As the reaction proceeds so $\xi$ increases. [NB The zero superscript signals ‘at time zero’.] At a given stage of the reaction and time $t$, (accepting $\mathrm{dt}$ is positive), Rate of Reaction = $\mathrm{d}\xi / \mathrm{dt}$ We now ask ‘why did chemical reaction proceed in this direction?’. The answer is ---- the chemical reaction was driven by the affinity for spontaneous change, symbol $\mathrm{A}$. By identifying these two ideas, affinity for spontaneous change and the rate of reaction $\mathrm{d}\xi / \mathrm{dt}$, we arrive at two important criteria for chemical equilibrium. Affinity for spontaneous change $\mathrm{A} = 0$ Rate of change $\mathrm{d}\xi / \mathrm{dt} = 0$ However we need to stand back a little and examine how we might advance generalizations concerning the direction of Spontaneous Chemical Reaction. What macroscopic property can be identified which accounts for the fact that alkaline hydrolysis of ethyl ethanoate is spontaneous? To make further progress we introduce two laws of thermodynamics. Actually these are not laws in the sense of being laid down by government or by religious doctrine. Rather these laws are AXIOMS. We explore these axioms in the context for which ξ refers to a change in composition resulting from chemical reaction [2]. Footnotes [1] For a discussion of the significance of extent of reaction $\xi$, see: 1. K. J. Laidler and N. Kallay, Kem. Ind. (Sofia) 1988, 37, 182. 2. F. R. Cruikshank, A. J. Hyde and D. Pugh, J.Chem.Educ., 1977, 54, 88. 3. P. G. Wright, Educ. Chem., 1986,23, 111. 4. M. J. Blandamer, Educ. in Chem.,1999,36,78. [2] The usefulness of the concept of extent of chemical reaction $\xi$ is further illustrated by the following examples. 1. A closed system (at fixed temperature $\mathrm{T}$ and pressure $\mathrm{p}$) is prepared using $\mathrm{n}_{\mathrm{x}}^{0}$ moles of chemical substance $\mathrm{X}$ in $\mathrm{w}_{1} \mathrm{~kg}$ of solvent, water. Spontaneous chemical reaction forms chemical substances $\mathrm{Y}$ and $\mathrm{Z}. Thus, \(2\mathrm{X}$ $\rightarrow 3\mathrm{Y}$ $+ \mathrm{~Z}$ At $t = 0$ $\mathbf{n}_{X}^{0}$ $0$ $0 \mathrm{~mol}$ After extent of reaction, $\xi$ with $\mathbf{n}_{X}^{0}$ $\mathrm{n}_{\mathrm{x}}^{0}-2 . \xi$ $3 . \xi$ $2 . \xi \mathrm{~mol}$ 2. If the chemical reaction in (A) proceeds to completion, $2 \mathrm{X}$ $\rightarrow 3 \mathrm{Y}$ $+ \mathrm{~Z}$ $0$ $3 \, n_{x}^{0} / 2$ $\mathrm{n}_{\mathrm{x}}^{0} / 2 \mathrm{~mol}$ 3. If the chemical reaction in examp1e (A) proceeds to chemical equilibrium, then with $\xi = \xi^{\mathrm{eq}$, $2\mathrm{X }\rightleftarrows$ $3\mathrm{Y } +$ $\mathrm{Z}$ Amounts $\mathrm{n}_{\mathrm{x}}^{0}-2 \, \xi^{\mathrm{eq}}$ $\mathrm{n}_{\mathrm{x}}^{0}-2 \, \xi^{\mathrm{eq}}$ $3 \, \xi^{\mathrm{eq}} \mathrm{~mol}$ Molalities $\left(n_{x}^{0}-2 \, \xi^{e q}\right) / w_{1}$ $3 \, \xi^{\mathrm{eq}} / \mathrm{w}_{1}$ $\xi^{e q} / w_{1} \mathrm{~mol kg}^{-1}$ 1.14.7: Extrathermodynamics - Background Essentially thermodynamics is used to analyze experimental data. In these terms, thermodynamics shows how properties of systems are related and how one can link measured properties with important thermodynamic variables. Nevertheless, there are cases where a pattern seems to emerge from measured variables which is not a consequence of the laws of thermodynamics. Furthermore, it is often discovered that the patterns can actually be accounted for if one or two additional postulates are made. These new postulates are therefore extra-thermodynamic and the analytical method is called extrathermodynamics [1,2]. The analysis has merit in that the new postulates point to patterns which can be developed for other systems. The essence of the argument can be understood by considering the molar volume of pure ethanol at ambient pressure and $298.2 \mathrm{~K}$. Clearly $\mathrm{V}^{*}\left(\ce{C2H5OH} ; \ell ; 298.2 \mathrm{~K} ; \left.101325 \mathrm{~N} \mathrm{~m}^{-2}\right)$ is a properly defined thermodynamic variable. But as chemists we might be tempted to explore an extrathermodynamic postulate in which $\mathrm{V}^{*}\left(\mathrm{C}_{2} \mathrm{H}_{5} \mathrm{OH} ; \ell\right)$ can be subdivided into group contributions. Thus $\mathrm{V}^{*}\left(\mathrm{C}_{2} \mathrm{H}_{5} \mathrm{OH} ; \ell\right)=\mathrm{V}\left(\mathrm{CH}_{3}\right)+\mathrm{V}\left(\mathrm{CH}_{2}\right)+\mathrm{V}(\mathrm{OH}) \label{a}$ This equation cannot be justified on thermodynamic grounds. Nevertheless we might examine molar volumes of several (liquid) alcohols at the same $\mathrm{T}$ and $\mathrm{p}$ and come up with a self-consistent set of group volumes. For example, $\left[\mathrm{V}^{*}\left(\mathrm{n}-\mathrm{C}_{3} \mathrm{H}_{7} \mathrm{OH} ; \ell\right)\right]=\mathrm{V}\left(\mathrm{CH}_{3}\right)+2 * \mathrm{~V}\left(\mathrm{CH}_{2}\right)+\mathrm{V}(\mathrm{OH}) \label{b}$ Hence comparison of Equations \ref{a} and \ref{b} yields directly $\mathrm{V}\left(\mathrm{CH}_{2}\right)$, the contribution of methylene groups to the molar volume of (liquid) alcohols at the same $\mathrm{T}$ and $\mathrm{p}$. Although such an analysis might be judged naïve, the general approach finds merit in several subject areas; e.g chemical equilibria and chemical kinetics. Footnotes [1] J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions, Wiley, London, 1963. [2] E. Grunwald, Thermodynamics of Molecular Species, Wiley, New York, 1997.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.6:_Extent_of_Reaction_-_General.txt
In aqueous solution at ambient pressure and $298.15 \mathrm{~K}$, benzoic acid exists in the form of a chemical equilibrium described in equation (a) $\mathrm{PhCOOH}(\mathrm{aq}) \Leftrightarrow=\mathrm{H}^{+}(\mathrm{aq})+\mathrm{PhCOO}^{-}(\mathrm{aq})$ At defined $\mathrm{T}$ and $\mathrm{p}$, $\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{PhCOOH} ; \mathrm{aq})=\mu^{0}\left(\mathrm{PhCOO}^{-} ; \mathrm{aq}\right)+\mu^{0}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)-\mu^{0}(\mathrm{PhCOOH} ; \mathrm{aq})$ In the case of a substituted benzoic acid, $\mathrm{XC}_{6} \mathrm{H}_{4} \mathrm{COOH} \quad[=\mathrm{XPhCOOH}]$, the corresponding description of the chemical equilibrium takes the following form. \begin{aligned} &\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{XPhCOOH} ; \mathrm{aq}) \ &=\mu^{0}(\mathrm{XPhCOO} ; ; \mathrm{aq})+\mu^{0}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)-\mu^{0}(\mathrm{XPhCOOH} ; \mathrm{aq}) \end{aligned} In aqueous solution at ambient pressure and $298.15 \mathrm{~K}$, the properties of an aqueous solution containing phenol can be described in terms of the following equilibrium. $\mathrm{PhOH}(\mathrm{aq}) \Leftrightarrow=\Longrightarrow \mathrm{H}^{+}(\mathrm{aq})+\mathrm{PhO}^{-}(\mathrm{aq})$ Then, (cf. equation (b)), $\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{PhOH} ; \mathrm{aq})=\mu^{0}\left(\mathrm{PhO}^{-} ; \mathrm{aq}\right)+\mu^{0}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)-\mu^{0}(\mathrm{PhOH} ; \mathrm{aq})$ In the case of a substituted phenol $\mathrm{XPhOH}$, the equation corresponding to equation (d) takes the following form. $\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{XPhOH} ; \mathrm{aq})=\mu^{0}\left(\mathrm{XPhO}^{-} ; \mathrm{aq}\right)+\mu^{0}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)-\mu^{0}(\mathrm{XPhOH} ; \mathrm{aq})$ In the following we compare situations where $\mathrm{X}$ is common to the substituted phenol and benzoic acid including position in the aromatic ring. The interesting point to emerge is that for a range of substituents, $\mathrm{X}$, the recorded dependence of $\Delta_{\Delta_{r}} \mathrm{G}^{0}(\mathrm{XPhOH} ; \mathrm{aq})$ on $\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{XPhCOOH} ; \mathrm{aq})$ is linear. Such a pattern is not a requirement of thermodynamics [1]. The challenge is to suggest a set of minimum relationships which account for this pattern [2,3]. Zone Model POSTULATE---Single Interaction Mechanism Consider the reference chemical potential for solute $\mathrm{RX}$ in aqueous solution at fixed $\mathrm{T}$ and $\mathrm{p}$, $\mu^{0}(\mathrm{RX} ; \mathrm{aq})$. As chemists we recognise that groups $\mathrm{R}$ and $\mathrm{X}$ do not make independent contributions to $\mu^{0}(\mathrm{RX} ; \mathrm{aq})$ [4]. The postulate, Single Interaction Mechanism, recognises that the groups $\mathrm{R}$ and $\mathrm{X}$ interact such that $\mu^{0}(\mathrm{RX} ; \mathrm{aq})$ is given by equation (g). $\mu^{0}(\mathrm{RX} ; \mathrm{aq})=\mu^{0}(\mathrm{R})+\mu^{0}(\mathrm{X})+\mathrm{I}(\mathrm{R}, \mathrm{X})$ Here symbol $\mathrm{R}$ identifies the substituent zone and $\mathrm{X}$ identifies the reaction zone so that $\mathrm{I}(\mathrm{R}, \mathrm{X})$ describes interaction between these two zones. Separability Postulate The interaction variable $\mathrm{I}(\mathrm{R}, \mathrm{X})$ is a function of scalar variables. Then $\mu^{0}(\mathrm{RX} ; \mathrm{aq})=\mu^{0}(\mathrm{R})+\mu^{0}(\mathrm{X})+\mathrm{I}(\mathrm{R}) \, \mathrm{I}(\mathrm{X})$ Hence for benzoic acid $\mathrm{PhCOOH}(\mathrm{aq})$, $\mu^{0}(\mathrm{PhCOOH} ; \mathrm{aq})=\mu^{0}(\mathrm{Ph})+\mu^{0}(\mathrm{COOH})+\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}(\mathrm{COOH})$ Similarly, $\mu^{0}(\mathrm{PhCOO} ; ; \mathrm{aq})=\mu^{0}(\mathrm{Ph})+\mu^{0}\left(\mathrm{COO}^{-}\right)+\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)$ Hence, $\begin{gathered} \Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{PhCOOH} ; \mathrm{aq})=\mu^{0}(\mathrm{Ph})+\mu^{0}\left(\mathrm{COO}^{-}\right)+\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)+\mu^{0}\left(\mathrm{H}^{+}\right) \ -\mu^{0}(\mathrm{Ph})-\mu^{0}(\mathrm{COOH})-\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}(\mathrm{COOH}) \end{gathered}$ Or, $\begin{gathered} \Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{PhCOOH} ; \mathrm{aq})=\mu^{0}\left(\mathrm{COO}^{-}\right)+\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)+\mu^{0}\left(\mathrm{H}^{+}\right) \ -\mu^{0}(\mathrm{COOH})-\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}(\mathrm{COOH}) \end{gathered}$ A similar equation emerges describing the acid dissociation of the substituted acid. Thus, $\begin{gathered} \Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{XPhCOOH} ; \mathrm{aq})=\mu^{0}\left(\mathrm{COO}^{-}\right)+\mathrm{I}(\mathrm{XPh}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)+\mu^{0}\left(\mathrm{H}^{+}\right) \ -\mu^{0}(\mathrm{COOH})-\mathrm{I}(\mathrm{XPh}) \, \mathrm{I}(\mathrm{COOH}) \end{gathered}$ By definition, $\Delta_{\mathrm{r}} \mathrm{G}^{0}=\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{XPhCOOH} ; \mathrm{aq})-\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{PhCOOH} ; \mathrm{aq})$ Hence, $\begin{gathered} \Delta \Delta_{\mathrm{r}} \mathrm{G}^{0}(\text { acids })=\left[\mathrm{I}(\mathrm{XPh}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)-\mathrm{I}(\mathrm{XPh}) \, \mathrm{I}(\mathrm{COOH})\right] \ -\left[\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}\left(\mathrm{COO}^{-}\right)-\mathrm{I}(\mathrm{Ph}) \, \mathrm{I}(\mathrm{COOH})\right] \end{gathered}$ Or, $\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0}(\text { acids })=[\mathrm{I}(\mathrm{XPh})-\mathrm{I}(\mathrm{Ph})] \,\left[\mathrm{I}\left(\mathrm{COO}^{-}\right)-\mathrm{I}(\mathrm{COOH})\right]$ Thus $\Delta_{r} G^{0}$ is given by the product of two terms; 1. a difference in substituent parameters, and 2. a difference in reaction zone parameters. We turn our attention to the acid strength of phenol and susbstituted phenols in aqueous solution at the same $\mathrm{T}$ and $\mathrm{p}$. A similar analysis to that set out above yields the following equation. $G (phenols) [I(XPh) I(Ph)] [I(O ) I(OH)] 0 ∆∆r = − ⋅ − − (q) Comparison of equations (p) and (q) yields equation (r). \[\left.\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0}(\text { phenols })=\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0} \text { (acids }\right) \,\left\{\left[\mathrm{I}\left(\mathrm{O}^{-}\right)-\mathrm{I}(\mathrm{OH})\right] /\left[\mathrm{I}\left(\mathrm{COO}^{-}\right)-\mathrm{I}(\mathrm{COOH})\right]\right\}$ The analysis rationalises the observation that $\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0} \text { (phenols) }$ is a linear function of $\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0} \text { (acids) }$. In other words we have not proved that such a linear function exists. Rather we have identified the minimum hypothesis required to account for the observation. In these terms the extrathermodynamic analysis has pointed to a reason for the recorded dependences of $\Delta \Delta_{\mathrm{r}} G^{0} \text { (phenols) }$ on $\Delta \Delta_{\mathrm{r}} \mathrm{G}^{0} \text { (acids) }$. The pattern is not a requirement of thermodynamics. Footnotes [1] See for example, 1. D. T. Y. Chen and K. J. Laidler, Trans. Faraday Soc., 1962, 58, 486.; and 2. C. M. Judson and M. L. Kilpatrick, J. Am. Chem. Soc., 1949, 71, 3115. [2] J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions, Wiley, London, 1963. [3] E. Grunwald, Thermodynamics of Molecular Species, Wiley, New York, 1997. [4] The superscript ‘0’ is retained although the meaning here is somewhat obscure. It effectively reminds us that we are dealing with the properties of a solute in its solution reference state.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.8:_Extrathermodynamics_-_Equilbrium_-_Acid_Strength.txt
A solution comprises at least two chemical substances, solvent and solute. The amount of solvent far exceeds the amount of solute so in this sense a solute is dispersed through a solvent [1,2]. Although the solvent molecules are in vast excess, our interest centres on the minor (solute) component because chemists attempt to understand how interactions between a solute molecule and surrounding solvent molecules control the properties of the solute molecule; e.g. control reactivity, solubility, and colour [3]. Out of this interest in solute - solvent interactions emerges the concept of solvent polarity which attempts to characterise this interaction [4]. The general concept of solvent polarity can be understood by considering developments in two subjects; 1. Chemical Kinetics, 2. Spectro-photometry. We review briefly each of these subject areas, indicating how the concept of polarity and/or solvent polarity emerged. We show how the intuitive concept of solvent polarity in these subject areas developed in quantitative terms. We used the word intuitive and this usage can be understood in the following terms. Asked to prepare a solution of sodium chloride (table salt), a first year freshman student would choose water as the solvent rather than (liquid) benzene or ethanol because (the student would argue) water is more "polar" than either ethanol or benzene. Here the term "polar" is little more than laboratory jargon. We seek a quantitative measure of solvent polarity. Y-values In 1862 Bertholet and Pean de Saint Gilles noted that the rate of chemical reaction depends on the solvent. In 1890, Menschutkin confirmed that finding in a very detailed study. So for more than 100 years chemists have attempted to describe quantitively these solvent effects. Perhaps not suprisingly the first attempts concentrated on the dependendence of rate constants on the relative permittivity of solvents. Many authors sought correlations using the treatments described by Kirkwood [5-8]. The quantity used in these correlations usually takes the form $\left(\varepsilon_{\mathrm{r}}-1\right) /\left(2 \, \varepsilon_{\mathrm{r}}+1\right)$. But as many authors point out, this Kirkwood function is little better than the relative permittivity for describing interactions at the molecular level. Nevertheless the challenge remained to describe kinetic solvent effects. A particular important stage was the growth of interest in physical organic chemistry [9]. Probably the ‘father’ of this subject was C.K.Ingold. In his classic monograph [9] , Ingold actually used the pharase ‘solvent polarity’when commenting on the rates of reactions through a series of - 1 - solvents of diminishing polarity; water, ethanol, propanone, benzene. But Ingold did not offer a polarity scale. One of the reactions discussed by Ingold was the hydrolysis of $\left(\mathrm{CH}_{3}\right)_{3}\mathrm{CCl}$. $\left(\mathrm{CH}_{3}\right)_{3} \mathrm{CCl}(\mathrm{aq})+\mathrm{H}_{2} \mathrm{O}(\mathrm{aq}) \rightarrow\left(\mathrm{CH}_{3}\right)_{3} \mathrm{COH}(\mathrm{aq})+\mathrm{H}^{+}(\mathrm{aq})+\mathrm{Cl}^{-}(\mathrm{aq})$ This classic reaction (although the mechanism is still debated) formed the basis of a quantitative description of solvent polarity described by Winstein and coworkers [10,11]. Using the rate of reaction described above they identified a reference solvent, a mixture formed by ethanol (80 vol%) and water(20 vol%). If the rate constant for this reaction is $\mathrm{k}^{0}$ in this solvent mixture and the rate constant is $\mathrm{k}$ in a new solvent, the Y-value of this new solvent is given by $\mathrm{Y}=\log \left(\mathrm{k} / \mathrm{k}^{0}\right)$ In effect $\mathrm{Y}$ measures the ionising power of the solvent – the extent to which the solvent favours charge separation within the neutral solute. Hence by measuring the rate constant for the above reaction in a given solvent, the polarity of the solvent is obtained as shown by its Y-value. This kinetic approach to the determination of solvent polarities has attracted attention, particularly in the context probing reaction mechanisms [9-12]. The Y-value approach can be rationalised using an extrathermodynamic analysis [13]. Nevetheless application of the solvent polarity scale based on Y-values is limited. The range of solvents for which Y-values can be measured is restricted. Z-values A feature of many dye molecules is the sensitivity of their colour to the solvent. This fact was exploited by Brooker and coworkers who used two dyes to define $\chi_{\mathrm{B}}$ and $\chi_{\mathrm{R}}$ values [14]. These scales have not found wide application. A polarity which has attracted attention was suggested by Kosower [15,16]. The scale is based on the uv/visible spectra of N-methyl pyridinium iodide. The low energy absorption band in the spectra characterises the charge transfer from iodide to the pyridinium ring. Kosower examined correlations between Z – and Y- values and between Z-values and other solvent sensitive partaneters . The consensus is that Z provides a reasonable sastifactory measure of solvent polarity. ET Values There can be little doubt that chemists find the concept of solvent polarity intuitively attractive . Granted the need there is an associated demand for a convenient, readily available method for measuring solvent polarity. Reichardt synthesised a betaine dye which - 2 - is particularly solvent sensitive as shown by the dependence on solvent of an intramolecular charge transfer band [17]. Reichardt expresses the energy of the energy band maximum of the absorption band in kilocalories per mol which defines the $\mathrm{E}_{\mathrm{T}}$ value for a given solvent. Solutions of the dye in methanol are red, violet in ethanol and green in propanone. So one has a striking visual indicator of solvent polarity. Foonotes [1] In a solution which is defined as ideal in a thermodynamic sense there are no (solute molecule) $\leftrightarrow$ (solute molecule) interactions. Hence the solute molecules are effectively infinitely far apart. [2] Some indication of the ratio of solute to solvent molecules is indicated by the following rough calculation. Dilute aqueous solutions used in a study of chemical kinetics have concentrations of approx. $10^{-3} \mathrm{mol dm}^{-3}$. In $1 \mathrm{dm}^{3}$ of water there are $55.5$ moles of water so the ratio of solute to solvent molecules is around $55000$. [3] Although the term is not used by chemists it may be helpful to imagine each solute molecule bathed in solvent molecules, implying a limitless expanse of solvent molecules around each solute molecule. [4] We confine attention to the properties of solvents (e.g. polarities) at ambient pressure and at $298.2 \mathrm{~K}$; i.e. 25 Celsius which is just above conventional room temperature. [5] J.G.Kirkwood, J. Chem. Phys.,1934,2,351. [6] Amis discusses treatments of kinetic data based on solvent permittivities; E. S. Amis, Solvent Effects on Reaction Rates and mechanisms, Academic Press, New York, 1966. [7] See comments by N. S. Isaacs, Physical Organic Chemistry, Longmans, London,1987. [8] See also comments concerning attempts to identify a single solvent property which accounts for solvent effects on rates of chemical reactions; J. B. F. N. Engberts, in Water-A Comprehensive Treatise, ed. F.Franks, Plenum Press, New York, 1979,Volume 6, chapter 4. [9] C.K.Ingold, Structure and Mechanism in Organic Chemistry, G. Bell, London, 1953; see page 347. [10] 1. E. Grunwald and S. Winstein, J. Am. Chem. Soc.,1948,70,841;846. 2. S. Winstein and A.H. Fainberg, J. Am. Chem.Soc.,1957,79,5937. 3. H.Langhals, Angew.Chem.Int.Ed.Engl.,1982,21,724. 4. O. Pytela, Collect. Czech. Chem.Commun.,1988,53,1333. [11] See also A. Streitweiser, Solvolytic Displacement Reactions, McGraw-Hill, New York, 1962 [12] Y-values have been used in the context of kinetics of reactions of inorganic solutes; M.J.Blandamer, J. Burgess, and S. Hamshere, Transit. Metals Chem.,1979,4, 291. [13] J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions,Wiley, New York, 1963; Dover Publications, New York, 1989. [14] L. G. S. Brooker, A. C. Craig, D.W. Heseltine, P.W.Jenkins and L. L. Lincoln, J. Am. Chem. Soc.,1965,87,2443. [15] E. Kosower, J.Am. Chem. Soc.1958,80,3253. [16] E. Kosower, Physical Organic Chemistry, Wiley, New York, 1968. [17] C.Reichardt, Chem. Rev.,1994,94,2319; Chem. Soc. Rev., 1992,147; Solvents and Solvent Effects in Organic Chemistry, VCH, Weinheim, 2nd. edn.,1988.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.14%3A_Excess_and_Extra_Thermodynamics/1.14.9:_Extrathermodynamics_-_Solvent_Polarity.txt
An aqueous solution molality $\mathrm{m}_{j}$, at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, contains a simple neutral solute, $j$. The chemical potential of the solute is given by equation (a). \begin{aligned} &\mu_{\mathrm{j}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)= \ &\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T}) \, \mathrm{dp} \end{aligned} $\mathrm{H}_{\mathrm{j}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)=\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \mathrm{T}^{2} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right]_{\mathrm{p}}$ where $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ \begin{aligned} &\mathrm{C}_{\mathrm{pj}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)= \ &\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-2 \, \mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right]_{\mathrm{p}}-\mathrm{R} \, \mathrm{T}^{2} \,\left[\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}^{2}\right]_{\mathrm{p}} \end{aligned} Here $\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\left[\partial \mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq}) / \partial \mathrm{T}\right]_{\mathrm{p}}$ For the solvent, the chemical potential is given by equation (f). $\mu_{1}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)=\mu_{1}^{0}\left(\ell ; \mathrm{T} ; \mathrm{p}^{0}\right)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{\mathrm{l}}^{*}(\ell ; \mathrm{T}) \, \mathrm{dp}$ $\mathrm{H}_{1}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)=\mathrm{H}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Hence, \begin{aligned} &\mathrm{C}_{\mathrm{pl} 1}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)= \ &\mathrm{C}_{\mathrm{pl} 1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})+2 \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial^{2} \phi}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}} \end{aligned} Where $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{C}_{\mathrm{p} 1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{C}_{\mathrm{pl} 1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})$ However, $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{p}_{1}}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}} \text { (aq) }$ Hence, from equations (d) and (h), \begin{aligned} &\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{T} ; \mathrm{p}\right)= \ &\mathrm{n}_{1} \,\left[\mathrm{C}_{\mathrm{pl}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})+2 \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi}{\partial \mathrm{T}}\right)_{\mathrm{p}}\right. \ &\left.+\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial^{2} \phi}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}\right] \ &+\mathrm{n}_{\mathrm{j}} \,\left[\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-2 \, \mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{p}}\right. \ &\left.-\mathrm{R} \, \mathrm{T}^{2} \,\left(\frac{\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}\right] \end{aligned} We rearrange the latter equation to describe the isobaric heat capacity of a solution prepared using $1 \mathrm{~kg}$ of water [1]. \begin{aligned} &\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{w}_{1}=1.0 \mathrm{~kg} ; \mathrm{T} ; \mathrm{p}\right)= \ &\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl}}^{*}(\ell ; \mathrm{T} ; \mathrm{p}) \ &+\mathrm{m}_{\mathrm{j}} \,\left[\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-2 \, \mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\mathrm{R} \, \mathrm{T}^{2} \,\left(\frac{\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}\right. \ &\left.\quad+2 \, \mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \phi}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\mathrm{R} \, \mathrm{T}^{2} \,\left(\frac{\partial^{2} \phi}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}\right] \end{aligned} The term inside the [….] brackets is the apparent molar isobaric heat capacity for the solute. Thus, $\begin{array}{r} \phi\left(C_{p j}\right)=C_{p j}^{\infty}(a q ; T ; p)-2 \, R \, T \,\left(\frac{\partial \ln \left(\gamma_{j}\right)}{\partial T}\right)_{p} \ -R \, T^{2} \,\left(\frac{\partial^{2} \ln \left(\gamma_{j}\right)}{\partial T^{2}}\right)_{p} \ +2 \, R \, T \,\left(\frac{\partial \phi}{\partial T}\right)_{p}+R \, T^{2} \,\left(\frac{\partial^{2} \phi}{\partial T^{2}}\right)_{p} \end{array}$ Hence, $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{w}_{1}=1.0 \mathrm{~kg} ; \mathrm{T} ; \mathrm{p}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ Then, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi\left(\mathrm{C}_{\mathrm{pj}}\right)=\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}=\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ Equation (m) shows that $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$ is a complicated property of a solution and that ‘the devil is in the detail’ [1]. A simplification in the algebra emerges if we define a set of J-properties which are excess properties [2,3]. Thus for a given solution prepared using $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ mole of solute $j$, $\mathrm{J}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{J}_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{J}_{\mathrm{j}}(\mathrm{aq})$ where $J(a q)=C_{p}(a q)-C_{p}(a q ; i d)$ $\mathrm{J}_{1}(\mathrm{aq})=\mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})-\mathrm{C}_{\mathrm{pl} 1}^{*}(\ell)$ $\mathrm{J}_{\mathrm{j}}(\mathrm{aq})=\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})-\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ $\phi\left(\mathrm{J}_{\mathrm{j}}\right)=\phi\left(\mathrm{C}_{\mathrm{p}_{\mathrm{j}}}\right)-\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}$ But $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})-\mathrm{C}_{\mathrm{p}}(\mathrm{aq} ; \mathrm{id})=\mathrm{n}_{\mathrm{j}} \,\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\phi\left(\mathrm{C}_{\mathrm{pj}}\right)^{\infty}\right]$ Then $\mathrm{J}(\mathrm{aq})=\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{J}_{\mathrm{j}}\right)$ An extensive literature reports the partial molar heat capacities of solutes in aqueous solution [2,3]. Further ${\mathrm{C}_{\mathrm{pj}}}^{\infty}(\mathrm{aq})$ for a range of related solutes can be analysed to yield group contributions [4-6]; e.g. at $298.15 \mathrm{~K}$ the contribution of a methyl group, $\mathrm{CH}_{3}$ to ${\mathrm{C}_{\mathrm{pj}}}^{\infty}$ for an aliphatic solute is $178 \mathrm{~J K}^{-1} \mathrm{~mol}^{-1}$. Granted that ${\mathrm{C}_{\mathrm{pj}}}^{\infty}(\mathrm{aq})$ has been obtained for solute $j$ and that the molar heat capacity of pure liquid $j$, $\mathrm{C}_{\mathrm{pj}}^{*}(\ell)$ is known , the isobaric heat capacity of solution $\Delta_{s \ln } \mathrm{C}_{\mathrm{pj}}^{0}$ is obtained [7]. $\Delta_{s \ln } C_{p j}^{0}=C_{p j}^{\infty}(a q)-C_{p j}^{*}(\ell)$ Footnotes [1] $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{m}_{\mathrm{j}} ; \mathrm{w}_{1}=1.0 \mathrm{~kg} ; \mathrm{T} ; \mathrm{p}\right)=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]$ $\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})=\left[\mathrm{kg} \mathrm{mol}^{-1}\right]^{-1} \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]$ $\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]$ $\mathrm{m}_{\mathrm{j}} \, 2 \, \mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,[1] \,\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right] \,[\mathrm{K}] \,[\mathrm{K}]^{-1}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]$ $\mathrm{m}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}^{2} \,\left(\frac{\partial^{2} \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}=\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right] \,[\mathrm{K}]^{2} \,\left[\mathrm{K}^{-2}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]\right.$ $\mathrm{m}_{\mathrm{j}} \, 2 \, \mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \phi}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left[\mathrm{mol} \mathrm{kg}^{-1}\right] \,[1] \,\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right] \,[\mathrm{K}] \,\left[\mathrm{K}^{-1}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]\right.$ $\mathrm{m}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}^{2} \,\left(\frac{\partial^{2} \phi}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}}=\left[\mathrm{mol} \mathrm{kg}{ }^{-1}\right] \,\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right] \,[\mathrm{K}]^{2} \,[\mathrm{K}]^{-2}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~kg}^{-1}\right]$ [2] Sucrose(aq); an early determination; F. T. Gucker and F. D. Ayres, J. Am. Chem.Soc.,1937,59,447. [3] For partial molar isobaric heat capacities of neutral solutes in aqueous solution see, 1. Simple hydrocarbons, G. Olofsson, A. A. Oshodj, E. Qvarnstrom and I. Wadso, J. Chem. Thermodyn., 1984,16,1041. 2. Amides, 2-methyl propan-2-ol, pentanol , R. Skold, J. Suurkuusk and I. Wadso, J. Chem. Thermodyn., 1976,8,1075. 3. Nitrogen and hydrogen, J. Alvarez, R. Crovetto and R. Fernandez-Prini, Ber. Bunsenges. Phys.Chem.,1988,92,935. 4. Amides, ketones, esters and ethers; G. Roux, G. Perron and J. E. Desnoyers, Can. J.Chem.,1978,56,2808. 5. Urea; O. D. Bonner, J. M. Bednarek and R. K. Arisman, J. Am. Chem.Soc.,1977,99,2898. 6. Urea + 2-methylpropan-2-ol(aq); C.de Visser, G.Perron and J.E.Desnoyers, J. Am. Chem. Soc., 1977,99,5894. 7. Alcohols; E. M. Arnett, W. B. Kover and J. V. Carter, J. Am. Chem.Soc.,1969,91,4028. 8. Sugars; O. D. Bonner and P. J. Cerutti, J. Chem. Thermodyn., 1976, 8,105. 9. Amino acids ($288 \mathrm{~K}$ to $328 \mathrm{~K}$); 1. A.W Hakin, M. M. Duke, S. A. Klassen, R. M. McKay and K E. Preuss, Can. J. Chem. 1994,72,362. 2. M. M. Duke, A.W.Hakin, R.M. McKay and K. M. Preuss, Can. J.Chem.,1994,72,1489. 3. A. W. Hakin, M. M. Duke, J. L. Marty and K. E. Preuss, J. Chem. Soc. Faraday Trans.,1994,90,14. 4. A. W. Hakin, M. M. Duke, L. L. Groft, J. L. Marty and M. L.Rushfeldt, Can. J.Chem.,1995,73,725. 5. A.W.Hakin, C. L. Beswick and M. M. Duke, J. Chem. Soc. Faraday Trans.,1996,92,207. 6. A. W. Hakin, A. K.Copeland, J. L. Liu, R. A. Marriott and K. E. Preuss, J. Chem. Eng. Data,1997,42,84. 7. A. W.Hakin and G.R Hedwig, J.Chem. Thermodyn.,2001,33,1709. 10. cyclic dipetides; 1. A. W. Hakin, B. Cavilla, J. L. Liu and B. Zorzetti, Phys.Chem.Chem.Phys.,2001, 3,3805. 2. A. W. Hakin, J. L. Liu, M. O’Shea and B. Zorzetti, Phys.Chem.Chem.,Phys.,2003,5,2653. 11. cyclic dipeptides; 1. C. J. Downes, A. W. Hakin and G. R. Hedwig, J.Chem.Thermodyn.,2001,33,873. 2. A.W. Hakin, M.G.Kowalchuck, J. L. Liu and R.A. Marriot, J. Solution Chem.2000, 29,131. 12. acetylamides; 1. J. L. Liu, A. W. Hakin and G. R. Hedwig, J. Solution Chem.,2001,30, 861. 2. A. W. Hakin and G. R. Hedwig, Phys. Chem. Chem. Phys.,2002,2,1795. 13. alkanolamines; Y. Maham, L. G. Helper, A. E. Mather, A.W. Hakin, and R. A. Marriott, J. Chem. Soc. Faraday Trans.,1997,93,1747. [4] ROH(aq); D. Mirejovsky and E. M. Arnett, J. Am. Chem.Soc.,1983,105,112. [5] Ph-X; group additivity; G. Perron and J. E. Desnoyers, Fluid Phase Equilib, 1979,2,239. [6] Amides(aq); R. Skold, J. Suurkuus and I. Wadso, J.Chem. Thermodyn., 1976,8,1075. [7] $\Delta_{s \ln } \mathrm{C}_{\mathrm{pj}}^{0}$ for gases(aq); G. Olofsson, A. A. Oshodj, E. pj Qvarstrom and I. Wadso, J. Chem. Thermodyn., 1984,16,1041.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.15%3A_Heat_Capacities/1.15.1%3A_Heat_Capacities%3A_Isobaric%3A_Neutral_Solutes.txt
We describe an excess enthalpy $\mathrm{H}^{\mathrm{E}}$ for a solution prepared using $1 \mathrm{~kg}$ of water and $\mathrm{m}_{j}$ moles of solute $j$ (at fixed $\mathrm{T}$ and $\mathrm{p}$) ion terms of solute-solute enthalpic interaction parameters. $\mathrm{H}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{h}_{\mathrm{ij}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}$ The corresponding excess isobaric heat capacity is defined by equation (b). $C_{p}^{E}\left(a q ; w_{1}=1 k g\right)=c_{p i j} \,\left(m_{j} / m^{0}\right)^{2}$ where $\mathrm{c}_{\mathrm{pij}}=\left(\frac{\partial \mathrm{h}_{\mathrm{ij}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Here $\mathrm{c}_{\mathrm{pjj}}$ is a pairwise solute-solute interaction isobaric heat capacity [1]. From $\mathrm{H}_{1}(\mathrm{aq})=\mathrm{H}_{1}^{*}(\ell)-\mathrm{M}_{1} \, \mathrm{h}_{\mathrm{ij}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}$ then, $\mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})=\mathrm{C}_{\mathrm{pl}^{2}}^{*}(\ell)-\mathrm{M}_{\mathrm{l}} \, \mathrm{c}_{\mathrm{pjj}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}$ From $\mathrm{H}_{\mathrm{j}}(\mathrm{aq})=\mathrm{H}_{\mathrm{j}^{\prime}}^{\infty}(\mathrm{aq})+2 \, \mathrm{h}_{\mathrm{jj}} \,\left(\mathrm{m}^{0}\right)^{-2} \, \mathrm{m}_{\mathrm{j}}$ then, $\mathrm{C}_{\mathrm{pj}}(\mathrm{aq})=\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+2 \, \mathrm{c}_{\mathrm{pjj}} \,\left(\mathrm{m}^{0}\right)^{-2} \, \mathrm{m}_{\mathrm{j}}$ Footnote [1] For a solution prepared using $1 \mathrm{~kg}$ of water and $\mathrm{m}_{j}$ moles of solute (at fixed $\mathrm{T}$ and $\mathrm{p}$) $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{p} 1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}(\mathrm{aq})$ Hence $\begin{gathered} \mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \,\left[\mathrm{C}_{\mathrm{pl}}^{*}(\ell)-\mathrm{M}_{1} \, \mathrm{c}_{\mathrm{pjj}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}\right] \ +\mathrm{m}_{\mathrm{j}} \,\left[\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+2 \, \mathrm{c}_{\mathrm{pjj}} \,\left(\mathrm{m}^{0}\right)^{-2} \, \mathrm{m}_{\mathrm{j}}\right] \end{gathered}$ Then, $\mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})+\mathrm{c}_{\mathrm{pji}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}$ Since, $\begin{gathered} \mathrm{C}_{\mathrm{p}}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{C}_{\mathrm{pl} 1}^{*}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq}) \ \mathrm{C}_{\mathrm{p}}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{c}_{\mathrm{pjj}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2} \end{gathered}$ 1.15.3: Heat Capacities: Isobaric: Solutions: Unit Volume A given aqueous solution was prepared using $\mathrm{n}_{1}$ moles of water($\ell$) and $\mathrm{n}_{j}$ moles of solute $j$. Then, $\mathrm{C}_{\mathrm{p}}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{C}_{\mathrm{pl}}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ For the solution, by definition, the isobaric heat capacity per unit volume, (or heat capacitance) $\sigma(\mathrm{aq})=\mathrm{C}_{\mathrm{p}}(\mathrm{aq}) / \mathrm{V}(\mathrm{aq})$ Similarly for the solvent at the same temperature and pressure, $\sigma_{1}^{*}(\ell)=\mathrm{C}_{\mathrm{p} 1}^{*}(\ell) / \mathrm{V}_{1}^{*}(\ell)$ With reference to equations (b) and (c), the four experimentally determined quantities are $\sigma(\mathrm{aq}), \sigma_{1}^{*}(\mathrm{aq}), \rho(\mathrm{aq}) \text { and } \rho_{1}^{*}(\ell)$. The latter two quantities are the densities of the solution and solvent respectively. Hence $\sigma(\mathrm{aq})$ is related to the concentration of the solution, $\mathrm{c}_{j}$ [1,2]. $\sigma(\mathrm{aq})=\sigma_{1}^{*}(\ell)+\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \sigma_{1}^{*}(\ell)\right] \, \mathrm{c}_{\mathrm{j}}$ The latter equation relates $\sigma(\mathrm{aq})$ to the property for the pure solvent, $\sigma_{1}^{*}(\ell)$ and to the concentration of solute, $\mathrm{c}_{\mathrm{j}}$. Equation (d) relates $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$ to the measured quantities $\sigma(\mathrm{aq})$ and $\sigma_{1}^{*}(\ell)$ together with the apparent molar volume $\phi \left(\mathrm{V}_{\mathrm{j}}\right)$. Thus $\sigma(\mathrm{aq})$ and $\phi \left(\mathrm{V}_{\mathrm{j}}\right)$ for a given solution yields together with $\sigma_{1}^{*}(\ell)$, the apparent molar isobaric heat capacity of the solute, $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$. In cases where the composition of the solution is expressed using molalities, equation (e) is the equation for $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$ [3,4]. \begin{aligned} \phi\left(\mathrm{C}_{\mathrm{pj}}\right)=& {\left[\mathrm{m}_{\mathrm{j}} \, \rho(\mathrm{aq}) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell) \, \sigma(\mathrm{aq})-\rho(\mathrm{aq}) \, \sigma_{1}^{*}(\ell)\right] } \ &+\mathrm{M}_{\mathrm{j}} \, \sigma(\mathrm{aq}) / \rho(\mathrm{aq}) \end{aligned} Footnotes [1] From equations (a) and (b), $\sigma(\mathrm{aq})=\left[\mathrm{n}_{1} / \mathrm{V}(\mathrm{aq})\right] \,\left[\mathrm{V}_{1}^{*}(\ell) / \mathrm{V}_{1}^{*}(\ell)\right] \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\left[\mathrm{n}_{\mathrm{j}} / \mathrm{V}(\mathrm{aq})\right] \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ The term $\left[\mathrm{V}_{1}^{*}(\ell) / \mathrm{v}_{1}^{*}(\ell)\right]$ has been introduced with the definition of $\sigma_{1}^{*}(\ell)$ in mind. But, $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ or, $\mathrm{n}_{\mathrm{l}} \, \mathrm{V}_{1}^{*}(\ell)=\mathrm{V}(\mathrm{aq})-\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Further concentration, $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}$ Then, $\sigma(\mathrm{aq})=\left[\mathrm{V}(\mathrm{aq})-\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right] \,[\mathrm{V}(\mathrm{aq})]^{-1} \, \sigma_{1}^{*}(\ell)+\mathrm{c}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ Hence, $\sigma(\mathrm{aq})=\sigma_{1}^{*}(\ell) \,\left[1-\mathrm{c}_{\mathrm{j}}\left(\mathrm{V}_{\mathrm{j}}\right)\right]+\mathrm{c}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right)$ or $\sigma(\mathrm{aq})=\sigma_{1}^{*}(\ell)+\left[\phi\left(\mathrm{C}_{\mathrm{pj}}\right)-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \sigma_{1}^{*}(\ell)\right] \, \mathrm{c}_{\mathrm{j}}$ [2] \begin{aligned} &\phi\left(\mathrm{C}_{\mathrm{pj}}\right) \, \mathrm{c}_{\mathrm{j}}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,\left[\frac{\mathrm{mol}}{\mathrm{m}^{3}}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~m}^{-3}\right] \ &\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \sigma_{1}^{*}(\ell) \, \mathrm{c}_{\mathrm{j}}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{JK}^{-1} \mathrm{~m}^{-3}\right] \,\left[\mathrm{mol} \mathrm{m}^{-3}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~m}^{-3}\right] \end{aligned} [3] From [1], $\mathrm{V}(\mathrm{aq})=\left[\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \sigma_{1}^{*}(\ell) / \sigma(\mathrm{aq})\right]+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) / \sigma(\mathrm{aq})$ But, $\mathrm{V}(\mathrm{aq})=\left(\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right) / \rho(\mathrm{aq})$ Then, $\frac{\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}=\left[\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \sigma_{1}^{*}(\ell) / \sigma(\mathrm{aq})\right]+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{C}_{\mathrm{pj}}\right) / \sigma(\mathrm{aq})$ or (dividing by $\mathrm{n}_{\mathrm{j}}$), $\left[\frac{\mathrm{n}_{1} \, \mathrm{M}_{1}}{\rho(\mathrm{aq}) \, \mathrm{n}_{\mathrm{j}}}\right]+\left[\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}}{\mathrm{n}_{\mathrm{j}} \, \rho(\mathrm{aq})}\right]=\left[\frac{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \sigma_{1}^{*}(\ell)}{\mathrm{n}_{\mathrm{j}} \, \sigma(\mathrm{aq})}\right]+\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)}{\sigma(\mathrm{aq})}$ But molality $\mathrm{m}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{1} \, \mathrm{M}_{1}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell) \, \rho_{1}^{*}(\ell)$ Then, $\left[\frac{1}{\rho(\mathrm{aq}) \, \mathrm{m}_{\mathrm{j}}}\right]-\left[\frac{\sigma_{1}^{*}(\ell)}{\rho_{1}^{*}(\ell) \, \mathrm{m}_{\mathrm{j}} \, \sigma(\mathrm{aq})}\right]+\frac{\mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}=\frac{\phi\left(\mathrm{C}_{\mathrm{pj}}\right)}{\sigma(\mathrm{aq})}$ As an equation for $\phi \left(\mathrm{C}_{\mathrm{pj}}\right)$; $\phi\left(\mathrm{C}_{\mathrm{pj}}\right)=\frac{\sigma(\mathrm{aq})}{\rho(\mathrm{aq}) \, \mathrm{m}_{\mathrm{j}}}-\frac{\sigma_{1}^{*}(\ell)}{\rho_{1}^{*}(\ell) \, \mathrm{m}_{\mathrm{j}}}+\frac{\mathrm{M}_{\mathrm{j}} \, \sigma(\mathrm{aq})}{\rho(\mathrm{aq})}$ Hence $\begin{gathered} \phi\left(\mathrm{C}_{\mathrm{pj}}\right)=\left[\mathrm{m}_{\mathrm{j}} \, \rho(\mathrm{aq}) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell) \, \sigma(\mathrm{aq})-\rho(\mathrm{aq}) \, \sigma_{1}^{*}(\ell)\right] \ +\mathrm{M}_{\mathrm{j}} \, \sigma(\mathrm{aq}) / \rho(\mathrm{aq}) \end{gathered}$ [4] \begin{aligned} &{\left[\mathrm{m}_{\mathrm{j}} \, \rho(\mathrm{aq}) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell) \, \sigma(\mathrm{aq})-\rho(\mathrm{aq}) \, \sigma_{1}^{*}(\ell)\right]=} \ &{\left[\frac{\mathrm{kg}}{\mathrm{mol}}\right] \,\left[\frac{\mathrm{m}^{3}}{\mathrm{~kg}}\right] \,\left[\frac{\mathrm{m}^{3}}{\mathrm{~kg}}\right] \,\left[\frac{\mathrm{kg}}{\mathrm{m}^{3}}\right] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~m}^{-3}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]} \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.15%3A_Heat_Capacities/1.15.2%3A_Heat_Capacities%3A_Solutions%3A_Solutes%3A_Interaction_Parameters.txt
The excess enthalpy $\mathrm{H}^{\mathrm{E}}$ of an aqueous salt solution prepared using $1 \mathrm{~kg}$ of water and $\mathrm{m}_{j}$ moles of a 1:1 salt is related to $\mathrm{m}_{j}$ using the DHLL. Because ${\mathrm{C}_{\mathrm{p}}}^{\mathrm{E}}$ is the isobaric temperature dependence of $\mathrm{H}^{\mathrm{E}}$, then ${\mathrm{C}_{\mathrm{p}}}^{\mathrm{E}}$ for this aqueous solution is given by equation (a) [1]. $\mathrm{C}_{\mathrm{p}}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=-(4 / 3) \, \mathrm{R} \, \mathrm{m}_{\mathrm{j}}^{(3 / 2)} \,\left(\mathrm{m}^{0}\right)^{-1 / 2}\left[2 \, \mathrm{T} \, \mathrm{S}_{\mathrm{H}}+\mathrm{T}^{2} \,\left(\partial \mathrm{S}_{\mathrm{H}} / \partial \mathrm{T}\right)_{\mathrm{p}}\right]$ $\mathrm{S}_{\mathrm{Cp}_{\mathrm{p}}}=2 \, \mathrm{T} \, \mathrm{S}_{\mathrm{H}}+\mathrm{T}^{2} \,\left(\partial \mathrm{S}_{\mathrm{H}} / \partial \mathrm{T}\right)$ $\mathrm{S}_{\mathrm{Cp}}$ is the DHLL factor in the equation for the isobaric heat capacity. $C_{p}^{E}\left(a q ; w_{1}=1 \mathrm{~kg}\right)=-(4 / 3) \, R \, S_{C p} \, m_{j}^{(3 / 2)} \,\left(m^{0}\right)^{-1 / 2}$ Using equation (c) [2], $\phi\left(\mathrm{J}_{\mathrm{j}}\right)=-(4 / 3) \, \mathrm{R} \, \mathrm{S}_{\mathrm{Cp}_{\mathrm{p}}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ We could perhaps have anticipated that according to DHLL, $\phi \left(\mathrm{J}_{j}}\right)$ is a linear function of $\left(\mathrm{m}_{\mathrm{j}}\right)^{1 / 2}$. An extensive literature describes the limiting partial molar isobaric heat capacities of ions in aqueous solution. One of the earliest investigations of the isobaric heat capacities of salt solutions was made by Randall and Ramage[3] and later by Randall and Taylor [4]. The groups lead by Hepler [5,6] and by Desnoyers [7] have made significant contributions in this area. However no agreement has been reached on a scale of absolute values. Hepler reported relative estimates based on $\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq} ; 298 \mathrm{~K}\right)$ equal to zero. Perhaps most attention has been directed at salts formed by alkylammonium cation [8,9] and hydrophobic anions; e.g. amino acids [10], phenylcarboxylates, t-butylcarboxlates[11] and cryptates[12]. Data [7] for $\mathrm{R}_{4} \mathrm{~N}^{+} \mathrm{Br}^{-}(\mathrm{aq})$ show that $\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq})$ increases with increase in hydrophobic character of the R-group. French and Criss argue [13] in favour of a scale which sets $\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right)$ at $– 68 \mathrm{~J K}^{-1} \mathrm{~mol}^{-1}$. An attempt[14] has identified the various contributions to $\mathrm{C}_{\mathrm{p}}^{\infty}(\text { ion; aq })$. Certainly trends in $\mathrm{C}_{\mathrm{p}}^{\infty}(\text { ion; aq })$ point to characteristic features associated with the properties of ions in aqueous solution. Nevertheless, interpretation is not straightforward [14]. Footnotes [1] $\mathrm{S}_{\mathrm{Cp}}=2 \, \mathrm{T} \, \mathrm{S}_{\mathrm{H}}+\mathrm{T}^{2} \,\left(\partial \mathrm{S}_{\mathrm{H}} / \partial \mathrm{T}\right)_{\mathrm{p}}=[1] \,[\mathrm{K}] \,\left[\mathrm{K}^{-1}\right]+[\mathrm{K}]^{2} \,\left[\mathrm{K}^{-1}\right] \,[\mathrm{K}]^{-1}=[1]$ [2] $\phi\left(\mathrm{J}_{\mathrm{j}}\right)=[1] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[1] \,[1]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right]$ [3] M. Randall and W.D.Ramage, J.Am.Chem.Soc.,1927,49,93 [4] M.Randall and M.D.Taylor, J. Am. Chem. Soc.,1941,45,959. [5] I. K. Hovey, L. G. Hepler and P. R. Tremaine, J. Phys. Chem.,1988, 92, 1323; Thermochim. Acta, 1988, 126, 245; J. Chem. Thermodyn.,1988, 20, 595. [6] J. J. Spitzer, I. V. Oloffson, P. P. Singh and L. G. Hepler, Thermochim. Acta,1979, 28, 155. [7] J.-L. Fortier, P.-A. Leduc and J. E. Desnoyers, J. Solution Chem, 1974, 3, 323. [8] B. Chawla and J. C. Ahluwalia, J. Phys. Chem.,1972 76, 2582. [9] E. M. Arnett and J. J. Campion, J. Am. Chem. Soc.,1970, 92, 7097. [10] J. C. Ahluwalia, C. Ostiguy, G. Perron and J. E. Desnoyers, Can. J. Chem., 1977, 55, 3364, and 3368. [11] M. Lucas and H. Le Bail, J. Phys. Chem., 1976, 80, 2620. [12] N. Morel-Desrosiers and J. P. Morel, J. Phys. Chem., 1985, 89, 1541. [13] R. N. French and C. M. Criss, J. Solution Chem.,1982, 11, 625. [14] C. Shin, I. Worsley and C.M.Criss, J. Solution Chem.,1976, 5 , 867. [15] For further details of heat capacities of salt solutions see— 1. NaOH(aq); G. Conti, P. Gianni, A. Papini and E. Matteoli, J. Solution Chem.,1988,17,481. 2. Alkali metal halides(aq);also solutions in D2O; J.-L. Fortier, P. R. Philip and J.E.Desnoyers, J. Solution Chem.,1974,3,523. 3. NaBPh4(aq); and in urea(aq); B. Chawla, S. Subramanian and J. C. Ahluwalia, J.Chem.Thermodyn.,1972,4,575. 4. Bu4NBr(aq) and NaBPh4(aq); and in aqueous mixtures; S. Subramanian and J. C.Ahluwalia, Trans. Faraday Soc.,1971,67,305. 5. R4NBr(aq; 382 to 363) K; M.J.Mastroianni and C. M. Criss, J. Chem. Thermodyn, 1972,4,321. 6. R4NBr(aq); E.M.Arnett and J. Campion, J. Am. Chem. Soc., 1970, 92, 7097. 7. R4NCl(aq); K.Tamaki, S. Yoshikawa and M.Kushida, Bull. Chem.Soc.,Jpn, 1975,48,3018. 8. R4N+ R’COO- (aq); P.-A. Leduc and J. E. Desnoyers, Can. J.Chem. 1973,51,2993. 9. Amino acids(aq); G. C. Kresheck J.Chem.Phys.,1970,52,5966. 10. MCl(aq + 2-methylpropan-2-ol); G. T. Hefter, J.-P. E. Grolier and A. H. Roux, J. Solution Chem.,1989,18,229. 11. Am4NBr(aq + 2-methylpropan-2-ol); R. K. Mohanty, S. Sunder and J. C. Ahluwalia, J. Phys.Chem.,1972,76,2577. 12. CsI(aq; 273 – 373 K); R. E. Mitchell, and J. W. Cobble, J. Am. Chem. Soc., 1964,86,5401. 13. Amino acids(aq) and (aq. + urea); C. Jolicoeur, B. Riedl, D. Desrochers, L. L. Lemelin, R. Zamojska and O. Enea, J. Solution Chem.,1986,15,109. 14. Bu4NBr(aq); NaBPh4(aq); ( also binary aq. mixtures); R. K. Mohanty, T. S. Sarma, S. Subrahamian and J. C. Ahluwalia, Trans. Faraday Soc., 1971, 67, 305. 15. NaCl(aq; 274 to 318 K); G. Perron, J.-L. Fortier and J. E. Desnoyers, J. Chem. Thermodynamics, 1975, 7,1177. 16. Salts (aq .and mixed solvents); J. E. Desnoyers, O. Kiyohara , G. Perron and L. Avedikian, Adv. Chem. Series, No 155. 1976. 17. Bu4N+ butyrate(aq; 283 to 323 K); A. S. Levine and S. Lindenbaum, J. Solution Chem., 1973,2,445. 18. NaCl(aq; 320 to 600 K); D. Smith-Magowan and R. H. Wood, J. Chem. Thermodyn., 1981, 13,1047. 19. CaCl2(aq; 306 to 603 K; 17.4 Mpa); D. E. White, A. L. Doberstein, J. A. Gates, D. M. Tillett and R. H. Wood, J. Chem. Thermodyn., 1987,19,251. 20. NaCl(aq; 273 to 313 K; 0 to 1000 bar); C.-T. A. Chen, J. Chem. Eng. Data, 1982,27,356. 21. CH3COOH(aq); CH3COONa(aq), NH3(aq); NH4Cl(aq); 283, 298 and 313 K; G. Allred and E. M Woolley, J. Chem. Thermodyn., 1981, 13, 155. 22. Group III metal perchlorates; 23. R. A. Marriott, A.W. Hakin and J. A. Rard, J.Chem.Thermodyn.,2001,33,643. 1. A. W. Hakin, M. J. Lukacs, J. L. Liu, K. Erickson and A. Madhavji, J.Chem.Thermodyn.,2003,35,775. 1.15.5: Heat Capacities: Isochoric: Liquid Mixtures: Ideal For an ideal binary liquid mixture the molar isobaric heat capacity is given by the mole fraction weighted sum of the isobaric heat capacities of the pure liquid components. $\mathrm{C}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{C}_{\mathrm{p} 2}^{*}(\ell)$ Both $\mathrm{C}_{\mathrm{p} 1}^{*}(\ell)$ and $\mathrm{C}_{\mathrm{p} 2}^{*}(\ell)$ can be measured so that $\mathrm{C}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})$ can be calculated for a given mixture as a function of mole fraction composition. Further $\Delta_{\operatorname{mix}} \mathrm{C}_{\mathrm{p}}(\mathrm{id})=0$ The isochoric heat capacity of the corresponding ideal mixture is related to the isobaric heat capacity using equation (c) [1]. $\mathrm{C}_{\mathrm{V}_{\mathrm{m}}}(\text { mix } ; \mathrm{id})=\mathrm{C}_{\mathrm{pm}}(\text { mix } ; \mathrm{id})-\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{pm}}(\text { mix} ; \mathrm{id})\right]^{2}}{\mathrm{~K}_{\mathrm{Tm}}(\text { mix } ; \mathrm{id})}$ Equations (a) and (c) provide an equation for $\mathrm{C}_{\mathrm{Vm}}(\operatorname{mix} ; \mathrm{id})$ in terms of the isochoric heat capacities of the pure liquid components. \begin{aligned} &\mathrm{C}_{\mathrm{V}_{\mathrm{m}}}(\operatorname{mix} ; \mathrm{id})=\ &\mathrm{x}_{1} \,\left[\mathrm{C}_{\mathrm{V} 1}^{*}(\ell)+\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 1}^{*}(\ell)}\right]+\mathrm{x}_{2} \,\left[\mathrm{C}_{\mathrm{V} 2}^{*}(\ell)+\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 2}^{*}(\ell)}\right]\ &-\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{id})\right]^{2}}{\mathrm{~K}_{\mathrm{Tm}}(\mathrm{mix} ; \mathrm{id})} \end{aligned} In terms of forming an ideal binary liquid mixture from two pure components, $\begin{gathered} \Delta_{\operatorname{mix}} \mathrm{C}_{\mathrm{V}_{\mathrm{m}}}(\mathrm{id})=\mathrm{x}_{1} \,\left[\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{pl}}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 1}^{*}(\ell)}\right]+\mathrm{x}_{2} \,\left[\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 2}^{*}(\ell)}\right] \ -\frac{\mathrm{T} \,\left[\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{id})\right]^{2}}{\mathrm{~K}_{\mathrm{Tm}}(\mathrm{mix} ; \mathrm{id})} \end{gathered}$ The equations become more complicated as we switch conditions from the intensive variables, $\mathrm{T}$ and $\mathrm{p}$, to extensive variables such as entropy and volume. The equations become even more complicated when we turn to a description of real mixtures. Footnote [1] Consider a closed system subjected to a change in temperature, the system remaining at equilibrium where the affinity for spontaneous change is zero. Then $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0} \quad \text { and } \quad \mathrm{C}_{\mathrm{V}}(\mathrm{A}=0)=\left(\frac{\partial \mathrm{U}}{\partial \mathrm{T}}\right)_{\mathrm{V}, \mathrm{A}=0}$ In the following we drop the condition ‘$\mathrm{A}=0$’ and take it as implicit in the following analysis. [A similar set of equations can be written for the condition ‘at fixed ξ’.] Then $\mathrm{C}_{\mathrm{p}}-\mathrm{C}_{\mathrm{V}}=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\left(\frac{\partial \mathrm{U}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ but by definition, $\mathrm{H}=\mathrm{U}+\mathrm{p} \, \mathrm{V}$ Then $C_{p}-C_{V}=\left(\frac{\partial H}{\partial T}\right)_{p}-\left(\frac{\partial H}{\partial T}\right)_{v}+V \,\left(\frac{\partial p}{\partial T}\right)_{v}$ Using a calculus operation, $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{v}}=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{v}}$ Then, $\mathrm{C}_{\mathrm{p}}-\mathrm{C}_{\mathrm{V}}=\left[\mathrm{V}-\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right] \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ By definition $\mathrm{H}=\mathrm{G}+\mathrm{T} \, \mathrm{S}$; then $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right)_{\mathrm{T}}+\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ A Maxwell equation requires that $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Then, $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\mathrm{V}-\mathrm{T} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Hence, $\mathrm{C}_{\mathrm{p}}-\mathrm{C}_{\mathrm{V}}=\mathrm{T} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ A calculus operation requires that $\left(\frac{\partial p}{\partial T}\right)_{V} \,\left(\frac{\partial T}{\partial V}\right)_{p} \,\left(\frac{\partial V}{\partial p}\right)_{T}=-1$ Then $C_{p}-C_{V}=-T \,\left[\left(\frac{\partial V}{\partial T}\right)_{p}\right]^{2} \,\left[\left(\frac{\partial V}{\partial p}\right)_{T}\right]^{-1}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.15%3A_Heat_Capacities/1.15.4%3A_Heat_Capacities%3A_Isobaric%3A_Salt_Solutions.txt
The term ‘ strong electrolyte’ has a long and honourable history in the development of an understanding of the properties of salt solutions. This term describes salt solutions where each ion contributes to the properties almost independently of all other ions in a given solution. The word ‘almost’ signals that the properties of a given salt solution are determined in part by charge –charge interactions between ions through the solvent separating ions in solution. Otherwise the ions can be regarded as free. Such is the case for aqueous salt solutions at ambient temperatures and pressures prepared using 1:1 salts such as $\mathrm{Na}^{+} \mathrm{Cl}^{-}$, $\mathrm{Et}_{4}\mathrm{N}^{+} \mathrm{Br}^{-}$ … However with decrease in relative permittivity of the solvent, the properties of salt solutions indicate that not all the ions can be regarded as free; a fraction of the ions are associated. For dilute salts solutions in apolar solvents such as propanone a fraction of the salt is described as being present as ion pairs formed by association of cations and anions. With further decrease in the permittivity of the solvent higher clusters are envisaged; e.g. triple ions, quadruple ions…. Here we concentrate attention on ion pair formation building on the model proposed by N. Bjerrum [1,2]. The analysis identifies a given $j$ ion in a salt solution as the reference ion such that at distant $\mathrm{r}$ from this ion the electric potential equals $\psi_{j}$ whereby the potential energy of ion $\mathrm{i}$ with charge number $\mathrm{z}_{\mathrm{i}}$ equals $\mathrm{z}_{i} \, e \, \psi_{j}. The solvent is a structureless continuum and each ion is a hard non-polarisable sphere characterised by its charge, \(\mathrm{z}_{j} \, e$, and radius $\mathrm{r}_{j}$. If the bulk number concentration of $\mathrm{i}$ ions is $\mathrm{p}_{\mathrm{i}}$, the average local concentration of $\mathrm{i}$ ions' $\mathrm{p}_{\mathrm{i}}$ is given by equation (a) [3]. $\mathrm{p}_{\mathrm{i}}^{\prime}=\mathrm{n}_{\mathrm{i}} \, \exp \left(-\mathrm{z}_{\mathrm{i}} \, \mathrm{e} \, \psi_{\mathrm{j}} / \mathrm{k} \, \mathrm{T}\right)$ The number of $\mathrm{i}$-ions, $\mathrm{dn}_{\mathrm{i}}$ in a shell thickness $\mathrm{dr}$ distance $\mathrm{r}$ from the reference $j$ ion is given by equation (b) [4]. $\mathrm{p}_{\mathrm{j}}=\mathrm{n}_{\mathrm{i}} \, \exp \left(-\mathrm{z}_{\mathrm{i}} \, \mathrm{e} \, \psi_{\mathrm{j}} / \mathrm{k} \, \mathrm{T}\right) \, 4 \, \pi \, \mathrm{r}^{2} \, \mathrm{dr}$ At small $\mathrm{r}$, the electric potential arising from the $j$ ion is dominant. Hence [5], $\psi_{\mathrm{j}}=\frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}}$ Hence, $\mathrm{dn}_{\mathrm{i}}=\mathrm{p}_{\mathrm{i}} \, \exp \left(-\frac{\mathrm{z}_{\mathrm{i}} \, \mathrm{e}}{\mathrm{k} \, \mathrm{T}} \, \frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}}\right) \, 4 \, \pi \, \mathrm{r}^{2} \, \mathrm{dr}$ Or [6], $\mathrm{dn}_{\mathrm{i}}=\mathrm{p}_{\mathrm{i}} \, \exp \left(-\frac{\mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}} \, \mathrm{e}^{2}}{4 \, \pi \, \mathrm{k} \, \mathrm{T} \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}}\right) \, 4 \, \pi \, \mathrm{r}^{2} \, \mathrm{dr}$ Using equation (e), the number of ions in a shell, thickness $\mathrm{dr}$ and distance $\mathrm{r}$ from the $j$ ion, at temperature $\mathrm{T}$ in a solvent having relative permittivity $\varepsilon_{\mathrm{r}}$ is obtained for ions with charge numbers $\mathrm{z}_{\mathrm{i}}$ and $\mathrm{z}_{j}$. For two ions having the same sign $\mathrm{dn}_{\mathrm{i}}$ increases with increase in $\mathrm{r}$, a pattern intuitively predicted. However for ions of opposite sign an interesting pattern emerges in which $\mathrm{dn}_{\mathrm{i}}$ decreases with increase in $\mathrm{r}$, passes through a minimum and then increases. In other words there exists a distance $\mathrm{q}$ at which there is a minimum in the probability of finding a counterion. Thus [7] $\mathrm{q}=\frac{\left|\mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}}\right| \, \mathrm{e}^{2}}{8 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T}}$ For a given salt, $\mathrm{q}$ increases with decrease in $\varepsilon_{\mathrm{r}}$ at fixed $\mathrm{T}$. Bjerrum suggested that the term ‘ion pair’ describes two counter ions where their distance apart is less than $\mathrm{q}$ [8]. In other words the proportion of a given salt in solution in the form of ions pairs increases with decrease in $\varepsilon_{\mathrm{r}}$. The interplay between solvent permittivity and ion size $\mathrm{a}_{j}$ as determined by the sum of cation and anion radii is important. For a fixed $\mathrm{a}_{j}$, the fraction of ions present as ion pairs increases with decrease in relative permittivity of the solvent. Thus high $\varepsilon_{\mathrm{r}}$ favours description of a salt as present as only ‘free’ cations and anions. The properties of such a real solutions might therefore be described using the Debye-Huckel Limiting Law. By way of contrast as $\varepsilon_{\mathrm{r}}$ decreases the extent of ion pair formation increases with decrease in ion size [9]. Ion Association The fraction of salt in solution $\theta$ in the form of ion pairs is given by the integral of equation (e) within the limits $\mathrm{a}$ and $\mathrm{q}$ where $\mathrm{a}$ is the distance of closest approach of cation and anion. Thus $\theta=4 \, \pi \, p_{i} \, \int_{a}^{q} \exp \left(-\frac{z_{+} \, z_{-} \, e^{2}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{r} \, k \, T \, r}\right) \, r^{2} \, d r$ Hence [10], for a solution where the concentration of salt $\mathrm{c}_{j}$ expressed using the unit, $\mathrm{mol dm}^{-3}$, $\theta$ is given by equation (h). $\theta=\frac{4 \, \pi \, N}{10^{3}} \,\left(\frac{\left|\mathrm{z}_{+} \, \mathrm{z}_{-} \, \mathrm{e}^{2}\right|}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T}}\right)^{3} \, \mathrm{Q}(\mathrm{b})$ where $Q(b)=\int_{2}^{b} x^{-4} \, e^{x} \, d x$ with $\mathrm{b}=\frac{\left|\mathrm{z}_{+} \, \mathrm{z}_{-}\right| \, \mathrm{e}^{2}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T} \, \mathrm{a}}$ and $\mathrm{x}=-\frac{\mathrm{z}_{+} \, \mathrm{z}_{-} \, \mathrm{e}^{2}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T} \, \mathrm{r}}$ The integral $\mathrm{Q}(\mathrm{b})$ has been tabulated as a function of $\mathrm{b}$ [1,10]. According to equation (h), $\theta$ increases with increase in $\mathrm{b}$; i.e. with increase in $\mathrm{a}$ and decrease in $\varepsilon_{\mathrm{r}}$. Ion Pair Association Constants The analysis leading to equation (h) is based on concentrations of salts in solution. Therefore the equilibrium between ions and ion pairs is described using concentration units. Here we consider the case of a 1:1 salt (e.g $\mathrm{Na}^{+} \mathrm{Cl}^{-}$) in the form of the following equilibrium describing the dissociation of ion pairs. [A common convention in this subject is to consider ‘dissociation’.] For a 1:1 salt $j$ in solution the chemical potential $\mu_{j}(s \ln )$ is given by equation (l). $\mu_{\mathrm{j}}(\mathrm{s} \ln )=\mu_{\mathrm{j}}^{0}(\mathrm{~s} \ln )+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{c}_{\mathrm{j}} \, \mathrm{y}_{\pm} / \mathrm{c}_{\mathrm{r}}\right)$ The mean ionic activity coefficient (concentration scale) is defined by equation (m) $\operatorname{limit}\left(c_{j} \rightarrow 0\right) y_{\pm}=1.0 \text { at all T and } p$ The thermodynamic properties of the neutral (dipolar) ion pair are treated as ideal. Then, $\mu_{\mathrm{ip}}(\mathrm{s} \ln )=\mu_{\mathrm{ip}}^{0}(\mathrm{~s} \ln )+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{c}_{\mathrm{ip}} / \mathrm{c}_{\mathrm{r}}\right)$ The equilibrium between ‘free’ ions (i.e. salt $j$) and ion pairs is described by the following equation. $\mathrm{M}^{+} \mathrm{X}^{-}(\mathrm{s} \ln ) \Leftarrow \Rightarrow \mathrm{M}^{+}(\mathrm{s} \ln )+\mathrm{X}^{-}(\mathrm{s} \ln )$ Then, $\mu_{i p}(s \ln )=\mu_{j}(s \ln )$ Hence the ion pair dissociation constants $\mathrm{K}_{\mathrm{D}}$ is given by equation (q). $\Delta_{\text {diss }} G^{0}=-R \, T \, \ln \left(K_{D}\right)$ where $\Delta_{\text {diss }} G^{0}=\mu_{j}^{0}(s \ln )-\mu_{\mathrm{ip}}^{0}(\mathrm{~s} \ln )$ Hence, $K_{D}=\frac{\left(c_{j} \, y_{\pm} / c_{r}\right)^{2}}{\left(c_{i p} / c_{r}\right)}$ But $\mathrm{c}_{\mathrm{j}}=\theta \, \mathrm{c}_{\mathrm{s}}$ and $\mathrm{c}_{\mathrm{ip}}=(1-\theta) \, \mathrm{c}_{\mathrm{s}}$ where $\mathrm{c}_{\mathrm{s}}$ is the total concentration of salt $\mathrm{M}^{+} \mathrm{X}^{-}$. Then, $\mathrm{K}_{\mathrm{D}}=\frac{\theta^{2} \, \mathrm{y}_{\pm}^{2} \, \mathrm{c}_{\mathrm{s}}}{(1-\theta) \, \mathrm{c}_{\mathrm{r}}}$ $\mathrm{K}_{\mathrm{D}}$ is dimensionless. The long-established convention in this subject defines a quantity $\mathrm{K}_{\mathrm{D}}^{\prime}$. Thus $\mathrm{K}_{\mathrm{D}}^{\prime}=\frac{\theta^{2} \, \mathrm{y}_{\pm}^{2} \, \mathrm{c}_{\mathrm{s}}}{(1-\theta)}$ For very dilute solutions, the assumption is made that $\theta = 1$ and $\mathrm{y}_{\pm} = 1$. Hence using equation (h), \begin{aligned} &\frac{1}{\mathrm{~K}_{\mathrm{D}}^{\prime}} \cong \frac{1-\theta}{\mathrm{C}_{\mathrm{S}}} \ &\theta=\frac{4 \, \pi \, N}{10^{3}} \,\left(\frac{\left|\mathrm{z}_{+} \, \mathrm{z}_{-}\right| \, \mathrm{e}^{2}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T}}\right)^{3} \, \mathrm{Q}(\mathrm{b}) \end{aligned} Conductivities of Salt Solutions The molar conductances of salt solutions at fixed $\mathrm{T}$ and $\mathrm{p}$ can be precisely measured. To a first approximation the molar conductance of a given solution offers a method of counting the number of free ions. For salt solutions in solvents of low permittivity the molar conductance offers a direct method for assessing the fraction of salt present as free ions and hence the fraction present as ion pairs. Hence electrical conductivities of salt solutions in solvents of low relative permittivity have been extensively studied in order to probe the phenomenon of ion pair formation. The classic study was reported [10] by Fuoss and Kraus in 1933 who studied the electrical conductivities of tetra-iso-amylammonium nitrate in dioxan + water mixtures [11] at $298.15 \mathrm{~K}$ over the range $2.2 \leq \varepsilon_{\mathrm{r}} \leq 78.6$. The dependence of measured dissociation constants followed the pattern required by Bjerrum’s theory. Following the publication of the study by Fuoss and Kraus [10], many papers were published confirming the general validity of the Bjerrum ion-pair model. We note below a few examples of these studies which lead in turn to developments of the theory. For example in solvents of very low relative pemitivities triple ions are formed of the ++- and +-- type [12,13]. Many experimental techniques have been used to support the Bjerrum model; e.g. cryoscopic studies [14], electric permittivities of solutions [15,16] and Wien effects [17]. Following the Bjerrum model, other models were suggested and developed. Denison and Ramsey [18] suggested that the term ‘ion pair’ describes ions in contact, all other ions being free. Sadek and Fuoss [19] proposed that association of free ions to form contact ion ion pairs involved formation of solvent separated ion pairs, although they later withdrew the proposal[20]. Gilkerson [21] modified equations describing ion-pair formation to include parameters describing ion-solvent interaction. In 1957 Fuoss [22] restricted the definition of the term ‘ion pair’ to ions in contact. The dipolar nature of an ion pair was confirmed by dielectric relaxation studies [23,24]. In the development of theories of ion pair formation Hammett notes the models of ion pair formation which involve charged spheres in a continuous dielectric may only be relevant under especially favourable circumstances [25]. General Comments The initial proposal by Bjerrum concerning ion pair formation has had an enormous impact in many branches of chemistry including mechanistic organic chemistry [26,27]. Spectroscopic studies identified ion-pairs in solution using charge transfer to solvent spectra [28]. Electron spin resonance identified the presence of ion pairs in solution. Particularly interesting are those solutions where the counterion hops between two sites in an organic radical anion [29]. Returning to the context of thermodynamics, the Bjerrum model of ion association has been extended to descriptions of partial molar volumes [30], apparent molar heat capacities and compressibilities of salts in non-aqueous solutions including cyanomethane [31]. Nevertheless the debate concerning ion association in solution has continued particularly with the development of statistical thermodynamic treatments of salt solutions. Grunwald [32] comments on the debate. To some extent the question arises as to the extent to which formation of ions pairs is either assumed from the outset or emerges from a given theoretical model for a salt solution. Footnotes [1] N. Bjerrum, K. Danske Vidensk Selskab, 1926,7, No. 9. [2] For more recent accounts see— 1. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths, London, 2nd ed. Revised,1965, chapter 14. 2. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions, Reinhold, New York, 2nd. edition, revised and enlarged, 1950, section 3-7. 3. S. Glasstone, An Introduction to Electrochemistry, D. Van Nostrand, New York, 1942. 4. J. O’M. Bockris and A. J K. N. Reddy, Modern Electrochemistry: Ionics, Plenum Press, New York, 2nd. edn.,1998,chapter 3. [3] $\begin{gathered} \frac{\mathrm{z}_{\mathrm{i}} \, \mathrm{e} \, \psi_{\mathrm{j}}}{\mathrm{k} \, \mathrm{T}}=\frac{[1] \,[\mathrm{C}] \,[\mathrm{V}]}{\left[\mathrm{J} \mathrm{K}^{-1}\right] \,[\mathrm{K}]}=\frac{[\mathrm{A} \mathrm{s}] \,\left[\mathrm{J} \mathrm{A} \mathrm{s}^{-1}\right]}{[\mathrm{J}]}=[1] \ \mathrm{p}_{\mathrm{i}}=\left[\mathrm{m}^{-3}\right] \quad \mathrm{p}_{\mathrm{i}}^{\prime}=\left[\mathrm{m}^{-3}\right] \end{gathered}$ [4] $\mathrm{p}_{\mathrm{i}} \, \exp \left(-\mathrm{z}_{\mathrm{i}} \, \mathrm{e} \, \Psi_{\mathrm{j}} / \mathrm{k} \, \mathrm{T}\right) \, 4 \, \pi \, \mathrm{r}^{2} \, \mathrm{dr}=\frac{1}{\left[\mathrm{~m}^{3}\right]} \,[1] \,[1] \,\left[\mathrm{m}^{3}\right]=[1]$ [5] $\psi_{\mathrm{j}}=\frac{[1] \,[\mathrm{C}]}{[1] \,[1] \,\left[\mathrm{F} \mathrm{m}^{-1}\right] \,[1] \,[\mathrm{m}]}=\frac{[\mathrm{A} \mathrm{s}]}{\left[\mathrm{As} \mathrm{V}^{-1}\right]}=[\mathrm{V}]$ [6] \begin{aligned} &\left(-\frac{\mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}} \, \mathrm{e}^{2}}{4 \, \pi \, \mathrm{k} \, \mathrm{T} \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}}\right) \ &=\frac{[1] \,[1] \,[\mathrm{C}]^{2}}{[1] \,[1] \,\left[\mathrm{J} \mathrm{K}^{-1}\right] \,[\mathrm{K}] \,\left[\mathrm{Fm}^{-1}\right] \,[1] \,[\mathrm{m}]}=\frac{[\mathrm{As}]^{2}}{[\mathrm{~J}] \,[\mathrm{F}]} \ &=\frac{[\mathrm{As}]^{2}}{[\mathrm{~J}] \,\left[\mathrm{As} \mathrm{A} \mathrm{s} \mathrm{J}^{-1}\right]}=[1] \end{aligned} Hence, $\mathrm{p}_{\mathrm{j}}=\left[\mathrm{mol} \mathrm{m}^{-3}\right] \,[1] \,[1] \,[1] \,\left[\mathrm{m}^{2}\right] \,[\mathrm{m}]=[1]$ [7] $\mathrm{q}=\frac{[1] \,[1] \,[\mathrm{C}]^{2}}{[1] \,[1] \,\left[\mathrm{Fm}^{-1}\right] \,[1] \,\left[\mathrm{J} \mathrm{K}^{-1}\right] \,[\mathrm{K}]}=\frac{[\mathrm{A} \, \mathrm{s}]^{2}}{\left[\mathrm{~A}^{2} \mathrm{~s}^{2} \mathrm{~J}^{-1} \mathrm{~m}^{-1}\right] \,[\mathrm{J}]}=[\mathrm{m}]$ [8] Distance $\mathrm{q}$ corresponds to the distance where $\frac{\left|\mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}}\right| \, \mathrm{e}^{2}}{4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{q}}=2 \, \mathrm{k} \, \mathrm{T}$ [9] We stress the distinction between association of cation $\mathrm{M}^{+}$ and anion $\mathrm{X}^{-}$ to form an ion pair and association in solution of $\mathrm{H}^{+}$ and $\mathrm{CH}_{3}\mathrm{COO}^{-}$ ions to form undissociated ethanoic acid. In the later case the cohesion is discussed in quantum mechanical terms. [10] R. M. Fuoss and C. A. Kraus, J. Am. Chem. Soc.,1933,55,1019. [11] The liquid mixture dioxan + water is notable for being completely miscible and ambient $\mathrm{T}$ and $\mathrm{p}$, the relative permitivities having a remarkable range. No other water + organic liquid offers such a range. [12] R. M. Fuoss and C. A. Kraus, J. Am. Chem. Soc.,1933,55,2387. [13] R. M. Fuoss, Chem. Rev.,1935,17,227. [14] F. M. Batson and C. A. Kraus, J. Am. Chem. Soc.,1934,56,2017. [15] G. S. Hooper and C. A. Kraus, J. Am. Chem. Soc.,1934,56,2265. [16] R. M. Fuoss, J.Am.Chem.Soc.,1934,56,1031. [17] D. J. Mead and R. M. Fuoss, J. Am. Chem.Soc.,1939,61,2047. [18] J. T. Denison and J. B. Ramsey, J.Chem.Phys.,1950,18,770. [19] H. Sadek and R. M. Fuoss, J. Am. Chem.Soc.,1954,76,5897,5902,5905. [20] H. Sadek and R. M. Fuoss, J. Am. Chem. Soc.,1959,81,4511. [21] W. Gilkerson, J. Chem. Phys.,1956,25,1199. [22] R. M. Fuoss, J. Am. Chem. Soc.,1957,79,3304. [23] A. H. Sharbaugh, H. C. Eckstrom and C. A. Kraus, J. Chem, Phys.,1947.15,54. [24] G. Williams, J. Phys.Chem.,1959,63,532. [25] L. P. Hammett, Physical Organic Chemistry, McGraw-Hill, New York, 2nd. edition, 1970. [26] See for example, S. Winstein, P. E.Klinedinst and E. Clippinger, J Am. Chem. Soc., 1961,83,4986. [27] E. A. Moelwyn-Hughes, Chemical Statics and Kinetics of Solutions, Academic Press, London, 1971,p.405. [28] For further references see M. J. Blandamer, and M. F. Fox., Chem. Revs., 1970.70, 59. [29] T. A. Claxton, J. Oakes and M. C. R. Symons, Trans. Faraday Soc., 1968, 64, 596. [30] J.-F. Cote, J. E. Desnoyers and J-C. Justice, J Solution Chem.,1996,25,113. [31] J.-F. Cote and J. E. Desnoyers, J. Solution Chem..,1999,28,395. [32] E. Grunwald, Thermodynamics of Molecular Species, Wiley, New York, 1997,chapter 12.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.16%3A_Ion_Interactions/1.16.1%3A_Ion_Association.txt
A classic subject in physical chemistry concerns the electric conductivities of salt solutions, most interest being centred on the aqueous salt solutions. Although the electric conductivities of these systems, transport properties, do not come under the heading ‘thermodynamic properties’, these conductivities have played a major part in the task of understanding the thermodynamic properties of salt solutions. Generally however interest in the electric conductivities of salt solutions has waned as spectroscopic properties in all its form have moved to a dominant position in physical chemistry. Nevertheless the contributions made by research into the electric properties of salt solutions have been and remain enormously important. Conductivities At this point there is merit in commenting on the technique, mass spectrometry. In this important experimental technique, ions are produced in an ion source and then subjected to an electric field gradient, where (usually) cations are accelerated. The ions pass through a magnetic field, the path of a given ion depending on the charge and mass of the ion. Descriptions of the electrical conductivities of salt solutions start out from a quite different basis. To understand the point we consider a reasonably concentrated aqueous solution of sodium chloride; i.e. $0.1 \mathrm{~mol dm}^{-3} \equiv 0.1 \mathrm{~mol}$ salt in water, mass $1 \mathrm{~kg} \equiv 0.1 \mathrm{~mol}$ salt in $(1.0/0.018) \mathrm{~mol} \text{ water } \equiv 0.1 \mathrm{~mol Na}^{+} \text { ions } + 0.1 \mathrm{~mol Cl}^{-} \text { ions } + 55.6 \mathrm{~mol} \text { water}(\ell)$. In other words, for every sodium ion there are 556 molecules of water in this aqueous solution. The contrast with the mass spectrometer experiment could not be more dramatic. Further in conventional experiments studying the electric conductivities of salt solutions, the effect of a modest electric potential gradient is simply to bias the otherwise Brownian motion of the ion in a direction depending on the sign of the charge on a given ion. As each ion makes its way through the solution it is jostled and impeded by the large number of solvent molecules. Nevertheless in theoretical treatments of the electric conductivities of salt solutions the theory envisages a slow direct progress through the solution, in the case of, for example, a cation down the electric potential gradient. The key experimental fact is that the electric properties of salt solutions at low electric currents and low electric potential gradients obey the phenomenological law, Ohm’s Law. Deviations from this law are observed for example at high electrical field gradients; e.g. Wien Effects. Molar Conductivities The key term in the context of the electric conductivities of a salt solution, concentration of salt $\mathrm{c}_{j}$ is the molar conductivity $\Lambda$ defined by equation (a) where $\kappa$ is the electrolytic conductivity [1,2]. $\Lambda=\kappa / \mathrm{c}_{j}$ For a salt solution prepared using a 1:1 salt , the molar conductivity can be expressed as the sum of ionic conductivities , $\lambda_{+}$ and $\lambda_{-}$. Thus $\Lambda=\lambda_{+}+\lambda_{-}$ Using equation (a), the electrolytic conductivity $\kappa$ is related to the ionic conductivities using equation (c) $\kappa=\mathrm{c}_{\mathrm{j}} \,\left(\lambda_{+}+\lambda_{-}\right)$ The electric mobility of a given ion, $\mathrm{u}_{j}$ is related to the mobility $v_{j}$ using equation (d) [3]. $\mathrm{u}_{\mathrm{j}}=\mathrm{v}_{\mathrm{j}} / \mathrm{E}$ Footnotes [1] \begin{aligned} \kappa &=(\text { electric current density) } / \text { (electric field strength }) \ &=[\mathrm{j}] /[\mathrm{E}] \ &=\left[\mathrm{A} \mathrm{m}^{-2}\right] /\left[\mathrm{V} \mathrm{m}^{-1}\right]=\left[\mathrm{S} \mathrm{m}^{-1}\right] \end{aligned} [2] $\Lambda=\left[\mathrm{S} \mathrm{m}^{-1}\right] /\left[\mathrm{mol} \mathrm{m}^{-3}\right]=\left[\mathrm{S} \mathrm{} \mathrm{m}^{2} \mathrm{~mol}^{-1}\right]$ [3] $\mathrm{u}_{\mathrm{j}}=\left[\mathrm{m} \mathrm{s}^{-1}\right] /\left[\mathrm{V} \mathrm{m}^{-1}\right]=\left[\mathrm{m}^{2} \mathrm{~s}^{-1} \mathrm{~V}^{-1}\right]$ 1.16.3: Ionic Strength: Ional Concentration The ionic strength of a salt solution I containing $\mathrm{i}$-ionic substances is defined by equation (a); $\mathrm{m}_{j}$ is the molality of ionic substance-$j$, charge number $\mathrm{z}_{j}$ [1]. $\mathrm{I}=(1 / 2) \, \sum_{\mathrm{j}=\mathrm{i}}^{\mathrm{j}=\mathrm{i}} \mathrm{m}_{\mathrm{j}} \, \mathrm{z}_{\mathrm{j}}^{2}$ The sum is taken over all $\mathrm{i}$-ionic substances in the solution. The situation is slightly complicated by the fact some authors use the term ‘ionic strength’ where the concentration $\mathrm{c}_{j}$ (expressed using the unit, $\mathrm{mol dm}^{-3}$) replaces $\mathrm{m}_{j}$. The substitution is reasonably satisfactory for dilute salt solutions at ambient $\mathrm{T}$ and $\mathrm{p}$ where the mass of water, volume $1 \mathrm{~dm}^{3}$, is approx. $1 \mathrm{~kg}$. The ional concentration of a salt solution $\Gamma$ is defined by equation (b) where $\mathrm{c}_{j}$ is expressed using the unit, $\mathrm{mol dm}^{-3}$ [2]. $\Gamma=\sum_{j=i}^{j=i} c_{j} \, z_{j}^{2}$ Footnotes [1] An aqueous solution contains $\mathrm{K}_{2}\mathrm{SO}_{4}$ ($0.1 \mathrm{~mol}$) in $1 \mathrm{~kg}$ of water($\ell$). $\mathrm{m}\left(\mathrm{K}_{2} \mathrm{SO}_{4}\right)=0.1 \mathrm{~mol} \mathrm{~kg}^{-1} ; \mathrm{m}\left(\mathrm{K}^{+}\right)=0.2 \mathrm{~mol} \mathrm{~kg}^{-1}$ and $\mathrm{m}\left(\mathrm{SO}_{4}{ }^{2-}\right)=0.1 \mathrm{~mol} \mathrm{~kg}^{-1}$ Hence $\mathrm{I}=(1 / 2) \,\left[(0.2)+\left(2^{2} \mathrm{X} 0.1\right)\right]=0.3 \mathrm{~mol} \mathrm{~kg}^{-1}$ [2] H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions, , Reinhold, New York, 2nd. revised edn.1950, p.33.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.16%3A_Ion_Interactions/1.16.2%3A_Ionic_Mobilities%3A_Aqueous_Solutions.txt
The seminal paper in this subject was published in 1933 by Bernal and Fowler [1]. These authors drew attention to the possible impact of ions on water - water interactions in aqueous solutions beyond nearest-neighbour water molecules. The paper is also notable for the fact that the authors in their examination of the properties of ice, water and salt solutions did not use the term ‘hydrogen bond’ [2]. At that time there was much debate concerning the nature of this interaction bearing in mind that the ‘valency’ of hydrogen is unity. Rather Bernal and Fowler concluded that ‘the unique properties of water are due to a structure of an extended complex characterized by tetrahedral co-ordination’. Verwey drew attention to the importance of the interactions between an ion and near-neighbor water molecules in aqueous solution [3,4], these water molecules often being described as ‘electrostricted ‘ by strong ion- solvent dipole interactions. Solvent water plays an important role in the control of partial molar entropies of ions in aqueous solutions [5] and transfer entropies of ions from aqueous to non-aqueous solvents [6]. A major landmark was a paper published by Frank and Evans who developed the concept that solutes, polar and apolar, have an important impact on water - water interactions in aqueous solutions [7]. In the development of models for ionic hydration, a distinction is drawn between hydrophilic and hydrophobic ions. Hydrophilic ions (alkali metal cations and halide anions) have strong attractive interactions with neighbouring dipolar water molecules. Neutron scattering data reveal important information concerning the arrangement of water molecules contiguous to ions [8-10]. For example in the case of chloride ions, the $\mathrm{Cl} – \mathrm{~H} - 0$ configuration is essentially linear. Nevertheless, there is clear evidence, albeit often secondary, that strong water - ion interactions have an impact on water - water interactions beyond the immediate hydration sheath. Viscosity data indicate that ions, such as iodide and potassium, have a structure breaking effect. The cospheres, for these ions, are drawn, showing two parts [11,12]; an inner zone A and an outer zone B [11]. In zone A, ion - water dipole interactions are strong, leading to the general description ionichydration [12]. An indication of the structure of hydrated ions in solution emerges from X-ray crystallographic studies [13]. In the case of $\mathrm{KF}.4\mathrm{H}_{2}\mathrm{O}$, the structure comprises $\mathrm{K}^{+} \left(\mathrm{H}_{2}\mathrm{O}\right)_{6}$ and $\mathrm{F}^{-} \left(\mathrm{H}_{2}\mathrm{O}\right)_{6}$ octahedra; the $\mathrm{K}^{+} -\mathrm{O}$ distance is $0.279 \mathrm{~nm}$. Kebarle showed that mass spectrometry could be used to study ion - water interactions and, interestingly, step-wise hydration in the gas phase [14]. If a given ion in aqueous solution is indeed surrounded by two zones identified as zones A band B, the expectation is that ion - ion interactions in solution will reflect the impact of these structural features [15]. In the context of the impact of zone B, the suggestion was that with increase in size of ions so zone B should increase. Hence the expectation was that, for example, the partial molar isobaric heat capacity of tetra-n-butylammonium bromide in aqueous solution would be large in magnitude and negative in sign. Such not the case; the sign is positive [16,17]. A link was therefore established between the hydration characteristic of tetra-alkylammoniun ions and the structures of the corresponding salt hydrates [18]; e.g. tetra-iso-amylammonium fluoride hydrate, $(\text{iso}-\mathrm{Am})_{4}\mathrm{N}^{+} \mathrm{~F}^{-} 38 \mathrm{~H}_{2}\mathrm{O}$ [19]. Generally, therefore, tetra-alkylammonium ions of $\mathrm{C}_{4} - \mathrm{~C}_{9}$ carboxylates and tri-alkylsulphonium ions are often identified as hydrophobic where the interaction between these ions and neighboring water molecules is weak. [20-32]. Interestingly, constricting the alkyl chains to form azoniaspiroalkane cations diminishes the hydrophobic character [33,34]. The impact of replacing a hydrophobic terminal group in $\mathrm{R}_{4}\mathrm{N}^{+}$ ions by a hydrophilic group on the properties of aqueous solutions is dramatic and offers an interesting insight into the role of ion – water interactions [35]. In contrast Finney and co-workers report that neutron diffraction data for aqueous solutions, containing $\mathrm{Me}_{4}\mathrm{N}^{+} \mathrm{~Cl}^{-}$, show no evidence for increased ice-like structure compared to pure water [36]. Nevertheless thermodynamic and transport properties generally point to the conclusion that the ion $\mathrm{Me}_{4} \mathrm{~N}^{+}$ does not promote near-neighbor water-water hydrogen bonding [37,38]. Footnotes [1] J. D. Bernal and R. H. Fowler, J.Chem.Phys.,1933,1,515. [2] See for example, P. A. Kollman and L. C. Allen, Chem. Rev.,1972, 72,283. [3] E. J. Verwey, Rec. Trav. Chim.,1942,61,127. [4] See also F. Vaslow, J. Phys.Chem.,1973,67,2773. [5] C. M. Criss, J. Phys. Chem.,1974,78,1000. [6] K. K. Kundu, Pure Appl. Chem.,1994,66,411. [7] H. S. Frank and M. W. Evans, J. Chem. Phys.,1945,13,507. [8] J. E. Enderby, Chem.Soc.Rev.,1995,24,159. [9] G. W. Neilson and J. E. Enderby, Adv.Inorg.Chem.,1989,34,195; and references therein. [10] J. E. Enderby and G.W. Neilson in Water A Comprehensive Treatise; ed.F. Franks, Plenum Press, New York, 1973,volume 6, chapter 1. [11] H. S. Frank and W.-Y. Wen, Discuss Faraday Soc.,1957,24,756. [12] W.-Y.Wen,in Ions and Molecules in Solution; ed. N. Tanaka, H. Otaki and R. Tamamushi, Elsevier, Studies in Physical and Theoretical Chemistry, Amsterdam, 1983, p.45. [13] G. Beurskens and G. A. Jeffrey, J.Chem.Phys.,1964,41,917 and 924. [14] P. Kebarle in Modern Aspects of Electrochemistry, ed.B.E.Conway and J. O’M. Bockris, 1974,9,1; and references therein [15] H. S. Frank, J. Phys Chem.,1963,67,1554. [16] E. M. Arnett and J. J. Campion, J.Am.Chem.Soc.,1970,92,7097. [17] K. Tamaki, S. Yoshikawa and M. Kushida, Bull. Chem. Soc. Jpn., 1975,48,3018. [18] G. A. Jeffrey, Prog. Inorg. Chem., 1967,8,43; and references therein [19] D. Feil and G. A. Jeffrey, J.Chem.Phys.,1961,35,1863. [20] C. Shin, I. Worsley and C.M. Criss, J. Solution Chem.,1976,5,867. [21] S. Lindebaum, J. Phys. Chem.,1971,75,3733; and references therein [22] P.-A. Leduc and J. E. Desnoyers, Can. J. Chem.,1973,51,2993. [23] S. Lindenbaum, J.Chem.Thermodyn.,1971,3,625. [24] A. H. Narten and S. Lindenbaum., J. Chem. Phys.,1969,51,1108. [25] R. H. Boyd, J. Chem.Phys.,1969,51,1470. [26] O. D. Bonner and C. F. Jumper, Infrared Physics, 1973,13,233. [27] R. L. Kay, Adv. Chem. 1968, 73,1 [28] T. S. Sarma and J. C. Ahluwalia, J. Phys. Chem.,1970,74,3547. [29] T. S. Sarma, R. K. Mohanty and J. C. Ahluwalia, Trans. Faraday Soc.,1969, 65,2333. [30] D. A. Johnson and J. F. Martin, J. Chem. Soc. Dalton, Trans.,1973,1585. [31] T. S. Sunder, B. Chawla and J. C. Ahluwalia, J. Phys. Chem.,1974,78, 738. [32] E. M. Arnett, M. Ho and L. L. Schaleyer, J.Am. Chem. Soc.,1970,93,77039. [33] A. LoSurdo, W.-Y. Wen, C. Jolicoeur and J.-L.Fortier,J.Phys.Chem.,1977, 81, 1813;and references therein. [34] W.-Y. Wen and S. Saito, J.Phys.Chem.,1965,69,3569. [35] G.P. Cunningham, D.F. Evans and R.L. Kay, J.Phys.Chem.,1966,70,3998. [36] J. Turner, A.K. Soper and J.L. Finney, Mol.Phvs.,1992,77,411. [37] R. L. Kay, D. F. Evans and M.A. Matesich, in Solute-Solute Interactions, ed. J. F. Coetzee and C. D. Ritchie, Marcel Dekker, New York, 1976,voume.2, chapter 10. [38] R. L. Kay, Water A Comprehensive Treatise, ed. F. Franks, Plenum Press, New York, 1973, volume 3, chapter 4.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.16%3A_Ion_Interactions/1.16.4%3A_Ion-Water_Interactions.txt
The term ‘adiabatic’ means that for a closed system no heat passes between system and surroundings; $\mathrm{q} = 0$. The term ‘isentropic’ introduces the further constraint that the system remains at equilibrium with the surroundings; i.e. the affinity for spontaneous change is zero. From the Second Law, $\mathrm{T} \, \mathrm{dS}=\mathrm{q}+\mathrm{A} \, \mathrm{d} \xi \quad \text { where } \mathrm{A} \, \mathrm{d} \xi \geq 0$ The isentropic condition means that both $\mathrm{A}$ and $\mathrm{q}$ are zero. Hence $\mathrm{dS}$ is zero, indicating that the entropy of the system remains constant. In other words, ‘isentropic’ describes an adiabatic change along an equilibrium and therefore reversible pathway. 1.17.2: Isentropic Thermal Pressure Coefficient The volume of a given closed system is defined by the following set of independent variables where $\xi$ is the general composition variable. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \xi^{\mathrm{eq}} ; \mathrm{A}=0\right]$ We have rather over-defined the system. The aim is to identify the composition variable at equilibrium and the condition that the affinity for spontaneous change is zero. The dependent variable entropy for this system is defined in analogous fashion; equation (b). $\mathrm{S}=\mathrm{S}\left[\mathrm{T}, \mathrm{p}, \xi^{\mathrm{eq}} ; \mathrm{A}=0\right]$ The system is perturbed by a change in temperature along a path for which the affinity for spontaneous change is zero. Moreover the entropy of the system remains the same as that given in equation (b). In order to hold the latter condition the equilibrium pressure must change. In the state defined by the independent variables $\left[\mathrm{T}, \mathrm{p}, \xi^{e q} ; \mathrm{A}=0\right]$ the (equilibrium) isentropic differential dependence of pressure $\mathrm{p}$ on temperature is the isentropic thermal pressure coefficient, $\beta_{\mathrm{S}}$; equation.(c). $\beta_{\mathrm{s}}=(\partial \mathrm{p} / \partial \mathrm{T})_{\mathrm{s}}$ Further [1] $\beta_{\mathrm{S}}=(\partial \mathrm{S} / \partial \mathrm{V})_{\mathrm{T}}$ Also [2], $\beta_{\mathrm{s}}=\sigma /\left(\mathrm{T} \, \alpha_{\mathrm{p}}\right)$ Here $\sigma$ is the isobaric heat capacity for unit volume (heat capacitance) of the system, $\mathrm{C}_{\mathrm{p}} / \mathrm{~V}$. The three isentropic properties $\alpha_{\mathrm{S}}$, $\kappa_{\mathrm{S}}$ and $\beta_{\mathrm{S}}$ are related using equation (f); [3]. $\beta_{\mathrm{s}}=-\alpha_{\mathrm{s}} / \kappa_{\mathrm{s}}$ With reference to the (equilibrium) thermal expansivity, $\alpha_{\mathrm{S}}$, we envisage that the temperature is changed to produce a change in volume along a path for which the entropy remains the same as in equation (b) and the affinity for spontaneous change remains at zero. $\alpha_{\mathrm{s}}(\mathrm{A}=0)=\frac{1}{\mathrm{~V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s} ; \mathrm{A}=0}$ In analogous fashion, $\kappa_{\mathrm{S}}$ is a measure of the change in volume produced by a change in pressure. $\kappa_{\mathrm{S}}(\mathrm{A}=0)=-\frac{1}{\mathrm{~V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{S} ; \mathrm{A}=0}$ Footnotes [1] From $\left[\frac{\partial}{\partial \mathrm{p}}\left(\frac{\partial \mathrm{H}}{\partial \mathrm{S}}\right)_{\mathrm{p}}\right]_{\mathrm{s}}=\left[\frac{\partial}{\partial \mathrm{S}}\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{s}}\right]_{\mathrm{p}}$ But at equilibrium where $\mathrm{A}=0$, $T=\left(\frac{\partial H}{\partial S}\right)_{P}$ and $\mathrm{V}=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{s}}$ Then $\left(\frac{\partial \mathrm{T}}{\partial \mathrm{p}}\right)_{\mathrm{s}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{S}}\right)_{\mathrm{p}}$ . From $\beta_{\mathrm{s}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}}$, Using the above Maxwell Relation, $\beta_{\mathrm{S}}=\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{T}}$ [2] From the definition, $\beta_{\mathrm{s}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}}$ Using a calculus operation $\beta_{\mathrm{s}}=-\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{S}}\right)_{\mathrm{T}}$ From the Gibbs - Helmholtz Equation, $\left(\frac{\partial S}{\partial T}\right)_{p}=\frac{C_{p}}{T}$ From a Maxwell equation, $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$. Then $\beta_{\mathrm{s}}=\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{T}} \, \frac{1}{\mathrm{E}_{\mathrm{p}}}$ But $E_{p}=V \, \alpha_{p}$ Then, $\beta_{\mathrm{s}}=\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{V}} \, \frac{1}{\mathrm{~T} \, \alpha_{\mathrm{p}}}$ Or, $\beta_{\mathrm{s}}=\sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ [3] From the definition, $\beta_{\mathrm{s}}=\left(\frac{\partial p}{\partial T}\right)_{\mathrm{s}}$, then, $\beta_{\mathrm{s}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}}\right)_{\mathrm{s}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{s}}$ Then, $\beta_{\mathrm{S}}=-\mathrm{E}_{\mathrm{S}} / \mathrm{K}_{\mathrm{s}}=-\left(\mathrm{E}_{\mathrm{S}} / \mathrm{V}\right) /\left(\mathrm{K}_{\mathrm{s}} / \mathrm{V}\right)=-\alpha_{\mathrm{S}} / \kappa_{\mathrm{S}}$ Also from [2] and [3], $\mathrm{E}_{\mathrm{s}} / \mathrm{K}_{\mathrm{s}}=-\frac{\mathrm{C}_{\mathrm{p}}}{\mathrm{V}} \, \frac{1}{\mathrm{~T} \, \alpha_{\mathrm{p}}}$ Then $\alpha_{\mathrm{s}} / \kappa_{\mathrm{s}}=-\sigma / \mathrm{T} \, \alpha_{\mathrm{p}}$ 1.17.3: Iso-Variables Isobaric: A given system is held at constant pressure. Isothermal: A given system is held at constant temperature. Isochoric: A given closed system is held at constant volume. Isentropic: This condition, linked to the adiabatic constraint, requires that during a reversible change the entropy of a system remains constant in a particular thermodynamic process; e.g. compression. Adiabatic + Reversible = Isentropic. We can find isentropic processes which are irreversible. In this case they are not adiabatic. Isolated System: The boundary insulates a given system from the surroundings . This is not really an iso-variable in the thermodynamic sense. Isoperibol: In the vast majority of calorimetric experiments, the surroundings and the reaction vessel (the system+ container) are at constant temperature. When the experiment is initiated the composition of the closed system changes resulting from, for example, chemical reaction, mixing of liquids….. The temperature of the closed system changes albeit by a small amount because the processes taking place in the calorimeter are either exo- or endo-thermic. A sensitive detector is used to measure the change in temperature of the system. In pedantic terms the system is not constrained to be ‘isothermal’. So the calorimeter being used in such an experiment is an isoperibol calorimeter.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.17%3A_Isentropic_and_Iso-Variables/1.17.1%3A_Isentropic.txt
A given closed system is characterised by a given intensive variable $\mathrm{X}$. In this section we have in mind an intensive property such as the relative permittivity of a liquid. The variable $\mathrm{X}$ may also refer to an equilibrium constant and related parameters such as the enthalpy of reaction, $\Delta_{\mathrm{r}}\mathrm{H}(\mathrm{T},\mathrm{p})$. In all cases we assert that the closed system is at thermodynamic equilibrium where the affinity for spontaneous change is zero. Thus we may define $\mathrm{X}$ for a given system in terms of the temperature and pressure. $\mathrm{X}=\mathrm{X}[\mathrm{T}, \mathrm{p}]$ The molar volume of the system is defined in analogous fashion. $\mathrm{V}_{\mathrm{m}}=\mathrm{V}_{\mathrm{m}}[\mathrm{T}, \mathrm{p}]$ Then $\mathrm{dV}_{\mathrm{m}}=\left(\frac{\partial \mathrm{V}_{\mathrm{m}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \, \mathrm{dT}+\left(\frac{\partial \mathrm{V}_{\mathrm{m}}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \, \mathrm{dp}$ In other words the dependence of molar volume on $\mathrm{T}$ and $\mathrm{p}$ is characterised by the partial derivatives $\left(\frac{\partial \mathrm{V}_{\mathrm{m}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ and $\left(\frac{\partial V_{m}}{\partial p}\right)_{T}$. With equation (b) and (c) in mind we return the intensive property $\mathrm{X}$ described in equation (a). The dependence of $\mathrm{X}$ on $\mathrm{T}$ and $\mathrm{p}$ is similarly characterized by the two partial derivatives, $\left(\frac{\partial X}{\partial T}\right)_{p}$ and $\left(\frac{\partial \mathrm{X}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$. A calculus operation yields an equation for the partial derivative $\left(\frac{\partial \mathrm{X}}{\partial T}\right)_{\mathrm{V}(\mathrm{m})}$. Thus $\left(\frac{\partial \mathrm{X}}{\partial \mathrm{T}}\right)_{\mathrm{V}(\mathrm{m})}=\left(\frac{\partial \mathrm{X}}{\partial \mathrm{T}}\right)_{\mathrm{p}}+\left(\frac{\partial \mathrm{X}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}(\mathrm{m})}$ The property $\left(\frac{\partial \mathrm{X}}{\partial T}\right)_{\mathrm{V}(\mathrm{m})}$ is the isochoric differential dependence of $\mathrm{X}$ on $\mathrm{T}$. Now (cf. equation (c)) volume $\mathrm{V}_{\mathrm{m}$ depends on $\mathrm{T}$. Hence to hold $\mathrm{V}_{\mathrm{m}}$ constant, the pressure has to change. In fact equation (c) is used to find the required change in pressure for a given change in $\mathrm{T}$; equation (e). $\mathrm{dp}=-\left(\frac{\partial \mathrm{V}_{\mathrm{m}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}_{\mathrm{m}}}\right)_{\mathrm{T}} \, \mathrm{dT}$ In other words the required change in pressure is determined by the equation of state for the system and is characteristic of the system, $\mathrm{T}$ and $\mathrm{p}$. For a given change in temperature, $\delta \mathrm{T}(\exp )$ there is a defined change in pressure, $\delta \mathrm{p}(\operatorname{def})$. The isochoric condition takes the following form granted that in the experiment we decide to change the temperature by an amount $\delta \mathrm{T}$. $\mathrm{V}_{\mathrm{m}}[\mathrm{T}, \mathrm{p}]=\mathrm{V}_{\mathrm{m}}[\mathrm{T}+\delta \mathrm{T}(\exp ) ; \mathrm{p}+\delta \mathrm{p}(\operatorname{def})]$ We now return to the property $\mathrm{X}$ defined in equation (a). We consider the property $\mathrm{X}$ at the two conditions highlighted in equation (f); $\mathrm{X}[\mathrm{T}, \mathrm{p}] ; \quad \mathrm{X}[\mathrm{T}+\delta \mathrm{T}(\exp ) ; \mathrm{p}+\delta \mathrm{p}(\operatorname{def})]$ The term $\left(\frac{\partial \mathrm{X}}{\partial \mathrm{T}}\right)_{\mathrm{V}(\mathrm{m})[\mathrm{T}, \mathrm{p}]}$ defines an isochoric dependence of $\mathrm{X}$ on $\mathrm{T}$ at pressure $\mathrm{p}$ and temperature $\mathrm{T}$. At each temperature the isochoric dependence of $\mathrm{X}$ on $\mathrm{T}$ reflects the dependence of $\mathrm{V}_{\mathrm{m}}$ on $\mathrm{T}$. The analysis outlined above is repeated but in terms of the isochoric dependence of $\mathrm{X}$ on pressure. In order that the volume of a system does not change when the pressure is changed by $\delta \mathrm{p}(\exp )$, the temperature must be changed by an amount $\delta \mathrm{T}(\operatorname{def})$ determined by the equation of state for the system. $\mathrm{V}_{\mathrm{m}}[\mathrm{T}, \mathrm{p}]=\mathrm{V}_{\mathrm{m}}[\mathrm{T}+\delta \mathrm{T}(\operatorname{def}) ; \mathrm{p}+\delta \mathrm{p}(\exp )]$ We compare property $\mathrm{X}$ under the isochoric condition given in equation (h); $\mathrm{X}[\mathrm{T}, \mathrm{p}] ; \quad \mathrm{X}[\mathrm{T}+\delta \mathrm{T}(\operatorname{def}) ; \mathrm{p}+\delta \mathrm{p}(\exp )]$ $\left(\frac{\partial \mathrm{X}}{\partial \mathrm{p}}\right)_{\mathrm{v}_{(\mathrm{m})[\mathrm{T}, \mathrm{p}]}}$ describes the isochoric dependence of $\mathrm{X}$ on pressure. We have carefully examined the concept of an isochoric dependence of a given variable on either $\mathrm{T}$ or \(\mathrm{p}. The reason for this care emerges from the observation that the literature describes a number of isochoric parameters. In some cases the analysis is recognized as extrathermodynamic. In other cases a patina of thermodynamics is introduced into an analysis leading to further debate. 1.17.5: Isochoric Thermal Pressure Coefficient The equilibrium volume of a given closed system is defined by the following set of independent variables where $\xi$ is the general composition variable. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \xi^{\mathrm{eq}} ; \mathrm{A}=0\right]$ We have rather over-defined the system. The aim is to identify the composition variable at equilibrium and under the condition that the affinity for spontaneous change is zero. The system is perturbed by a change in temperature but we require that the system travels a path where the volume remains constant (and at equilibrium). The pressure must be changed in order to satisfy these conditions. By definition the isochoric differential dependence of pressure on temperature defines the isochoric thermal pressure coefficient. $\beta_{V}=\left(\frac{\partial p}{\partial T}\right)_{v}$ Three interesting equations follow [1-3]. $\beta_{\mathrm{V}}=\alpha_{\mathrm{p}} / \kappa_{\mathrm{T}}$ $\beta_{\mathrm{V}}=-\mathrm{C}_{\mathrm{V}} / \mathrm{T} \, \mathrm{V} \, \alpha_{\mathrm{s}}$ $\alpha_{\mathrm{p}} / \kappa_{\mathrm{T}}=-\mathrm{C}_{\mathrm{V}} / \mathrm{T} \, \mathrm{V} \, \alpha_{\mathrm{s}}$ Footnotes [1] From equation (a) $\beta_{\mathrm{V}}=-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}}\right)_{\mathrm{T}}$ Or, $\beta_{\mathrm{v}}=\mathrm{E}_{\mathrm{p}} / \mathrm{K}_{\mathrm{T}}=\alpha_{\mathrm{p}} / \kappa_{\mathrm{T}}$ [2] Using a Maxwell relationship $\beta_{V}=\left(\frac{\partial S}{\partial V}\right)_{T}=-\left(\frac{\partial S}{\partial T}\right)_{V} \,\left(\frac{\partial T}{\partial V}\right)_{S}$ But $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\mathrm{C}_{\mathrm{V}} / \mathrm{T}$ Then $\beta_{\mathrm{v}}=-\mathrm{C}_{\mathrm{v}} / \mathrm{T} \, \mathrm{E}_{\mathrm{s}}=-\mathrm{C}_{\mathrm{v}} / \mathrm{T} \, \mathrm{V} \, \alpha_{\mathrm{s}}$ [3] From [1] and [2], $\mathrm{E}_{\mathrm{p}} / \mathrm{K}_{\mathrm{T}}=-\mathrm{C}_{\mathrm{V}} / \mathrm{T} \, \mathrm{E}_{\mathrm{S}}$ Or, $\alpha_{\mathrm{p}} / \kappa_{\mathrm{T}}=-\mathrm{C}_{\mathrm{V}} / \mathrm{T} \, \mathrm{V} \, \alpha_{\mathrm{S}}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.17%3A_Isentropic_and_Iso-Variables/1.17.4%3A_Isochoric_Properties.txt
The isotonic method is ‘beautifully simple’[1]. The technique, described as both Isotonic and Isopiestic, leads to osmotic coefficients for solvents and activity coefficients for solutes in solution, generally aqueous solutions. Authors reporting their results describe apparatus and procedures which often differ marginally from those of other authors. Scatchard and coworkers describe how six small platinum cups, volume approx. $15 \mathrm{~cm}^{3}$, are held in a gold-plated copper block, the cups being fitted with hinged lids [1]. The cups and copper block, filled with solutions (see below) are held in a partially evacuated thermostatted chamber. The copper block is rocked gently. Over a period of time the cups are removed, weighed and replaced.The experiment ends when the masses of solutions in the cups are constant. The development of the isopiestic method can be traced to the experiments reported in 1917 by Bousfield[2] (who used the word, iso-piestic). A closed system was set up containing several solid salts in separate sample cells together with a little water($\ell$) in a separate sample cell. A little more water($\ell$) was added to the separate sample cell and the sample cells containing salts reweighed over a period of many days. The uptake of water by the salts was monitored, eventually forming salt solutions. A quantity $\mathrm{h}$, the number of moles of water taken up by a mole of salt was calculated; e.g. $\mathrm{h } = 12.43(\mathrm{KCl}), 14.23 (\mathrm{NaCl}) \text { and } 17.18(\mathrm{LiCl})$. The system is isopiestic, meaning that all samples have equal vapour pressure. [One cannot help but feel sorry for Bousfield after reading the Discussion after the paper was presented at a meeting. The critics clearly did not appreciate what Bousfield was attempting to do.] Modern techniques developed from this approach[3]. To illustrate the technique, consider the case where just two cups, $\mathrm{A}$ and $\mathrm{B}$, are used containing aqueous solutions of two salts, $\mathrm{i}$ and $\mathrm{j}$. Spontaneous transfer of solvent water occurs through the vapour phase until eventually (often after many hours) equilibrium is attained and no change in mass occurs [4,5]. At equilibrium the chemical potentials of water in the two dishes are equal. Thus, $\mu_{\mathrm{j}}^{\mathrm{eq}}(\operatorname{dish} \mathrm{A}, \mathrm{T}, \mathrm{p})=\mu_{\mathrm{i}}^{\mathrm{cq}}(\operatorname{dish} \mathrm{B}, \mathrm{T}, \mathrm{p})$ Granted that the masses of salts used to prepare the solutions in the two cups are accurately known, the mass of cups at equilibrium yields the equilibrium molalities. In most studies one dish (e.g. dish $\mathrm{A}$) holds a standard [e.g. $\mathrm{KCl}(\mathrm{aq})$] for which the dependence of practical osmotic coefficient on composition is accurately known. If, for example, the two cups contain aqueous salt solutions, equation (a) is rewritten as follows granted that the pressure is close to ambient. $\left[\mu_{1}^{*}(\ell)-\phi_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T} \, \mathrm{v}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right]_{\mathrm{A}}=\left[\mu_{1}^{*}(\ell)-\phi_{\mathrm{i}} \, \mathrm{R} \, \mathrm{T} \, \mathrm{v}_{\mathrm{i}} \, \mathrm{m}_{\mathrm{i}}\right]_{\mathrm{B}}$ Here $\mathrm{m}_{\mathrm{i}}$ and $\mathrm{m}_{\mathrm{j}}$ are the equilibrium molalities, where the word ‘equilibrium’ refers to the solvent water. Hence $\left(\phi_{\mathrm{j}} \, \mathrm{v}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right)_{\mathrm{A}}=\left(\phi_{\mathrm{i}} \, \mathrm{v}_{\mathrm{i}} \, \mathrm{m}_{\mathrm{i}}\right)_{\mathrm{B}}$ The isopiestic ratio $\mathrm{R}_{\text{iso}}$ is defined by equation (d). $\mathrm{R}_{\text {iso }}=\left(\mathrm{v}_{\mathrm{i}} \, \mathrm{m}_{\mathrm{i}}\right)_{\mathrm{B}} /\left(\mathrm{v}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right)_{\mathrm{A}}$ Hence, $\phi_{\mathrm{B}}=\phi_{\mathrm{A}} / \mathrm{R}_{\mathrm{iso}}$ Therefore $\phi_{\mathrm{B}}$ is obtained from the experimentally determined $\mathrm{R}_{\text{iso}}$ and a known (i.e. previously published standard) $\phi_{\mathrm{B}}$. In general terms, an ‘isopiestic experiment’ is based around the properties of the solvent water in a given solution. But the aim of the experiment is to gain information about the activity coefficient of the solute. The calculation therefore relies on the Gibbs - Duhem Equation . According to the Gibbs – Duhem equation the dependence of chemical potentials of salt and solvent are linked . If a given solution comprises $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{\mathrm{j}}$ moles of solute, then $\mathrm{n}_{1} \,\left(\mathrm{d} \mu_{1} / \mathrm{dn} \mathrm{n}_{\mathrm{j}}\right)+\mathrm{n}_{\mathrm{j}} \,\left(\mathrm{d} \mu_{\mathrm{j}} / \mathrm{dn}_{\mathrm{j}}\right)=0$ For a solution molality $\mathrm{m}_{\mathrm{j}}$ in a solvent, molar mass $\mathrm{M}_{1}$ $\left(1 / \mathrm{M}_{1}\right) \, \mathrm{d} \mu_{\mathrm{1}} / \mathrm{dm} \mathrm{m}_{\mathrm{j}}+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \mu_{\mathrm{j}} / \mathrm{dm}_{\mathrm{j}}=0$ Then if pressure $\mathrm{p}$ is close to the standard pressure, $\left(1 / \mathrm{M}_{1}\right) \, \mathrm{d}\left[\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right] / \mathrm{dm}_{\mathrm{j}}+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)\right] / \mathrm{dm}_{\mathrm{j}}=0$ Or, $-\phi-\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi / \mathrm{dm}_{\mathrm{j}}+1+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm}_{\mathrm{j}}=0$ Thus $-\left(\phi / \mathrm{m}_{\mathrm{j}}\right)-\mathrm{d} \phi / \mathrm{dm} \mathrm{m}_{\mathrm{j}}+\left(1 / \mathrm{m}_{\mathrm{j}}\right)+\mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm} \mathrm{m}_{\mathrm{j}}=0$ No further progress can be made until we have determined in a series of experiments the dependence of $\phi$ on $\mathrm{m}_{\mathrm{j}}$. The dependence of $\phi_{\mathrm{B}}$ on molality $\mathrm{m}_{\mathrm{B}}$ is obtained after many experiments. In a common procedure the dependence is fitted to a polynomial in mj such that integration yields the activity coefficient for the solute $\gamma_{\mathrm{j}}$. Suppose for example we find that for a given system $\phi$ is a linear function of molality $\mathrm{m}_{\mathrm{j}}$. Thus $\phi=1+\mathrm{a} \, \mathrm{m}_{\mathrm{j}}$ Or, $\mathrm{d} \phi / \mathrm{dm}_{\mathrm{j}}=\mathrm{a}$ Hence, $\ln \left(\gamma_{\mathrm{j}}\right)=2 \, \mathrm{a} \, \mathrm{m}_{\mathrm{j}}$ We note how the analysis relies on the fact the solute and solvent ‘communicate with each other’. Footnotes [1] G. Scatchard, W.J.Hamer and S.E.Wood, J.Am.Chem.Soc.,1938,60,3061. [2] W. R. Bousfield, Trans. Farady Soc.,1917,13,401. [3] J. A.Rard and R. F. Platford, Activity Coefficients in Electrolyte Solutions, ed. K. S. Pitzer, CRC Press, Boca Raton, 2nd edition, 1991. [4] In fact one can regard the phenomenon as osmosis, the vapour phase being a perfect semi-permeable membrane. [5] The system is partially evacuated so that equilibrium vapour pressure is reasonably rapidly attained.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.17%3A_Isentropic_and_Iso-Variables/1.17.6%3A_Isotonic_Method%3B_Isopiestic_Method.txt
An extensive literature reports applications of the isopiestic technique to the determination of osmotic coefficients and ionic activity coefficients for salt solutions [1-11]. In effect the technique probes the role of ion-ion interactions in determining the properties of real salt solutions. Several approaches have been reported for analyzing isopiestic results. A common method starts with the isopiestic ratio $\mathrm{R}_{\text{iso}}$. For solutions in dishes $\mathrm{A}$ and $\mathrm{B}$ at equilibrium, the isopiestic equilibrium conditions is given by equation (a). $\left(\phi_{j} \, V_{j} \, m_{j}\right)_{A}=\left(\phi_{i} \, V_{i} \, m_{i}\right)_{B}$ The isopiestic ratio, $\mathrm{R}_{\text {iso }}=\left(\mathrm{v}_{\mathrm{i}} \, \mathrm{m}_{\mathrm{i}}\right)_{\mathrm{B}} /\left(\mathrm{v}_{\mathrm{j}} \, \mathrm{m}_{\mathrm{j}}\right)_{\mathrm{A}}$ An important task formulates an equation relating the osmotic coefficient for a given salt solution and the mean ionic coefficient $\gamma_{\pm}$ If the salt solution contains a single salt, then according to the Gibbs-Duhem Equation, $\left(1 / M_{1}\right) \, d \mu_{1}(a q)=-m_{j} \, d \mu_{j}(a q)$ Hence (where pressure $\mathrm{p}$ is close to the standard pressure) \begin{aligned} \left(1 / \mathrm{M}_{1}\right) \, \mathrm{d}\left[\mu_{1}^{*}(\mathrm{l})\right.&\left.-\left(\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{v} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right)\right]=\ &-\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\left(\mathrm{v} \, \mathrm{Q} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\pm} / \mathrm{m}^{0}\right)\right]\right. \end{aligned} Then, $\left.\mathrm{d}\left[\phi \, \mathrm{m}_{\mathrm{j}}\right)\right]=-\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\left(\ln \left(\mathrm{m}_{\mathrm{j}}\right)+\ln \left(\gamma_{\pm}\right)\right]\right.$ Or, $\left.-\phi \, \mathrm{dm}_{\mathrm{j}}-\mathrm{m}_{\mathrm{j}} \, \mathrm{d}[\phi)\right]=-\mathrm{m}_{\mathrm{j}} \,\left[\mathrm{d}\left(\mathrm{m}_{\mathrm{j}}\right) / \mathrm{m}_{\mathrm{j}}+\mathrm{d} \ln \left(\gamma_{\pm}\right)\right]$ Equation (f) is integrated between the limits ‘$\mathrm{m}_{\mathrm{j}} = 0$’ and $\mathrm{m}_{\mathrm{j}}$ [3,4]. Then, $\ln \left(\gamma_{\pm}\right)=(\phi-1)+\int_{0}^{m(j)}(\phi-1) \, d \ln \left(m_{j}\right)$ And, $\phi=1+\frac{1}{m_{j}} \, \int_{0}^{m(j)} m_{j} \, d \ln \left(\gamma_{\pm}\right)$ Hence the dependences of both $\gamma_{\pm}$ and $\phi$ are obtained [1] for salt solutions and of both $\gamma_{\mathrm{j}} and \(\phi$ for solutions containing neutral solutes [5]. An important challenge at this stage is to express the experimentally determined dependence of $\phi$ on $\mathrm{m}_{\mathrm{j}}$. Having expressed this dependence quantitively, the dependence of $\gamma_{\pm}$ on $\mathrm{m}_{\mathrm{j}}$ is obtained using equation (g). The integration can be done graphically [6] or numerically using a computer- based analysis. The Debye-Huckle Limiting Law plus extended form can be used to express the dependence of $\phi$ on $\mathrm{m}_{\mathrm{j}}$. $\phi=1-\left(\mathrm{S}_{\gamma} / 3\right) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+\sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{j}} \mathrm{A}_{\mathrm{i}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{\mathrm{r}(\mathrm{i})}$ The parameter $\mathrm{r}(\mathrm{i})$ increases in quarter powers. Then [7,8], $\ln \left(\gamma_{\pm}\right)=-\mathrm{S}_{\mathrm{\gamma}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}+\sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{j}} \mathrm{A}_{\mathrm{i}} \,\left(\frac{\mathrm{r}_{\mathrm{i}}+1}{\mathrm{r}_{\mathrm{i}}}\right) \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{\mathrm{r}(\mathrm{i})}$ In more recent accounts, Pitzer’s equations have been used to represent the dependence of $\phi$ on ionic strength [9,10]. If the isopiestic experiments are repeated at several temperatures, the relative partial molar enthalpy of the solvent $\mathrm{L}_{1}(\mathrm{aq})$ is obtained [10]. In summary a large scientific literature reports thermodynamic data for aqueous solutions containing salts [11] and mixed salt [12] systems. Footnotes [1] G. Scatchard, W. J. Hamer and S. E. Wood, J.Am.Chem.Soc.,1938,60,3061. [2] For reviews and further data compilations see 1. R. N. Goldberg and R. L. Nuttall, J.Phys.Chem.Ref.Data, 1978,7,263. 2. E. C. W. Clarke, J.Phys.Chem.Ref.Data, 1985,14,489. [3] R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths, London, 2nd edn. (revised), 1965, p. 34. [4] A. K. Covington and R. A. Matheson, J. Solution Chem., 1977, 6, 263; $\mathrm{NH}_{4} \mathrm{~CNS}(\mathrm{aq})$. [5] 1. G . Barone, E. Rizzo and V. Volpe, J. Chem. Eng Data. 1976,21,59; alkyureas(aq) 2. O. D. Bonner, C. F. Jordan, R. K. Arisman and J. Bednarek, J. Chem. Thermodyn., 1976,8,1173; thioureas(aq) 3. O. D. Bonner and W. H. Breazeale, J. Chem. Eng. Data,1965,10,325; dextrose(aq); dimethylurea(aq). 4. H. D. Ellerton and P. J. Dunlop, J. Phys.Chem.,1966,70,1831; sucrose(aq). [6] J. A. Rard and D. J. Miller, J. Chem.Eng. Data 1982,27,169; CsCl(aq) and $\mathrm{SrCl}_{2}(\mathrm{aq})$. [7] J. A. Rard, J.Chem.Eng.Data,1987,32,92. $\mathrm{La}\left(\mathrm{NO}_{3}\right)_{3}(\mathrm{aq}) \text { and } \mathrm{Eu}\left(\mathrm{NO}_{3}\right)_{3}(\mathrm{aq})$. [8] J. B. Maskill and R. G. Bates, J.Solution Chem.,1986,15,418 Tris(aq). [9] L.M. Mukherjee and R. G. Bates, J.Solution Chem.,1985,14,255; $\mathrm{R}_{4}\mathrm{N}^{+} \mathrm{Br}^{-} \left(\mathrm{D}_{2}\mathrm{O}\right)$. [10] S. Lindenbaum, L. Leifer, G. E. Boyd and J. W. Chase, J. Phys. Chem., 1970,74, 761; $\mathrm{R}_{4}\mathrm{NX}(\mathrm{aq})$ [11] 1. KCl(aq) at 45 Celsius; T.M.Davis, L.M.Duckett, J.F.Owen, C.S.Patterson and R.Saleeby, J.Chem.Eng. Data, 1985, 30,432. 2. $\mathrm{NH}_{4}\mathrm{Br}(\mathrm{aq})$; A.K.Covington and D.Irish, J.Chem.Eng.Data, 1972,17,175. 3. Sodium benzoate and hydroxybenzoates; J.E.Desnoyers, R.Page, G. Perron, J.-L.Fortier, P.-A.Leduc and R.F.Platford, Can. J.Chem.,1973,51,2129. 4. $\mathrm{CaCl}_{2}(\mathrm{aq})$; L. M. Duckett, J. M. Hollifield and C. S. Patterson, J.Chem. Eng. Data, 1986, 31, 213. 5. $\mathrm{CaCl}_{2}(\mathrm{aq})$; J.A.Rard and F.H.Spedding, J.Chem.Eng.Data, 1977,22,56. 6. Borates(aq); R. F. Platford, Can. J.Chem.,1969,47,2271. 7. $\mathrm{Pr}\left(\mathrm{NO}_{3}\right)_{3}(\mathrm{aq}) \text { and } \mathrm{Lu}\left(\mathrm{NO}_{3}\right)_{3}$; J.A.Rard, J. Chem..Eng.Data, 1987,32,334. 8. Alkali metal trifluoroethanoates(aq); O.D.Bonner, J.Chem.Thermodyn.,1982,14,275. [12] 1. J.A Rard and D.G.Miller, J.Chem.Eng. Data, 1987,32,85; and references therein. 2. G. E. Boyd, J. Solution Chem.,1977,6,95; NaCl+Na p-ethylbenzenesulfonate. 3. A. K.Covington, T. H. Lilley and R. A. Robinson, J.Phys.Chem.,1968,72,2579; $\mathrm{M}^{+}\mathrm{X}^{-}\text { pairs}(\mathrm{aq})$. 4. C. C. Briggs, R. Charlton and T. H. Lilley, J. Chem.Thermodyn., 1973, 5, 445; $\mathrm{HClO}_{4} + \mathrm{~NaClO}_{4} + \mathrm{~LiClO}_{4}(\mathrm{aq})$. 5. C. P. Bezboruah, A. K. Covington and R. A. Robinson, J. Chem.Thermodyn., 1970, 2, 431; $\mathrm{KCL} + \mathrm{~NaNO}_{3}(\mathrm{aq})$. 6. S. Lindenbaum, R. M. Rush and R. A. Robinson, J. Chem.Thermodyn., 1972,4,381. 7. D. Rosenzweig, J. Padova and Y. Marcus, J.Phys.Chem.,1976,80,601; $\mathrm{NaBr}+ \mathrm{~R}_{4}\mathrm{NBr}(\mathrm{aq})$. 8. I.R.Lantzke, A.K.Covington and R.A.Robinson, J.Chem. Eng. Data, 1973,18,421; $\mathrm{Na}_{2}\mathrm{S}_{2}\mathrm{O}_{6}(\mathrm{aq}), \mathrm{~Na}_{2}\mathrm{SO}_{3}(\mathrm{aq})$. 9. W.-Y. Wen, S.Saito and C-m. Lee, J. Phys. Chem.,1966,70,1244; $\mathrm{R}_{4}\mathrm{NF}(\mathrm{aq})$. 10. A.K.Covington, R.A.Robinson and R.Thomson, J.Chem.Eng. Data, 1973,18,422; methane sulfonic acid(aq).
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.17%3A_Isentropic_and_Iso-Variables/1.17.7%3A_Isopiestic%3A_Aqueous_Salt_Solutions.txt
A given binary liquid mixture is prepared using liquid-1 and liquid –2 at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the latter being close to the standard pressure. The chemical potentials, $\mu_{1}\left(\operatorname{mix} ; x_{1}\right)$ and $\mu_{2}\left(\operatorname{mix} ; x_{2}\right)$ are related to the mole fraction composition, $x_{1}$ and $x_{2} (= 1 - x_{1})$ using equations (a) and (b) where $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquid components at the same $\mathrm{T}$ and $\mathrm{p}$; $\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1}\right)=\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ $\mu_{2}\left(\operatorname{mix} ; x_{2}\right)=\mu_{2}^{*}(\ell)+R \, T \, \ln \left(x_{2} \, f_{2}\right)$ Here for both $\mathrm{i} = 1,2$ at all $\mathrm{T}$ and $\mathrm{p}$, $\operatorname{limit}\left(x_{i} \rightarrow 1\right) f_{i}=1$ The term “regular mixture” describes a liquid mixture for which the rational activity coefficients $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$ are given by equations (d) and (e) where the property $\mathrm{w}$ is independent of temperature and liquid mixture composition [1-5]. $\ln \left(f_{1}\right)=(w / R \, T) \, x_{2}^{2}$ $\ln \left(f_{2}\right)=(w / R \, T) \,\left(1-x_{2}\right)^{2}$ Then, for example, at all $\mathrm{T}$ and $\mathrm{p}$ [6], $\operatorname{limit}\left(x_{2} \rightarrow 0\right) \ln \left(f_{1}\right)=0 ; f_{1}=1$ Similarly, $\operatorname{limit}\left(x_{2} \rightarrow 1\right) \ln \left(f_{2}\right)=0 ; f_{2}=1$ Interest in regular liquid mixtures stems from the observation that the properties of such real (as opposed to ideal) mixtures are simply described. Of course the term “real” only means that the dependences of rational activity coefficients on mole fraction composition are defined by equations (d) and (e) and that the thermodynamic properties of the liquid mixture are not ideal. For example, according to equation (d), $d \ln \left(f_{1}\right) / d T=-\left(w / R \, T^{2}\right) \, x_{2}^{2}$ With reference to the dependence of the properties of binary liquid mixtures on temperature (at fixed pressure), equation (a) yields equation (i). $\frac{\mathrm{d}\left[\mu_{1}(\mathrm{mix}) / \mathrm{T}\right]}{\mathrm{dT}}=\frac{\mathrm{d}\left[\mu_{1}^{*}(\ell) / \mathrm{T}\right]}{\mathrm{dT}}+\mathrm{R} \,\left[\frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dT}}\right]$ From the Gibbs - Helmholtz equation, $-\frac{\mathrm{H}_{1}(\mathrm{mix})}{\mathrm{T}^{2}}=-\frac{\mathrm{H}_{1}^{*}(\ell)}{\mathrm{T}^{2}}-\mathrm{R} \,\left(\frac{\mathrm{w}}{\mathrm{R} \, \mathrm{T}^{2}}\right) \, \mathrm{x}_{2}^{2}$ Hence $\mathrm{H}_{1}(\operatorname{mix})=\mathrm{H}_{1}^{*}(\ell)+\mathrm{w} \, \mathrm{x}_{2}^{2}$ Here $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 0\right) \mathrm{H}_{1}(\mathrm{mix})=\mathrm{H}_{1}^{*}(\ell)$ According to equation (k), $\mathrm{H}_{\mathrm{l}}(\mathrm{mix})$ is a quadratic function of the mole fraction composition. Further [7], $\mathrm{S}_{1}(\operatorname{mix})=\mathrm{S}_{1}^{*}(\ell)-\mathrm{R} \, \ln \left(\mathrm{x}_{1}\right)$ Hence entropic properties of regular mixtures do not deviate from the properties of an ideal liquid mixture. In terms of excess properties for regular mixtures, $S_{m}^{E}=0$ and therefore $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}$. Equations (d) and (e) can be written as explicit equations for $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$ respectively. $f_{1}=\exp \left[\left(\frac{w}{R \, T}\right) \, x_{2}^{2}\right]$ $f_{2}=\exp \left[\left(\frac{w}{R \, T}\right) \,\left(1-x_{2}\right)^{2}\right]$ The partial pressures of the two chemical substances are given by Raoult’s Law. $\mathrm{p}_{1}=\mathrm{p}_{1}^{*} \, \mathrm{x}_{1} \, \exp \left[\left(\frac{\mathrm{W}}{\mathrm{R} \, \mathrm{T}}\right) \, \mathrm{x}_{2}^{2}\right]$ $\mathrm{p}_{2}=\mathrm{p}_{2}^{*} \, \mathrm{x}_{2} \, \exp \left[\left(\frac{\mathrm{w}}{\mathrm{R} \, \mathrm{T}}\right) \,\left(1-\mathrm{x}_{2}\right)^{2}\right]$ For both liquid components, deviations from ideal thermodynamic properties increase with increase in the magnitude of $(\mathrm{w} / \mathrm{R} \, \mathrm{T})$. If $\mathrm{w} >0$, the deviations are called positive whereas if $\mathrm{w} <0$ the deviations are called negative. In the event that $(\mathrm{w} / \mathrm{R} \, \mathrm{T})$ equals 2, the plots of $\mathrm{p}_{1}$ and $\mathrm{p}_{2}$ against mole fraction composition are horizontal when $x_{1} = x_{2} = 0.5$. But in the event that $(\mathrm{w} / \mathrm{R} \, \mathrm{T})$ equals 3, a range of binary liquid mixtures exist having intermediate mole fraction compositions and are unstable. These mixtures separate into two liquid mixtures, one rich in component 1 and the other rich in component 2. Footnotes [1] J. Hildebrand, J Am. Chem. Soc.,1929, 51,69. Accounts of this class of mixtures are given in references [2]-[5]. [2] E. A. Guggenheim, Thermodynamics, North Holland Publishing Company, Amsterdam, 1950, chapter 5; note that Guggenheim uses the symbol $x$ to represent the mole fraction composition of a binary liquid mixture $x_{2}$; see page 173. [3] E. A. Guggenheim, Mixtures, Clarendon Press, Oxford, 1952, chapter IV. [4] M.L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979, chapter 16. [5] G. N. Lewis and M. L. Randall, Thermodynamics, revised by K. S. Pitzer and L. Brewer, McGraw-Hill, New York, 1961, chapter 21. [6] $\mathrm{R} \, \mathrm{T}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]=\left[\mathrm{J} \mathrm{mol}^{-1}\right]$ Then, $\mathrm{w}=\left[\mathrm{J} \mathrm{mol}^{-1}\right]$, a molar energy. [7] From, $\mu_{1}(\operatorname{mix})=\mathrm{H}_{1}(\operatorname{mix})-\mathrm{T} \, \mathrm{S}_{1}(\operatorname{mix})$ $\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)+\mathrm{w} \, \mathrm{x}_{2}^{2}=\mathrm{H}_{1}^{*}(\ell)+\mathrm{w} \, \mathrm{x}_{2}^{2}-\mathrm{T} \, \mathrm{S}_{1}(\mathrm{mix})$ Or, $\mu_{1}^{*}(\ell)-\mathrm{H}_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)=-\mathrm{T} \, \mathrm{S}_{1}(\operatorname{mix})$ Or, $-\mathrm{T} \, \mathrm{S}_{1}^{\prime \prime}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)=-\mathrm{T} \, \mathrm{S}_{1}(\mathrm{mix})$ [8] From equation (p) with $(w / R \, T)=2$, \begin{aligned} &\mathrm{p}_{1}=\mathrm{p}_{1}^{*} \, \mathrm{x}_{1} \, \exp \left(2 \, \mathrm{x}_{2}^{2}\right)\ &p_{1}=p_{1}^{*} \, x_{1} \, \exp \left[2 \,\left(1-x_{1}\right)^{2}\right]\ &\frac{\mathrm{d}\left(\mathrm{p}_{1} / \mathrm{p}_{1}^{*}\right)}{\mathrm{dx}_{1}}=\exp \left[2 \,\left(1-\mathrm{x}_{1}\right)^{2}\right]-\mathrm{x}_{1} \, 4 \,\left(1-\mathrm{x}_{1}\right) \, \exp \left[2 \,\left(1-\mathrm{x}_{1}\right)^{2}\right]\ &=\exp \left[2 \,\left(1-x_{1}\right)^{2}\right] \,\left[1-4 \, x_{1} \,\left(1-x_{1}\right)\right]\ &=\exp \left[2 \,\left(1-x_{1}\right)^{2}\right] \,\left[1-4 \, x_{1}+4 \, x_{1}^{2}\right]\ &\frac{\mathrm{d}\left(\mathrm{p}_{1} / \mathrm{p}_{1}^{*}\right)}{\mathrm{dx_{1 }}}=\exp \left[2 \,\left(1-\mathrm{x}_{1}\right)^{2}\right] \,\left(1-2 \, \mathrm{x}_{1}\right)^{2} \ &\frac{\mathrm{d}\left(\mathrm{p}_{2} / \mathrm{p}_{2}^{*}\right)}{\mathrm{dx}_{2}}=\exp \left[2 \,\left(1-\mathrm{x}_{2}\right)^{2}\right]+\mathrm{x}_{2} \,(-4) \,\left(1-\mathrm{x}_{2}\right) \, \exp \left[2 \,\left(1-\mathrm{x}_{2}\right)^{2}\right] \ &\frac{\mathrm{d}\left(\mathrm{p}_{2} / \mathrm{p}_{2}^{*}\right)}{\mathrm{dx_{2 }}}=\exp \left[2 \,\left(1-\mathrm{x}_{2}\right)^{2}\right] \,\left(1-2 \, \mathrm{x}_{2}\right)^{2} \end{aligned}
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.18%3A_Liquid_Mixtures/1.18.1%3A_Liquid_Mixtures%3A_Regular_Mixtures.txt
A given binary liquid mixture is prepared using liquid-1 and liquid –2 at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the latter being close to the standard pressure. The chemical potentials, $\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1}\right)$ and $\mu_{2}\left(\operatorname{mix} ; \mathrm{x}_{2}\right)$ are related to the mole fraction composition, $x_{1}$ and $x_{2} (= 1 - x_{1})$ using equations (a) and (c) where $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquid components at the same $\mathrm{T}$ and $\mathrm{p}$; $\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1}\right)=\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ where $\operatorname{limit}\left(x_{1} \rightarrow 1\right) f_{1}=1$ $\mu_{2}\left(\operatorname{mix} ; \mathrm{x}_{2}\right)=\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)$ where $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 1\right) \mathrm{f}_{2}=1$ A general equation for activity coefficient $\mathrm{f}_{1}$ takes the following form [1]. $\ln \left(f_{1}\right)=\sum_{k=1}^{k=\infty} \alpha_{k} \, x_{s}^{\lambda(k)}$ Equation (e) satisfies the condition, $\operatorname{limit}\left(x_{2} \rightarrow 0\right) \ln \left(f_{1}\right)=0 ; f_{1}=1$ The parameter $\alpha_{\mathrm{k}}$ is characteristic of the mixture, temperature and pressure. The property $\lambda_{\mathrm{k}}$ is a real number. In the limit that the liquid mixture is dilute in chemical substance liquid-2, equation (e) simplifies to equation (g). $\ln \left(f_{1}\right)=\alpha \, x_{2}^{\lambda}$ In general terms [2], $x_{1} \, d \ln \left(f_{1}\right)+x_{2} \, d \ln \left(f_{2}\right)=0$ We combine equations (e) and (h) with $\lambda_{k} \geq 2$ [3]. $\frac{\mathrm{d} \ln \left(\mathrm{f}_{1} / \mathrm{f}_{2}\right)}{\mathrm{dx}_{2}}=\frac{1}{\mathrm{x}_{2}} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dx} \mathrm{x}_{2}}=\sum_{\mathrm{k}=1}^{\mathrm{k}=\infty} \alpha_{\mathrm{k}} \, \lambda_{\mathrm{k}} \, \mathrm{x}_{2}^{\lambda(\mathrm{k})-2}$ Equation (i) is integrated to yield equation (j) where $\mathrm{I}$ is the constant of integration. $\ln \left(f_{2}\right)=\ln \left(f_{1}\right)-\sum_{k=1}^{k=\infty} \frac{\alpha_{k} \, \lambda_{k} \, x_{2}^{\lambda(k)-1}}{\lambda_{k}-1}-I$ Hence [4,5] \begin{aligned} \ln \left(f_{2}\right) &=\ln \left(f_{1}\right)-\sum_{k=1}^{k=\infty} \frac{\alpha_{k} \, \lambda_{k}}{\lambda_{k}-1} \,\left(x_{2}^{\lambda(k)-1}-1\right)-\sum_{k=1}^{k=\infty} \alpha_{k} \ &=\ln \left(f_{1}\right)-\sum_{k=1}^{k=\infty} \alpha_{k} \,\left[\frac{\lambda_{k}}{\lambda_{k}-1} \,\left(x_{2}^{\lambda-(k-1)}-1\right)-1\right] \end{aligned} In other words, granted that $\ln \left(f_{1}\right)$ is known as a function of $x_{2}$, then $\ln \left(f_{2}\right)$ can be calculated. Footnotes [1] I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Green, London, 1954. [2] For a binary liquid mixture , the Gibbs-Duhem equation relates activity coefficients $\mathrm{f}_{1}$ and $\mathrm{f}_{2}$. Thus, $-S \, d T+V \, d p+n_{1} \, d \mu_{1}+n_{2} \, d \mu_{2}=0$ At fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{n}_{1} \, \mathrm{d} \mu_{1}+\mathrm{n}_{2} \, \mathrm{d} \mu_{2}=0$ Divide by $\left(n_{1}+n_{2}\right)$; $x_{1} \, d \mu_{1}+x_{2} \, d \mu_{2}=0$ $\begin{gathered} \mathrm{x}_{1} \, \mathrm{d}\left[\mu_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)\right]+\mathrm{x}_{2} \, \mathrm{d}\left[\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)\right]=0 \ \mathrm{x}_{1} \, \mathrm{d} \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)+\mathrm{x}_{2} \, \mathrm{d} \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)=0 \ \mathrm{x}_{1} \, \mathrm{d} \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{1} \, \mathrm{d} \ln \left(\mathrm{f}_{1}\right)+\mathrm{x}_{2} \, \mathrm{d} \ln \left(\mathrm{x}_{2}\right)+\mathrm{x}_{2} \, \mathrm{d} \ln \left(\mathrm{f}_{2}\right)=0 \end{gathered}$ But $\mathrm{x}_{1} \, \mathrm{d} \ln \left(\mathrm{x}_{1}\right)+\mathrm{x}_{2} \, \mathrm{d} \ln \left(\mathrm{x}_{2}\right)=\left(\mathrm{x}_{1} / \mathrm{x}_{1}\right) \, \mathrm{dx} \mathrm{x}_{1}+\left(\mathrm{x}_{2} / \mathrm{x}_{2}\right) \, \mathrm{dx} \mathrm{x}_{2}$ Also $\mathrm{x}_{1}+\mathrm{x}_{2}=1$ so that $\mathrm{dx}_{1}+\mathrm{dx}_{2}=0$ [3] From equation (h) for a binary liquid mixture at fixed $\mathrm{T}$ and $\mathrm{p}$, $\begin{array}{r} \left(1-x_{2}\right) \, \frac{d \ln \left(f_{1}\right)}{d x_{2}}+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{2}}=0 \ \frac{d \ln \left(f_{1}\right)}{d x_{2}}-x_{2} \, \frac{d \ln \left(f_{1}\right)}{d x_{2}}+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{2}}=0 \end{array}$ We divide by $x_{2}$ and rearrange the equation. $\frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dx} \mathrm{x}_{2}}-\frac{\mathrm{d} \ln \left(\mathrm{f}_{2}\right)}{\mathrm{dx}_{2}}=\frac{1}{\mathrm{x}_{2}} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dx} \mathrm{x}_{2}}$ Or, $\frac{\mathrm{d} \ln \left(\mathrm{f}_{1} / \mathrm{f}_{2}\right)}{\mathrm{dx} \mathrm{x}_{2}}=\frac{1}{\mathrm{x}_{2}} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dx_{2 }}}$ [4] From equations (e) and (j), $\ln \left(f_{2}\right)=\sum_{k=1}^{k=\infty} \alpha_{k} \, x_{2}^{\lambda(k)}-\sum_{k=1}^{k=\infty} \frac{\alpha_{k} \, \lambda_{k} \, x_{2}^{\lambda(k)-1}}{\lambda_{k}-1}-I$ But at $x_{2} = 1, f_{2} = 1$. Then, $0=\sum_{\mathrm{k}=1}^{\mathrm{k}=\infty} \alpha_{\mathrm{k}}-\sum_{\mathrm{k}=1}^{\mathrm{k}=\infty} \frac{\alpha_{\mathrm{k}} \, \lambda_{\mathrm{k}}}{\lambda_{\mathrm{k}}-1}-\mathrm{I}$ [5] J. N. Bronsted and P. Colmart, Z. Phys. Chem.,1934,A168, 381 ( as quoted in reference 1). 1.18.3: Liquid Mixtures: Series Functions for Activity Coefficients A given binary liquid mixture is prepared using liquid-1 and liquid-2 at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the latter being close to the standard pressure. The chemical potentials, $\mu_{1}\left(\operatorname{mix} ; x_{1}\right)$ and $\mu_{2}\left(\operatorname{mix} ; \mathrm{x}_{2}\right)$ are related to the mole fraction composition, $\mathrm{x}_{1}$ and $\mathrm{x}_{2}\left(=1-\mathrm{x}_{1}\right)$ using equations (a) and (c) where $\mu_{1}^{*}(\ell)$ and $\mu_{2}^{*}(\ell)$ are the chemical potentials of the two pure liquid components at the same $\mathrm{T}$ and $\mathrm{p}$; $\mu_{1}\left(\operatorname{mix} ; \mathrm{x}_{1}\right)=\mu_{1}^{*}(\ell)+R \, T \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ where $\operatorname{limit}\left(x_{1} \rightarrow 1\right) f_{1}=1$ $\mu_{2}\left(\operatorname{mix} ; \mathrm{x}_{2}\right)=\mu_{2}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{2} \, \mathrm{f}_{2}\right)$ where $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 1\right) \mathrm{f}_{2}=1$ A quite general approach to understanding the properties of binary liquid mixtures expresses, for example, $\ln \left(f_{1}\right)$ as a series function in terms of mole fraction $x_{2}$ at fixed $\mathrm{T}$ and $\mathrm{p}$. Using only three terms we obtain equation (e). $\ln \left(f_{1}\right)=\alpha_{2} \, x_{2}^{2}+\alpha_{3} \, x_{2}^{3}+\alpha_{4} \, x_{2}^{4}$ As required, $\operatorname{limit}\left(x_{2} \rightarrow 0\right) \ln \left(f_{1}\right)=0 ; f_{1}=1$ Hence [1], \begin{aligned} \ln \left(f_{2}\right) &=\left[\alpha_{2}+(3 / 2) \, \alpha_{3}+2 \, \alpha_{4}\right] \, x_{1}^{2} \ &-\left[\alpha_{3}+(8 / 3) \, \alpha_{4}\right] \, x_{1}^{3}+\alpha_{4} \, x_{1}^{4} \end{aligned} As required, $\operatorname{limit}\left(x_{1} \rightarrow 0\right) \ln \left(f_{2}\right)=0 ; f_{2}=1$ Footnotes [1] From, $\ln \left(f_{1}\right)=\alpha_{2} \, x_{2}^{2}+\alpha_{3} \, x_{2}^{3}+\alpha_{4} \, x_{2}^{4}$ $\ln \left(f_{1}\right)=\alpha_{2} \,\left(1-x_{1}\right)^{2}+\alpha_{3} \,\left(1-x_{1}\right)^{3}+\alpha_{4} \,\left(1-x_{1}\right)^{4}$ Then $\frac{\mathrm{d} \ln \left(f_{1}\right)}{d x_{1}}=-2 \, \alpha_{2} \,\left(1-x_{1}\right)-3 \, \alpha_{3} \,\left(1-x_{1}\right)^{2}-4 \, \alpha_{4} \,\left(1-x_{1}\right)^{3}$ But from the Gibbs-Duhem equation (at fixed $\mathrm{T}$ and $\mathrm{p}$) $x_{1} \, \frac{d \ln \left(f_{1}\right)}{d x_{1}}+x_{2} \, \frac{d \ln \left(f_{2}\right)}{d x_{1}}=0$ Or, $\frac{\mathrm{d} \ln \left(f_{2}\right)}{\mathrm{dx}_{1}}=-\frac{\mathrm{x}_{1}}{\mathrm{x}_{2}} \, \frac{\mathrm{d} \ln \left(\mathrm{f}_{1}\right)}{\mathrm{dx}_{1}}$ Or, $\frac{d \ln \left(f_{2}\right)}{d x_{1}}=-\frac{x_{1}}{\left(1-x_{1}\right)} \, \frac{d \ln \left(f_{1}\right)}{d x_{1}}$ Then, $\frac{\mathrm{d} \ln \left(\mathrm{f}_{2}\right)}{\mathrm{dx}_{1}}=2 \, \alpha_{2} \, \mathrm{x}_{1}+3 \, \mathrm{x}_{1} \, \alpha_{3} \,\left(1-\mathrm{x}_{1}\right)+4 \, \alpha_{4} \, \mathrm{x}_{1} \,\left(1-\mathrm{x}_{1}\right)^{2}$ $\frac{\mathrm{d} \ln \left(f_{2}\right)}{d x_{1}}=2 \, \alpha_{2} \, x_{1}+3 \, x_{1} \, \alpha_{3}-3 \, \alpha_{3} \, x_{1}^{2}+4 \, \alpha_{4} \, x_{1}-8 \, \alpha_{4} \, x_{1}^{2}+4 \, \alpha_{4} \, x_{1}^{3}$ Or, $\frac{\mathrm{d} \ln \left(\mathrm{f}_{2}\right)}{\mathrm{dx}_{1}}=\left[2 \, \alpha_{2}+3 \, \alpha_{3}+4 \, \alpha_{4}\right] \, \mathrm{x}_{1}-\left[3 \, \alpha_{3}+8 \, \alpha_{4}\right] \, \mathrm{x}_{1}^{2}+4 \, \alpha_{4} \, \mathrm{x}_{1}^{3}$ The latter equation is integrated. $\ln \left(f_{2}\right)=\left[\alpha_{2}+(3 / 2) \, \alpha_{3}+2 \, \alpha_{4}\right] \, x_{1}^{2}-\left[\alpha_{3}+(8 / 3) \, \alpha_{4}\right] \, x_{1}^{3}+\alpha_{4} \, x_{1}^{4}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.18%3A_Liquid_Mixtures/1.18.2%3A_Liquid_Mixtures%3A_General_Equations.txt
For an ideal liquid mixture containing i-liquid components four important molar properties are related to the corresponding properties of the pure liquid components using the following equations. $\mathrm{C}_{\mathrm{Vm}_{\mathrm{m}}}(\operatorname{mix} ; \mathrm{id})=\sum_{\mathrm{i}} \mathrm{x}_{\mathrm{i}} \,\left\{1-\left[\frac{\mathrm{E}_{\mathrm{pi}}^{*}(\ell)}{\mathrm{C}_{\mathrm{V}_{\mathrm{i}}}^{*}(\ell)}\right] \,\left[\beta_{\mathrm{v}}(\operatorname{mix} ; \mathrm{id})-\beta_{\mathrm{V}_{\mathrm{i}}}^{*}(\ell)\right] \, \mathrm{C}_{\mathrm{V}_{\mathrm{i}}}^{*}(\ell)\right\}$ \begin{aligned} &\mathrm{E}_{\mathrm{Sm}}(\operatorname{mix} ; \mathrm{id}) \ &=\sum_{\mathrm{i}} \mathrm{x}_{\mathrm{i}} \,\left\{1-\left[\frac{\mathrm{C}_{\mathrm{pi}}^{*}(\ell)}{\mathrm{E}_{\mathrm{Si}}^{*}(\ell)}\right] \,\left[\left[\beta_{\mathrm{v}}(\mathrm{mix} ; \mathrm{id})\right]^{-1}-\left[\beta_{\mathrm{vi}_{\mathrm{i}}}^{*}(\ell)\right]^{-1}\right] \, \mathrm{E}_{\mathrm{Si}}^{*}(\ell)\right\} \end{aligned} $\mathrm{E}_{\mathrm{Sm}}(\operatorname{mix} ; \mathrm{id})=\sum_{\mathrm{i}} \mathrm{x}_{\mathrm{i}} \,\left\{1-\left[\frac{\mathrm{K}_{\mathrm{pi}}^{*}(\ell)}{\mathrm{E}_{\mathrm{Si}}^{*}(\ell)}\right] \,\left[\beta_{\mathrm{s}}(\operatorname{mix} ; \mathrm{id})-\beta_{\mathrm{Si}}^{*}(\ell)\right] \, \mathrm{E}_{\mathrm{Si}}^{*}(\ell)\right\}$ \begin{aligned} &\mathrm{K}_{\mathrm{Sm}}(\text { mix; id }) \ &\quad=\sum_{\mathrm{i}} \mathrm{x}_{\mathrm{i}} \,\left\{1-\left[\frac{\mathrm{E}_{\mathrm{pl}}^{*}(\ell)}{\mathrm{K}_{\mathrm{Si}}^{*}(\ell)}\right] \,\left[\left[\beta_{\mathrm{s}}(\text { mix; id) }]^{-1}-\left[\beta_{\mathrm{si}}^{*}(\ell)\right]^{-1}\right] \, \mathrm{K}_{\mathrm{Si}}^{*}(\ell)\right\}\right. \end{aligned} With reference to these four equations, interesting features emerge. If $\mathrm{V}_{1}^{*}(\ell)$ and $\mathrm{V}_{2}^{*}(\ell)$ for the two components of a binary liquid mixture having ideal thermodynamic properties are linearly related at different temperatures and pressures then at fixed liquid mixture composition, $\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{v}_{1}^{*}(1)}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{v}_{2}^{*}(1)}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{v}_{(\operatorname{mix} ; \mathrm{d})}}$ Or, $\beta_{\mathrm{v}_{1}}^{*}(\ell)=\beta_{\mathrm{v}_{2}}^{*}(\ell)=\beta_{\mathrm{v}}(\mathrm{mix} ; \mathrm{id})$ Under these conditions the two properties described in equations (a) and (b) are given by the mole fraction weighted sums of the properties of the pure liquids. The internal pressure pint is given by $\left[\mathrm{T} \, \beta_{\mathrm{V}}-1\right]$. Hence the same condition holds with respect to the two properties defined by equations (a) and (b) if the internal pressures are equal. In practice liquids have different internal pressures. However this difference is often small for chemically similar liquids. An interesting feature emerges if the molar entropies of the two liquids are linearly related over a range of temperatures and pressures. Thus, $\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}_{1}^{*}(\theta)}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{s}_{2}^{*}(\theta)}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{S}(\text { mid; } ; \mathrm{d})}$ Or, $\beta_{\mathrm{s} 1}^{*}(\ell)=\beta_{\mathrm{S} 2}^{*}(\ell)=\beta_{\mathrm{s}}(\operatorname{mix} ; \mathrm{id})$ Therefore a liquid mixture where the components have identical isentropic thermal pressure coefficients, $\mathrm{K}_{\mathrm{Sm}}(\mathrm{mix} ; \mathrm{id})$ and $\mathrm{E}_{\mathrm{Sm}}(\mathrm{mix} ; \mathrm{id})$ are given by the mole fraction weighted sums of the properties of the pure components [1]. In the case of an ideal binary liquid mixture the following three equations relate the isochoric heat capacities, isentropic compressions and isentropic expansions to the properties of the component pure liquids. \begin{aligned} &\mathrm{C}_{\mathrm{Vm}}(\operatorname{mix} ; \mathrm{id}) \ &=\mathrm{x}_{1} \, \mathrm{C}_{\mathrm{v} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{C}_{\mathrm{V} 2}^{*}(\ell) \ &+\mathrm{T} \,\left\{\left[\frac{\mathrm{x}_{1} \,\left[\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 1}^{*}(\ell)}\right]+\left[\frac{\mathrm{x}_{2} \,\left[\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{~K}_{\mathrm{T} 2}^{*}(\ell)}\right]-\left[\frac{\left[\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{x}_{1} \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell)}\right]\right\} \end{aligned} \begin{aligned} &\mathrm{K}_{\mathrm{Sm}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{K}_{\mathrm{S} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{K}_{\mathrm{S} 2}^{*}(\ell) \ &+\mathrm{T} \,\left\{\left[\frac{\mathrm{x}_{1} \,\left[\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)\right]^{2}}{\mathrm{C}_{\mathrm{p} 1}^{*}(1 \ell)}\right]+\left[\frac{\mathrm{x}_{2} \,\left[\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{C}_{\mathrm{p} 2}^{*}(\ell)}\right]-\left[\frac{\left[\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{pl}}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\right]^{2}}{\mathrm{x}_{1} \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{C}_{\mathrm{p} 2}^{*}(\ell)}\right]\right\} \end{aligned} \begin{aligned} &\mathrm{E}_{\mathrm{Sm}}(\operatorname{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{Sl}}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{S} 2}^{*}(\ell)\ &+\mathrm{T} \,\left\{\begin{array}{l} {\left[\frac{\mathrm{x}_{1} \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)}{\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)}\right]+\left[\frac{\mathrm{x}_{2} \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell) \, \mathrm{C}_{\mathrm{p} 2}^{*}(\ell)}{\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}\right]} \ -\left[\frac{\left[\mathrm{x}_{1} \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell)\right] \,\left[\mathrm{x}_{1} \, \mathrm{C}_{\mathrm{pl} 1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{C}_{\mathrm{p} 2}^{*}(\ell)\right]}{\mathrm{x}_{1} \, \mathrm{E}_{\mathrm{pl}}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{E}_{\mathrm{p} 2}^{*}(\ell)}\right] \end{array}\right\} \end{aligned} Inspection shows that in each case the condition for simple additivity requires that the sum inside the brackets {……} vanishes. In the case of $\mathrm{C}_{\mathrm{Vm}}(\operatorname{mix} ; \mathrm{id})$ a sufficient condition ( and most probably also necessary) is that $\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)=\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\left[=\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{id})\right]$ and that $\mathrm{K}_{\mathrm{T} 1}^{*}(\ell)=\mathrm{K}_{\mathrm{T} 2}^{*}(\ell)\left[=\mathrm{K}_{\mathrm{Tm}}(\mathrm{mix} ; \mathrm{id})\right]$ at a given $\mathrm{T}$ and $\mathrm{p}$. In the case of $\mathrm{K}_{\mathrm{Sm}}(\mathrm{mix} ; \mathrm{id})$ the required conditions are that $\mathrm{E}_{\mathrm{pl}}^{*}(\ell)=\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\left[=\mathrm{E}_{\mathrm{pm}}(\text { mix; } \mathrm{id})\right]$ and that $\mathrm{C}_{\mathrm{p} 1}^{*}(\ell)=\mathrm{C}_{\mathrm{p} 2}^{*}(\ell)\left[=\mathrm{C}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})\right]$ at a given $\mathrm{T}$ and $\mathrm{p}$. But since $\mathrm{E}_{\mathrm{p}}=-(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}$ and $\mathrm{C}_{\mathrm{p}}=\mathrm{T} \,(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{p}}$, the condition can be restated as follows. Although molar entropies of liquid 1 and 2 may differ, they should have identical isobaric dependences on temperature and isothermal dependence on pressure. In the case of $\mathrm{E}_{\mathrm{Sm}}(\mathrm{mix} ; \mathrm{id})$ the three conditions are that $\mathrm{E}_{\mathrm{p} 1}^{*}(\ell)=\mathrm{E}_{\mathrm{p} 2}^{*}(\ell)\left[=\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{id})\right]$, $\mathrm{K}_{\mathrm{T} 1}^{*}(\ell)=\mathrm{K}_{\mathrm{T} 2}^{*}(\ell)\left[=\mathrm{K}_{\mathrm{Tm}}(\mathrm{mix} ; \mathrm{id})\right]$, and $\mathrm{C}_{\mathrm{pl}}^{*}(\ell)=\mathrm{C}_{\mathrm{p} 2}^{*}(\ell)\left[=\mathrm{C}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{id})\right]$. If we extend the foregoing analysis to the variables isentropic compressibilities $\kappa_{\mathrm{S}}$ and isentropic expansibilities $\alpha_{\mathrm{S}}$ we find that because $\kappa_{\mathrm{S}}(\operatorname{mix} ; \mathrm{id})=\mathrm{K}_{\mathrm{Sm}}(\operatorname{mix} ; \mathrm{id}) / \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})$, the condition described above for $\mathrm{K}_{\mathrm{Sm}}(\text { mix; } \mathrm{id})$ requires that $\kappa_{\mathrm{S}}(\operatorname{mix} ; \mathrm{id})$ is given by the volume weighted sum of $\kappa_{\mathrm{sl}}^{*}(\ell)$ and $\kappa_{S 2}^{*}(\ell)$. Similarly we find that the three conditions described above in the context of $\mathrm{E}_{\mathrm{Sm}}(\mathrm{mix} ; \mathrm{id})$ is necessary in order that $\alpha_{S}(\operatorname{mix} ; \mathrm{id})$ is given by the volume weighted sum of $\alpha_{\mathrm{s} 1}^{*}(\ell)$ and $\alpha_{\mathrm{s} 2}^{*}(\ell)$. The conditions described above are expressed in thermodynamic terms but we note that in no case can the properties of real pure liquids comply with these conditions. Nevertheless they provide useful pointers in the task of understanding the properties of real liquid mixtures. Even for a mixture prepared using $\mathrm{H}_{2}\mathrm{O}(\ell)$ and $\mathrm{D}_{2}\mathrm{O}(\ell)$, $\mathrm{C}_{\mathrm{Vm}}(\operatorname{mix} ; \mathrm{id})$ would depart from mole fraction additivity. Only for mixtures of ideal gases would the condition hold for $\mathrm{C}_{\mathrm{Vm}}(\operatorname{mix} ; \mathrm{id})$. Indeed for a monatomic gas the energy is entirely translational and $\mathrm{C}_{\mathrm{pm}}=(5 / 2) \ldot \mathrm{R}$. Then $\left(\partial \mathrm{S}_{\mathrm{m}} / \partial \mathrm{T}\right)_{\mathrm{p}}=\mathrm{C}_{\mathrm{pm}} / \mathrm{T}=(5 / 2) \, \mathrm{R} / \mathrm{T}$ for both pure gases and the mixture, a consequence of the Sackur-Tetrode equation for the molar entropy of ideal gases. Footnote [1] G. Douheret, M. I. Davis, J. C. R. Reis and M. J. Blandamer, Chem. Phys. Chem. Phys.,2001,2,148. 1.18.5: Liquid Mixtures: Binary: Pseudo-Excess Properties At defined $\mathrm{T}$ and $\mathrm{p}$, a thermodynamic (molar) property $\mathrm{P}$ of an ideal binary liquid mixture (e.g. volume) can be expressed as a function of the mole fraction composition using equation (a). $\mathrm{P}(\mathrm{mix} ; \mathrm{id})=\mathrm{x}_{1} \, \mathrm{P}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{P}_{2}^{*}(\ell)$ Here $\mathrm{P}_{1}^{*}(\ell)$ and $\mathrm{P}_{2}^{*}(\ell)$ are the properties of the two pure liquids at the same $\mathrm{T}$ and $\mathrm{p}$. In many cases equation (a) is taken as a pattern on which to base a description of other properties of liquid mixtures; e.g. relative permitivities, surface tensions and viscosities. There is often no thermodynamic basis for this description although it has to be admitted that such an equation has an intuitively attractive form. In the next stage the difference between measured property $\mathrm{P}(\mathrm{mix})$ and $\mathrm{P}(\text { mix } ; \mathrm{id})$ leads to a defined pseudo-excess property. $\mathrm{P}^{\mathrm{E}}$. For the sake of completeness, the use of molar changes on mixing is recommended in the present context. Thus, $X_{m}(n o-\operatorname{mix})=x_{1} \, X_{1}^{*}(\ell)+x_{2} \, X_{2}^{*}(\ell)$ Then by definition at common temperature and pressure, $\Delta_{\text {mix }} X_{m}(\operatorname{mix})=X_{m}(\operatorname{mix})-X_{m}(n o-m i x)$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.18%3A_Liquid_Mixtures/1.18.4%3A_Liquid_Mixtures%3A_Binary%3A_Less_Common_Properties.txt
In a description of the properties of gases, the term ‘perfect’ means that there are no intermolecular forces, either attractive or repulsive. The equation of state for $\mathrm{n}_{\mathrm{j}}$ moles of perfect gas $\mathrm{j}$ takes the following form where $\mathrm{R}$ is the Gas Constant, $8.314 \mathrm{J K}^{-1} \mathrm{~mol}^{-1}$ [1]. $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ The chemical potential of a perfect gas $\mu_{j}^{\text {id }}$ at temperature $\mathrm{T}$ is related to pressure $\mathrm{p}_{j}$ using equation (b). $\mu_{\mathrm{j}}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}_{\mathrm{j}}\right)=\mu_{\mathrm{j}}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{p}_{\mathrm{j}} / \mathrm{p}^{0}\right)$ Thus $\mu_{j}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}_{\mathrm{j}}\right)$ is the chemical potential of gas $j$ at pressure $\mathrm{p}_{j}$ whereas $\mu_{\mathrm{j}}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}^{0}\right)$ is the corresponding chemical potential at the standard pressure $\mathrm{p}^{0}$ [2]. The ratio $\left(\mathrm{V}_{\mathrm{j}} / \mathrm{n}_{\mathrm{j}}\right)$ is the molar volume of gas $j$, $\mathrm{V}_{\mathrm{mj}}$. Equation (a) describing a perfect gas can be written as follows. $\mathrm{p}_{\mathrm{j}}^{\mathrm{id}} \, \mathrm{V}_{\mathrm{mj}}=\mathrm{R} \, \mathrm{T}$ No real gas is perfect at all temperatures and pressures although at high temperatures and low pressures the product $\mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{mj}}$ is arithmetically almost equal to the product, $\mathrm{R} \, \mathrm{T}$. Generally however equation (c) does not describe real gases. The properties of real gases are described in several ways. In one approach $\mu_{j}\left(T, p_{j}\right)$ is related to $\mu_{\mathrm{j}}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}^{0}\right)$ using equation (d) where $\mathrm{f}_{j}$ is the fugacity. $\mu_{\mathrm{j}}\left(\mathrm{T}, \mathrm{p}_{\mathrm{j}}\right)=\mu_{\mathrm{j}}\left(\mathrm{T}, \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{f}_{\mathrm{j}} / \mathrm{p}^{0}\right)$ Thus $\operatorname{limit}\left(\mathrm{p}_{\mathrm{j}} \rightarrow 0\right) \mathrm{f}_{\mathrm{j}}=\mathrm{p}_{\mathrm{j}}$ Another approach uses virial coefficients [3]. Thus pressure $\mathrm{p}_{j}$ is related to molar volume $\mathrm{V}_{\mathrm{mj}$ using a power series in the term $\mathrm{V}_{\mathrm{mj}}$. Thus, $\mathrm{p}_{\mathrm{j}}=\frac{\mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{mj}}}\left[1+\frac{\mathrm{B}}{\mathrm{V}_{\mathrm{mj}}}+\frac{\mathrm{C}}{\mathrm{V}_{\mathrm{mj}}^{2}}+\ldots \ldots\right]$ In the event that a given gas is only slightly imperfect the terms C, D,…. are negligibly small. Then, $\mathrm{p}_{\mathrm{j}}=\frac{\mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{mj}}}\left[1+\frac{\mathrm{B}}{\mathrm{V}_{\mathrm{mj}}}\right]$ At low temperatures $\mathrm{B}$ tends to be negative but at high temperatures $\mathrm{B}$ is positive. Footnotes [1] For equation (a), $\left[\mathrm{N} \mathrm{m}^{-2}\right] \,\left[\mathrm{m}^{3}\right]=[\mathrm{mol}] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]$ where $[\mathrm{J}]=[\mathrm{Nm}]$ [2] $\mu_{\mathrm{j}}^{\mathrm{id}}\left(\mathrm{T}, \mathrm{p}_{\mathrm{j}}\right)=\left[\mathrm{J} \mathrm{mol}^{-1}\right] \quad \mathrm{R} \, \mathrm{T}=\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right] \,[\mathrm{K}]=\left[\mathrm{J} \mathrm{mol}^{-1}\right]$ [3] I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Green, London, 1953, chapter 11. 1.19.2: Perfect Gas: The Gas Constant Throughout these Topics, the Gas Constant, symbol $\mathrm{R}$, plays an important role. Here we examine how such an important quantity emerges [1]. An important concept in chemical thermodynamics is the perfect gas. In practice the properties of real gases differ from those of the perfect gas but the concept provides a useful basis for understanding the properties of real gases and by extension the properties of liquid mixtures and solutions. After all, nothing is perfect. The starting point for the analysis is the following equation (see Topic 2500) for the change in thermodynamic energy of a closed system $\mathrm{dU}$ at temperature $\mathrm{T}$, pressure $\mathrm{p}$ and affinity for spontaneous change $\mathrm{A}$ [1]. $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi$ Then for processes at equilibrium where $\mathrm{A}$ is zero, $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}$ For one mole of chemical substance $\mathrm{j}$, equation (b) can be written in the following form. $\mathrm{dU}_{\mathrm{j}}=\mathrm{T} \, \mathrm{dS}_{\mathrm{j}}-\mathrm{p} \, \mathrm{dV_{ \textrm {j } }}$ Then, $\mathrm{dS}_{\mathrm{j}}=\frac{\mathrm{dU}_{\mathrm{j}}+\mathrm{p} \, \mathrm{dV} \mathrm{V}_{\mathrm{j}}}{\mathrm{T}}$ The molar isochoric heat capacity $\mathrm{C}_{\mathrm{Vj}}$ describes the differential dependence of molar thermodynamic energy $\mathrm{U}_{j}$ on temperature at fixed volume. Thus $\mathrm{C}_{\mathrm{Vj}}=\left(\partial \mathrm{U}_{\mathrm{j}} / \partial \mathrm{T}\right)_{\mathrm{V}(\mathrm{j})}$ Using equation (d), $\mathrm{dS}_{\mathrm{j}}=\frac{\mathrm{C}_{\mathrm{v}_{\mathrm{j}}}}{\mathrm{T}} \, \mathrm{dT}+\frac{\mathrm{p}}{\mathrm{T}} \, \mathrm{dV}_{\mathrm{j}}$ The latter equation emerges from an equation expressing the molar entropy of an ideal gas $j$ as a function of the independent variables $\mathrm{T}$ and $\mathrm{V}_{j}$. Thus, $\mathrm{S}_{\mathrm{j}}=\mathrm{S}_{\mathrm{j}}\left[\mathrm{T}, \mathrm{V}_{\mathrm{j}}\right]$ According to Joules Law [2]. The molar thermodynamic energy of a perfect gas depends only on temperature. Hence from equation (e) the molar isochoric heat capacity $\mathrm{C}_{\mathrm{Vj}}$ is solely a function of temperature. Therefore equation (f) yields the following two important equations [3]. $\left(\frac{\partial S_{j}}{\partial T}\right)_{v}=\frac{C_{v_{j}}}{T}$ $\left(\frac{\partial S_{j}}{\partial V}\right)_{T}=\frac{p}{T}$ According to Boyles Law, the molar volume of gas $j$ is inversely proportional to the pressure at fixed temperature. Thus $V_{j}=f(T) / p$ Alternatively $\mathrm{p}=\mathrm{f}(\mathrm{T}) / \mathrm{V}_{\mathrm{j}}$ Hence using equation (i), $\left(\frac{\partial S_{j}}{\partial V}\right)_{T}=\frac{f(T)}{T \, V_{j}}$ A calculus condition requires that $\frac{\partial}{\partial V}\left(\frac{\partial S}{\partial T}\right)=\frac{\partial}{\partial T}\left(\frac{\partial S}{\partial V}\right)$ In other words, $\frac{\partial\left(\mathrm{C}_{\mathrm{vj}} / \mathrm{T}\right)}{\partial \mathrm{V}}=\frac{\partial(\mathrm{p} / \mathrm{T})}{\partial \mathrm{T}}$ Or, using equation (k), $\frac{\partial\left(\mathrm{C}_{\mathrm{v}_{\mathrm{j}}} / \mathrm{T}\right)}{\partial \mathrm{V}}=\frac{\partial\left[\mathrm{f}(\mathrm{T}) / \mathrm{T} \, \mathrm{V}_{\mathrm{j}}\right]}{\partial \mathrm{T}}$ But the isochoric heat capacity $\mathrm{C}_{\mathrm{Vj}}$ is independent of volume. Hence $\frac{\partial\left(\mathrm{C}_{\mathrm{v}_{\mathrm{j}}} / \mathrm{T}\right)}{\partial \mathrm{V}}=0$ Then, $\frac{\partial\left[\mathrm{f}(\mathrm{T}) / \mathrm{T} \, \mathrm{V}_{\mathrm{j}}\right]}{\partial \mathrm{T}}=0$ In other words $[\mathrm{f}(\mathrm{T}) / \mathrm{T}]$ must be a constant, conventionally called the Gas Constant with symbol $\mathrm{R}$. As the name implies $\mathrm{R}$ is a constant used to describe the properties of all gases [3]. We can therefore rewrite equation (k) as follows (recalling that $\mathrm{V}_{\mathrm{j}}$ is the molar volume of a perfect gas) [4,5]. $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}$ The perfect gas is an artificial chemical substance having defined properties. The link with reality stems from the idea that the properties of real gases approach those of an ideal gas as the pressure is reduced. In addition to the definition given by equation (q), the ideal gas is defined by the following equation which requires that the thermodynamic energy of an ideal gas is independent of volume, being nevertheless a function of temperature. $\left(\partial \mathrm{U}_{\mathrm{j}} / \partial \mathrm{V}_{\mathrm{j}}\right)_{\mathrm{T}}=0$ Footnotes [1] I. Prigogine and R Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Green, London, 1954, chapter X. [2] Reference 1, page 116. [3] \begin{aligned} &\mathrm{R}=8.31450 \mathrm{~J} \mathrm{~K}^{-1} \mathrm{~mol}^{-1} \ &\mathrm{R}=\mathrm{N}_{\mathrm{A}} \, \mathrm{k} \end{aligned} where $\mathrm{N}_{\mathrm{A}} =$ Avogadro’s constant and k = Boltzmann’s constant \begin{aligned} &\mathrm{k}=1.380658 \times 10^{-23} \mathrm{~J} \mathrm{~K}^{-1} \ &\mathrm{~N}_{\mathrm{A}}=6.0221367 \times 10^{23} \mathrm{~mol}^{-1} \end{aligned} [4] G. N. Lewis and M. Randall, Thermodynamics, McGraw-Hill, 1923, page 63. [5] P. W. Atkins, Concepts in Physical Chemistry, Oxford University Press, Oxford,1995.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.19%3A_Perfect_and_Real_Gases/1.19.1%3A_Perfect_and_Real_Gases.txt
In the nineteenth century a major theme in science concerned the properties of gases and their liquefaction. The challenge offered by the liquefaction of gases also prompted the development of thermodynamics and the production of low temperatures. Michael Faraday is noteworthy in this context. In 1823 Faraday had liquefied chlorine, ammonia and sulfur dioxide using a combination of pressure and low temperatures [1]. Based on the observation that the densities of liquids are higher than gases the expectation was that liquefaction of a given gas would follow application of high pressure. However this turned out not to be the case. For example, Newton showed that application of 2790 atmospheres ($\equiv 2.8 \times 10^{8} \mathrm{~N} \mathrm{~m}^{-2}$) did not liquefy air. In 1877 Cailletet and Pictet working independently obtained a mist of oxygen by sudden expansion of gas compressed at 300 atmospheres ($\equiv 3 \times 10^{7} \mathrm{~N} \mathrm{~m}^{-2}$) and cooled by $\mathrm{CO}_{2}(\mathrm{s})$. However in developing the background to this subject we turn attention to the work of Joule. Experiments by Joule A gas, chemical substance $j$, is held in a closed system. The molar thermodynamic energy $\mathrm{U}_{j}$ is defined by equation (a) where $\mathrm{V}_{j}$ is the molar volume and $\mathrm{T}$, the temperature. $\mathrm{U}_{\mathrm{j}}=\mathrm{U}_{\mathrm{j}}\left[\mathrm{T}, \mathrm{V}_{\mathrm{j}}\right]$ The complete differential of equation (a) describes the change in $\mathrm{U}_{\mathrm{j}}, \mathrm{dU}_{\mathrm{j}}$, as a function of temperature and volume. $\mathrm{dU}_{\mathrm{j}}=\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{v}} \, \mathrm{dT}+\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}} \, \mathrm{dV}_{\mathrm{j}}$ The molar isochoric heat capacity $\mathrm{C}_{\mathrm{Vj}}$ is defined by equation (c). $\mathrm{C}_{\mathrm{Vj}}=\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ The first law of thermodynamics relates the change in $\mathrm{U}_{j}$ to the work done on the system $\mathrm{w}$ and heat $\mathrm{q}$ passing from the surroundings into the system. Thus, $\mathrm{dU}_{\mathrm{j}}=\mathrm{q}+\mathrm{w}$ Hence, $\mathrm{q}=\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{V}} \, \mathrm{dT}+\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}} \, \mathrm{dV} \mathrm{V}_{\mathrm{j}}-\mathrm{w}$ The apparatus used by Joule comprised two linked vessels having equal volumes. A tube joining the two vessels included a tap. In an experiment, one vessel was filled with gas $j$ at a known pressure whereas the second vessel was evacuated. When the tap was opened gas flowed into the second vessel, equalizing the pressure in the two vessels. By flowing into an evacuated vessel the gas did no work because there was no confining pressure; i.e. $\mathrm{w} =$ zero. The temperature of the gas in the containing vessel fell and that in the originally empty vessel rose by an equal amount. In other words $\mathrm{dT}$ for the two vessel system is zero. Hence $\left(\frac{\partial \mathrm{U}_{j}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=0$ The clear hope was that the temperature would fall dramatically leading to liquefaction of the gas. In fact and with the benefit of hindsight the overall change in temperature $\mathrm{dT}$ was too small to be measured. More sophisticated apparatus would show that $\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{V}} \neq 0$ because as the gas expands work is done against cohesive intermolecular interaction. It would only be zero for a perfect gas. Experiments by Joule and Thomson [2,3] In a series of famous experiments carried out in an English brewery, Joule and Thomson used an apparatus in which the gas under study passed through a porous plug from high to low pressures. The plug impeded the flow of the gas such that the pressure of the gas on the high pressure side and the pressure of gas on the low pressure side remained constant. It was observed that the temperature of the gas decreased as a consequence of the work done by the gas against intermolecular cohesion. A technological breakthrough was now made. A portion of the cooled gas was re-cycled to cool the gas on the input side. On passing through the plug the temperature of the gas fell to a lower temperature. As this process continues a stage was reached where a fraction of the gas is liquefied. As noted above, the cooling emerges because work is done on expansion of the gases against intermolecular interaction. This is a quite general observation. When the pressure drops the mean intermolecular distance increases with the result that the temperature decreases. However there are exceptions to this generalisation. If the pressure is high the dominant intermolecular force is repulsion. Consequently when the pressure drops, work is done by the repulsive forces increasing the intermolecular distances thereby raising the temperature. Footnotes [1] The account given here is based on that given by N. K. Adam, Physical Chemistry, Oxford, The Clarendon Press, 1956, chapter III. [2] Thomson ≡ Lord Kelvin [3] J. P. Joule and W. Thomson, Proc. Roy. Soc.,1853,143,3457. [4] G.N. Lewis and M. Randall, Thermodynamics, revised by K. S. Pitzer and L Brewer, McGraw-Hill, 2nd. edn.,1961, New York, pages 47-49.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.19%3A_Perfect_and_Real_Gases/1.19.3%3A_Real_Gases%3A_Liquefaction_of_Gases.txt
A given plane surface phase contains i-chemical substances having amounts $\mathrm{n}_{j}$ for each $j$-chemical substance. The plane surface phase is perturbed leading to a change in thermodynamic energy $\mathrm{dU}^{\sigma}$ where the symbol $\sigma$ identifies the surface phase. Using the Master Equation as a guide we set down the corresponding fundamental equation for the plane surface phase [1-4]. $\mathrm{dU}^{\sigma}=\mathrm{T} \, \mathrm{dS}^{\sigma}-\mathrm{p} \, \mathrm{dV}^{\sigma}+\gamma \, \mathrm{dA}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mu_{\mathrm{j}} \, \mathrm{dn}_{\mathrm{j}}^{\sigma}$ With reference to equation (a), $\mathrm{T}, \mathrm{~p}, \gamma \text { and } \mu_{j}$ are intensive variables whereas $\mathrm{~U}, \mathrm{~S}, \mathrm{~V}, \mathrm{~A} \text { and } \mathrm{n}_{j}$ are extensive variables. We integrate equation (a) to yield equation (b). $\mathrm{U}^{\sigma}=\mathrm{T} \, \mathrm{S}^{\sigma}-\mathrm{p} \, \mathrm{V}^{\sigma}+\gamma \, \mathrm{A}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mu_{\mathrm{j}} \, \mathrm{n}_{\mathrm{j}}^{\sigma}$ The general differential of equation (b) is equation (c). $\begin{gathered} \mathrm{dU}^{\sigma}=\mathrm{T} \, \mathrm{dS}{ }^{\sigma}+\mathrm{S}^{\sigma} \, \mathrm{dT}-\mathrm{p} \, \mathrm{dV}{ }^{\sigma}-\mathrm{V}^{\sigma} \, \mathrm{dp}+\gamma \, \mathrm{dA}+\mathrm{A} \, \mathrm{d} \gamma \ +\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mu_{\mathrm{j}} \, \mathrm{dn}{ }_{\mathrm{j}}^{\sigma}+\sum_{\mathrm{j}=1}^{\mathrm{o}} \mathrm{n}_{\mathrm{j}}^{\sigma} \, \mathrm{d} \mu_{\mathrm{j}} \end{gathered}$ Using equations (a) and (c). $0=\mathrm{S}^{\sigma} \, \mathrm{dT}-\mathrm{V}^{\sigma} \, \mathrm{dp}+\mathrm{A} \, \mathrm{d} \gamma+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}}^{\sigma} \, \mathrm{d} \mu_{\mathrm{j}}$ Hence for a surface phase $\sigma$ at fixed $\mathrm{T}$ and $\mathrm{p}$, $0=\mathrm{A} \, \mathrm{d} \gamma+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}}^{\sigma} \, \mathrm{d} \mu_{\mathrm{j}}$ We restrict attention to two component systems, comprising components labelled 1 and 2 wherein there are two phases $\alpha$ and $\beta$. In practice the boundary between phases $\alpha$ and $\beta$ comprises a region across which the compositions of small sample volumes change from pure $\mathrm{i}$ to pure $\mathrm{j}$. We imagine that the boundary layer can be replaced by a surface. For component $j$, $\Gamma_{j}$ is the amount of chemical substance $j$ adsorbed per unit area; i.e. the surface concentration expressed in $\mathrm{mol m}^{-2}$. Then, $0=\mathrm{d} \gamma+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \Gamma_{\mathrm{j}} \, \mathrm{d} \mu_{\mathrm{j}}$ Consider the case of a system prepared using two chemical substances, 1 and 2. We set the interphase between the two phases by a mathematical plane where $\Gamma_{1}$ is zero. Then $\Gamma_{2}=-\mathrm{d} \gamma / \mathrm{d} \mu_{2}$ The definition based on $\Gamma_{1}$ defines the surface excess per unit area of the surface separating the two phases. Chemical substance 1 is the solvent (e.g. water) and chemical substance 2 is the solute. The surface divides the liquid and vapor phases. Equation (g) describes the surface excess of the solute. We assume that the surface and bulk aqueous phases are in (thermodynamic) equilibrium. Moreover we assume that the thermodynamic properties of the solutions are ideal. Then in terms of the concentration scale (where $\mathrm{c}_{r} = 1 \mathrm{~mol dm}^{-3}$), $\mu_{2}(a q)=\mu_{2}^{0}(a q)+R \, T \, \ln \left(c_{2} / c_{r}\right)$ Then, $\mathrm{d} \mu_{2}(\mathrm{aq})=\mathrm{R} \, \mathrm{T} \, \mathrm{d} \ln \left(\mathrm{c}_{2} / \mathrm{c}_{\mathrm{r}}\right)$ Hence, $\Gamma_{2}=-\frac{1}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{d} \gamma}{\mathrm{d} \ln \left(\mathrm{c}_{2} / \mathrm{c}_{\mathrm{r}}\right)}$ Equation (j) is the ‘Gibbs adsorption equation’ for a two-component system using the Gibbs definition of surface excess. The validity of the Gibbs treatment was confirmed in 1932 by McBain who used an automated fast knife to remove a layer between 5 and 1 mm thick from the surface of a solution [5]. The compositions of this layer and the solution were than analyzed. As N. K. Adam comments [1] ‘in every case so far examined’, the measured adsorption agreed with that predicted by Equation (j). If $\gamma$ decreases with increase in $\mathrm{c}_{2}$, $\Gamma_{2}$ is positive as is the case for organic solutes then these solutes are positively adsorbed at the air-water interface. The reverse pattern is observed for salt solutions. Footnotes [1] N. K. Adam, The Physics and Chemistry of Surfaces, Dover, New York, 1968; a corrected version of the third edition was published in 1941 by Oxford University Press. [2] N. K. Adam, Physical Chemistry, Oxford, 1956, chapter XVII. [3] S. E. Glasstone, Physical Chemistry, MacMillan, London, 1948, chapter XIV. As forcefully expressed to undergraduates, N. K. Adam did not like the symbols used in this reference. [4] G. N. Lewis and M. Randall, Thermodynamics, revised by K. S. Pitzer and L. Brewer, McGraw-Hill, New York, 2nd edn., 1961, chapter 29. [5] J. W. McBain and C. W. Humphreys, J. Phys.Chem.,1932,36,300.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.20%3A_Surfactants/1.20.1%3A_Surface_Phase%3A_Gibbs_Adsorption_Isotherm.txt
‘Nonsense, McBain’. The story is told that with these words an eminent scientist, chairman of a meeting of the Royal Society in London in the early 20th Century, reacted to the proposal by J. W. McBain that surfactants [ = surface active agents] might aggregate in aqueous solution [1]. Subsequent events confirmed that the nub of McBain’s model is correct. The enormous industry based on ‘soaps’ and detergents [2] prompts intensive studies of these complicated systems, supported by monographs [3-5] and detailed reviews [6-11]. In the context of aqueous solutions, surfactant molecules are amphipathic meaning that the solutes have dual characteristics : amphi ≡ dual and pathi ≡ sympathy. These dual characteristics emerge because a solute molecule contains both hydrophobic and hydrophilic parts. The subject is complicated because there is no general agreement concerning the nature-structure of these solute aggregates in aqueous solutions. Amphipathic molecules are broadly classified as either ionic or non-polar. Ionic surfactants are typified by salts such as sodium dodecylsulfate ($\mathrm{C}_{12}\mathrm{H}_{25}\mathrm{OSO}_{3}^{-}\mathrm{~Na}^{+}$) and hexyltrimethylammonium bromide ($\mathrm{C}_{16}\mathrm{H}_{33}\mathrm{N}^{+} \mathrm{~Me}_{3}\mathrm{Br}^{-}$; CTAB). Non-ionic surfactants are typified by those based on ethylene oxide; e.g. hexaethylene glycol dodecylether, ($\mathrm{C}_{16}\mathrm{H}_{25}\left(\mathrm{OCH}_{2}\mathrm{CH}\right)_{6}\mathrm{OH}$). When small amounts of a given surfactant are gradually added to a given volume of water ($\ell$), the properties of the aqueous solutions are unexceptional until the concentration of the surfactant exceeds a characteristic concentration of surfactant (at defined $\mathrm{T}$ and $\mathrm{p}$) called the critical micellar concentration, cmc. At this point further added surfactant exists in solution as aggregates of generally 20 to 100 monomers, which are called micelles. The formation of micelles is often signalled by a change in the pattern of the dependence of a given property $\mathrm{P}$ of a solution on surfactant concentration. The property can be surface tension, molar conductance of an ionic surfactant, uv-visible absorption spectra of water soluble dyes (or, an iron complex [12]). When more surfactant is added, the micelles cluster to form more complicated aggregates [13]. Micelles are not formed by the gradual association of monomers, forming dimers, trimers…..[14] . Rather micelles are examples of organized structures spontaneously formed by simple molecules [15]. A quoted aggregation number is not a stoichiometric number. The quoted number (e.g. approx. 90 for CTAB at $298 \mathrm{~K}$) is taken as an ‘average’ over the micelles in a given system. In terms of the structure of micelles in aqueous solutions, key questions centre on 1. the extent to which water penetrates into a micelle, and 2. the organization of alkyl chains in a given micelle [16]. The two questions are linked by the question – to what extent does the terminal group in, for example, the hexadecyl chains of CTAB come into contact with the aqueous solution? If the answer is ‘never’, micelles have a structure in which there is a well organized hydrophobic core. If the answer is ‘frequently’ the micelles are very dynamic assemblies with continuous changes in organization/structure. We do not become involved in this debate. However we note that there is such a debate over what precisely the thermodynamic analysis is asked to describe. Description of micellar-surfactant systems emerges from the Phase Rule. The number of phases = 3; vapor, aqueous solution and micelle. The number of components = 2; water and surfactant. Hence having defined the temperature, the remaining intensive variables are defined; i.e. vapor pressure, mole fraction of surfactant monomer in aqueous solution, and mole fraction of surfactant in micellar phase. The starting point for a thermodynamic analysis is the assumption of an equilibrium between surfactant monomers and micelles in solution (at defined $\mathrm{T}$ and $\mathrm{p}$). A key feature of these systems is that above the cmc when more surfactant is added the concentration of monomers remains essentially constant, the added surfactant existing in micellar form. Two models for micelle formation are discussed; 1. phase equilibrium [17], and 2. closed association model, often called the mass-action model. Here we assume that the closed system containing solvent and surfactant is at equilibrium at defined $\mathrm{T}$ and $\mathrm{p}$. Each system is at a unique minimum in Gibbs energy. The system is at ambient pressure, which for our purposes is effectively the standard pressure $\mathrm{p}^{0}$. We characterize micelle formation using thermodynamic variables describing monomers and micelles. Then $\Delta_{\text {mic }} G^{0}$ is the standard Gibbs energy of micelle formation which is dependent on temperature. The latter dependence is characterized by the standard enthalpy of micelle formation, $\Delta_{\text {mic }} H^{0}$, where, $\Delta_{\text {mic }} \mathrm{G}^{0}=\Delta_{\text {mic }} \mathrm{H}^{0}-\mathrm{T} \, \Delta_{\text {mic }} \mathrm{S}^{0}$ Clearly the definition of $\Delta_{\text {mic }} \mathrm{G}^{0}$ is directly associated with the definition of standard states for both the simple salt in solution and the micelles in the aqueous system. These thermodynamic variables together with aggregation numbers are extensively documented [18]. Footnotes [1] F. M. Menger, Acc. Chem.Res.,1979,12,111; and references therein. [2] E. M. Kirshner, Chem. Eng. News, 1998,76,39 [3] J. H. Clint, Surfactant Aggregation, Blackie, Glasgow, 1992. [4] Y. Mori, Micelles, Plenum Press, New York, 1992. [5] D. F. Evans and H. Wennerstrom, The Colloidal Domain, VCH, New York, 1994. [6] B. Lindman and H. Wennerstrom, Top. Curr. Chem., 1980, 87,1. [7] G. C. Kresheck in, Water: A Comprehensive Treatise, ed. F. Franks, Plenum Press, New York, 1973, volume 4,chapter 2. [8] J. E. Desnoyers, 1. J. Surface Sci. Technol., 1989,5,289; 2. Pure Appl. Chem.,1982,54,1469. [9] Polymer-surfactant interactions; 1. E.D.Goddard, JAOCS, 1994,71,1. 2. J. Kevelam, J. F. L. van Breemen, W. Blokzijl and J. B. F. N. Engberts, Langmuir, 1996, 12, 4709. [10] Solubilisation by micelles; C. Treiner, Chem. Soc. Rev.,1994,23,349. [11] Alcohol-micelle interactions; R. E. Verrall, Chem. Soc. Rev.,1995,24,79. [12] 1. N. M. van Os, G. J. Daane and G. Haandrikman, J.Colloid Interface Sci.,1991,141,199. 2. M. J. Blandamer, B. Briggs, J. Burgess, P. M. Cullis and G. Eaton, J. Chem. Soc. Faraday Trans.,1991,87,1169. [13] S. Backlund, H. Hoiland, O. J. Krammen and E. Ljosland, Acta Chem. Scand,,Ser. A, 1982, 87,1169. [14] But see D. Schuhman, Prog. Colloid Polym. Sci.,1989,71,338. [15] See for example, K. Shinoda, Langmuir, 1991,7,2877. [16] F. Menger and D. W. Doll , J. Am. Chem.Soc.,1984,106,1109. [17] S. Puvvada and D. Blankschtein, J.Phys.Chem.,1992,96, 5567. [18] N. M. van Os, J. R. Haak and L A. M. Rupert, Physico-Chemical Properties of Selected Anionic, Cationic and Non-Ionic Surfactants, Elsevier, Amsterdam,1993.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.20%3A_Surfactants/1.20.2%3A_Surfactants_and_Micelles.txt
The properties of aqueous solutions containing non-ionic surfactants [1-6] can be described using two models. Model Phase Equilibrium: Dry Micelle We envisage a non-ionic surfactant $\mathrm{X}$. When chemical substance $\mathrm{X}$ is added to $\mathrm{n}_{1}$ moles of water (at fixed $\mathrm{T}$ and $\mathrm{p}$), solute $\mathrm{X}$ exists as a simple solute $\mathrm{X}(\mathrm{aq})$ until the concentration of solute $\mathrm{X}$, $\mathrm{c}_{\mathrm{X}}$ reaches a characteristic concentration $\mathrm{cmc}_{\mathrm{X}}$ when a trace amount of the micellar phase appears. Each micelle comprises $\mathrm{n}$ molecules of surfactant $\mathrm{X}$. The equilibrium between monomer surfactant $\mathrm{X}(\aq) and surfactant in the micelles is described by the following equation. $\mu_{\mathrm{X}}^{\mathrm{eq}}(\mathrm{aq})=\mu_{\mathrm{X}}^{*}(\mathrm{mic})$ If \(\mathrm{X}(\mathrm{aq})$ is a typical neutral solute in aqueous solution $\mu_{\mathrm{x}}^{\mathrm{eq}}(\mathrm{aq})$ is related to the cmc of $\mathrm{X}(\mathrm{aq})$ in solution at the point where only a trace amount of micellar phase exists. Hence, $\mu_{\mathrm{X}}^{0}(\mathrm{aq} ; \mathrm{c}-\mathrm{scale})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc} \, \mathrm{y}_{\mathrm{X}} / \mathrm{c}_{\mathrm{r}}\right)=\mu_{\mathrm{X}}^{\mathrm{*}}(\mathrm{mic})$ Here $\mathrm{y}_{\mathrm{X}}$ is the solute activity coefficient for $\mathrm{X}(\mathrm{aq})$ taking account of solute-solute interactions in the aqueous solution. Therefore, by definition, $\Delta_{\text {mic }} \mathrm{G}^{0}(\mathrm{aq})=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{\text {mic }}^{0}\right)=\mu_{\mathrm{X}}^{*}(\mathrm{mic})-\mu_{\mathrm{x}}^{0}(\mathrm{aq} ; \mathrm{c}-\text { scale })$ Equilibrium constant $\mathrm{K}_{\text {mic }}^{0}$ describes the phase equilibrium involving surfactant X in aqueous solution and micellar phase. $\Delta_{\text {mic }} \mathrm{G}^{0}(\mathrm{aq})=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{\mathrm{mic}}^{0}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc} \, \mathrm{y}_{\mathrm{X}} / \mathrm{c}_{\mathrm{r}}\right)$ If $\mathrm{X}(\mathrm{aq})$ is a neutral solute and the cmc is low, a useful approximation sets $\mathrm{y}_{\mathrm{X}}$ at unity. Therefore $\Delta_{\text {mic }} \mathrm{G}^{0}(\mathrm{aq})$ is the standard increase in Gibbs energy when one mole of surfactant $\mathrm{X}(\mathrm{aq})$ forms one mole of $\mathrm{X}$ in the micellar phase. Combination of equations (c) and (d) yields equation (e). $\mathrm{K}_{\text {mic }}^{0}=\left(\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right)^{-1}$ Here $\mathrm{K}_{\text {mic }}^{0}$ describes the equilibrium between surfactant in the micellar phase and the aqueous solution. A famous equation suggested by Harkin relates the cmc to the number of carbon atoms in the alkyl chain, $\mathrm{n}_{\mathrm{C}}$; equation (f) $\log \left(\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right)=\mathrm{A}-\mathrm{B} \, \mathrm{n}_{\mathrm{C}}$ With equation (e), $\log \left(\mathrm{K}_{\text {mic }}^{0}\right)=A-\left(B \, n_{C}\right)$ The above analysis is also used for ionic surfactants if it can be assumed the degree of counter ion binding by the micelles is small, the thermodynamic properties of the solution are ideal and the aggregation number is high. Non-Ionic Surfactant: Phase Equilibrium: Wet Micelle The aqueous phase comprises an aqueous solution of solute $\mathrm{X}$, $\mathrm{X}(\mathrm{aq})$. The micellar phase comprises both water and surfactant $\mathrm{X}$ such that the mole fraction of surfactant in the micellar phase equals $\mathbf{X}_{\mathrm{X}}^{\mathrm{eq}}$. We treat the micellar phase using the procedures used to describe the properties of a binary liquid mixture. For the micellar phase the chemical potential of $\mathrm{X}$ is given by the following equation. $\mu_{\mathrm{x}}(\mathrm{mic})=\mu_{\mathrm{x}}^{*}(\text { mic })+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{\mathrm{x}} \, \mathrm{f}_{\mathrm{x}}\right)^{\mathrm{eq}}$ where $\operatorname{limit}\left(x_{x} \rightarrow 1\right) f_{x}=1 \text { at all } \mathrm{T} \text { and } \mathrm{p}$ But at equilibrium for a system containing a trace of the micellar phase, $\mu_{\mathrm{x}}^{\mathrm{eq}}(\mathrm{mic})=\mu_{\mathrm{x}}^{\mathrm{eq}}(\mathrm{aq})$ Then, $\mu_{\mathrm{X}}^{\circ}(\text { mic })+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{\mathrm{X}} \, \mathrm{f}_{\mathrm{X}}\right)^{\mathrm{cq}}=\mu_{\mathrm{X}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc}_{\mathrm{x}} \, \mathrm{y}_{\mathrm{X}} / \mathrm{c}_{\mathrm{r}}\right)$ By definition $\Delta_{\text {mic }} \mathrm{G}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{\text {mic }}^{0}\right)=\mu_{\mathrm{x}}^{*}(\mathrm{mic})-\mu_{\mathrm{x}}^{0}(\mathrm{aq})$ Then, $\mathrm{K}_{\text {mic }}^{0}=\left[\mathrm{x}_{\mathrm{X}} \, \mathrm{f}_{\mathrm{X}}\right]^{\mathrm{eq}} /\left[\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right]$ If the micelle is only ‘damp’ rather than wet , a reasonable assumption sets $\mathrm{f}_{\mathrm{X}}$ equal to unity although it is not obvious how $\mathrm{x}_{\mathrm{X}}$ might be determined. Non-Ionic Surfactant: Mass Action Model: Dry Micelle Micelle formation is described as an equilibrium between $\mathrm{X}(\mathrm{aq})$ as a solute in aqueous solution and a micellar aggregate in aqueous solution formed by n molecules of the monomer $\mathrm{X}(\mathrm{aq})$. Then at the point where micelles are first formed, the following equilibrium is established. $\mathrm{nX}(\mathrm{aq}) \Leftrightarrow \mathrm{X}_{\mathrm{n}}(\mathrm{aq})$ The total amount of surfactant in the system equals $\mathrm{N}(\mathrm{X} ; \mathrm{aq})+\mathrm{n} \, \mathrm{N}\left(\mathrm{X}_{\mathrm{n}} ; \mathrm{mic}\right)$ where $\mathrm{N}(\mathrm{X} ; \mathrm{aq})$ is the amount of monomer surfactant and where $\mathrm{N}\left(\mathrm{X}_{\mathrm{n}} ; \mathrm{mic}\right)$ is the total amount of micelles , each micelle containing n surfactant molecules. But for the micellar aggregate $\mathrm{X}_{\mathrm{n}}(\mathrm{aq})$ treated as a single solute, $\mu\left(\mathrm{X}_{\mathrm{n}} ; \mathrm{aq}\right)=\mu^{0}\left(\mathrm{X}_{\mathrm{n}} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{c}\left(\mathrm{X}_{\mathrm{n}}\right) \, \mathrm{y}\left(\mathrm{X}_{\mathrm{n}}\right) / \mathrm{c}_{\mathrm{r}}\right]$ Here $c\left(X_{n}\right)\left[=N\left(X_{n} ; a q\right) / V\right.$ where $\mathrm{V}$ is the volume of the system] is the concentration of micelles in the system, activity coefficient $\mathrm{y}\left(\mathrm{X}_{\mathrm{n}}\right)$.The latter can be assumed to be unity if there are no micelle-micelle interactions and no micelle-monomer interactions in the aqueous system. Although this approach seems similar to that used to describe chemical equilibria, the procedure has problems in the context of determining $\mathrm{c}\left(\mathrm{X}_{\mathrm{n}}\right)$. Footnotes [1] J. E. Desnoyers, G. Caron, R. DeLisi, D. Roberts, A. Roux and G. Perron, J. Phys. Chem.,1983,87, 1397. [2] G. Olofsson, J.Phys.Chem.,1985, 89,1473. [3] J. E. Desnoyers, Pure Appl.Chem.,1982, 54,1469. [4] M. J. Blandamer, J. M. Permann, J. Kevelam, H. A. van Doren, R. M. Kellogg and J. B. F. N. Engberts, Langmuir, 1999,15,2009. [5] M. J. Blandamer, K. Bijma and J. B. F. N. Engberts, Langmuir, 1998,14,79. [6] M. J. Blandamer, B. Briggs, P. M. Cullis, J. B. F. N. Engberts and J. Kevelam, Phys.Chem.Chem.Phys.,2000,2,4369.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.20%3A_Surfactants/1.20.3%3A_Surfactants_and_Micelles%3A_Non-Ionics.txt
An intense debate concerns the structure of micelles, particularly those formed by ionic surfactants such as SDS and CTAB. It seems generally agreed that micelles are essentially spherical in shape. The polar head groups ( e.g. $– \mathrm{N}^{+} \mathrm{Me}_{3}$) are at the surface of each micelle, having strong interactions with the surrounding solvent. In close proximity in the Stern layer are counterions (e.g. bromide ions in the case of CTAB); the aggregation number $\mathrm{n}$ describes the number of cations which form each micelle. The total charge on the micelle is determined the aggregation number and a quantity $\beta$, the latter being the fraction of charge of aggregated ions forming the micelle neutralized by the micelle bound counter ions. The remaining fraction of counter ions exists as ‘free’ ions in aqueous solution. Both $mathrm{n}$ and $\beta$ are characteristic of a given surfactant system, and are obtained from analysis of experimental data [1]. The properties of ionic surfactants have been extensively studied [2-14]. Here we examine four thermodynamic descriptions of these systems. Ionic Surfactant:1:1 salt: Phase Equilibrium: Dry Neutral Micelle We consider a dilute aqueous solution of an ionic surfactant; e.g. $\mathrm{AM}^{+} \mathrm{Br}^{-}$. As more surfactant is added a trace amount of micelles appear in the solution when the concentration of surfactant just exceeds the cmc. The trace amount of surfactant is present as micelles constituting a micellar phase. At defined $\mathrm{T}$ and $\mathrm{p}$, the following equilibrium is established in the case of the model surfactant $\mathrm{AM}^{+} \mathrm{Br}^{-}$; $\mathrm{AM}^{+} \mathrm{Br}^{-}(\mathrm{aq}) \Leftrightarrow \mathrm{AM}^{+} \mathrm{Br}^{-}(\mathrm{mic})$ Then, $\mu^{\mathrm{cq}}\left[\mathrm{AM}^{+} \mathrm{Br}^{-}(\mathrm{aq})\right]=\mu^{\mathrm{cq}}\left[\mathrm{AM}^{+} \mathrm{Br}^{-}(\mathrm{mic})\right]$ We assume that the micelles carry no charge. The chemical potential of the surfactant in aqueous solution is related to the cmc using the following equation where $\mathrm{y}_{\pm}$ is the mean ionic activity coefficient. We set $\mu^{\mathrm{eq}}\left[\mathrm{AM}^{+} \mathrm{Br}^{-} \text {(mic) }\right]$ equal to the chemical potential of the surfactant in the pure micellar state, $\mu^{*}\left[\mathrm{AM}^{+} \mathrm{Br}^{-} \text {(mic) }\right]$. \begin{aligned} \mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \mathrm{aq} ; \mathrm{c}-\mathrm{scale}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc} \, \mathrm{y}_{\pm} \, / \mathrm{c}_{\mathrm{r}}\right) \ &=\mu^{*}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \text {micellar phase }\right) \end{aligned} Here $\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \mathrm{aq} ; \mathrm{c}-\text { scale }\right)$ is the chemical potential of the salt $\mathrm{AM}^{+} \mathrm{Br}^{-}$ in aqueous solution at unit concentration where the properties of the salt are ideal. Thus $\mathrm{y}_{\pm}$ describes the role of ion-ion interactions in the solution having salt concentration cmc. Because the model states that there is only a trace amount of micelles in the system, we do not take account of salt-micelle interactions. Then $\Delta_{\text {mic }} \mathrm{G}^{0}=\mu^{*}\left(\text { micellar phase; } \mathrm{AM}^{+} \mathrm{Br}^{-}\right)-\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \text {aq; } \mathrm{c}-\text { scale }\right)$ Hence, $\Delta_{\text {mic }} G^{0}(\mathrm{aq} ; \mathrm{c}-\text { scale })=2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc} \, \mathrm{y}_{\pm} / \mathrm{c}_{\mathrm{r}}\right)$ If the salt concentration in the aqueous solution at the cmc is quite low, a useful assumption sets $\mathrm{y}_{\pm}$ equal to unity. Then, $\Delta_{\text {mic }} \mathrm{G}^{0}(\mathrm{aq} ; \mathrm{c}-\mathrm{scale})=2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right)$ The latter equation leads to the calculation of the standard increase in Gibbs energy when one mole of salt $\mathrm{AM}^{+} \mathrm{Br}^{-}$ passes from the ideal solution, concentration $1 \mathrm{mol dm}^{-3}$ to the micellar phase. There is a modest problem with the latter equation which can raise conceptual problems. As normally stated the cmc for a given salt is expressed using the unit ‘$\mathrm{mol dm}^{-3}$‘ so that $\mathrm{c}_{\mathrm{r}} = 1 \mathrm{~mol dm}^{-3}$. This means that when $\mathrm{cmc} > 1 \mathrm{~mol dm}^{-3}$, $\Delta_{\operatorname{mic}} G^{0}(\mathrm{aq} ; \mathrm{c}-\text { scale })$ is positive. For solutes where $\mathrm{cmc} < 1 \mathrm{~mol dm}^{-3}$, the derived quantity is negative. Another approach expresses the cmc using the mole fractions, cmx such that equation (c) is written as follows. $\begin{gathered} \mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \text {aq } ; \mathrm{x}-\mathrm{scale}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{cmx} \, \mathrm{f}_{\pm}^{*}\right) \ =\mu^{*}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \text {micellar phase }\right) \end{gathered}$ Here $\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \mathrm{aq} ; \mathrm{x}-\text { scale }\right)$ is the chemical potential of the salt $\mathrm{AM}^{+} \mathrm{Br}^{-}$ in an ideal solution where the (asymmetric) activity coefficient $\mathrm{f}_{\pm}^{*}=1.0$ and $\mathrm{cmx} = 1.0$. . By definition $\operatorname{limit}\left[x\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right) \rightarrow 0\right] \mathrm{f}_{\pm}^{*}=1.0 \text { at all } \mathrm{T} \text { and } \mathrm{p}$.} The analogue of equation (f) takes the following form. $\Delta_{\text {mic }} \mathrm{G}^{0}(\mathrm{aq} ; \mathrm{x}-\text { scale })=2 \, \mathrm{R} \, \mathrm{T} \, \ln (\mathrm{cm} \mathrm{x})$ Because cmx is always less than unity, $\Delta_{\text {mic }} G^{0}(a q ; x-\text { scale })$ is always negative. It is important in these calculations to note the definitions of reference and standard states for solutes and micelles otherwise false conclusions can be drawn [14]. The analysis proceeds to use the Gibbs-Helmholtz equation. Hence, $\Delta_{\text {mic }} \mathrm{H}^{0}(\mathrm{aq} ; \mathrm{x}-\text { scale })=-2 \, \mathrm{R} \, \mathrm{T}^{2} \,\{\partial \ln (\mathrm{cmx}) / \partial \mathrm{T}\}_{\mathrm{p}}$ The term $\left\{\partial \ln (\mathrm{cmx}) / \partial \mathrm{T}_{\mathrm{P}}\right.$ is conveniently obtained by expressing the dependence of cmx on temperature using the following polynomial. $\ln (c m x)=a_{1}+a_{2} \, T+a_{3} \, T^{2}+\ldots$ Equation (h) is straightforward, the stoichiometric factor ‘2’ emerging from the fact that each mole of salt $\mathrm{AM}^{+} \mathrm{Br}^{-}$ produces on complete dissociation 2 moles of ions. A key assumption in this analysis is that the micelles carry no electric charge. In other words a micelle is formed by $\mathrm{n}$ moles of cation $\mathrm{AM}^{+}$, $\mathrm{n}$ moles of counter ions $\mathrm{Br}^{-}$ being bound within the Stern layer such that the charge on each micelle is zero. This model is a little unrealistic. Ionic Surfactant: 1:1 salt: Phase Equilibrium: Dry Charged Micelle A cationic surfactant $\mathrm{AM}^{+} \mathrm{Br}^{-}$ in aqueous solution forms micelles when $\mathrm{n}$ cations come together to form a micellar phase. Bearing in mind that $\mathrm{n}$ might be greater than 20, the idea that there exists micro-phases of macro-cations in a system with an electric charge at least +20 is not attractive. In practice the charge is partially neutralised by bromide ions in the Stern layer. The quantity $\beta$ refers to the fraction of counter ions bound to cations. Thus the formal charge number on each micelle is $[n \,(1-\beta)]$. In the model developed here we represent the formation of the micro-phase comprising the micelles as follows where n is the number of cation monomers which cluster, the remaining bromide ions being present in the aqueous solution (phase). \begin{aligned} \mathrm{nAM}^{+}(\mathrm{aq}) &+\mathrm{n} \,(1-\beta+\beta) \mathrm{Br}^{-}(\mathrm{aq}) \ & \Leftrightarrow\left[\mathrm{nAM}^{+} \mathrm{n} \, \beta \mathrm{Br}^{-}\right]^{\mathrm{n} \,(1-\beta)}(\mathrm{mic})+\mathrm{n} \,(1-\beta) \mathrm{Br}^{-}(\mathrm{aq}) \end{aligned} We re-express this equilibrium in terms of equilibrium chemical potentials for a system at fixed $\mathrm{T}$ and $\mathrm{p}$. \begin{aligned} &\mathrm{n} \, \mu^{\mathrm{cq}}\left(\mathrm{AM}^{+} ; \mathrm{aq}\right)+\mathrm{n} \,(1-\beta+\beta) \, \mu^{\mathrm{eq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right) \ &=\mu^{\mathrm{eq}}\left\{\left[\mathrm{nAM}^{+} \mathrm{n} \, \beta \mathrm{Br}^{-}\right]^{\mathrm{n}(1-\beta)} ; \text { micelle }\right\}+\mathrm{n} \,(1-\beta) \mu^{\mathrm{cq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right) \end{aligned} We define the chemical potential of the micelle microphase which contains 1 mole of $\mathrm{AM}^{+}$. This is a key extrathermodynamic step. We also describe the micelle as a pure ‘phase’. \begin{aligned} \mu^{\mathrm{eq}}\left\{\left[\mathrm{AM}^{+} \beta \mathrm{Br}^{*}\right]^{(1-\beta)} ; \text { micelle }\right\} \ &=\mu^{\mathrm{cq}}\left\{\left[\mathrm{nAM}^{+} \mathrm{n} \, \beta \mathrm{Br}^{-}\right]^{\mathrm{n}(1-\beta)} ; \text { micelle }\right\} / \mathrm{n} \end{aligned} Hence, \begin{aligned} &\mu^{\mathrm{eq}}\left(\mathrm{AM}^{+} ; \mathrm{aq}\right)+\mu^{\mathrm{eq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right) \ &=\mu^{*}\left\{\left[\mathrm{AM}^{+} \beta \mathrm{Br}^{*}\right]^{(1-\beta)} ; \text { micelle }\right\}+(1-\beta) \mu^{\mathrm{eq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right) \end{aligned} Or, \begin{aligned} &\mu^{\mathrm{cq}}\left(\mathrm{AM}^{+} \mathrm{Br}^{-1} ; \mathrm{aq}\right)= \ &\mu^{\mathrm{eq}}\left\{\left[\mathrm{AM}^{+} \beta \mathrm{Br}^{-}\right]^{(1-\beta)} ; \text { micelle\} }+(1-\beta) \mu^{\mathrm{cq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right)\right. \end{aligned} The term $\mu^{थ}\left(\mathrm{AM}^{+} \mathrm{Br}^{-1} ; \mathrm{aq}\right)$ is the equilibrium chemical potential of a 1:1 salt in solution at the cmc. The term $\mu^{\mathrm{eq}}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right)$ is the equilibrium chemical potential of the bromide ion in the solution at the cmc of the surfactant. In any event the system is electrically neutral. \begin{aligned} &\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-1} ; \mathrm{aq}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{cmc} \, \mathrm{y}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right) / \mathrm{c}_{\mathrm{r}}\right] \ &=\mu^{*}\left\{\left[\mathrm{AM}^{+} \beta \mathrm{Br}^{*}\right]^{(1-\beta)} ; \text { micelle }\right\} \ &+(1-\beta) \,\left\{\mu^{0}\left(\mathrm{Br}{ }^{-} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{cmc} \, \mathrm{y}\left(\mathrm{Br}^{-}\right) / \mathrm{c}_{\mathrm{r}}\right]\right\} \end{aligned} By definition, \begin{aligned} \Delta_{\text {mic }} \mathrm{G}^{0}=\mu^{*} &\left\{\left[\mathrm{AM}^{+} \beta \mathrm{Br}^{*}\right]^{(1-\beta)} ; \text { micelle }\right\} \ &+(1-\beta) \, \mu^{0}\left(\mathrm{Br}^{-} ; \mathrm{aq}\right)-\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-1} ; \mathrm{aq}\right) \end{aligned} Assuming both $\mathrm{y}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right)$ and $\mathrm{y}\left(\mathrm{Br}^{-}\right)$ are unity, $\Delta_{\text {mic }} \mathrm{G}^{0}=2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right]-(1-\beta) \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right]$ or $\Delta_{\text {mic }} \mathrm{G}^{0}=(1+\beta) \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{cmc} / \mathrm{c}_{\mathrm{r}}\right]$ The latter equation closely resembles that for non-ionic surfactants for which $\beta$ is unity. For ionic surfactants it is not justified to assume that $\beta$ is also unity. Ionic Surfactant: 1:1 salt: Dry Charged Micelle:Mixed Salt Solutions As more ionic surfactant is added to a solution having the concentration of surfactant equal to the cmc, so the solution increasingly resembles a mixed salt solution, simple salt, charged micelles and counter ions. Analysis of the properties of such solutions was described by Burchfield and Woolley [2-5]. We might develop the analysis from equation (k). An advantage of writing the equation in this form stems from the observation that both sides of the equation describe an electrically neutral system. Woolley and co-- workers [4,5] prefer a form which removes a contribution $\mathrm{n} \,(1-\beta) \mathrm{Br}^{-}(\mathrm{aq})$ from each side of equation (k). $\mathrm{nAM}^{+}(\mathrm{aq})+\mathrm{n} \, \beta \mathrm{Br}^{-}(\mathrm{aq}) \Leftrightarrow\left[\mathrm{nAM}^{+} \mathrm{n} \, \beta \mathrm{Br}^{-}\right]^{\mathrm{a} \,(1-\beta)}(\mathrm{aq})$ Nevertheless one might argue that equation (k) does have the merit in comparing two salts whereas equation (t) describes the links between three ions. In terms of equation (k) , there are two salts in solution. 1. $\mathrm{AM}^{+} \mathrm{Br}^{-}$ where $v_{+}=1, v_{-}=1, v=2, Q=\left(v_{+}^{v+} \, v_{-}^{v-}\right)^{1 / v}=1, y_{\pm}^{v}=y_{+}^{v+} \, y_{-}^{v-}, \text { or } y_{\pm}^{2}=y_{+} \, y_{-}$ But \begin{aligned} &\mu\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right)= \ &\mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{c}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right) \, \mathrm{y}_{\pm}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right) / \mathrm{c}_{\mathrm{r}}\right) \end{aligned} 2. For the micellar salt, $\left[\mathrm{nAM}^{+} \mathrm{n} \, \beta \mathrm{Br}^{-}\right]^{\mathrm{n} \,(1-\beta)} \mathrm{n} \,(1-\beta) \mathrm{Br} \mathrm{r}^{-}$ $\mathrm{v}_{+}=1, \mathrm{v}_{-}=\mathrm{n} \,(1-\beta),$ and $v=n \,(1-\beta)+1$ and $\mathrm{Q}^{\mathrm{n}(1-\beta)+1}=\left[1 \,\{\mathrm{n} \,(1-\beta)\}^{\mathrm{n}(1-\beta)}\right]$ with $\mathrm{y}_{\pm}^{\mathrm{n} \,(1-\beta)+1}=\mathrm{y}_{+}^{1} \, \mathrm{y}_{-}^{\mathrm{n} \,(1-\beta)}$ Then, \begin{aligned} &\mu(\text { mic. salt }) \ &\left.=\mu^{0} \text { (mic. salt }\right)+[n \,(1-\beta)+1] \, R \, T \, \ln \left[Q \, c(\text { mic.salt }) \, y_{\pm} / c_{r}\right] \end{aligned} At equilibrium, $\mathrm{n} \, \mu^{\mathrm{eq}}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \mathrm{aq}\right)=\mu^{\mathrm{eq}}(\text { mic. salt; aq })$ Hence, $\Delta_{\text {mics slt }} \mathrm{G}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{0}\right)=\mu^{0}(\text { mic.salt })-\mathrm{n} \, \mu^{0}\left(\mathrm{AM}^{+} \mathrm{Br}^{-}\right)$ The total concentration of salt ctot in the system is given by equation (y). $\operatorname{ctot}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \text {system }\right)=\mathrm{n} \, \mathrm{c}(\text { ch arg ed micelles })+\mathrm{c}\left(\mathrm{AM}^{+} \mathrm{Br}^{-} ; \mathrm{aq}\right)$ The analysis makes no explicit reference to a cmc. Instead the micellar system is described as a mixed salt solution. Application of these equations requires careful computer –based curve fitting for multi-parametric equations. The latter include equations relating mean ionic activity coefficients for salts to the composition of a given solution. A shielding factor $\delta$ was use by Burchfield and Woolley to reduce the impact of micellar charge of the cationic micelles on calculated ionic strength [2]. Thus the effective charge on the cationic micelles was written as $n \,(1-\beta) \, \delta$ where $\delta$ is approx. 0.5. Ionic Surfactant: Mass Action Model In general terms the equilibrium between surfactant monomers $\mathrm{Z}^{+}$, counter anions $\mathrm{X}^{-}$ and micelles $\mathrm{M}$ can be represented by the following equation. $\mathrm{n} Z^{+}(\mathrm{aq})+\mathrm{mX} \mathrm{X}^{-}(\mathrm{aq}) \Leftrightarrow \mathrm{M}^{(\mathrm{n}-\mathrm{m})+}(\mathrm{aq})$ Then in terms of the mass action model, the concentration equilibrium constant, $\mathrm{K}_{\mathrm{c}}^{0}=\left[\mathrm{M}^{(\mathrm{n}-\mathrm{m})+}\right] /\left\{\left[\mathrm{Z}^{+}\right]^{\mathrm{n}} \,\left[\mathrm{X}^{-}\right]^{\mathrm{m}}\right\}$ By definition, $\Delta_{\text {mic }} \mathrm{G}^{0}=-(\mathrm{n})^{-1} \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{\mathrm{c}}^{0}\right)$ Then, $\Delta_{\text {mic }} \mathrm{G}^{0} /(\mathrm{R} \, \mathrm{T})=-(\mathrm{n})^{-1} \, \ln \left[\mathrm{M}^{(\mathrm{n}-\mathrm{m})+}\right]+\ln \left[\mathrm{Z}^{+}\right]+(\mathrm{m} / \mathrm{n}) \, \ln \left[\mathrm{X}^{-}\right]$ Footnotes [1] N. M. van Os, J. R. Haak and L. A. M. Rupert, Physico – Chemical Properties of Selected Anionic, Cationic and Non-ionic Surfactants, Elsevier, Amsterdam 1993. [2] T. E. Burchfield, and E. M. Woolley, J. Phys. Chem.,1984,88,2149. [3] T. E. Burchfield and E. M. Woolley, in Surfactants in Solution, ed. K. L. Mittal and P. Bothorel, Plenum Press, New Yok, 1987, volume 4, 69. [4] E. M. Woolley and T. E. Burchfield, J. Phys. Chem.,1984,88,2155. [5] T. E. Burchfield and E.M.Wooley, Fluid Phase Equilib., 1985,20,207. [6] D. F.Evans, M. Allen, B.W. Ninham and A. Fouda, J. Solution Chem.,1984,13,87. [7] D. G. Archer, J. Solution Chem.,1986,15,727 [8] 1. L. Espada, M. N. Jones and G. Pilcher, J.Chem. Thermodyn., 1970, 2,1, 333; and references therein. 2. M. N. Jones, G. Pilcher and L.Espada, J. Chem.Thermodyn,.,1970,2,333 [9] M. J. Blandamer, P. M. Cullis, L. G. Soldi and M. C. S. Subha, J. Therm. Anal.,1996,46,1583. [10] R. Zana, Langmuir, 1996,12,1208. [11] M. J. Blandamer, K. Bijma, J. B. F. N. Engberts, P. M. Cullis, P. M. Last, K. D. Irlam and L. G. Soldi, J. Chem.Soc. Faraday Trans.,1997,93,1579; and references therein. [12] M. J. Blandamer, W. Posthumnus, J. B. F. N. Engberts and K. Bijma, J. Mol. Liq., 1997, 73-74,91. [13] R. DeLisi, E. Fiscaro, S. Milioto, E. Pelizetti and P. Savarino, J. Solution Chem.,1990,19, 247. [14] M. J. Blandamer, P. M. Cullis, L. G. Soldi, J. B. F. N. Engberts, A. Kacperska, N. M. van Os and M. C. S. Subha, Adv. Colloid Interface Sci.,1995,58,171. [15] For further references concerning the Stern Layer, see N. J. Buurma, P. Serena, M. J. Blandamer and J. B. F. N. Engberts, J. Org. Chem., 2004, 69, 3899.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.20%3A_Surfactants/1.20.4%3A_Surfactants_and_Miceles%3A_Ionics.txt
In many industrial and commercial applications, mixed surfactant systems are used [1]. An extensive literature examines the properties of these systems [2-8]. A given aqueous solution contains two surfactants $\mathrm{X}$ and $\mathrm{Y}$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The critical micellar concentrations are $\mathrm{cmc}_{\mathrm{X}}^{0}$ and $\mathrm{cmc}_{\mathrm{Y}}^{0}$. For a solution containing both surfactants $\mathrm{X}$ and $\mathrm{Y}$, the critical micellar concentration of the mixed surfactant is cmc(mix). Here we use a pseudo-separate phase model for the micelles. The system under consideration comprises $\mathrm{n}_{\mathrm{X}}^{0}$ and $\mathrm{n}_{\mathrm{Y}}^{0}$ moles of the two surfactants. [Here the superscript ‘zero’ refers to the composition of the solution as prepared using the two pure surfactants.] A property $\mathrm{r}$ is the ratio of the concentration of surfactant $\mathrm{Y}$ to the total concentration of the two surfactants in the solution. Thus, $\mathrm{r}=\frac{\mathrm{n}_{\mathrm{Y}}^{0}}{\mathrm{n}_{\mathrm{X}}^{0}+\mathrm{n}_{\mathrm{Y}}^{0}}=\frac{\mathrm{c}_{\mathrm{Y}}^{0}}{\mathrm{c}_{\mathrm{X}}^{0}+\mathrm{c}_{\mathrm{Y}}^{0}}$ Here $\mathrm{c}_{\mathrm{X}}^{0}$ and $\mathrm{c}_{\mathrm{Y}}^{0}$ are the concentrations of the two surfactants in solution. We define a model system where cmc(mix) is a linear function of the property $\mathrm{r}$; equation (b). $c m c(\operatorname{mix})=\left(\mathrm{cmc}_{\mathrm{Y}}^{0}-\mathrm{cmc}_{\mathrm{X}}^{0}\right) \, \mathrm{r}+\mathrm{cmc}_{\mathrm{X}}^{0}$ Or, $\mathrm{cmc}(\mathrm{mix})=\mathrm{r} \, \mathrm{cmc}_{\mathrm{Y}}^{0}+\mathrm{cmc}_{\mathrm{X}}^{0} \,(1-\mathrm{r})$ Hence the critical micellar concentration of the two surfactants in a given solution, $\mathrm{cmc}_{\mathrm{X}}$ and $\mathrm{cmc}_{\mathrm{Y}}$, depend on parameter $\mathrm{r}$. $\mathrm{cmc}_{\mathrm{Y}}=\mathrm{cmc}_{\mathrm{Y}}^{0} \, \mathrm{r}$ $\mathrm{cmc}_{\mathrm{X}}=\mathrm{cmc}_{\mathrm{X}}^{0} \,(1-\mathrm{r})$ Clearly in the absence of surfactant $\mathrm{Y}$, micelles are not formed by surfactant $\mathrm{X}$ until the concentration exceeds $\mathrm{cmc}_{\mathrm{X}}^{0}$. If surfactant $\mathrm{Y}$ is added to the solution, the $\mathrm{cmc}_{\mathrm{X}}$ changes. In other words the properties of surfactants $\mathrm{X}$ and $\mathrm{Y}$ in a given solution are linked. We anticipate that for real system cmc(mix) is a function of $\mathrm{c}_{\mathrm{X}}^{0}$ and $\mathrm{c}_{\mathrm{Y}}^{0}$ so that cmc(mix) is a function of ratio $\mathrm{r}$ and a quantity $\theta$. The latter takes account of surfactant-surfactant interactions in the micellar pseudophase. Then $\mathrm{cmc}(\operatorname{mix})=\mathrm{r} \,\left(\mathrm{cmc}_{\mathrm{Y}}^{0}-\mathrm{cmc}_{\mathrm{X}}^{0}\right) \, \exp [-\theta \,(1-\mathrm{r})]+\mathrm{cmc}_{\mathrm{X}}^{0}$ For the surfactants $\mathrm{X}$ and $\mathrm{Y}$, $\mathrm{cmc}_{\mathrm{Y}}(\mathrm{mix})=\mathrm{r} \, \mathrm{cmc}_{\mathrm{Y}}^{0} \, \exp [-\theta \,(1-\mathrm{r})]$ and $\mathrm{cmc}_{\mathrm{x}}(\mathrm{mix})=\mathrm{cmc}_{\mathrm{x}}^{0} \,\{1-\mathrm{r} \, \exp [-\theta \,(1-\mathrm{r})]\}$ Therefore we envisage that the cmc of solutions containing a mixture of surfactants differs from that for model systems. We turn attention to the enthalpies of mixed surfactant solutions. In the case of a mixed aqueous solution containing surfactants $\mathrm{X}$ and $\mathrm{Y}$, the partial molar enthalpies of the surfactants are anticipated to depend on their concentrations. We characterize a given system by a single enthalpic interaction parameter, $\mathrm{h}(\text {int})$. $\mathrm{H}_{\mathrm{x}}(\mathrm{aq})=\mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\mathrm{h}(\text {int}) \,\left(\mathrm{c}_{\mathrm{x}} / \mathrm{c}_{\mathrm{r}}\right)$ $\mathrm{H}_{\mathrm{Y}}(\mathrm{aq})=\mathrm{H}_{\mathrm{Y}}^{\infty}(\mathrm{aq})+\mathrm{h}(\text { int }) \,\left(\mathrm{c}_{\mathrm{Y}} / \mathrm{c}_{\mathrm{r}}\right)$ Here $\mathrm{H}_{\mathrm{x}}^{\infty}(\mathrm{aq})$ and $\mathrm{H}_{\mathrm{Y}}^{\infty}(\mathrm{aq})$ are the ideal (infinite dilution) partial molar enthalpies of the two monomeric surfactants in aqueous solutions at defined $\mathrm{T}$ and $\mathrm{p}$. The micellar pseudo-separate phase comprises two surfactants amounts $\mathrm{n}_{X}(\text { mic })$ and $\mathrm{n}_{Y}(\text { mic })$. The mole fractions $\mathrm{x}_{\mathrm{X}}(\mathrm{mic})$ and $\mathrm{x}_{\mathrm{X}}(\mathrm{mic}) \left[=\left(1-\mathrm{x}_{\mathrm{X}}(\mathrm{mic})\right]$ are given by equation (k). $\mathrm{x}_{\mathrm{X}}(\mathrm{mic})=\mathrm{n}_{\mathrm{X}}(\mathrm{mic}) /\left[\mathrm{n}_{\mathrm{x}}(\mathrm{mic})+\mathrm{n}_{\mathrm{Y}}(\mathrm{mic})\right]=1-\mathrm{x}_{\mathrm{Y}}(\mathrm{mic})$ We relate the partial molar enthalpies $\mathrm{H}_{\mathrm{X}}(\text {mic})$ and $\mathrm{H}_{\mathrm{Y}}(\text {mic})$ in the mixed pseudo-separate phase to the molar enthalpies of surfactants $\mathrm{X}$ and $\mathrm{Y}$ in pure pseudo-separate micellar phases, $\mathrm{H}_{\mathrm{X}}^{*}(\mathrm{mic})$ and $\mathrm{H}_{\mathrm{Y}}^{*} \text { (mic) }$ using equations (l) and (m) where $\mathrm{U}$ is a surfactant-surfactant interaction parameter. $\mathrm{H}_{\mathrm{X}}(\text { mic })=\mathrm{H}_{\mathrm{X}}^{*}(\mathrm{mic})+\left[1-\mathrm{x}_{\mathrm{X}}(\mathrm{mic})\right]^{2} \, \mathrm{U}$ $\mathrm{H}_{\mathrm{Y}}(\mathrm{mic})=\mathrm{H}_{\mathrm{Y}}^{*}(\text { mic })+\left[\mathrm{x}_{\mathrm{X}}(\mathrm{mic})\right]^{2} \, \mathrm{U}$ Footnotes [1] J. H. Clint, J. Chem. Soc. Faraday Trans.,1, 1975,71,1327. [2] P. M .Holland, Adv. Colloid Interface Sci.,1986,26,111; and references therein. [3] A. H. Roux, D. Hetu, G. Perron and J. E. Desnoyers, J. Solution Chem.,1984,13,1. [4] M. J. Hey, J. W. MacTaggart and C. H. Rochester, J. Chem. Soc. Faraday Trans.1, 1985,81,207. [5] J. L. Lopez-Fontan, M. J. Suarez, V. Mosquera and F. Sarmiento, Phys. Chem. Chem. Phys.,1999,1,3583. [6] R. DeLisi, A. Inglese, S. Milioto and A. Pellerito, Langmuir, 1997,13,192. [7] M. J. Blandamer, B. Briggs, P. M. Cullis and J. B. F. N. Engberts, Phys. Chem. Chem. Phys.,2000,2,5146. [8] J. F. Rathman and J. F. Scamehorn, Langmuir, 1988,4,474. [9] A. H. Roux, D. Hetu, G. Perron and J. E. Desn oyers, J. Solution Chem.,1984,13,1.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.20%3A_Surfactants/1.20.5%3A_Surfactants_and_Micelles%3A_Mixed.txt
Thermodynamics provides a basis for the mathematical description of important phenomena such as chemical equilibria, solubilities, densities, and heats of reaction. Chemists have confidence in this approach to chemistry. However for many chemists it is somewhat of a shock to discover that at the heart of mathematics there is serious flaw. In 1931 K. Godel showed that there is a fundamental inconsistency in mathematics [1]. In other words mathematics is incomplete [2]. Nevertheless chemists do not ‘throw out the baby with the bathwater’. Atkins [2] notes that it would be foolish to discard mathematics even though there are treacherous regions deep inside its structure. Footnotes [1] An interesting account is given by D. R. Hofstadter in Godel, Esher and Bach;An Extended Golden Braid, Vintage, New York, 1980. [2] P. W. Atkins, Galileo’s Finger, Oxford, 2003, chapter 10. 1.21.2: Thermodynamic Energy The thermodynamic energy $\mathrm{U}$ of a closed system increases when work $\mathrm{w}$ is done by the surroundings on the system and heat $\mathrm{q}$ flows from the surroundings into the system. $\Delta \mathrm{U}=\mathrm{q}+\mathrm{w}$ Equation (a) uses the acquisitive convention. In effect we record all changes from the point of view of the system. 1.21.3: Thermodynamic Energy: Potential Function The Master Equation states that the change in thermodynamic energy of a closed system is given by equation (a). $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi ; \quad \mathrm{A} \, \mathrm{d} \xi \geq 0$ At constant entropy (i.e. $\mathrm{dS} = 0$) and constant volume (i.e. $\mathrm{dV} = 0$), equation (a) leads to equation (b). $\mathrm{dU}=-\mathrm{A} \, \mathrm{d} \xi$ But $A \, d \xi \geq 0$ Therefore all spontaneous processes at constant $\mathrm{S}$ and constant $\mathrm{V}$ take place in a direction for which the thermodynamic energy decreases. The latter statement shows the power of thermodynamics in that it is quite general; we have not stated the nature of the spontaneous process. Of course chemists are interested in those cases where the spontaneous process is chemical reaction. Thus we have a signal of what happens to the energy of the system; the key word here is spontaneous. In the context of most chemists interests, equation (b) is not terribly helpful. Chemists do not normally run their experiments at constant $\mathrm{S}$ and constant $\mathrm{V}$. In fact it is not obvious how one might do this. Nevertheless equation (b) is important finding its application when we turn to other thermodynamic variables which can be used as thermodynamic potentials; e.g. Gibbs energy. 1.21.4: Thermodynamic Potentials The following important equations describe changes in thermodynamic energy, enthalpy, Helmholtz energy and Gibbs energy of a closed system. $p\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mu_{\mathrm{j}} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mu_{\mathrm{j}} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ $\mathrm{dF}=-\mathrm{S} \, \mathrm{dT}-\mathrm{p} \, \mathrm{dV}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mu_{\mathrm{j}} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mu_{\mathrm{j}} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ These four differential equations relate, for example, the change in $\mathrm{U}, \mathrm{~H}, \mathrm{~F} \text { and } \mathrm{G}$ with the change in amount of each chemical substance, $\mathrm{dn}_{j}$. These four equations are integrated [1] to yield the following four equations. $\mathrm{U}=\mathrm{T} \, \mathrm{S}-\mathrm{p} \, \mathrm{V}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ $\mathrm{H}=\mathrm{T} \, \mathrm{S} +\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ $\mathrm{F}=-\mathrm{p} \, \mathrm{V}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ $G=\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ The latter equation is particularly useful because it signals that the total Gibbs energy of a system is given by the sum of the products of amounts and chemical potentials of all substances in the system. In the case of an aqueous solution containing $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ moles of chemical substance $j$, the Gibbs energy of the solution is given by equation (i). $\mathrm{G}(\mathrm{aq})=\mathrm{n}_{1} \, \mu_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}(\mathrm{aq})$ In conjunction with equation (i) we do not have to attach the phrase ‘at temperature $\mathrm{T}$ and pressure $\mathrm{p}$’. Similarly the volume of the solution is given by equation (j) where $\mathrm{V}_{1}(\mathrm{aq})$ and $\mathrm{V}_{j}(\mathrm{aq})$ are the partial molar volumes of solvent and solute respectively. $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ The same argument applies in the case of a solution prepared using $\mathrm{n}_{1}$ moles of solvent water, $\mathrm{n}_{\mathrm{x}}$ moles of solute $\mathrm{X}$ and $\mathrm{n}_{\mathrm{y}}$ moles of solute $\mathrm{Y}$. Then, for example, $\mathrm{G}(\mathrm{aq})=\mathrm{n}_{1} \, \mu_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{x}} \, \mu_{\mathrm{x}}(\mathrm{aq})+\mathrm{n}_{\mathrm{y}} \, \mu_{\mathrm{y}}(\mathrm{aq})$ The analogue of equation (j) also follows but only if $\mathrm{n}_{\mathrm{x}}$ and $\mathrm{n}_{\mathrm{y}}$ are independent of pressure. If these two solutes are in chemical equilibrium [eg. $\mathrm{X}(\mathrm{aq}) \Leftrightarrow \mathrm{Y}(\mathrm{aq})$, amounts $\mathrm{n}_{\mathrm{x}}^{\mathrm{eq}}$ and $\mathrm{n}_{\mathrm{y}}^{\mathrm{eq}}$ respectively], then account must be taken of the dependences of $\mathrm{n}_{\mathrm{x}}^{\mathrm{eq}}$ and $\mathrm{n}_{\mathrm{y}}^{\mathrm{eq}}$ on pressure at fixed temperature and (with reference to entropies and enthalpies) on temperature at fixed pressure. The simple form of equations (i) and (k) emerge from equation (h) because other than the composition variables, the other differential terms $\mathrm{dT}$ and $\mathrm{dp}$ in equation (d) refer to change in intensive variables. For this reason chemists find it advantageous to describe chemical properties in the $\mathrm{T}-\mathrm{p}$-composition domain. The relationships between thermodynamic potentials are described as Legendre transforms [2]. The product term $\mathrm{T} \, \mathrm{S}$ may be called bounded energy. Then the Helmholtz energy ($\mathrm{F}=\mathrm{U}-\mathrm{T} \, \mathrm{S}$) is the free internal energy and the Gibbs energy ($\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$) is the free enthalpy. These comments help to understand the old designations of free Helmholtz energy (together with symbol $\mathrm{F}$), free Gibbs energy and the still currently used (in the French scientific literature) free enthalpy. Footnotes [1] The term “integrated” in this context deserves comment. Within the set of variables, $\mathrm{p}-\mathrm{V}-\mathrm{T}-\mathrm{S}$, $\mathrm{p}$ and $\mathrm{T}$ are intensive whereas $\mathrm{V}$ and $\mathrm{S}$ are extensive variables. Similarly $\mu_{j}$ is intensive whereas $\mathrm{n}_{j}$ is extensive. These are the conditions for operating Euler’s integration method. Still the word “integrate” in the present context has been used in subtle arguments when Euler’s theorem is not invoked. E. F. Caldin [Chemical Thermodynamics, Oxford, 1958, p. 166] identifies $\mathrm{T}, \mathrm{~p} \text { and } \mu_{j}$ as intensive and then integrates by gradual increments of the amount of each chemical substance, keeping the relative amounts constant. K. Denbigh [The Principles of Chemical Equilibrium, Cambridge, 1971, 3rd edn. p. 93] uses a similar argument, but comments that development of the equations (e) to (h) is not mathematical in the sense that the variables are simple. Rather we use our physical knowledge in that intensive variables do not depend on the state of the system. E. A. Guggenheim [Thermodynamics, North-Holland, Amsterdam, 1950, 2nd edn. p. 23] states that the equations (a) to (d) can be integrated by following the artifice when $\mathrm{dT} = 0, \mathrm{~dp = 0$ and each $\mathrm{n}_{j}$ is changed by the same proportions as are the extensive variables $\mathrm{S}$ and $\mathrm{V}$. {The term artifice is used here to mean a ‘device’, skill rather than “trickery” or “something intended to deceive”; Pocket Oxford Dictionary, Oxford 1942, 4th edn. and Cambridge International Dictionary of English, Cambridge, 1995. [2] There is a pleasing internal consistency between the definitions advanced at this stage 1. $\mathrm{U}=\mathrm{T} \, \mathrm{S}-\mathrm{p} \, \mathrm{V}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ $G=\sum_{j=1}^{j=k} n_{j} \, \mu_{j}$ Then (cf. definition of $\mathrm{G}$) $\mathrm{G}=\mathrm{U}-\mathrm{T} \, \mathrm{S}+\mathrm{p} \, \mathrm{V}$ 2. $F=-p \, V+\sum_{j=1}^{j=k} n_{j} \, \mu_{j}$ Then from (a), $\mathrm{F}=\mathrm{U}-\mathrm{T} \, \mathrm{S}$ 3. $\mathrm{H}=\mathrm{T} \, \mathrm{S}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ Then from (a), $\mathrm{H}=\mathrm{U}+\mathrm{p} \, \mathrm{V}$ 4. Combining equations. relating $\mathrm{G}$ and $\mathrm{U}$ to $\mathrm{H}$ yields $\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$. Similarly the connections of $\mathrm{G}$ and $\mathrm{F}$ to $\mathrm{U}$ give $\mathrm{G} = \mathrm{~F} + \mathrm{~p} \, \mathrm{~V}$ 1.21.5: Thermodynamic Stability: Chemical Equilibria At fixed temperature and pressure, all spontaneous processes lower the Gibbs energy of a closed system. Thermodynamic equilibrium corresponds to the state where $\mathrm{G}$ is a minimum and the affinity for spontaneous change is zero. The equilibrium is stable. The condition for chemical thermodynamic stability is that $(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}<0$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.21%3A_Thermodynamics/1.21.1%3A_Thermodynamics_and_Mathematics.txt
On the bench in front of a chemist is a stoppered flask containing a liquid mixture, ethanol + water, at temperature $\mathrm{T}$ and ambient pressure $\mathrm{p}$. The chemist might wonder why the homogeneous liquid does not spontaneously separate into two liquids, say water-rich and alcohol-rich mixtures. The chemist might also wonder why the mixture does not spontaneously produce a system which comprises a warm liquid mixture and a cold liquid mixture. Yet the fact that these changes do not occur spontaneously leads to the conclusion that the conditions in operation which forbid these changes can be traced to the Second Law of Thermodynamics, prompted by the word ‘spontaneously’ used above [1]. Thermal Stability Initially a given closed system has thermodynamic energy $2\mathrm{U}$ and volume $2\mathrm{V}$. We imagine that the mixture does in fact separate into two liquids, both at equilibrium, having energy $\mathrm{U} + \delta \mathrm{U}$ with volume $\mathrm{V}$, and energy $\mathrm{U} - \delta \mathrm{U}$ also with volume $\mathrm{V}$. The overall change in entropy at constant overall composition is given by equation (a). $\delta \mathrm{S}=\mathrm{S}(\mathrm{U}+\delta \mathrm{U}, \mathrm{V})+\mathrm{S}(\mathrm{U}-\delta \mathrm{U}, \mathrm{V})-\mathrm{S}(2 \mathrm{U}, 2 \mathrm{~V})$ The change in entropy can be understood in terms of a Taylor expansion for a change at constant volume $\mathrm{V}$. Thus $\delta S=\left(\frac{\partial^{2} S}{\partial U^{2}}\right)_{V, \xi} \,(\delta U)^{2}$ However at constant $\mathrm{V}$ and composition $\xi$, the Second Law of Thermodynamics requires that $\delta \mathrm{S}$ is positive for all spontaneous processes. The fact that such a change is not observed requires that $\left(\frac{\partial^{2} \mathrm{~S}}{\partial \mathrm{U}^{2}}\right)$ is negative. But $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{U}}\right)_{\mathrm{V}, \xi}=\mathrm{T}^{-1}$ Then $\left(\frac{\partial^{2} \mathrm{~S}}{\partial \mathrm{U}^{2}}\right)_{\mathrm{V}, \xi}=\frac{\partial}{\partial \mathrm{U}}\left(\mathrm{T}^{-1}\right)=-\frac{1}{\mathrm{~T}^{2}} \,\left(\frac{\partial \mathrm{T}}{\partial \mathrm{U}}\right)_{\mathrm{V}, \xi}=-\frac{1}{\mathrm{~T}^{2} \, \mathrm{C}_{\mathrm{V} \xi}}$ In order for the latter condition to hold, $\mathrm{C}_{\mathrm{V}\xi}$ must be positive. This is therefore the condition for thermal stability. In other words we will not observe spontaneous separation into hot and cold domains in that heat capacities are positive variables. Diffusional Stability A given system at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ contains $2 \, \mathrm{n}_{\mathrm{i}}$ moles of each i-chemical substance, for $\mathrm{i} = 1, 2, 3 \ldots$. The system is divided into two parts such that each part at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ contains $\mathrm{n}_{\mathrm{i}}$ moles of each chemical substance for $\mathrm{i} = 2, 3, 4, \ldots$. However one part contains $\mathrm{n}_{1} + \Delta \mathrm{n}_{1}$ moles and the other part contains $\mathrm{n}_{1} - \Delta \mathrm{n}_{1}$ moles of chemical substance 1. Then the change in Gibbs energy $\delta \mathrm{G}$ is given by equation (e). $\begin{gathered} \delta \mathrm{G}=\mathrm{G}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}+\Delta \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{n}_{3} . .\right]+\mathrm{G}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}-\Delta \mathrm{n}_{1}, \mathrm{n}_{2}, \mathrm{n}_{3} . .\right] \ -\mathrm{G}\left[\mathrm{T}, \mathrm{p}, 2 \, \mathrm{n}_{1}, 2 \, \mathrm{n}_{2}, 2 \, \mathrm{n}_{3} . .\right] \end{gathered}$ In terms of a Taylor expansions, $\delta G=\left(\frac{\partial^{2} G}{\partial n_{1}^{2}}\right)_{T, p, n(2), n(3) . .} \,\left(\delta n_{1}\right)^{2}$ But chemical potential, $\mu_{1}=\left(\frac{\partial \mathrm{G}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2), \mathrm{n}(3) \ldots}$ Then $\left(\frac{\partial^{2} \mathrm{G}}{\partial \mathrm{n}_{1}^{2}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2), \mathrm{n}(3) \ldots}=\left(\frac{\partial \mu_{1}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2), \mathrm{n}(3) \ldots .}$ We conclude that spontaneous separation of the system into parts rich and depleted in chemical substance-1 would occur if $\left(\frac{\partial \mu_{1}}{\partial n_{1}}\right)_{T, p, n(2), n(3)}$ is negative. But this process is never observed. Hence the condition for diffusional (or material) stability is that $\left(\frac{\partial \mu_{1}}{\partial n_{1}}\right)_{T, p, n(2), n(3) \ldots .}>0$. Further if we add $\delta \mathrm{n}_{1}$ moles of chemical substance to a closed system at fixed $\mathrm{T}, \mathrm{p}, \mathrm{n}_{2}, \mathrm{n}_{3} \ldots$, chemical potential µ1 must increase. Hydrostatic Stability A given system at temperature $\mathrm{T}$ and chemical composition $\xi$ has volume $2\mathrm{V}$. We imagine that a infinitely thin partition exists separating the system into two parts having equal volumes $\mathrm{V}$. The Helmholtz energy of the system volume $2\mathrm{V}$ is given by equation (i). $\mathrm{F}=\mathrm{F}[\mathrm{T}, 2 \mathrm{~V}, \xi]$ The Helmholtz energy of the two parts, volume $\mathrm{V}$ is given by equation (j). $\mathrm{F}=\mathrm{F}[\mathrm{T}, \mathrm{V}, \xi]$ The partition is envisaged as moving to produce two parts having volumes $(\mathrm{V}+\delta \mathrm{V})$ and $(\mathrm{V}-\delta \mathrm{V})$. Then at constant composition the change in Helmholtz energy is given by equation (k). $\delta \mathrm{F}=\mathrm{F}[\mathrm{T}, \mathrm{V}+\delta \mathrm{V}]+\mathrm{F}[\mathrm{T}, \mathrm{V}-\delta \mathrm{V}]-\mathrm{F}[\mathrm{T}, 2 \mathrm{~V}]$ Then using Taylor’s theorem, $\delta \mathrm{F}=\left(\partial^{2} \mathrm{~F} / \partial \mathrm{V}^{2}\right)_{\mathrm{T}} \,(\partial \mathrm{V})^{2}$ Hence [2] $\left(\partial^{2} \mathrm{~F} / \partial \mathrm{V}^{2}\right)_{\mathrm{T}}=-(\partial \mathrm{p} / \partial \mathrm{V})_{\mathrm{T}, \xi}$ Irrespective of the sign of ($\delta \mathrm{V}$), we conclude that $\delta \mathrm{F}$ would be negative in the event that $(\partial \mathrm{p} / \partial \mathrm{V})_{\mathrm{T}, \xi}$ is positive. But we never witness such a spontaneous separation of a system into two parts. In other words $(\partial \mathrm{p} / \partial \mathrm{V})_{\mathrm{T}, \xi}<0$. Hence, $\kappa_{\mathrm{T}}=-\left(\frac{1}{\mathrm{~V}}\right) \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \xi}>0$ Therefore for a system at fixed composition and temperature, if we as observers of this system increase the pressure $\mathrm{p}$, the volume of the system decreases. This is the condition of hydrostatic (or, mechanical) stability [1]. Physical Consequences of Stability Taken together, conditions for thermal stability $\left(\mathrm{C}_{\mathrm{V} \xi}>0\right)$ and for mechanical stability $\left(\kappa_{\mathrm{T}}>0\right)$ have further consequences [3]. Since, $\sigma=\mathrm{C}_{\mathrm{V} \xi} / \mathrm{V}+\frac{\mathrm{T} \,\left(\alpha_{\mathrm{p}}\right)^{2}}{\kappa_{\mathrm{T}}}$ Then isobaric heat capacities and heat capacitances , $\sigma\left(=\mathrm{C}_{\mathrm{pg}} / \mathrm{V}\right)$ are positive for stable phases. Furthermore, heat capacities and compressibilities are related by equation (p). $\frac{\mathrm{K}_{\mathrm{S}}}{\mathrm{K}_{\mathrm{T}}}=\frac{\mathrm{C}_{\mathrm{V}}}{\mathrm{C}_{\mathrm{p}}}$ Hence the isentropic compressibility $\kappa_{\mathrm{S}}$ of a stable phase must also be positive. In summary, $\mathrm{C}_{\mathrm{pm}} \geq \mathrm{C}_{\mathrm{Vm}_{\mathrm{m}}}>0$ $\kappa_{\mathrm{T}} \geq \kappa_{\mathrm{S}}>0$ Footnotes [1] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979, chapter 7. [2] By definition, $\mathrm{F}=\mathrm{F}[\mathrm{T}, \mathrm{V}, \xi]$. The total differential of the latter equation is as follows. $\mathrm{dF}=\left(\frac{\partial \mathrm{F}}{\partial \mathrm{T}}\right)_{\mathrm{V}, \xi} \, \mathrm{dT}+\left(\frac{\partial \mathrm{F}}{\partial \mathrm{V}}\right)_{\mathrm{T}, \xi} \, \mathrm{dV}+\left(\frac{\partial \mathrm{F}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{V}} \, \mathrm{d} \xi$ But (the ‘all-minus’ equation) $\mathrm{dF}=-\mathrm{S} \, \mathrm{dT}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi$ Then, $\left(\frac{\partial \mathrm{F}}{\partial \mathrm{V}}\right)_{\mathrm{T}, \xi}=-\mathrm{p}$ Or, $\left(\frac{\partial^{2} \mathrm{~F}}{\partial \mathrm{V}^{2}}\right)_{\mathrm{T}, \xi}=-\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}}\right)_{\mathrm{T}, \xi}$ [3] H. B. Callen, Thermodynamics and an Introduction to Thermostatistics, Wiley, New York, 2nd edn., 1985, pp.209-210.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.21%3A_Thermodynamics/1.21.6%3A_Thermodynamic_Stability%3A_Thermal%2C_Diffusional_and_Hydrostatic.txt
A given system at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ is prepared using $\mathrm{n}_{1}$ moles of water($\ell$), the solvent, together with $\mathrm{n}_{\mathrm{X}}^{0}$ and $\mathrm{n}_{\mathrm{Y}}^{0}$ moles of chemical substances $\mathrm{X}$ and $\mathrm{Y}$ respectively at time ‘$\mathrm{t} = 0$’. The molalities of these solutes are $\mathrm{m}_{\mathrm{X}}^{0}\left(=\mathrm{n}_{\mathrm{X}}^{0} / \mathrm{n}_{1} \, \mathrm{M}_{1}=\mathrm{n}_{\mathrm{X}}^{0} / \mathrm{w}_{1}\right)$ and $\mathrm{m}_{\mathrm{Y}}^{0}\left(=\mathrm{n}_{\mathrm{Y}}^{0} / \mathrm{n}_{1} \, \mathrm{M}_{1}=\mathrm{n}_{\mathrm{Y}}^{0} / \mathrm{w}_{1}\right)$ respectively at time ‘$\mathrm{t} = 0$’; the concentrations are $\mathrm{c}_{\mathrm{X}}^{0}\left(=\mathrm{n}_{\mathrm{XA}}^{0} / \mathrm{V}\right)$ and $\mathrm{c}_{\mathrm{Y}}^{0}\left(=\mathrm{n}_{\mathrm{Y}}^{0} / \mathrm{V}\right)$ respectively. Spontaneous chemical reaction leads to the formation of product $\mathrm{Z}$. Here we consider this spontaneous change from the standpoints of chemical thermodynamics and chemical kinetics. Thermodynamics The spontaneous chemical reaction is driven by the affinity for chemical reaction, $\mathrm{A}$ [1]. At each stage of the reaction the composition is described by the extent of reaction $\xi$. The affinity $\mathrm{A}$ is defined by the thermodynamic independent variables, $\mathrm{T}, \mathrm{~p} \text { and } \xi$. Thus $\mathrm{A}=\mathrm{A}[\mathrm{T}, \mathrm{p}, \xi]$ Therefore $\mathrm{dA}=\left(\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \xi} \, \mathrm{dT}+\left(\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \xi} \, \mathrm{dp}+\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}} \, \mathrm{d} \xi$ At constant $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{dA}=\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}} \, \mathrm{d} \xi$ In terms of thermodynamics, the reference point is thermodynamic equilibrium where the affinity for spontaneous change is zero and the composition is $\xi^{\mathrm{eq}}$. Chemical Kinetics In the context of chemical reaction in solution, the system under study is, conventionally, a very dilute solution so that from a macroscopic standpoint the system at ‘$\mathrm{t} = 0$’ is slightly displaced from equilibrium where $\mathrm{A}$ is zero. Thus chemists exploit their skill in monitoring for a solution the change with time of the absorbance at fixed wavelength, electrical conductivity, pH…. In a key assumption, the rate of change of composition $\mathrm{d}\xi / \mathrm{~dt}$ is proportional to the affinity $\mathrm{A}$ for spontaneous change[2]; $\mathrm{d} \xi / \mathrm{dt}=\mathrm{L} \, \mathrm{A}$ Here $\mathrm{L}$ is a phenomenological constant describing, in the present context, the phenomenon of spontaneous chemical reaction. In general terms for processes at fixed temperature and pressure, the phenomenological property $\mathrm{L}$ is related the isobaric – isothermal dependence of affinity $\mathrm{A}$ on extent of chemical reaction by a relaxation time $\tau_{\mathrm{T}, \mathrm{p}}$. Thus $\mathrm{L}^{-1}=-\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}} \, \tau_{\mathrm{T}, \mathrm{p}}$ Relaxation time $\tau_{\mathrm{T},\mathrm{p}$ is a macroscopic property of a given system (at defined $\mathrm{T}$ and $\mathrm{p}$) which chemists understand in terms of spontaneous chemical reaction (in a closed system). The task for chemists is to identify the actual chemical reaction in a given closed system. Thermodynamics and Chemical Kinetics In most treatments of chemical reactions the reference state is chemical equilibrium [3] where away from equilibrium the property $\mathrm{dA}$ equals the affinity $\mathrm{A}$ on the grounds that at equilibrium, $\mathrm{A}$ is zero; $\mathrm{A}=\mathrm{A}-\mathrm{A}^{\mathrm{eq}}=\mathrm{A}-0$. Hence combination of equations (c), (d) and (e) yields the key kinetic-thermodynamic equation. $\frac{\mathrm{d} \xi}{\mathrm{dt}}=-\mathrm{A} \,\left(\tau_{\mathrm{T}, \mathrm{p}}\right)^{-1} \,\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}^{-1}$ Equation (6) relates the rate of change of composition to the affinity for chemical reaction and relaxation time $\tau_{\mathrm{T}, \mathrm{p}}$. Equation (f) is therefore the key equation describing spontaneous chemical reaction in a closed system. In this context we stress the importance of equation (f). Law of Mass Action Equation (f) is an interesting and important description of the kinetics of chemical reaction. In fact the link between the rate of chemical reaction ($\mathrm{d} \xi / \mathrm{dt}$) and the affinity for spontaneous change $\mathrm{A}$ is intuitively attractive. However while one may monitor the dependence of composition on time, $\mathrm{d} \xi / \mathrm{dt}$, it is not immediately obvious how one might estimate the affinity $\mathrm{A}$ and the property $\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}$ at time $\mathrm{t}$. The Law of Mass Action offers a way forward although this law does not emerge from either the First or Second Laws of Thermodynamics. As Hammett [4] notes the Law of Mass Action was ‘first derived from limited observations’ and became ‘established through accumulation of observations with the principle and in the absence of contradictory evidence’. After the ‘Dark Ages’, came the renaissance and ‘Bartlett and Ingold and Peterson… accepting without question or comment the validity of the law of mass action’ [4]. The link back to thermodynamics was constructed using Transition State Theory developed by Eyring and described by Glasstone, Laidler and Eyring [5]. Therefore the phenomenological Law of Mass Action was brought into the fold of thermodynamics by offering a language which allowed activation parameters to be understood in terms of, for example, standard enthalpies and standard isobaric heat capacities of activation. Footnotes [1] I. Prigogine and R. Defay, Chemical Thermodynamics, (transl. D. H. Everett) Longmans Green, London, 1954. [2] See for example, M. J. Blandamer, Introduction to Chemical Ultrasonics, Academic Press, London, 1973. [3] E. F. Caldin, Fast Reactions in Solution, Blackwell, Oxford, 1964. [4] L. P. Hammett, Physical Organic Chemistry, McGraw-Hill, New York, McGraw-Hill, New York, 2nd. edition, 1970, p.94. [5] S. Glasstone, K. J. Laidler and H. Eyring, The Theory of Rate Processes, McGraw-Hill, New York,1941. 1.21.8: Third Law of Thermodynamics The Third Law of Thermodynamics states that the entropy and the heat capacity of a perfect crystal vanish in the limit of zero kelvin [1]. When a system is heated, energy is stored in the form of molecular vibrations leading to an increase in entropy. The isochoric heat capacity is related to the differential dependence of energy on temperature at equilibrium. A more detailed analysis is required if phase changes and inter-component mixing is involved. Here there are additional contributions to the change in entropy accompanying irreversible processes which Gurney describes as the cratic part of the entropy [2]. In a certain sense an isentropic process is ‘a draw between Maxwell’s demon and a natural process’ [3]. Footnotes. [1] K. S. Pitzer, Thermodynamics, McGraw-Hill, New York, 3rd. edn.,1965. [2] R. W. Gurney, Ionic Processes in Solution, McGraw-Hill, New York, 1953. [3] A. B. Pippard, The Elements of Classical Thermodynamics, University Press, Cambridge ,1957, p.99.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.21%3A_Thermodynamics/1.21.7%3A_Thermodynamics_and_Kinetics.txt
At temperature $\mathrm{T}$, pressure $\mathrm{p}$ and equilibrium, the volume of a closed system containing i-chemical substances where the amounts can be independently varied, is defined by the following equation. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \ldots \ldots \mathrm{n}_{\mathrm{i}}\right]$ Or, in general terms according to Euler’s theorem, $\mathrm{V}=\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}$ where $\mathrm{V}_{\mathrm{j}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}_{\mathrm{i} \neq \mathrm{j}}}$ The general differential of equation (b) has the following form. $\mathrm{dV}=\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mathrm{dV} \mathrm{V}_{\mathrm{j}}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{V}_{\mathrm{j}} \, \mathrm{dn} \mathrm{n}_{\mathrm{j}}$ The general differential of equation (a) has the following form $\mathrm{dV}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{n}_{\mathrm{i}}} \, \mathrm{dT}+\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{n}_{\mathrm{i}}} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}}\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}_{\mathrm{i} \neq \mathrm{j}}} \, \mathrm{dn}_{\mathrm{j}}$ Comparison of equations (d) and (e) shows that $0=-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{n}_{\mathrm{i}}} \, \mathrm{dT}-\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{n}_{\mathrm{i}}} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mathrm{dV} \mathrm{j}_{\mathrm{j}}$ Equation (f) is the Gibbs-Duhem Equation with respect to the volumetric properties of a closed system at equilibrium. Application A given closed system contains $\mathrm{n}_{1}$ moles of solvent (water) and $\mathrm{n}_{j}$ moles of solute $j$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The system is at equilibrium where $\mathrm{G}$ is a minimum, the affinity for spontaneous change $\mathrm{A}$ is zero and the composition-organisation $\xi^{\mathrm{eq}}$. The dependent variable volume $\mathrm{V}$ is defined using a set of independent variables; equation (g). $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ Equation (k) has an interesting property. If we multiply the extensive variables $\mathrm{n}_{1}$ and $\mathrm{n}_{j}$ by a factor $\mathrm{k}$, the volume of the system equals ($\mathrm{V}. \mathrm{~k}$). In terms of Euler’s Theorem [1], the variable $\mathrm{V}$ linked to the variables $\mathrm{n}_{1}$ and $\mathrm{n}_{j}$ is a homogeneous function of the first degree. The important consequence is the following key relation. $\mathrm{V}=\mathrm{n}_{1} \, \mathrm{V}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}$ where $\mathrm{V}_{1}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{j})}$ and $\mathrm{V}_{\mathrm{j}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(1)}$ We do not have to specify the conditions ‘at constant $\mathrm{T}$ and $\mathrm{p}$’ in conjunction with equation (h) which is a mathematical identity. Footnote [1] Degree of Homogeneity At temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the volume of a closed system containing $\mathrm{n}_{j}$ moles of each chemical substance $j$ is given by $\mathrm{V}=\mathrm{V}\left[\mathrm{n}_{1}, \mathrm{n}_{2} \ldots \ldots \ldots \ldots . \ldots \mathrm{n}_{\mathrm{k}}\right]$ The property volume has unit degree of homogeneity. That is to say – if the amount of each substance is increased by a factor $\lambda$ then the volume increases by the same factor. Thus $\mathrm{V}\left[\lambda \mathrm{n}_{1}, \lambda \mathrm{n}_{2} \ldots \ldots \ldots \ldots . . \lambda \mathrm{n}_{\mathrm{k}}\right]=\lambda \, \mathrm{V}\left[\mathrm{n}_{1}, \mathrm{n}_{2} \ldots \ldots \ldots \ldots . \ldots \mathrm{n}_{\mathrm{k}}\right]$ 1.22.10: Volume: Aqueous Binary Liquid Mixtures For binary aqueous mixtures (at ambient pressure and fixed temperature) there are two interesting reference points. 1. The molar volume of the pure liquid component 2, $\mathrm{V}_{2}^{*}(\lambda)$. In the latter case we imagine that each molecule of liquid 2 is surrounded by an infinite expanse of water. With gradual increase in $\mathrm{x}_{2}$, so (on average) the molecules of liquid 2 move closer together. Typically Aqueous Mixtures For these systems $\left[\mathrm{V}_{2}^{\infty}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ is negative. But this pattern is not unique to aqueous systems. The unique feature is the decrease in $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ with increase in $\mathrm{x}_{2}$ at low $\mathrm{x}_{2}$ [1]. In fact with increase in hydrophobicity of chemical substance 2, the decrease is more striking and the minimum in $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ occurs at lower $\mathrm{x}_{2}$. At mole fractions beyond $\mathrm{x}_{2}\left[\mathrm{~V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ increases with increase in $\mathrm{x}_{2}$. Many explanations have been offered for this complicated pattern. The following is one explanation. The negative $\left[\mathrm{V}_{2}^{\infty}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ is accounted for in terms of a liquid clathrate in which part of the hydrophobic group ‘occupies’ a guest site in the liquid water ‘lattice’. The decrease in $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ is accounted for in terms of an increasing tendency towards a liquid clathrate hydrate structure. With increase in $\mathrm{x}_{2}$ there comes a point where there is insufficient water to construct the liquid clathrate host. Hence $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ increases [2,3]. Typically Non-Aqueous Systems Although $\left[\mathrm{V}_{2}^{\infty}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$ is negative no minimum is observed in $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\lambda)\right]$. Footnotes [1] See for example; fluoroalcohols(aq); C. H. Rochester and J. R. Symonds, J. Fluorine Chem.,1974,4,141. [2] F. Franks, Ann. N. Y. Acad. Sci.,1955,125,277. [3] For many binary aqueous mixtures the patterns in volume related properties often identify transition points at ‘structurally interesting compositions’; G. Roux, D. Roberts, G. Perron and J. E. Desnoyers, J. Solution Chem.,1980,9,629.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.1%3A_Volume%3A_Partial_Molar%3A_General_Analysis.txt
The ‘Method of Tangents’ is an important technique which is readily illustrated using the volumetric properties of binary liquid mixtures. The starting point is (as always?) the Gibbs - Duhem Equation which leads to equation (a) for systems at fixed temperature and pressure. $\mathrm{n}_{1} \, \mathrm{dV}_{1}(\operatorname{mix})+\mathrm{n}_{2} \, d \mathrm{dV}_{2}(\operatorname{mix})=0$ Dividing by $\left(\mathrm{n}_{1} + \mathrm{~n}_{2}\right)$, $\mathrm{x}_{1} \, d \mathrm{~V}_{1}(\operatorname{mix})+\mathrm{x}_{2} \, \mathrm{dV}_{2}(\operatorname{mix})=0$ The molar volume is given by equation (c). $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})=\mathrm{x}_{1} \, \mathrm{V}_{1}(\mathrm{mix})+\mathrm{x}_{2} \, \mathrm{V}_{2}(\mathrm{mix})$ Hence (at equilibrium, fixed temperature and pressure) the differential dependence of $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})$ on mole fraction $\mathrm{x}_{1}$ is given by equation (d). \begin{aligned} \frac{\mathrm{dV}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{dx}_{1}}=\mathrm{V}_{1}(\mathrm{mix}) &+\mathrm{x}_{1} \,\left[\frac{\mathrm{dV}_{1}(\mathrm{mix})}{\mathrm{dx}_{1}}\right] \ &+\mathrm{V}_{2}(\mathrm{mix}) \,\left[\frac{\mathrm{dx}_{2}}{\mathrm{dx}}\right]+\mathrm{x}_{2} \,\left[\frac{\mathrm{dV}_{2}(\mathrm{mix})}{\mathrm{dx}_{1}}\right] \end{aligned} From the Gibbs-Duhem equation, $\mathrm{x}_{1} \, \frac{\mathrm{dV}_{1}(\mathrm{mix})}{\mathrm{dx}_{1}}+\mathrm{x}_{2} \, \frac{\mathrm{dV}_{2}(\mathrm{mix})}{\mathrm{dx}_{1}}=0$ [Note the common denominator.] Also $\mathrm{x}_{1} + \mathrm{~x}_{2} = 1$. And so, $\mathrm{dx}_{1}=-\mathrm{dx}_{2}$ Therefore, $\frac{\mathrm{dV} \mathrm{m}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{dx}_{1}}=\mathrm{V}_{1}(\operatorname{mix})-\mathrm{V}_{2}(\operatorname{mix})$ Combination of equations (c) and (g) yields the following equation. $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})=\mathrm{x}_{1} \, \mathrm{V}_{1}(\mathrm{mix})+\mathrm{x}_{2} \,\left[\mathrm{V}_{1}(\mathrm{mix})-\frac{\mathrm{dV}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{dx}_{1}}\right]$ Further, $\mathrm{x}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{x}_{2} \, \mathrm{V}_{1}(\mathrm{mix})=\mathrm{V}_{1}(\mathrm{mix})$ Hence, $\mathrm{V}_{1}(\mathrm{mix})=\mathrm{V}_{\mathrm{m}}(\mathrm{mix})+\left(1-\mathrm{x}_{1}\right) \, \frac{\mathrm{dV}_{\mathrm{m}}(\mathrm{mix})}{\mathrm{dx}_{1}}$ At a given mole fraction, we determine the molar volume of the mixture $\mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ and its dependence on mole fraction. $\left[\mathrm{dV} \mathrm{m}_{\mathrm{m}}(\operatorname{mix}) / \mathrm{dx}_{1}\right]$ is the gradient of the tangent at mole fraction $\mathrm{x}_{1}$ to the curve recording the dependence of $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})$ on $\mathrm{x}_{1}$; hence the name of this method of data analysis. This analysis is relevant because, as commented above, we can determine the variables $\mathrm{V}_{1}^{*}(\ell), \mathrm{V}_{2}^{*}(\ell) \text { and } \mathrm{V}_{\mathrm{m}}(\mathrm{mix})$. Another approach is based on excess molar volumes $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ and their dependence on mole fraction at fixed temperature and pressure. Since $\mathrm{V}_{\mathrm{m}}(\mathrm{id})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ And $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{V}_{\mathrm{m}}(\mathrm{mix})-\mathrm{V}_{\mathrm{m}}(\mathrm{id})$ From equations (c), and (k), $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \,\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]-\mathrm{x}_{2} \,\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ We define excess partial molar volumes; $\mathrm{V}_{1}^{\mathrm{E}}(\operatorname{mix})=\mathrm{V}_{1}(\operatorname{mix})-\mathrm{V}_{1}^{*}(\ell)$ and $\mathrm{V}_{2}^{\mathrm{E}}(\operatorname{mix})=\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)$ Hence the excess molar volume of the mixture is related to two excess partial molar volumes. $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \, \mathrm{V}_{1}^{\mathrm{E}}(\operatorname{mix})+\mathrm{x}_{2} \, \mathrm{V}_{2}^{\mathrm{E}}(\operatorname{mix})$ We use equation (m) to obtain the differential dependence of $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ on mole fraction $\mathrm{x}_{1}$. \begin{aligned} \mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}=& {\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]+\mathrm{x}_{1} \,\left[\frac{\mathrm{dV}_{1}(\mathrm{mix})}{\mathrm{dx} \mathrm{x}_{1}}\right] } \ &+\left[\frac{\mathrm{dx}_{2}}{\mathrm{dx} \mathrm{x}_{1}}\right] \,\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]+\mathrm{x}_{2} \,\left[\frac{\mathrm{dV}_{2}(\mathrm{mix})}{\mathrm{dx}_{1}}\right] \end{aligned} We write the Gibbs - Duhem equation in the form shown in equation (e) together with equation (p). Hence, $\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}=\mathrm{V}_{1}^{\mathrm{E}}-\mathrm{V}_{2}^{\mathrm{E}}$ or, $\mathrm{V}_{2}^{\mathrm{E}}=\mathrm{V}_{1}^{\mathrm{E}}-\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ Hence using equation (o), $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \, \mathrm{V}_{1}^{\mathrm{E}}+\mathrm{x}_{2} \, \mathrm{V}_{1}^{\mathrm{E}}-\mathrm{x}_{2} \, \mathrm{dV} \mathrm{V}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ Thus, $\mathrm{V}_{1}^{\mathrm{E}}=\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}+\left(1-\mathrm{x}_{1}\right) \, d \mathrm{~V}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ Equation (t) is the excess form of equation (j). A plot of $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ against $\mathrm{x}_{1}$ shows a curve passing through '$\mathrm{V}_{\mathrm{m}}^{\mathrm{E}} = 0$' at $\mathrm{x}_{1} = 0$ and $\mathrm{x}_{1} = 1$. Other than these two reference points, thermodynamics does not define the shape of the plot of $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ against $\mathrm{x}_{1}$. Thermodynamics does not define the shape of the plot of $\mathrm{V}_{1}^{\mathrm{E}}$ against $\mathrm{x}_{1}$ other than to require that at $\mathrm{x}_{1} = 1$, $\mathrm{V}_{1}^{\mathrm{E}}$ is zero. An interesting feature is the sign and magnitude of $\mathrm{V}_{1}^{\mathrm{E}}$ in the limit that $\mathrm{x}_{1} = 0$; i.e. at $\mathrm{x}_{2} = 1$. The volumetric properties of a binary liquid (homogeneous) mixture is summarized by a plot of excess molar volume $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ against, for example, mole fraction $\mathrm{x}_{1}$. In fact this type of plot is used for many excess molar properties including $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}$ and $\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}$. Here we consider a general excess molar property $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}$. The corresponding excess partial molar property of chemical substance 1 is $\mathrm{X}_{\mathrm{1}}^{\mathrm{E}}$ which is related to $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}$ and the dependence of $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}$ on $\mathrm{x}_{1}$ at mole fraction $\mathrm{x}_{1}$, $\mathrm{X}_{1}^{\mathrm{E}}=\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}+\left(1-\mathrm{x}_{1}\right) \, \mathrm{dX} \mathrm{X}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ Calculation of $\mathrm{X}_{\mathrm{1}}^{\mathrm{E}}$ requires the gradient $\mathrm{dX} \mathrm{m} / \mathrm{dx}_{1}$ as a function of mole fraction composition. The way forward involves fitting the dependence of $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}$ on $\mathrm{x}_{1}$ to a general equation and then calculating $\mathrm{dX} \mathrm{m} / \mathrm{dx}_{1}$ using the derived parameters [1-4]. Footnotes [1] C. W. Bale and A. D. Pelton, Metallurg. Trans.,1974,5,2323. [2] C. Jambon and R. Philippe, J.Chem.Thermodyn.,1975,7,479. [3] M. J. Blandamer, N. J. Blundell, J. Burgess, H. J. Cowles and I. M. Horn, J. Chem. Soc. Faraday Trans.,1990,86,277. [4] A description of a useful procedure for non-linear least squares analysis is given by W. E. Wentworth, J.Chem.Educ.,1965,42,96.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.11%3A_Volumes%3A_Liquid_Mixtures%3A_Binary%3A_Method_of_Tangents.txt
Consider a chemical equilibrium between two solute $\mathrm{X}(\mathrm{aq})$ and $\mathrm{Y}(\mathrm{aq})$ in aqueous solution at fixed $\mathrm{T}$ and $\mathrm{p}$. We assume that the thermodynamic properties of the two solutes are ideal. The chemical equilibrium is be expressed as follows. $\mathrm{X}(\mathrm{aq}) \Leftrightarrow \mathrm{Y}(\mathrm{aq})$ The (dimensionless intensive) degree of reaction $\alpha$ is related to the equilibrium constant $\mathrm{K}^{0}$ using equation (b) [1]. $\alpha=\mathrm{K}^{0} /\left(1+\mathrm{K}^{0}\right)$ At fixed temperature, $\frac{\mathrm{d} \alpha}{\mathrm{dp}}=\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}} \, \frac{\mathrm{d} \ln \left(\mathrm{K}^{0}\right)}{\mathrm{dp}}$ Or, $\frac{\mathrm{d} \alpha}{\mathrm{dp}}=-\frac{\mathrm{K}^{0}}{\left(1+\mathrm{K}^{0}\right)^{2}} \, \frac{\Delta_{\mathrm{r}} \mathrm{V}^{0}(\mathrm{aq})}{\mathrm{R} \, \mathrm{T}}$ $\Delta_{\mathrm{r}} \mathrm{V}^{0}(\mathrm{aq})$ is the limiting volume of reaction. The (equilibrium) volume of the system at a defined $\mathrm{T}$ and $\mathrm{p}$ is given by equation (e). $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{\mathrm{x}} \, \mathrm{V}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\mathrm{n}_{\mathrm{Y}} \, \mathrm{V}_{\mathrm{Y}}^{\infty}(\mathrm{aq})+\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)$ $\mathrm{V}_{1}^{*}(\ell)$ is the molar volume of solvent, water. If $\mathrm{n}_{\mathrm{x}}^{0}$ is total amount of solute, (i.e. $\mathrm{X}$ and $\mathrm{Y}$) in the system, $\mathrm{V}(\mathrm{aq})=(1-\alpha) \, \mathrm{n}_{\mathrm{X}}^{0} \, \mathrm{V}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\alpha \, \mathrm{n}_{\mathrm{X}}^{0} \, \mathrm{V}_{\mathrm{Y}}^{\infty}(\mathrm{aq})+\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)$ Or, $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{\mathrm{X}}^{0} \, \mathrm{V}_{\mathrm{x}}^{\infty}(\mathrm{aq})+\alpha \, \mathrm{n}_{\mathrm{X}}^{0} \, \Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq})+\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)$ $\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq})$ is the limiting volume of reaction. We assume that at temperature $\mathrm{T}$, the properties $\mathrm{V}_{\mathrm{x}}^{\infty}(\mathrm{aq}), \Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq}) \text { and } \mathrm{V}_{1}^{*}(\ell)$ are independent of pressure. Hence using equations (d) and (g) [2,3], $\left(\frac{\partial \mathrm{V}(\mathrm{aq})}{\partial \mathrm{p}}\right)_{\mathrm{T}}=-\mathrm{n}_{\mathrm{X}}^{0} \, \frac{\left[\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq})\right]^{2}}{\mathrm{R} \, \mathrm{T}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}$ We have taken account of the fact that, $\frac{\mathrm{d} \ln \left(\mathrm{K}^{0}\right)}{\mathrm{dp}}=-\frac{\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq})}{\mathrm{R} \, \mathrm{T}}$ Similarly [2] $\left(\frac{\partial \mathrm{V}(\mathrm{aq})}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\mathrm{n}_{\mathrm{X}}^{0} \, \frac{\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq}) \, \Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}$ Equation (h) shows that irrespective of the sign of $\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq})$, the contribution to $\left(\frac{\partial V(a q)}{\partial p}\right)_{\mathrm{T}}$ is always negative. No such generalisation emerges with respect to equation (j) [4]. A closely related subject concerns the dependence of rate constants on pressure leading to volumes of activation [5]. Footnotes [1] From equation (a) $\mathrm{X}(\mathrm{aq})$ $\Leftrightarrow$ $\mathrm{Y}(\mathrm{aq})$ At $t = 0$, $\mathrm{n}_{\mathrm{X}}^{0}$   0 $\mathrm{mol}$ At equilib; $\mathrm{n}_{\mathrm{X}}^{0} - \xi$   $\xi$ $\mathrm{mol}$ In volume $\mathrm{V}$ $\mathrm{n}_{\mathrm{X}}^{0}$   $\xi / \mathrm{V}$ $\mathrm{mol m}^{-3}$ [2] \begin{aligned} &\mathrm{dV} / \mathrm{dp}=\left[\mathrm{m}^{3}\right] /\left[\mathrm{N} \mathrm{m}^{-2}\right]=\left[\mathrm{m}^{5} \mathrm{~N}^{-1}\right]\ &=\frac{\left[\mathrm{m}^{6}\right]}{[\mathrm{N} \mathrm{m}]}=\left[\mathrm{m}^{5} \mathrm{~N}^{-1}\right] \end{aligned} [3] \begin{aligned} &\frac{\mathrm{dV}}{\mathrm{dT}}=\frac{\left[\mathrm{m}^{3}\right]}{[\mathrm{K}]}\ &\mathrm{n}_{\mathrm{X}}^{0} \, \frac{\Delta_{\mathrm{r}} \mathrm{V}^{\infty}(\mathrm{aq}) \, \Delta_{\mathrm{r}} \mathrm{H}^{\infty}(\mathrm{aq})}{\mathrm{R} \, \mathrm{T}^{2}} \, \frac{\mathrm{K}^{0}}{\left[1+\mathrm{K}^{0}\right]^{2}}\ &=[\mathrm{mol}] \, \frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{J} \mathrm{mol}^{-1}\right]}{\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]^{2}}=\left[\mathrm{m}^{3} \mathrm{~K}^{-1}\right] \end{aligned} [4] See for example, J. E. Desnoyers, Pure Appl.Chem.,1982, 54,1469. [5] (i) W. J. leNoble, J. Chem. Educ.,1967,44,729. (ii) B. S. El’yanov and S. D. Hamann, Aust. J Chem.,1975,28,945. (iii) B.S. El’yanov and M. G. Gonikberg, Russian J. Phys. Chem., 1972, 46, 856. 1.22.2: Volume: Components One Component For a system containing one chemical substance we define the volume as follows, $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}\right]$ The variables in the square brackets are called the INDEPENDENT VARIABLES. The term independent means that within limits, we can change $\mathrm{T}$ independently of the pressure and $\mathrm{n}_{1}$; change $\mathrm{p}$ independently of $\mathrm{T}$ and $\mathrm{n}_{1}$; change $\mathrm{n}_{1}$ independently of $\mathrm{T}$ and $\mathrm{p}$. There are some restrictions in our choice of independent variables. At least one of the variables must define the amount of all chemical substances in the system and one variable must define the degree of ‘hotness’ of the system. Two Chemical Substances If the composition of a given closed system is specified in terms of the amounts of two chemical substances, 1 and 2, four independent variables $\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}\right]$ define the independent variable $\mathrm{V}$. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}\right]$ Volume i - Chemical Substances For a system containing i - chemical substances where the amounts can be independently varied, the dependent variable $\mathrm{V}$ is defined by the following equation. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \ldots . \mathrm{n}_{\mathrm{i}}\right]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.12%3A_Volume_of_Reaction%3A_Dependence_on_Pressure.txt
In descriptions of the volumetric properties of solutions, two terms are extensively used. We refer to the partial molar volume of solute $j$ in, for example, an aqueous solution $\mathrm{V}_{j}(\mathrm{aq})$ and the corresponding apparent molar volume $\phi\left(\mathrm{V}_{j}\right)$. Here we explore how these terms are related. We consider an aqueous solution prepared using water, $1 \mathrm{~kg}$, and $\mathrm{m}_{j}$ moles of solute $j$. The volume of this solution at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ is given by equation (a). $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ The chemical potential of solvent, water, in the aqueous solution $\mu_{1}(\mathrm{aq})$ is related to the molality $\mathrm{m}_{j}$ using equation (b) where $\mu_{1}^{*}(\ell)$ is the chemical potential of pure water($\ell$), molar mass $\mathrm{M}_{1}$, at the same $\mathrm{T}$ and $\mathrm{p}$. $\mu_{1}(\mathrm{aq})=\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ Here practical osmotic coefficient $\phi$ is defined by equation (c). $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi=1 \quad \text { at all T and } \mathrm{p}$ But $\mathrm{V}_{1}(\mathrm{aq})=\left[\partial \mu_{1}(\mathrm{aq}) / \partial \mathrm{p}\right]_{\mathrm{T}}$ Then[1] $\mathrm{V}_{1}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{V}_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,(\partial \phi / \partial \mathrm{p})_{\mathrm{T}}$ For the solute, the chemical potential $\mu_{j}(\mathrm{aq})$ is related to the molality of solute $\mathrm{m}_{j}$ using equation (f) where pressure $\mathrm{p}$ is close to the standard pressure. $\mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)$ where $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \gamma_{j}=1 \quad \text { at all } \mathrm{T} \text { and } \mathrm{p}$ Then[2] $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})=\mathrm{V}_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}$ $\operatorname{Limit}\left(m_{j} \rightarrow 0\right) V_{j}(a q)=V_{j}^{0}(a q)=V_{j}^{\infty}(a q)$ Combination of equations (a), (e) and (h) yields equation (j). $\begin{gathered} \mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \,\left[\mathrm{V}_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \,(\partial \phi / \partial \mathrm{p})_{\mathrm{T}}\right] \ \quad+\mathrm{m}_{\mathrm{j}} \,\left\{\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}\right\} \end{gathered}$ An important point emerges if we re-arrange equation (j). \begin{aligned} &\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell) \ &\quad+\mathrm{m}_{\mathrm{j}} \,\left\{\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}-\mathrm{R} \, \mathrm{T} \,(\partial \phi / \partial \mathrm{p})_{\mathrm{T}}\right\} \end{aligned} Equation (k) has an interesting form in that the brackets {….} contain $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$ and terms describing the extent to which the volumetric properties of the solution are not ideal in a thermodynamic sense. It is therefore convenient to define an apparent molar volume of solute $j$, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ using equation (l). $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}-\mathrm{R} \, \mathrm{T} \,(\partial \phi / \partial \mathrm{p})_{\mathrm{T}}$ Then $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi\left(\mathrm{V}_{\mathrm{j}}\right)=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$ Therefore we obtain equation (n). $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Interest in equation (n) arises from the fact that the for a given solution $\mathrm{V}(\mathrm{aq})$ can be measured {using the density $\rho(\mathrm{aq})$} and hence knowing $\mathrm{V}_{1}^{*}(\ell)$, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ is obtained. If we measure $\phi\left(\mathrm{V}_{j}\right)$ as a function of $\mathrm{m}_{j}$, equation (m) indicates how one obtains $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$. Moreover the difference $\left[\phi\left(\mathrm{V}_{\mathrm{j}}\right)-\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})\right]$ signals the role of solute - solute interactions. Footnotes [1] $\begin{array}{r} \mathrm{R} \, \mathrm{T} \,\left[\frac{\partial \phi}{\partial \mathrm{p}}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}] \,\left[\frac{1}{\left[\mathrm{~N} \mathrm{~m}^{-2}\right]}\right] \ =\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \end{array}$ [2] \begin{aligned} \mathrm{R} \, \mathrm{T} \,\left[\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right]=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}] \,\left[\frac{1}{\left[\mathrm{~N} \mathrm{~m}^{-2}\right]}\right] \ =& {[\mathrm{N} \mathrm{m} \mathrm{mol}] \,\left[\frac{1}{\left[\mathrm{~N} \mathrm{~m}^{-2}\right.}\right]=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] } \end{aligned} 1.22.4: Volume: Partial Molar: Frozen and Equilibrium Consider the volume of a closed system defined by equation (a). $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ This system is displaced to a neighboring state by addition of a small amount of substance $j$, $\delta\mathrm{n}_{j}$. The change in volume at fixed affinity $\mathrm{A}$ is related to the change in volume at fixed composition or organization. At fixed temperature, fixed pressure, and fixed $\mathrm{n}_{1}$, $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{A}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\xi}-\left(\frac{\partial \mathrm{A}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\xi} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{n}_{\mathrm{j}}} \,\left(\frac{\partial \mathrm{V}}{\partial \xi}\right)_{\mathrm{n}_{\mathrm{j}}}$ For a system at equilibrium ($\mathrm{A} = 0 \text { and } \xi = \xi^{\mathrm{eq}}$), the triple product term on the R.H.S. of equation (b) is not zero. Hence we distinguish between two properties; $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{~A}=0} \text { and } \left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{eq}^{\mathrm{eq}}}$. By convention the first of these two terms is called the partial molar volume of substance $j$ in the system. Example 1 A given aqueous solution is prepared by dissolving $\mathrm{n}_{j}$ moles of urea in $\mathrm{n}_{1}$ moles of water at $298.2 \mathrm{~K}$ and ambient pressure. This system has volume $\mathrm{V}(\mathrm{aq})$ which is determined in part by water-water, water-urea and urea-urea interactions. We add $\delta \mathrm{n}_{j}$ moles of urea to this system but stipulate that the water-water, water-urea and urea-urea interactions remain unchanged; i.e. frozen. The property $\left(\partial \mathrm{V} / \partial \mathrm{n}_{\mathrm{j}}\right)_{\mathrm{T} ; \mathrm{p} ; \mathrm{n}(1), \xi}$ is a frozen partial molar volume of urea in the aqueous solution. On the other hand, if we stipulate that the water-water, water-urea and urea-urea interactions re-adjust in order that the system is at a minimum in Gibbs energy, the property $\left(\partial \mathrm{V} / \partial \mathrm{n}_{\mathrm{j}}\right)_{\mathrm{T}, \mathrm{p} ; \mathrm{n}(1), \mathrm{A}=0}$ is the equilibrium partial molar volume for urea in this aqueous solution. Example 2 We consider an aqueous solution containing $\mathrm{n}_{j}$ moles of ethanoic acid in $\mathrm{n}_{1}$ moles of water at defined temperature and defined pressure. Conventionally, the chemical equilibrium operating in the system is expressed in the following form. $\mathrm{HA}(\mathrm{aq}) \leftrightarrow \mathrm{H}^{+}(\mathrm{aq})+\mathrm{A}^{-}(\mathrm{aq})$ The volume of this system $\mathrm{V}(\mathrm{HA} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is a state variable. We add $\delta \mathrm{δ}(\mathrm{HA})$ moles of substance $\mathrm{HA}$ to the system. In the frozen limit, the amounts of $\mathrm{H}^{+}(\mathrm{aq})$ and $\mathrm{A}^{-}(\mathrm{aq})$ in the solution do not change. In terms of composition all that happens is the amount of $\mathrm{HA}(\mathrm{aq})$ increases. Hence, $\left(\frac{\delta \mathrm{V}}{\delta \mathrm{n}(\mathrm{HA})}\right)$ is a measure of the ‘frozen partial molar volume’ of $\mathrm{HA}$ in the system. If we remove the frozen restriction and allow chemical equilibrium to be re-established, the derived quantity is the equilibrium partial molar volume for $\mathrm{HA}$ in this aqueous solution, part of added $\delta\mathrm{P}mathrm{n}(\mathrm{HA})$ having dissociated in order that the resulting solution has zero affinity for spontaneous change. We use quotation marks ‘….’ Around the phrase ‘frozen partial molar volume’ to make the point that this property is not a proper equilibrium thermodynamic property.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.3%3A_Volume%3A_Partial_and_Apparent_Molar.txt
An aqueous solution is prepared using $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ moles of solute. Thus, $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{\mathrm{j}}\right]$ The density of this solution $\rho(\mathrm{aq})$ can be accurately measured at the specified temperature and pressure together with the density of the pure solvent, $\rho_{1}^{*}(\ell)$. The molar mass of the solute is $\mathrm{M}_{j} \mathrm{~kg mol}^{-1}$. Two equations [1-3] are encountered in the literature depending on the method used to describe the composition of the solution [4]. Molality Scale [1] $\phi\left(V_{j}\right)=\left[m_{j} \, \rho(\mathrm{aq}) \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell)-\rho(\mathrm{aq})\right]+\mathrm{M}_{\mathrm{j}} / \rho(\mathrm{aq})$ Concentration Scale [2,3] $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\left[\mathrm{c}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell)-\rho(\mathrm{aq})\right]+\mathrm{M}_{\mathrm{j}} / \rho_{1}^{*}(\ell)$ Equation (b) using molalities and (c) using concentrations yield the same property of the solute, namely the apparent molar volume of solute $j$, $\phi\left(\mathrm{V}_{j}\right)$. Equations (b) and (c) are exact. The equations are readily distinguished by the difference in the denominators of the last terms. In any event the trick in deriving these equations is to seek an equation having the form, {[Property of solvent] minus [Property of solute]}. The subject is slightly complicated because the concentration of solute j can be expressed using either the unit ‘$\mathrm{mol m}^{-3}$’ or the unit ‘$\mathrm{mol dm}^{-3}$’, the latter being the most common. There is also a problem over the unit used for densities. Some authors use the unit ‘$\mathrm{kg m}^{-3}$‘ whereas other authors use the unit ‘$\mathrm{g cm}^{-3}$‘. The latter practice accounts for the numerical factor $10^{3}$ which often appears in many published equations of the form shown in equations (b) and (c). Partial molar and partial molal properties are often identified. The two terms are synonymous in the case of partial molar volumes and partial molal volumes of solutes in aqueous solutions. IUPAC recommends the use of the term ‘partial molar volume’ [5]. Significantly we can never know the absolute value of the chemical potential of a solute in a given solution but we can determine the partial molar volume, the differential dependence of chemical potential on pressure. Indeed the challenge of understanding patterns in partial molar volumes seems less awesome than the task of understanding other thermodynamic properties of solutes. Equations (b) and (c) do not describe how $\phi\left(\mathrm{V}_{j}\right)$ for a given solute depends on either $\mathrm{m}_{j}$ or $\mathrm{c}_{j}$. This dependence is characteristic of a solute (at fixed $\mathrm{T}$ and $\mathrm{p}$) and reflects the role of solute - solute interactions. In many cases where solute $j$ is a simple neutral solute, $\phi\left(\mathrm{V}_{j}\right)$ for dilute solutions is often satisfactorily accounted for by an equation in which $\phi\left(\mathrm{V}_{j}\right)$ is a linear function of $\mathrm{m}_{j}$. The slope $\mathrm{S}$ is characteristic of the solute (at fixed $\mathrm{T}$ and $\mathrm{p}$) [4d,6]. $\phi\left(V_{\mathrm{j}}\right)=\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}+\mathrm{S} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi\left(\mathrm{V}_{\mathrm{j}}\right)=\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$ In the case of urea(aq) at $298.2 \mathrm{~K}$ and ambient pressure the dependence of $\phi\left(\mathrm{V}_{j}\right)$ on $\mathrm{m}_{j}$ is described by the following quadratic equation [7,8]. $\phi\left(\mathrm{V}_{\mathrm{j}}\right) / \mathrm{cm}^{3} \mathrm{~mol}^{-1}=44.20+0.126 \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)-0.004 \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{2}$ In general terms therefore $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq})$ {and $\mathrm{V}_{\mathrm{j}}^{\infty}$ for solute $j$ in other solvents [9]} characterizes solute - solvent interactions and the dependences of $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ on $\mathrm{m}_{j}$ characterizes solute - solute interactions. Of course the partial molar volume of solute-$j$ in solution is not the actual volume of solute-$j$. Instead $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ measures the differential change in the volume of an aqueous solution when $\delta \mathrm{n}_{j}$ moles of substance-$j$ are added. We emphasize the importance of an approach using the molalities of solutes. The reasons are straightforward. If we compare $\phi\left(\mathrm{V}_{j}\right)$ for a solute in solutions containing $0.1$ and $0.01 \mathrm{~mol kg}^{-1}$, in this comparison, the mass of solvent remains the same. If on the other hand we compare $\phi\left(\mathrm{V}_{j}\right)$ for solute in solutions where $\mathrm{c}_{j} / \mathrm{~mol dm}^{-3} = 0.1$ and $0.01$, the amounts of solvent are not defined. Nevertheless many treatments of the properties of solutions examine $\phi\left(\mathrm{V}_{j}\right)$ as a function of concentration. In fact chemists tend to think in terms of concentrations and hence in terms of distances between solute molecules. So in these terms concentration might be thought of as the 'natural' scale. Just as in life, one is more interested in the distance between two people rather than their mass. No rule forbids one to fit the dependences of $\phi\left(\mathrm{V}_{j}\right)$ on $\mathrm{c}_{j}$ using an equation of the following form. $\phi\left(V_{j}\right)=a_{1}+a_{2} \, c_{j}+a_{3} \, c_{j}^{2}+\ldots$ But if $\phi\left(\mathrm{V}_{j}\right)$ is a linear function of $\mathrm{m}_{j}$, $\phi\left(\mathrm{V}_{j}\right)$ is not a linear function of $\mathrm{c}_{j}$ [10]. Of course $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}$ in both $\operatorname{limit}\left(\mathrm{m}_{j} \rightarrow 0 \right)$ and $\operatorname{limit} \left(\mathrm{c}_{j} \rightarrow 0 \right)$. Granted the outcome of an experiment is the dependence of $\phi\left(\mathrm{V}_{j}\right)$ on $\mathrm{m}_{j}$, the partial molar volume $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})$ of solute $j$ is readily calculated; equation (h) [11]. $\mathrm{V}_{\mathrm{j}}(\mathrm{aq})=\phi\left(\mathrm{V}_{\mathrm{j}}\right)+\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)$ We note important features in the context of two plots; 1. $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1} / \mathrm{kg}=1\right)$ and 1. $\left[\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)\right]$ against molality $\mathrm{m}_{j}$. Then $\mathrm{V}_{j}(\mathrm{aq})$ is the gradient of the tangent to the curve in plot (i) at the specified molality; $\phi\left(\mathrm{V}_{j}\right)$ is the gradient of the line in plot(ii) joining the origin and $\left[\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)\right]$ at the specified molality. Footnotes [1] For the solution volume $\mathrm{V}(\mathrm{aq})$, $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ If the molar mass of the solvent is $\mathrm{M}_{1} \mathrm{~kg mol}^{-1}$, $\mathrm{V}(\mathrm{aq})=\left[\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right] / \rho(\mathrm{aq})$ and $\mathrm{V}_{1}^{*}(\ell)=\mathrm{M}_{1} / \rho_{1}^{*}(\ell)$. From equation (a), $\frac{\mathrm{n}_{1} \, \mathrm{M}_{1}}{\rho(\mathrm{aq})}+\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}=\mathrm{n}_{1} \, \frac{\mathrm{M}_{1}}{\rho_{1}^{*}(\ell)}+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ Divide by $\mathrm{n}_{1} \, \mathrm{~M}_{1}$ and rearrange; $\frac{n_{j}}{n_{1} \, M_{1}} \, \phi\left(V_{j}\right)=\frac{1}{\rho(a q)}-\frac{1}{\rho_{1}^{*}(\ell)}+\frac{n_{j} \, M_{j}}{n_{1} \, M_{1} \, \rho(a q)}$ But molality $\mathrm{m}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{1} \, \mathrm{M}_{1}$. Then, $\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)=\left[\frac{1}{\rho(\mathrm{aq})}-\frac{1}{\rho_{1}^{*}(\ell)}\right]+\frac{\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}$ or, $\phi\left(V_{j}\right)=\left[m_{j} \, \rho(a q) \, \rho_{1}^{*}\right]^{-1} \,\left[\rho_{1}^{*}(\ell)-\rho(a q)\right]+\frac{M_{j}}{\rho(a q)}$ Thus, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\left[\frac{1}{\left[\mathrm{~mol} \mathrm{~kg}^{-1}\right] \,\left[\mathrm{kg} \mathrm{m}^{-3}\right]^{2}}\right] \,\left[\mathrm{kg} \mathrm{m}^{-3}\right]+\frac{\left[\mathrm{kg} \mathrm{mol}^{-1}\right]}{\left[\mathrm{kg} \mathrm{m}^{-3}\right.}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]$ [2] At fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}(\mathrm{aq})=\mathrm{n}_{\mathrm{j}} /\left[\mathrm{n}_{\mathrm{1}} \, \mathrm{V}_{1}^{*}(\mathrm{l})+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)\right]$ But, $\mathrm{m}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{\mathrm{l}} \, \mathrm{M}_{1}$. Then, $\mathrm{c}_{\mathrm{j}}=\frac{\mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}}{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}$ Invert. $\frac{1}{\mathrm{c}_{\mathrm{j}}}=\frac{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}}+\frac{\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}}$ But $\mathrm{n}_{\mathrm{j}} / \mathrm{m}_{\mathrm{j}} \, \mathrm{n}_{1} \, \mathrm{M}_{1}=1$ Then, $\frac{1}{\mathrm{c}_{\mathrm{j}}}=\frac{1}{\mathrm{~m}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)}+\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ or, $\frac{1}{\mathrm{~m}_{\mathrm{j}}}=\frac{\rho_{1}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}}}-\rho_{1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ The latter equation links $\mathrm{m}_{j}$ and $\mathrm{c}_{j}$. [3] From [1], $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\frac{1}{\mathrm{~m}_{\mathrm{j}}} \,\left[\frac{1}{\rho(\mathrm{aq})}-\frac{1}{\rho_{1}^{*}(\ell)}\right]+\frac{\mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}$ Then, from [2], $\phi\left(V_{j}\right)=\left[\frac{\rho_{1}^{*}(\ell)}{c_{j}}-\rho_{1}^{*}(\ell) \, \phi\left(V_{j}\right)\right] \,\left[\frac{1}{\rho(\mathrm{aq})}-\frac{1}{\rho_{1}^{*}(\ell)}\right]+\frac{M_{j}}{\rho(\mathrm{aq})}$ Hence, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\frac{\rho_{1}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}} \, \rho(\mathrm{aq})}-\frac{1}{\mathrm{c}_{\mathrm{j}}}-\frac{\rho_{1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\rho(\mathrm{aq})}+\phi\left(\mathrm{V}_{\mathrm{j}}\right)+\mathrm{M}_{\mathrm{j}} / \rho(\mathrm{aq})$ Then, $\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \frac{\rho_{1}^{*}(\ell)}{\rho(\mathrm{aq})}=\frac{\rho_{1}^{*}(\ell)}{\mathrm{c}_{\mathrm{j}} \, \rho(\mathrm{aq})}-\frac{1}{\mathrm{c}_{\mathrm{j}}}+\frac{\mathrm{M}_{\mathrm{j}}}{\rho(\mathrm{aq})}$ $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\frac{1}{\mathrm{c}_{\mathrm{j}}}-\frac{1}{\mathrm{c}_{\mathrm{j}}} \, \frac{\rho(\mathrm{aq})}{\rho_{1}^{*}(\ell)}+\frac{\mathrm{M}_{\mathrm{j}}}{\rho_{1}^{*}(\ell)}$ or, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\left[\mathrm{c}_{\mathrm{j}} \, \rho_{1}^{*}(\ell)\right]^{-1} \,\left[\rho_{1}^{*}(\ell)-\rho(\mathrm{aq})\right]+\mathrm{M}_{\mathrm{j}} / \rho_{1}^{*}(\ell)$ Thus, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\frac{1}{\left[\mathrm{~mol} \mathrm{~m}^{-3}\right]} \, \frac{1}{\left[\mathrm{~kg} \mathrm{~m}^{-3}\right]^{-1}} \,\left[\mathrm{kg} \mathrm{m}^{-3}\right]+\frac{\left[\mathrm{kg} \mathrm{mol}^{-1}\right]}{\left[\mathrm{kg} \mathrm{m}^{-3}\right]}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]$ [4] The following publications use equation (b) based on the molality composition scale. 1. Dipeptides(aq); J. E. Reading, I. D. Watson and G .R. Hedwig, J. Chem. Thermodyn., 1990, 22, 159. 2. N-alkoxyethanols(aq); G. Roux, G. Peron and J. E. Desnoyers, J. Solution Chem.,1978,7,639. 3. Cyclic organic chemical substhances in 1-ocytanol; P. Berti, S. Cabani and V. Mollica, Fluid Phase Equilib.,1987,32,195. 4. Alkali halides in urea + water; N. Desrosiers, G. Perron, J. G. Mathieson, B. E. Conway and J. E. Desnoyers, J. Solution Chem.,1974,3,789. 5. HCl(aq), HBr(aq) and HI(aq); T. M. Herrington, A.D. Pethybridge and M. G. Roffey, J. Chem. Eng. Data,1985,30,264. 6. LiOH(aq), NaOH(aq), KOH(aq); T. M. Herrington, A.D. Pethybridge and M. G. Roffey, J. Chem. Eng. Data,1986,31,31. 7. $\mathrm{R}_{4}\mathrm{NI}(\mathrm{aq})$;B. M. Lowe and H. M. Rendall, Trans. Faraday Soc.,1971,67,2318. 8. HCl(aq) and $\mathrm{HClO}_{4}(\mathrm{aq})$; R. Pogue and G. Atkinson, J. Chem. Eng. Data, 1988,33,495. 9. $\mathrm{MCl}_{2}(\mathrm{aq})$ where M= Mn, Co, Ni, Zn and Cd; T. M. Herrington, M. G. Roffey, and D. P. Smith, J. Chem.Eng. Data,1986,31,221. 10. $\mathrm{NiCl}_{2}(\mathrm{aq}), \mathrm{~Ni}\left(\mathrm{ClO}_{4}\right)_{2}(\mathrm{aq}), \mathrm{~CuCl}_{2}(\mathrm{aq}) \text { and } \mathrm{Cu}\left(\mathrm{ClO}_{4}\right)_{2}(\mathrm{aq})$; R. Pogue and G. Atkinson, J. Chem. Eng. Data, 1988,33,370. 11. $\mathrm{RMe}_{3}\mathrm{NBr}(\mathrm{aq})$; R. De Lisi, S. Milioto and R. Triolo, J. Solution Chem.,1988,17,673. 12. $\mathrm{Ph}_{4}\mathrm{AsCl}(\mathrm{aq})$; F. J. Millero, J. Chem. Eng. Data, 1971,16,229. 13. $\mathrm{R}_{4}\mathrm{NBr}(\mathrm{aq} + \mathrm{BuOH})$; L. Avedikian, G. Perron and J. E. Desnoyers, J. Solution Chem.,1975,4,331. Applications of equation. (c) include 1. $\mathrm{Bu}_{4}\mathrm{N}^{+} \text { carboxylates}(\mathrm{aq})$; P.-A.Leduc, and J. E. Desnoyers, Can. J. Chem.,1973,51,2993. 2. N-Alkylamine hydrobromides(aq); P.-A.Leduc, and J. E. Desnoyers, J. Phys. Chem., 1974, 78, 1217. 3. $\mathrm{R}_{4}\mathrm{N}^{+} \mathrm{~Cl}^{-} (\mathrm{aq} + \mathrm{~DMSO})$; D. D. Macdonald and J.B.Hyne, Can. J.Chem.,1970,48,2416. 4. $\mathrm{R}_{4}\mathrm{N}^{+} \mathrm{~Cl}^{-} (\mathrm{aq} + \mathrm{~EtOH})$; I. Lee and J. B. Hyne, Can. J.Chem.,1968,46,2333. [5] Manual of Symbols and Terminology for Physicochemical Quantities and Units, IUPAC, Pergamon, Oxford, 1979. [6] F. Franks and H.T. Smith, Trans. Faraday Soc., 1968, 64, 2962. [7] D. Hamilton and R. H. Stokes, J. Solution Chem., 1972,1, 213. [8] R. H. Stokes, Aust . J. Chem., 1967,20, 2087. [9] Solvent $\mathrm{V}^{infty}$ (urea; sln; $298 \mathrm{~K}$; ambient p)/cr $\mathrm{mol}^{-1}$ $\mathrm{H}_{2}\mathrm{O}$ 44.24 $\mathrm{CH}_{3}\mathrm{OH}$ 36.97 $\mathrm{C}_{2}\mathrm{H}_{5}\mathrm{OH}$ 40.75 Formamide 44.34 $\mathrm{DMF}$ 39.97 $\mathrm{DMSO}$ 41.86 [10] From [2], $\mathrm{m}_{\mathrm{j}}=\mathrm{c}_{\mathrm{j}} /\left[\rho_{1}^{*}(\ell)-\rho_{1}^{*}(\ell) \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \mathrm{c}_{\mathrm{j}}\right]$. If $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\mathrm{a}_{1}+\mathrm{a}_{2} \, \mathrm{m}_{\mathrm{j}}$ Then, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\mathrm{a}_{1}+\left\{\mathrm{a}_{2} / \rho_{1}^{*}(\ell) \,\left[1-\phi\left(\mathrm{V}_{\mathrm{j}}\right) \, \mathrm{c}_{\mathrm{j}}\right]\right\} \, \mathrm{c}_{\mathrm{j}}$ i.e. the slope depends on the product $\phi\left(\mathrm{V}_{j}\right) \, \mathrm{c}_{j}$. [11] From $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ At constant $\mathrm{n}_{1}$, $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)=\phi\left(\mathrm{V}_{\mathrm{j}}\right)+\mathrm{n}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{n}_{\mathrm{j}}}\right)$ Or, $\mathrm{V}_{\mathrm{j}}=\phi\left(\mathrm{V}_{\mathrm{j}}\right)+\mathrm{m}_{\mathrm{j}} \,\left(\frac{\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right)$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.5%3A_Volumes%3A_Solutions%3A_Apparent_and_Partial_Molar_Volumes%3A_Determination.txt
For a solution prepared using $1 \mathrm{~kg}$ of water, the volume is related to the apparent molar volume of the solute $\phi \left(\mathrm{V}_{j}\right)$ using equation (a). $\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ If the thermodynamic properties of this solution are ideal, $\mathrm{V}\left(\mathrm{aq} ; \mathrm{id} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}$ Here $V_{\mathrm{j}}^{\infty}(\mathrm{aq}) \equiv \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}$. The difference between $\mathrm{V}(\mathrm{aq})$ and $\mathrm{V}(\mathrm{aq}: \mathrm{id})$ defines an excess volume $\mathrm{V}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)$. Thus, $\mathrm{V}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)-\mathrm{V}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}: \mathrm{id}\right)$ Hence, $\mathrm{V}^{\mathrm{E}}\left(\mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mathrm{m}_{\mathrm{j}} \,\left[\phi\left(\mathrm{V}_{\mathrm{j}}\right)-\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}\right]$ 1.22.7: Volumes: Neutral Solutes: Limiting Partial Molar Volumes A given aqueous solution at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ contains solute $j$, having molality $\mathrm{m}_{j}$. The chemical potential of solute $j$, $\mu_{j}(\mathrm{aq})$ is related to $\mathrm{m}_{j}$ using equation (a). \begin{aligned} &\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) \ &=\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T}) \, \mathrm{dp} \end{aligned} But $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\left(\frac{\partial \mu_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ Also $\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{0}\right)$ is, by definition, independent of pressure. From equation (a), $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \ln \left(\gamma_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ In equation (c), there is no term explicitly in terms of molaity $\mathrm{m}_{j}$. From the definition of $\gamma_{j}$, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is the limiting partial molar volume of solute $j$ in aqueous solution at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. In other words $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is the partial molar volume of solute $j$ in the (ideal) solution where there are no solute-solute interactions and characterizes solute-water interactions. Because $\gamma_{j}$ tends to unity as $\mathrm{m}_{j}$ tends to zero, $\gamma_{j}$ is sometimes called an asymmetric activity coefficient [1]. [Contrast rational activity coefficients where $\mathrm{f}_{1} \rightarrow 1 \text { as } \mathrm{x}_{1} \rightarrow 1$.] At the risk of being repetitive we distinguish between the two possible reference states for substance $j$ such as urea. One reference state is the pure solid chemical substance $j$ at ambient pressure and $298.2 \mathrm{~K}$. Another reference state is the ideal solution where $\mathrm{m}_{\mathrm{j}}=1 \mathrm{~mol} \mathrm{~kg}^{-1}$ at ambient pressure and $298.2 \mathrm{~K}$. The properties of urea in the two states, pure solid and solution standard state are clearly quite different. Indeed, we can compare $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; 298.2 \mathrm{~K} ; \text { ambient p) }$ and $\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~s} ; 298.2 \mathrm{~K} ; \text { ambient } \mathrm{p})$. We can also compare, for example, $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{j}=\text { urea; sln} 298.2 \mathrm{~K} ; \text { ambient p) }$ in a range of solvents [2]. These points are also nicely illustrated by the volumetric properties of water [3,]. At $298.2 \mathrm{~K}$ and ambient pressure $\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{H}_{2} \mathrm{O}\right)$ is $18.07 \mathrm{~cm}^{3} \mathrm{~mol}^{-1}$ but for water as a solute in three solvents, $\mathrm{V}^{\infty}\left(\mathrm{H}_{2} \mathrm{O} ; \operatorname{sln}\right)=18.47(\mathrm{MeOH}), 14.42(\mathrm{EtOH}) \text { and } 17.00(\mathrm{THF}) \mathrm{cm}^{3} \mathrm{~mol}^{-1}$ [4]. There is, of course, no reason why we should expect anything different. A water molecule in liquid water is surrounded by many millions of other water molecules. But a water molecule at infinite dilution in solvent ethanol is surrounded by many millions of ethanol molecules [5,6]. In the analysis of experimental results , we may express the composition of the solution in terms of mole fraction of solute $\mathrm{x}_{j}$. Then \begin{aligned} &\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) \ &=\mu_{\mathrm{j}}^{0}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{0} ; \mathrm{x}-\mathrm{scale}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{\mathrm{j}} \, \mathrm{f}_{\mathrm{j}}^{*}\right)+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T}) \, \mathrm{dp} \end{aligned} But mole fraction $\mathrm{x}_{j}$ is independent of pressure. $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \,\left(\frac{\partial \ln \left(\mathrm{f}_{\mathrm{j}}^{*}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ From the definition of $\mathrm{f}_{\mathrm{j}}^{*}$, $\operatorname{limit}\left(\mathrm{x}_{\mathrm{j}} \rightarrow 0\right) \mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ The limiting value of $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is identical on the molality and mole fraction scales. If we use the concentration scale a problem arises in that the concentration of solute $j$, $\mathrm{c}_{j}$ is dependent on pressure because the volume of the solution is pressure dependent. Footnote [1] W. L. Masterton and H. K. Seiler, J. Phys. Chem., 1968,72, 4257. [2] For $j =$ urea at $298.2 \mathrm{~K}$ and ambient pressure, $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{s} \ln ) / \mathrm{cm}^{3} \mathrm{~mol}^{-1}=$ 44.24 (water), 36.97 (methanol), 40.75 (ethanol) and 41.86 (DMSO). [3] $\mathrm{V}_{\mathrm{j}}^{\infty}(298.15 \mathrm{~K} ; \mathrm{j}=\text { water })=18.57 \mathrm{~cm}^{3} \mathrm{~mol}^{-1}(\text { solvent }=\text { octan-1-ol })$ and $31.3 \mathrm{cm}^{3} \mathrm{~mol}^{-1} \text { (solvent }= \mathrm{CCl}_{4})$; P. Berti, S. Cabani and V. Mollica, Fluid PhaseEquilib., 1987,32 , 1. [4] M. Sakurai and T. Nakagawa, Bull. Chem. Soc. Jpn., 1984, 55, 195; J. Chem. Thermodyn., 1982, 14, 269; 1984, 16 , 171. [5] A similar contrast exists (H. Itsuki, S. Terasawa, K. Shinohara and H. Ikezwa, J. Chem. Thermodyn., 1987,19, 555) between the molar volume of a hydrocarbon and its limiting partial molar volume in another hydrocarbon; $\mathrm{V}^{*}\left(\ell ; \mathrm{C}_{6} \mathrm{H}_{14}\right)=131.61 \mathrm{~cm}^{3} \mathrm{~mol}^{-1}$, but $\mathrm{V}^{\infty}\left(\mathrm{C}_{6} \mathrm{H}_{14} ; \text { sln; solvent }=\mathrm{C}_{16} \mathrm{H}_{34}\right)=130.2 \mathrm{~cm}^{3} \mathrm{~mol}^{-1}$ at $298 \mathrm{~K}$ and ambient pressure. In this context the limiting enthalpies of solution water in monohydric alcohols depend on the alcohol at $298.2 \mathrm{~K}$; (S.-O. Nilsson, J. Chem. Thermodyn., 1986, 18, 1115). [6] The partial molar volumes of fullerene in solution is $401 \mathrm{~cm}^{3} \mathrm{~mol}^{-1}$ in cis-decalin and $389 \mathrm{~cm}^{3} \mathrm}{mol}^{-1}$ in 1,2-dichlorobenzene both values being significantly less than the predicted volume of the pure liquid $\mathrm{C}_{60}$; (P. Ruelle, A. Farina-Cuendet and U. W. Kesselring, J. Chem. Soc. Chem. Commun., 1995, 1161).
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.6%3A_Volumes%3A_Apparent_Molar_and_Excess_Volumes.txt
Differentiation of the Born Equation with respect to pressure (at fixed temperature) yields the Born-Drude-Nernst Equation which describes the difference in partial molar volumes of ion $j$ in the gas phase and in solution. The simplest model assumes that the radius $\mathrm{r}_{j}$ is independent of pressure [1]. \begin{aligned} \Delta(\mathrm{pfg}&\rightarrow \mathrm{s} \ln ) \mathrm{V}_{\mathrm{j}}\left(\mathrm{c}_{\mathrm{j}}=1 \mathrm{~mol} \mathrm{dm} \mathrm{m}^{-3} ; \mathrm{id} ; \mathrm{p}, \mathrm{T}\right)=\ &-\mathrm{N}_{\mathrm{A}} \,\left(\mathrm{z}_{\mathrm{j}} \, \mathrm{e}\right)^{2} \,\left[\frac{1}{\varepsilon_{\mathrm{r}}} \,\left(\frac{\partial \varepsilon_{\mathrm{r}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}\right] \, \frac{1}{8 \, \pi \, \mathrm{r}_{\mathrm{j}} \, \varepsilon_{0}} \end{aligned} A more complicated equation emerges if radius $\mathrm{r}_{j}$ is assumed to depend on pressure, but there seems little merit in taking account of such a dependence. Footnote [1] $\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]=\left[\mathrm{mol}^{-1}\right] \,\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1} \,[1]^{-1} \,[1]^{-1} \,\left[\mathrm{m}^{-1}\right] \,\left[\mathrm{F} \mathrm{m}^{-1}\right]^{-}$ where, $\left[\mathrm{F} \mathrm{m}^{-1}\right]=\left[\mathrm{A}^{2} \mathrm{~s}^{4} \mathrm{~kg}^{-1} \mathrm{~m}^{-3}\right]$ 1.22.9: Volume: Liquid Mixtures A given binary liquid mixture is prepared using $\mathrm{n}_{1}$ moles of liquid 1 and $\mathrm{n}_{2}$ moles of liquid 2 at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The term ‘mixture’ usually means that a homogeneous single liquid phase is spontaneously formed on mixing characterized by a minimum in Gibbs energy $\mathrm{G}$ where the molecular organization is characterized by $\xi^{\mathrm{eq}}$. $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}$ measures the extent to which $\mathrm{G} /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ differs from this ratio in the event that the mixing is, in a thermodynamic sense, ideal. A given binary liquid mixture is displaced to a neighboring state by a change in pressure at constant temperature. The overall composition remains at $\left(\mathrm{n}_{1} + \mathrm{n}_{2}\right)$ but the organization changes to a new value for $\xi^{\mathrm{eq}}$ where ‘$\mathrm{A} = 0$’. The differential dependence of $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}$ on pressure at constant temperature $\mathrm{T}$ is the excess molar volume $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$. $\mathrm{V}(\operatorname{mix})=\left(\frac{\partial \mathrm{G}(\operatorname{mix})}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}=0}$ $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\left(\frac{\partial \mathrm{G}_{\mathrm{m}}^{\mathrm{E}}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}=0}$ For the molar volume, $\mathrm{V}_{\mathrm{m}}=\left(\frac{\partial \mathrm{G}_{\mathrm{m}}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}=0}$ The quantities, $\mathrm{V}(\mathrm{mix})$, $\mathrm{V}_{\mathrm{m}}$ and $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ are interesting because they can be determined [1] whereas the same cannot be said for $\mathrm{G}(\mathrm{mix})$ and $\mathrm{G}_{\mathrm{m}}$ although $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}}$ can be obtained m from vapor pressures of mixtures and pure components. $\mathrm{V}_{\mathrm{m}}=\mathrm{V}(\mathrm{mix}) /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ Density, $\rho(\operatorname{mix})=\left(\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{2} \, \mathrm{M}_{2}\right) / \mathrm{V}(\operatorname{mix})$ $\mathrm{M}_{1}$ and $\mathrm{M}_{2} are the molar masses of the two liquid components. By measuring \(\rho(\mathrm{mix})$ as a function of mixture composition, we form a plot of molar volume Vm as a function of mole fraction composition. The plot has two limits; $\operatorname{limit}\left(\mathrm{x}_{1} \rightarrow 1\right) \mathrm{V}_{\mathrm{m}}=\mathrm{V}_{1}^{*}(\ell)$ $\operatorname{limit}\left(\mathrm{x}_{2} \rightarrow 1\right) \mathrm{V}_{\mathrm{m}}=\mathrm{V}_{2}^{*}(\ell)$ If the thermodynamic properties of the binary liquid mixture are ideal (i.e. $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=0$), $\mathrm{V}_{\mathrm{m}}(\mathrm{id})=\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ Or, $\mathrm{V}_{\mathrm{m}}(\mathrm{id})=\left(1-\mathrm{x}_{2}\right) \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ Hence, $\mathrm{V}_{\mathrm{m}}(\mathrm{id})=\mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \,\left[\mathrm{V}_{2}^{*}(\ell)-\mathrm{V}_{1}^{*}(\ell)\right]$ The latter is an equation for a straight line. The molar volume of a real binary liquid mixture is usually less than $\mathrm{V}_{m}(\mathrm{id})$. For a real binary liquid mixture, $\mathrm{V}_{\mathrm{m}}(\operatorname{mix})=\mathrm{x}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{x}_{2} \, \mathrm{V}_{2}(\mathrm{mix})$ The difference between the molar volume of real and ideal binary liquid mixture is the excess molar volume $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$. $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \,\left[\mathrm{V}_{1}(\operatorname{mix})-\mathrm{V}_{1}^{*}(\ell)\right]+\mathrm{x}_{2} \,\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ Or, $V_{m}^{E}=x_{1} \, V_{1}^{E}(\operatorname{mix})+x_{2} \, V_{2}^{E}(\operatorname{mix}$ A given mixture, mole fraction $\mathrm{x}_{2}$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, is perturbed by addition of $\delta \mathrm{n}_{2}$ moles of chemical substance 2. The system can be perturbed either at constant organization $\xi$ or constant affinity $\mathrm{A}$. Here we are concerned with $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{2}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(1), \mathrm{A}=0}$, the (equilibrium) partial molar volume of substance 2 in the mixture, $\mathrm{V}_{2}(\mathrm{mix})$.The condition '$\mathrm{A} = 0$' implies that there is a change in organization $\xi$ in order to hold the system in the equilibrium state. A similar argument is formulated for the (equilibrium) partial molar volume $\mathrm{V}_{1}(\mathrm{mix})$. Moreover according to the Gibbs-Duhem equation (at constant temperature and pressure), $\mathrm{n}_{1} \, d \mathrm{~V}_{1}(\operatorname{mix})+\mathrm{n}_{2} \, d \mathrm{~V}_{2}(\operatorname{mix})=0$ Further, $\mathrm{V}(\operatorname{mix})=\mathrm{n}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{n}_{2} \, \mathrm{V}_{2}(\operatorname{mix})$ The property $\mathrm{V}(\mathrm{mix})$ is directly determined from the density $\rho(\mathrm{mix})$. $\mathrm{V}(\operatorname{mix})=\left(\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{2} \, \mathrm{M}_{2}\right) / \rho(\operatorname{mix})$ The important point is that the thermodynamic extensive property $\mathrm{V}(\mathrm{mix})$ is directly determined by experiment whereas we cannot for example measure the enthalpy $\mathrm{H}(\mathrm{mix})$. The excess molar volume is given by equation (l). \begin{aligned} \frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{d \mathrm{x}_{1}}=& {\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]+\mathrm{x}_{1} \, \frac{\mathrm{dV} \mathrm{V}_{1}(\mathrm{mix})}{\mathrm{dx}_{1}} } \ &-\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]+\mathrm{x}_{2} \, \frac{\mathrm{dV}_{2}(\mathrm{mix})}{\mathrm{dx}_{1}} \end{aligned} Using the Gibbs -Duhem equation, $\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}=\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]-\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ Or, $\left[\mathrm{V}_{1}(\mathrm{mix})-\mathrm{V}_{1}^{*}(\ell)\right]=\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}+\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ From equation (l), $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{1} \, \frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}+\mathrm{x}_{1} \,\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]+\mathrm{x}_{2} \,\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ Hence, $\left[\mathrm{V}_{2}(\mathrm{mix})-\mathrm{V}_{2}^{*}(\ell)\right]=\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}-\mathrm{x}_{1} \, \frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ The derivation leading up to equation (s) is the 'Method of Tangents'. Moreover at the mole fraction composition where $\frac{\mathrm{dV}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}$ is zero, $\left[\mathrm{V}_{2}(\operatorname{mix})-\mathrm{V}_{2}^{*}(\ell)\right]$ equals $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$. For liquid component 1 the chemical potential in the liquid mixture is related to the mole fraction composition (at temperature $\mathrm{T}$ and pressure $\mathrm{p}$). $\mu_{1}(\operatorname{mix}, \mathrm{T}, \mathrm{p})=\mu_{1}^{*}(\ell, \mathrm{T})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{1}^{*}(\ell ; \mathrm{T}) \, \mathrm{dp}$ At temperature $\mathrm{T}$ and pressure $\mathrm{p}$, $\mathrm{V}_{1}(\operatorname{mix})=\mathrm{V}_{1}^{*}(\ell)+\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}$ But $\mathrm{V}_{1}^{\mathrm{E}}(\operatorname{mix})=\mathrm{V}_{1}(\operatorname{mix})-\mathrm{V}_{1}^{*}(\ell)$ Hence, $\mathrm{V}_{1}^{\mathrm{E}}(\operatorname{mix})=\mathrm{R} \, \mathrm{T} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}$ Then, $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}(\mathrm{mix})=\mathrm{R} \, \mathrm{T} \,\left\{\mathrm{x}_{1} \,\left[\partial \ln \left(\mathrm{f}_{1}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}+\mathrm{x}_{2} \,\left[\partial \ln \left(\mathrm{f}_{2}\right) / \partial \mathrm{p}\right]_{\mathrm{T}}\right\}$ Footnote [1] R. Battino, Chem.Rev.,1971,71,5. [2] 1. $\mathrm{C}_{\mathrm{m}}\mathrm{H}_{2\mathrm{m}+2}$ as a solute in $\mathrm{C}_{\mathrm{n}}\mathrm{H}_{2\mathrm{n}+2}$ solvent; H. Itsuki, S. Terasowa, K. Shinora and H. Ikezawa, J Chem. Thermodyn.,19897,19,555. 2. 60[Fullerene} in apolar solvents; P. Ruella, A. Farina-Cuendet and U.W. Kesselring, J. Chem. Soc. Chem. Commun.,1995,1161.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.22%3A_Volume/1.22.8%3A_Volume%3A_Salt_Solutions%3A_Born-Drude-Nernst_Equation.txt
There can be little doubt that water($\ell$) has been all-pervasive in the development of thermodynamics and, of course, solution chemistry. Chemical laboratories throughout the whole world have this liquid ‘on tap’. Speculation about intelligent life forms (cf. are we alone?) in the rest of the universe [1] often start with comments concerning the presence of water($\ell$) on distant planets. Footnotes [1] Bill Bryson, A Short History of Nearly Everything, Doubleday, New York, 2003, chapter 16. 1.23.2: Water: Molar Volume The molar volume $\mathrm{V}_{1}^{*}(\ell)$ and density $\rho_{1}^{*}(\ell)$ of water($\ell$) are intensive properties. The $\mathrm{p}-\mathrm{V}-\mathrm{T}$ properties of water are perhaps the most extensively studied. Two properties are almost universally known; 1. the molar volume of water at ambient pressure and $273.15 \mathrm{~K}$ is less than that of ice, and 2. the molar volume of water at $273.15 \mathrm{~K}$ decreases on heating to reach a TMD before increasing. At ambient pressure the temperature of maximum density for water, $\mathrm{TMD} = 3.98 \text { Celsius}$; for $\mathrm{D}_{2}\mathrm{O}$, $\mathrm{TMD} = 11.44 \text { Celsius}$. The $\mathrm{TMD}$ for $\mathrm{SiO}_{2}$ is around 15270 Celsius, the dependence of density on temperature being less marked about the $\mathrm{TMD}$ than that for water [1]. The dependence of $\mathrm{V}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ on temperature and pressure is reported by many laboratories. Most accounts cite the study reported by Kell and Whalley in 1965, later extended in 1978 [2,3]. Kell has examined the results in detail [4]. The $\mathrm{TMD}$ has, of course, attracted considerable attention. Nevertheless the $\mathrm{TMD}$ has no deep significance in the context of understanding the properties of water($\ell$). Other properties of water($\ell$) show extrema at other temperatures; e.g. isothermal compressibility near $300 \mathrm{~K}$ [5-7]. Footnotes [1] C. A. Angell and H. Kanno, Science, 1976,193,1121. [2] G. S. Kell and E. Whalley, Philos. Trans. R. Soc. London, 1965,258,565. [3] G. S. Kell, G. M. McLaurin and E.Whalley, Proc. R. Soc. London, Ser. A,1978,360,389. [4] G. S. Kell, J. Chem. Eng. Data, 1967,12,66; 1975,20,97; 1970,15,119. [5] R.A. Fine and F. J. Millero, J.Chem.Phys.,1975,63,89; 1973,59,5529. [6] D.-P. Wang and F. J. Millero, J.Geophys. Res., 1973,78,7122. [7] F. J. Millero, R. W. Curry and W. Drost-Hansen, J. Chem. Eng. Data, 1969, 14,422. 1.23.3: Water: Hydrogen Ions Chemists are often faced with the situation where on adding salt $\mathrm{MX}$ to water($\ell$) experimental evidence shows that the cation exists as a hydrate $\mathrm{M}\left(\mathrm{H}_{2} \mathrm{O}\right)_{\mathrm{n}}$. For example, adding $\mathrm{CuSO}_{4}(\mathrm{s})$, a white powder, to water produces a blue solution containing $\left[\mathrm{Cu}\left(\mathrm{H}_{2}\mathrm{O}\right)_{4} \right]^{2+}$. If solute molecules bind solvent molecules to produce new solute molecules, one can imagine a limiting situation where, as the depletion of solvent continues, there is little ‘solvent’ as such left in the system. An important example of the problems linked to description concerns hydrogen ions in aqueous solution [1]. Two common descriptions are 1. $\mathrm{H}^{+} (\mathrm{aq})$, and 2. $\mathrm{H}_{3}\mathrm{O}^{+} (\mathrm{aq})$. As a starting point, we assume that the system comprises $\mathrm{n}_{j}$ moles of solute $\mathrm{HX}$ and $\mathrm{n}_{1}$ moles of water A simple description of hydrogen ions is in terms of $\mathrm{H}^{+} (\mathrm{aq})$ although intuitively the idea of protons as ions in aqueous solutions is not attractive. Arguably a more satisfactory description of hydrogen ions in solution is in terms of $\mathrm{H}_{3}\mathrm{O}^{+}$ ions. Description of hydrogen ions in aqueous solution as $\mathrm{H}_{3}\mathrm{O}^{+}$ finds general support. Adam recalls [2] the experiment conducted by Bagster and Cooling [3]. The latter authors observed that the electrical conductivity of a solution of $\mathrm{HCl}$ in $\mathrm{SO}_{2} (\ell)$ is low but increases dramatically when 1 mole of water($\ell$) is added for each mole of $\mathrm{HCl}$. Moreover on electrolysis, hydrogen ions and water are liberated at the cathode; water($\ell$) drips from this electrode. In the chemistry of aqueous solutions, two ions $\mathrm{H}^{+}$ and $\mathrm{OH}^{-}$ command interest. Hydrogen ions are also called [1] hydronium ions when written as $\mathrm{H}_{3}\mathrm{O}^{+}$. The latter ion is a flat pyramid with $\mathrm{d}(\mathrm{O}-\mathrm{H}) = 96.3 \mathrm{~pm}$; the $\mathrm{HOH}$ angle = 110-1120 [4]. A case for writing the formula $\mathrm{H}_{3}\mathrm{O}^{+}$ is based on the existence of isomorphous solids, + − $\mathrm{NH}_{4}^{+}\mathrm{ClO}_{4}^{-} \text { and } \mathrm{H}_{3}\mathrm{O}^{+}\mathrm{ClO}_{4}^{-}$. The mass spectra of $\mathrm{H}^{+}\left(\mathrm{H}_{2} \mathrm{O}\right)_{\mathrm{n}}$ have been observed for $1 \leg \mathrm{n} \leq 8$ [5]. Neutron-scattering and X-ray scattering data show that for $\mathrm{D}_{3}\mathrm{O}^{+}$ ions in solution , $\mathrm{d}(\mathrm{O}-\mathrm{D}) = 101.7 \mathrm{~pm}$ [6]. In an ‘isolated’ $\mathrm{H}_{3}\mathrm{O}^{+}$ ion the $\mathrm{O}-\mathrm{H}$ bond length is $97 \mathrm{~pm}$ and the $\mathrm{HOH}$ angle is 110 - 112. In aqueous solution, $\mathrm{H}_{3}\mathrm{O}^{+}$ ions do not exist as solutes comparable to $\mathrm{Na}^{+}$ ions in $\mathrm{NaCl}(\mathrm{aq})$. Instead a given $\mathrm{H}_{3}\mathrm{O}^{+}$ ion transfers a proton to a neighboring water molecule. The time taken for the transfer is very short, approx. 10-13 seconds granted that the receiving water molecule has the correct orientation. But a given $\mathrm{H}_{3}\mathrm{O}^{+}$ ion has a finite lifetime, sufficient to be characterized by thermodynamic and spectroscopic properties The rate determining step in proton migration involves reorientation of neighboring water molecules, thereby accounting for the increase in molar conductance $\lambda^{\infty}\left(\mathrm{H}^{+};\mathrm{aq})$ with increase in $\mathrm{T}$ and $\mathrm{p}$ [7 - 10]. [In ice, proton transfer is rate determining and so the mobilities of $\mathrm{H}^{+}$ and $\mathrm{OH}^{-}$ ions are higher in ice than in water($\ell$).} The high mobility of protons in aqueous solution involves a series of isomerizations between $\mathrm{H}_{9}\mathrm{O}_{4}^{+}$ and $\mathrm{H}_{5}\mahtrm{O}_{2}^{+}$, the first triggered by hydrogen bond cleavage of a second shell water molecule and the second by the reverse, hydrogen bond formation [11]. An iconoclastic approach, expressed by Hertz and co-workers, argues against the existence of $\mathrm{H}^{+}$ ions as such except in so far as this symbol describes as dynamical property of a solution [12,13]. In their view $\mathrm{H}_{3}\mathrm{O}^{+}$ is ephemeral. Among other interesting ions discussed in this context are $\mathrm{H}_{5}\mathrm{O}_{2}^{+}$ [14]. Comparisons are often drawn between $\mathrm{NH}_{4}^{+}(\mathrm{aq})$ and $\mathrm{H}_{3}\mathrm{O}^{+} (\mathrm{aq})$ ions but they are not really related. For example the electrical mobility of $\mathrm{NH}_{4}^{+}(\mathrm{aq})$ is not exceptional. The chemical potential of $\mathrm{H}^{+} (\mathrm{aq})$ describes the change in Gibbs energy when $\delta \mathrm{n}\left(\mathrm{H}^{+}\right)$ moles are added at constant $\mathrm{n}\left(\mathrm{H}_{2}\mathrm{O}\right), \mathrm{~n}\left(\mathrm{X}^{-}\right)$, $\mathrm{T}$ and $\mathrm{p}$. The chemical potential of $\mathrm{H}^{+} (\mathrm{aq})$ is related to the molality $\mathrm{m}\left(\mathrm{H}^{+}\right) \left[= \mathrm{n}\left(\mathrm{H}^{+}\right) / \mathrm{n}_{1} \, \mathrm{M}_{1}\right]$ and the single activity coefficient $\gamma\left(\mathrm{H}^{+}\right)$. The Gibbs energy of an aqueous solution $\mathrm{G}(\mathrm{aq})$ prepared using $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ moles of acid $\mathrm{H}^{+} \mathrm{~X}^{-}$ is given by the equation (a) (for the solution at defined $\mathrm{T}$ and $\mathrm{p}$, which we assume is close to the standard pressure $\mathrm{p}^{0}$). $\mathrm{G}(\mathrm{aq})=\mathrm{n}_{1} \, \mu_{1}\left(\mathrm{H}_{2} \mathrm{O} ; \mathrm{aq}\right)+\mathrm{n}\left(\mathrm{H}^{+}\right) \, \mu\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mathrm{n}\left(\mathrm{X}^{-}\right) \, \mu\left(\mathrm{X}^{-} ; \mathrm{aq}\right)$ where $\mu\left(\mathrm{H}^{+}\right)=\left[\partial \mathrm{G} / \partial \mathrm{n}\left(\mathrm{H}^{+}\right)\right]_{\mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right), \mathrm{n}\left(\mathrm{X}^{-}\right)}$ and $\mu\left(\mathrm{H}^{+} ; \mathrm{aq}\right)=\mu^{0}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}\left(\mathrm{H}^{+}\right) \, \gamma\left(\mathrm{H}^{+}\right) / \mathrm{m}^{0}\right]$ For the electrolyte $\mathrm{H}^{+} \mathrm{~X}^{-}$ with $ν = 2$, $\mu\left(\mathrm{H}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right)=\mu^{0}\left(\mathrm{H}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \, \gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}^{0}\right]$ and, all $\mathrm{T}$ and $\mathrm{p}$, $\operatorname{limit}\left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \rightarrow 0\right] \gamma_{\pm}=1.0$ In a similar fashion, we define the partial molar volume, enthalpy and isobaric heat capacity for $\mathrm{H}^{+}$ in aqueous solution. $\mathrm{V}\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)=\left[\partial \mathrm{V} / \partial \mathrm{n}\left(\mathrm{H}^{+}\right)\right]_{\mathrm{T}, \mathrm{p}, \mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right), \mathrm{n}\left(\mathrm{X}^{-}\right)}$ $\mathrm{H}\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)=\left[\partial \mathrm{H} / \partial \mathrm{n}\left(\mathrm{H}^{+}\right)\right]_{\mathrm{T}, \mathrm{p}, \mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right), \mathrm{n}\left(\mathrm{X}^{-}\right)}$ For electrolyte $\mathrm{H}^{+} \mathrm{~X}^{-}$, $\mathrm{V}^{\infty}\left(\mathrm{H}^{+} \mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)=\mathrm{V}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)+\mathrm{V}^{\infty}\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)$ $\mathrm{H}^{\infty}\left(\mathrm{H}^{+} \mathrm{X} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)=\mathrm{H}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)+\mathrm{H}^{\infty}\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)$ and $\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{H}^{+} \mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)=\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)+\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)$ The chemical potential of $\mathrm{H}_{3}\mathrm{O}^{+}$ ions in aqueous solution describes the change in Gibbs energy when $\delta \mathrm{n}\left(\mathrm{H}_{3}\mathrm{O}^{+}\right)$ ions are added at constant $\mathrm{n}\left(\mathrm{H}_{2}\mathrm{O}\right), \mathrm{~n}\left(\mathrm{X}^{-}\right)$, $\mathrm{T}$ and $\mathrm{p}$. At defined $\mathrm{T}$ and $\mathrm{p}$, $\mu\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)=\left[\partial \mathrm{G} / \partial \mathrm{n}\left(\mathrm{H}_{3} \mathrm{O}^{+}\right)\right]_{\mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right) ; \mathrm{n}\left(\mathrm{X}^{-}\right)}$ The chemical potential of the electrolyte $\mathrm{H}_{3}\mathrm{O}^{+}\mathrm{X}^{-}$ in aqueous solution is described by the following equation. \begin{aligned} &\mu\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right) \ &=\mu^{0}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right) \, \gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right) / \mathrm{m}^{0}\right] \end{aligned} where $\mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)=\mathrm{n}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right) / \mathrm{n}\left(\mathrm{H}_{2} 0\right) \, \mathrm{M}_{1}$ $\mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right)=\mathrm{n}_{1}-\mathrm{n}\left(\mathrm{H}_{3} 0^{+}\right)$ and where $\operatorname{limit}\left[\mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right) \rightarrow 0\right) \gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)=1.0 \text { at all } \mathrm{T} \text { and } \mathrm{p}$ Thus the chemical potential $\mu\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{~X}^{-} ; \mathrm{aq} ; \mathrm{T} ; \mathrm{p}\right)$ of the electrolyte $\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{~X}^{-}$ is related to the molality $\mathrm{m}\left(\mathrm{H}_{3}\mathrm{O}^{+} \mathrm{~X}^{-}\right)$ and the mean ionic activity coefficient $\gamma_{\pm} \left(\mathrm{H}_{3}\mathrm{O}^{+} \mathrm{~X}^{-}\right)$. We compare two descriptions of the same system. In description I, the system is an aqueous solution containing $\mathrm{H}^{+}$ and $\mathrm{X}^{-}$ ions whereas in description II the system is an aqueous solution containing $\mathrm{H}_{3}\mathrm{O}^{+}$ and $\mathrm{X}^{-}$ ions. Description I; $\mathrm{n}_{j}$ moles of $\mathrm{H}^{+} \mathrm{~X}^{-}$; $\mathrm{G}(\mathrm{aq} ; \mathrm{I})=\mathrm{n}_{1} \, \mu_{1}(\mathrm{aq} ; \mathrm{I})+\mathrm{n}_{\mathrm{j}} \, \mu\left(\mathrm{H}^{+} ; \mathrm{aq} ; \mathrm{I}\right)+\mathrm{n}_{\mathrm{j}} \, \mu\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{I}\right)$ Description II $\mathrm{G}(\mathrm{aq} ; \mathrm{II})=\left(\mathrm{n}_{1}-\mathrm{n}_{\mathrm{j}}\right) \, \mu_{1}(\mathrm{aq} ; \mathrm{II})+\mathrm{n}_{\mathrm{j}} \, \mu\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)+\mathrm{n}_{\mathrm{j}} \, \mu\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{II}\right)$ At equilibrium, 1. $\mathrm{G}(\mathrm{I} ; \mathrm{aq})=\mathrm{G}(\mathrm{II} ; \mathrm{aq})$ 2. $\mu_{1}(\mathrm{aq} ; \mathrm{I})=\mu_{1}(\mathrm{aq} ; \mathrm{II})$ 3. $\mu\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{I}\right)=\mu\left(\mathrm{X}^{-} ; \mathrm{aq} ; \mathrm{II}\right)$ At equilibrium (at defined $\mathrm{T}$ and $\mathrm{p}$) $\mu\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mu_{1}(\mathrm{aq})=\mu\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)$ In effect we shifted a mole of water for each mole of $\mathrm{H}^{+}$ ions from consideration as part of the solvent in description I to part of the solute in description II forming $\mathrm{H}_{3}\mathrm{O}^{+}$ ions. The link between these descriptions is achieved through two formulations of $\mathrm{G}^{\mathrm{eq}}(\mathrm{aq})$ which is identical for both systems (as are $\mathrm{V}^{\mathrm{eq}}, \mathrm{~S}^{\mathrm{eq}} \text { and } \mathrm{H}^{\mathrm{eq}}$). The equality of the total Gibbs function and equilibrium chemical potentials of substances common to both descriptions leads to equation (u) relating $\mu\left(\mathrm{H}^{+} ; \mathrm{aq}\right)$ and $\mu\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)$. We take the analysis a stage further and use the equations relating chemical potential and composition for the two ions $\mathrm{H}^{+}$ and $\mathrm{H}_{3}\mathrm{O}^{+}$ in aqueous solution at defined $\mathrm{T}$ and $\mathrm{p}$. Description I $\mathrm{m}\left(\mathrm{H}^{+}\right)=\mathrm{n}\left(\mathrm{H}^{+}\right) / \mathrm{n}_{1} \, \mathrm{M}_{1} \quad \mathrm{~m}\left(\mathrm{X}^{-} ; \mathrm{I}\right)=\mathrm{n}\left(\mathrm{X}^{-}\right) / \mathrm{n}_{1} \, \mathrm{M}_{1}$ Description II $\mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+}\right)=\mathrm{n}\left(\mathrm{H}_{3} 0^{+}\right) /\left(\mathrm{n}_{1}-\mathrm{n}_{\mathrm{j}}\right) \, \mathrm{M}_{1}$ and $\mathrm{m}\left(\mathrm{X}^{-} ; \mathrm{II}\right)=\mathrm{n}\left(\mathrm{X}^{-}\right) /\left(\mathrm{n}_{1}-\mathrm{n}_{\mathrm{j}}\right) \, \mathrm{M}_{1}$ Also, $\mu^{e q}\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)=\mu^{e q}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mu_{1}^{\mathrm{eq}}(\mathrm{aq})$ Hence, $\begin{array}{r} \mu^{0}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X} ; \mathrm{aq}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}\right) \, \gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right) / \mathrm{m}^{0}\right] \ =\mu^{0}\left(\mathrm{H}^{+} \mathrm{X} ; \mathrm{aq}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \, \gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}\right) / \mathrm{m}^{0}\right] \ \left.\quad+\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})-2 \, \phi(\mathrm{I}) \, \mathrm{R} \, \mathrm{T} \, \mathrm{m}^{*} \mathrm{H}^{+} \mathrm{X}^{\circ}\right) \, \mathrm{M}_{1} \end{array}$ But \begin{aligned} &\operatorname{limit}\left(\mathrm{n}_{\mathrm{j}} \rightarrow 0\right) \mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+}\right)=1.0 \ &\gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)=1 \quad \phi(\mathrm{I})=1.0 \end{aligned} Hence, $\mu^{\prime \prime}\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)=\mu^{\prime \prime}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mu_{1}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ Also, $\mathrm{V}^{\infty}\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)=\mathrm{V}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mathrm{V}_{1}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ And $\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{H}_{3} \mathrm{O}^{+} ; \mathrm{aq}\right)=\mathrm{C}_{\mathrm{p}}^{\infty}\left(\mathrm{H}^{+} ; \mathrm{aq}\right)+\mathrm{C}_{\mathrm{pl}}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ Interestingly the difference in reference chemical potentials of $\mathrm{H}^{+}(\mathrm{aq})$ and $\mathrm{H}_{3}\mathrm{O}^{+}(\mathrm{aq})$ equals the chemical potential of water($\ell$) at the same $\mathrm{T}$ and $\mathrm{p}$. We combine equations (z) and (za) to obtain an equation relating the mean ionic activity coefficients $\gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right)$ and $\gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)$. Thus (at defined $\mathrm{T}$ and $\mathrm{p}$) \begin{aligned} \ln \left[\gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)\right] &=\ln \left[\gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right)\right]+\ln \left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)\right] \ &-\phi(\mathrm{I}) \, \mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \, \mathrm{M}_{1} \ +& {[1 / 2 \, \mathrm{R} \, \mathrm{T}] \,\left[\mu^{0}\left(\mathrm{H}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right)+\mu_{1}^{*}(\ell)-\mu^{0}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-} ; \mathrm{aq}\right)\right] } \end{aligned} Then \begin{aligned} \ln \left[\gamma_{\pm}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)\right] &=\ln \left[\gamma_{\pm}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right)\right]+\ln \left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)\right] \ &-\phi(\mathrm{I}) \, \mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \, \mathrm{M}_{1} \end{aligned} Also from equations (v) and (w), $\ln \left[\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}^{-}\right)=\ln \left[1-\mathrm{M}_{1} \, \mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right)\right]\right.$ Clearly in dilute aqueous solutions where $\left\{\mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{m}\left(\mathrm{H}_{3} \mathrm{O}^{+} \mathrm{X}-\right)\right\}$ is approximately unity and $\phi(\mathrm{I}) \, \mathrm{m}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) \, \mathrm{M}_{1}$ is negligibly small, the two mean ionic activity coefficients are equal but this approximation becomes less acceptable with increase in the ratio $\mathrm{n}\left(\mathrm{H}^{+} \mathrm{X}^{-}\right) / \mathrm{n}\left(\mathrm{H}_{2} \mathrm{O}\right)$. Footnotes [1] H. L. Clever, J. Chem.Educ.,1963,40,637. [2] Neil Kensington Adam, Physical Chemistry, Oxford, 1956, pp. 376. [3] L.S.Bagster and G.Cooling, J. Chem. Soc., 1920,117,693. [4] R. Triolo and A. H. Narten, J. Chem.Phys.,1975, 63, 3264. [5] A. J. C. Cunningham, J. D. Payzant and P. Kebarle, J. Am. Chem. Soc.,1972,94,7627. [6] P. A. Kollman and C. F. Bender, Chem. Phys. Lett., 1973,21,271. [7] M. Eigen and L. De Maeyer, Proc. R. Soc. London,Ser. A, 1958,247,505. [8] B. E. Conway, J. O’M. Bockris and H. Linton, J. Chem.Phys.,1956, 24 ,834. [9] G. J. Hills, P. J. Ovenden and D. R. Whitehouse. Discuss. Faraday Soc.. 1965,39, 207. [10] E.U. Franck, D. Hartmann and F. Hensel, Discuss. Faraday Soc., 1965,39, 200. [11] N. Agmon, Chem. Phys. Lett., 1995, 244,456. [12] H. G. Hertz, Chemica Scripta,1987,27,479. [13] H. G. Hertz, B. M. Braun, K. J. Muller and R. Maurer, J. Chem.Educ.,1987,64,777. [14] J.-O. Lundgren and I. Olovsson, Acta Cryst., 1967,23,966,971.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.23%3A_Water/1.23.1%3A_Water.txt
Parallel with interest in the $\mathrm{p}-\mathrm{V}-\mathrm{T}$ properties of water($\ell$), enormous interest has been shown in the relative permittivity of water($\ell$) as a function of $\mathrm{T}$ and $\mathrm{p}$. Owen and Brinkley [1] showed that a type of Tait equation could be used to express the $\mathrm{T}-\mathrm{p}$ dependence of the relative permittivity of many liquids, including water(l). In 1980, Uematsu and Franck reviewed published information concerning the permittivity of water($\ell$) and steam published over a period of 100 years [2,3]. A careful determination was reported by Deul in 1984 [4]. At $298.15 \mathrm{~K}$ and ambient pressure, the relative permittivities of water($\ell$) and $\mathrm{D}_{2}\mathrm{O}(\ell$) are 78.39 and 78.06 respectively [5]. A report by Owen and co-workers in 1961 stated a value of 78.358 for water($\ell$) at $298.15 \mathrm{~K}$. Bradley and Pitzer [7] survey published relative permittivities of water($\ell$) over an extensive temperature range. Footnotes [1] B. B. Owen and S. R. Brinkley, Phys. Rev.,1943,64,32. [2] M. Uematsu and E. U. Franck, J. Phys. Chem. Ref. Data, 1980, 9,1291. [3] K. Heger, M. Uematsu and E. U. Franck, Ber. Bunsenges, Phys.Chem.,1980,84,758. [4] R. Deul, PhD Thesis, Faculty of Chemistry, University of Karlsruhe, 1984. [5] G. A. Vidulich, D. F. Evans and R. L. Kay, J.Phys.Chem.,1967,71,656. [6] B. B. Owen, R. C. Miller, C. E. Milner and H. L. Cogan, J. Phys. Chem., 1961, 65, 2065. [7] D. J. Bradley and K. S. Pitzer, J Phys.Chem.,1979,83,1599. 1.23.5: Water: Self-Dissociation A significant contribution to the chemistry of aqueous solutions stems from the self dissociation of water($\ell$)(see also $\mathrm{D}_{2}\mathrm{O}(\ell$) [1]). At $298.15 \mathrm{~K}$ and ambient pressure, $\mathrm{pK}_{\mathrm{a}} equals 14.004 [2]. Olofsson and Hepler [3] recommended a ‘best value’ for the standard enthalpy of self dissociation at ambient pressure and \(298.15 \mathrm{~K}$ equal to $55.81 \mathrm{~kJ mol}^{-1}$ [4,5]. Hepler and colleagues recommend a best value for$\Delta_{\mathrm{d}} C_{\mathrm{p}}^{0}$ equal to $- 215 \mathrm{~J K}^{-1} \mathrm{~mol}^{-1}$ [5]; see also [6-8] together with [1] for details characterizing $\mathrm{D}_{2}\mathrm{O}$. The standard volume of self-dissociation for water($\ell$) at $298 \mathrm{~K}$ is negative, approx. $- 20 \mathrm{cm}^{3} \mathrm{~mol}^{-1}$, decreasing with increase in temperature [8,9]. An extensive literature describes the thermodynamics of self-dissociation of water in binary liquid mixtures [11]; see also [12] for $\mathrm{D}_{2}\mathrm{O}$. Footnotes [1] A. K. Covington, R. A. Robinson and R. G. Bates, J. Phys. Chem., 1966, 70, 3820. [2] A. K. Covington, M. I. A. Ferra and R. A. Robinson, J. Chem. Soc. Faraday Trans.,1,1977,73,1721. [3] G. Olofsson and L. G. Hepler, J. Solution Chem.,1975,4,127. [4] G. Olofsson and I. Olofsson, J. Chem. Thermodyn.,1973, 5, 533; 1977, 9, 65. [5] O. Enea, P. P. Singh, E. M. Woolley, K. G. McCurdy and L. G. Hepler, J. Chem. Thermodyn., 1977, 9,731. [6] G. C. Allred and E. M. Woolley, J. Chem. Thermodyn..,1981, 13, 147. [7] P. P. Singh, K. G. McCurdy, E. M. Woolley and L. G. Hepler, J. Solution Chem., 1977,6,327. [8] J. J. Christensen, G. L .Kimball, H. D. Johnston and R. M. Izatt, Thermochim. Acta,1972,4,141. [9] 1. D. A. Lown, H. R. Thirsk and Lord Wynne-Jones, Trans. Faraday Soc., 1968, 64,2073. 2. F. J. Millero, E. V. Hoff and L. Kahn, J. Solution Chem.,1972,1,309. [10] J. P. Hershey, R. Damasceno and F. J. Millero, J. Solution Chem.,1984,13,825. [11] 1. DMSO + water($\ell$); A. K. Das and K. K. Kundu, J. Chem. Soc. Faraday Trans.1,1973, 69,730. 2. $\left(\mathrm{CH}_{3}\right)_{3}\mathrm{COH} + \text { water}(\ell)$; H. Gillet., L. Avedikian and J.-P. Morel, Can. J.Chem.,1975, 53,455. 3. MeOH + water($\ell$); C.H.Rochester, J. Chem. Soc., Dalton Trans.,1972,5. 4. Alcohol + water($\ell$); E. M. Woolley, D. G. Hurkot and L. G. Hepler, J. Phys. Chem., 1970,74,3908. 5. $\mathrm{CH}_{3}\mathrm{CN} + \text { water}(\ell)$; U. Mandal, S. Bhattacharya and K. K. Kundu, Indian J. Chem. Sect. A,1985,24,191. 6. E. M. Woolley, and L.G.Hepler, Anal. Chem.,1972,44,1520. 7. Urea+water; A. K. Das and K. K. Kundu, J. Phys.Chem.,1975,79,2604. [12] R. E.George and E. M. Woolley, J. Solution Chem.,1972,1,279. 1.23.6: Water: (Shear) Viscosity When the properties of water($\ell$) are reviewed, general practice is to identify the importance of intermolecular hydrogen bonding as a molecular cohesive force. In these terms it is perhaps a surprise to discover that water($\ell$) pours quite smoothly and freely, certainly more freely than, say, glycerol($\ell$). Indeed the viscosity of water($\ell$) is quite modest; $0.8903 \mathrm{~cP}$ at $298.15 \mathrm{~K}$ [1,2]. Nevertheless there are indications of complexity because below $230 \mathrm{~K}$ the viscosity of water($\ell$) decreases with increase in pressure before increasing. Good agreement exists between the results reported by many laboratories for viscosities of water($\ell$) at low pressures but disagreement at high pressure [3]. Footnotes [1] $\mathrm{P}=10^{-1} \mathrm{~Pa} \mathrm{~s}=10^{-1} \mathrm{~J} \mathrm{~m}^{-3} \mathrm{~s}$ [2] G. S. Kell, in Water; A Comprehensive Treatise, ed. F. Franks, McGraw-Hill, New York, 1973, volume 1, chapter 10. [3] S. D. Hamann, Physics and Chemistry of the Earth, 1981,13,89.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.23%3A_Water/1.23.4%3A_Water%3A_Relative_Permittivity.txt
For a closed system undergoing a change in thermodynamic properties under the adiabatic constraint, heat does not cross between system and surroundings. The system is thermally insulated (i.e. adiabatically enclosed) from the surroundings. The First Law for closed systems has the following form. $\Delta \mathrm{U}=\mathrm{q}+\mathrm{w}$ For changes under the adiabatic constraint, $\mathrm{q}$ is zero [1]. Then for adiabatic changes, $\Delta \mathrm{U}=\mathrm{w} .$ Footnote [1] ‘Adiabatic’ means impassable, from the Greek: ‘a = not’ + ‘dia = through’ + ‘bathos = deep’. 1.14.02: Adsorption- Langmuir Adsorption Isotherm- One Adsorbate The theoretical basis of Adsorption Isotherms is customarily described in terms of a balance of rates of adsorption and desorption. [1] Three important assumptions are made. 1. The adsorbate covers the surface up to complete coverage as a monolayer on the substrate (adsorbent). 2. There are no adsorbate-adsorbate interactions on the surface of the host substrate. 3. On the substrate all binding sites are equivalent. IUPAC recommends the use of variables expressed in terms of unit mass of adsorbent; $\mathrm{n}_{j}^{\mathrm{ad}} / \mathrm{w}$ is the specific reduced adsorption of chemical substance $j$ where $\mathrm{n}_{j}^{\mathrm{ad}}$ is the amount of adsorbate bound to $j$ mass $\mathrm{w}$ of adsorbent. In other words equations describing the process are transformed into intensive variables [2,3]. In many interesting cases, small molecules (guest, adsorbate) bind to larger polymeric molecules (host, adsorbent) which provide a surface on which the smaller (guest) molecules are adsorbed. Adsorption data for such systems often follow a Langmuir pattern. Here we use a thermodynamic approach in descriptions of a thermodynamic equilibrium. Most texts describing Langmuir adsorption use a `kinetic model’, as indeed did Langmuir is his description of adsorption [1]. We consider the case where water is the solvent and substance $j$ is the adsorbate. In the absence of adsorbate $j$, the surface of the adsorbent is covered with water. When adsorbate $j$ is added to the system, the adsorption is described by the following equation. $\mathrm{j}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}\right)+\mathrm{H}_{2} \mathrm{O}\left(\mathrm{ad} ; \mathrm{x}_{1}^{\mathrm{ad}}\right) \rightarrow \mathrm{j}\left(\mathrm{ad} ; \mathrm{x}_{\mathrm{j}}^{\mathrm{ad}}\right)+\mathrm{H}_{2} \mathrm{O}\left(\mathrm{aq} ; \mathrm{x}_{1}\right)$ The latter equation describes a physical process rather than a ‘chemical’ reaction but the symbolism is common. Thus $x_{j}$ is the mole fraction of solute $j$ in the aqueous phase; $\mathrm{x}_{\mathrm{1}}^{\mathrm{ad}}$ is the mole fraction of water in a thin solution adjacent to the surface of the adsorbate; $x_{1}$ is the mole fraction of water in the aqueous solution.; $\mathrm{x}_{\mathrm{j}}^{\mathrm{ad}}$ is the mole fraction of the substance $j$ in the adsorbed layer. The process represented by equation (a) involves displacement of water from the ‘thin’ solution next to the adsorbent into the bulk solution. The reverse process describes the displacement of substance $j$ from this layer by water($\lambda$). At equilibrium, the two driving forces are balanced. We use a simple on/off model for the adsorption in a closed aqueous system at fixed temperature $\mathrm{T}$ and fixed pressure $p$ ($\cong p^{0}$). $\begin{array}{lc} \text { Then, } \quad \mathrm{j}(\mathrm{aq}) \Leftarrow \,s \mathrm{j}(\mathrm{ad}) \ \text { At } \mathrm{t}=0, \quad \mathrm{n}_{\mathrm{j}}^{0} \quad 0 \quad \mathrm{~mol} \ \text { At } t=\infty \quad n_{j}^{0}-\xi \quad \xi \quad \mathrm{mol} \end{array}$ The latter condition refers to the equilibrium state; $\xi$ is the extent of binding of guest solute $j$ to the host adsorbate. To describe this equilibrium we need equations for the chemical potentials of $j(\mathrm{aq})$ and $j(\mathrm{ad}$). The fraction of adsorbent surface covered by chemical substance $j$ is defined as $\theta$. If there are $\mathrm{N}$ sites on the adsorbate for adsorption, the amount of sites occupied equals $\mathrm{N} \, \theta$ and the amount of vacant sites equals $[\mathrm{N} \,(1-\theta)]$. The aim of the analysis is a plot showing the degree of occupancy of the surface of the adsorbent $\theta$ as a function of the equilibrium composition of the system. If the experiment involves calorimetry, we require equations which describe the heat $\mathrm{q}$ associated with injection of a small aliquot of a solution containing the adsorbate into a solution containing the adsorbent. The chemical potential of adsorbed chemical substance $j$, $\mu_{j}(\mathrm{ad})$ is related to $\theta$ using a general equation. $\mu_{\mathrm{j}}(\mathrm{ad})=\mu_{\mathrm{j}}^{0}(\mathrm{ad})+\mathrm{R} \, \mathrm{T} \, \ln [\mathrm{f}(\theta)]$ In order to make progress we need an explicit equation for $\mathrm{f}(\theta)$; $\mu_{\mathrm{j}}^{0}(\mathrm{ad})$ is the chemical potential of an ideal adsorbate on an ideal adsorbent where $\theta =1/2$ at the same $\mathrm{T}$ and $p$ [4]. At equilibrium the chemical potential of solute $j$ in solution equals the chemical potential of the adsorbate. We express the composition of the solution in terms of the concentration of chemical substance $j$, $\mathrm{c}_{j}$. $\text { At equilibrium. } \quad \mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}(\mathrm{ad})$ $\text { For solute, } \mathrm{j}(\mathrm{aq}), \quad \mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{c}_{\mathrm{j}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{j}}(\mathrm{aq}) / \mathrm{c}_{\mathrm{r}}\right]$ By definition, $\mathrm{c}_{\mathrm{r}} = 1 \mathrm{~mol dm}^{−3}$; $\gamma_{j}$ is the solute activity coefficient where, $\lim \operatorname{it}\left(c_{j} \rightarrow 0\right) y_{j}(a q)=1 \quad \text { at all } T \text { and } p$ $\mu_{\mathrm{j}}^{0}(\mathrm{aq})$ is the chemical potential of solute $j$ in solution having unit concentration $\mathrm{c}_{j}$, the thermodynamic properties of solute $j$ being ideal. From equations (c) and (e) the equilibrium condition (d) requires, by definition, that \begin{aligned} &\Delta_{\mathrm{ad}} \mathrm{G}^{0}=\mu_{\mathrm{j}}^{0}(\mathrm{ad})-\mu_{\mathrm{j}}^{0}(\mathrm{aq}) \ &=-\mathrm{R} \, \mathrm{T} \, \ln [\mathrm{f}(\theta)]+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{c}_{\mathrm{j}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{j}}(\mathrm{aq}) / \mathrm{c}_{\mathrm{r}}\right] \ &=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}_{\mathrm{ad}}\right) \end{aligned The dimensionless property $\mathrm{K}_{\mathrm{ad}$ is the equilibrium adsorption constant which depends on both $\mathrm{T}$ and $p$ [5]. $\text { Then, } \mathrm{K}_{\mathrm{ad}}=\mathrm{f}(\theta) \, \mathrm{c}_{\mathrm{r}} / \mathrm{c}_{\mathrm{j}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{j}}(\mathrm{aq})$ We envisage a solution volume $\mathrm{V}$ prepared using $n_{j}{}^{0}$ moles of chemical substance $j$. [The assumption is usually made that the volume of the system is the volume of the solution.] The equilibrium concentration of substance $j$, $\mathrm{c}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})$ is given by equation (i) where $\xi^{\mathrm{eq}$ is the equilibrium extent of adsorption $\mathrm{c}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})=\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right) / \mathrm{V}$ $\mathrm{K}_{\mathrm{ad}}=\left[\mathrm{f}(\theta) /\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi\right)\right] \,\left[\mathrm{V} \, \mathrm{c}_{\mathrm{r}} / \mathrm{y}_{\mathrm{j}}(\mathrm{aq})\right]$ Equation (j) describes the composition of the system. Granted therefore the applicability of equation (i) we anticipate specific applications of this equation will, at minimum, yield two pieces of information. (i) Dependence of $\theta$ on total concentration of $j$ in the system, $\mathrm{n}_{\mathrm{j}}^{0} / \mathrm{V}$. (ii) Dependence of percentage of chemical substance $j$ bound on total concentration of $j$ in the system, $n_{j}^{0} / V$. If the system is prepared using $n_{1}$ moles of water, the enthalpy of the system is given by equation (k). $\mathrm{H}(\text { system })=\mathrm{n}_{1} \, \mathrm{H}_{1}(\mathrm{aq})+\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right) \, \mathrm{H}_{\mathrm{j}}(\mathrm{aq})+\xi^{\mathrm{eq}} \, \mathrm{H}_{\mathrm{j}}(\mathrm{ad})$ The dependence of enthalpy on the extent of adsorption is given by equation ($\lambda$). \begin{aligned} &{[\partial \mathrm{H}(\text { system }) / \partial \xi \xi]=} \ &\begin{aligned} \mathrm{n}_{1} \,\left[\partial \mathrm{H}_{1}(\mathrm{aq}) / \partial \xi\right]+\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right) \,\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{aq}) / \partial \xi\right]-\mathrm{H}_{\mathrm{j}}(\mathrm{aq}) \ &+\mathrm{H}_{\mathrm{j}}(\mathrm{ad})+\xi^{\mathrm{eq}} \,\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{ad}) / \partial \xi\right] \end{aligned} \end{aligned} As a working hypothesis we assume that the properties of substance $j$ are ideal both in solution and as adsorbate. In other words there are no solute $j \rightleftarrows$ solute $j$ interactions in the aqueous solution, no bound $j \rightleftarrows$ bound $j$ interactions between adsorbed molecules and no solute $j \rightleftarrows$ bound $j$ interactions. In summary the adsorbed $j$ molecules form monolayers and the adsorbed molecules are non-interacting with other adsorbed $j$ molecules and with $j$ molecules in solution [6]. 1. $\mathrm{y}_{\mathrm{j}}(\mathrm{aq})=1.0$ 2. $\left[\partial \mathrm{H}_{1}(\mathrm{aq}) / \partial \xi\right]=0.0$ 3. $\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{aq}) / \partial \xi\right]=0.0$ 4. $\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{ad}) / \partial \xi\right]=0.0$ $\text { Therefore, }[\partial \mathrm{H}(\text { system }) / \partial \xi \xi]=\mathrm{H}_{\mathrm{j}}(\mathrm{ad})^{0}-\mathrm{H}_{\mathrm{j}}(\mathrm{aq})^{\infty}=\Delta_{\mathrm{ad}} \mathrm{H}^{0}$ Here $\mathrm{H}_{\mathrm{j}}(\mathrm{aq})^{\infty}$ is the limiting partial molar enthalpy of solute $j$ meaning that in effect the solute molecules in solution are infinitely apart. $\mathrm{H}_{\mathrm{j}}(\mathrm{ad})^{0}$ is the standard partial molar enthalpy of the adsorbate, implying that on the surface of the host adsorbent the adsorbate molecules are infinitely far apart; i.e. there are no adsorbate-adsorbate interactions. $\Delta_{\mathrm{ad}} \mathrm{H}^{0}$ is the standard molar enthalpy for the adsorption of substance $j$ from aqueous solution on to the adsorbate. In the next stage we require an equation for $\mathrm{f}(\theta)$ in order to obtain an explicit equation for the chemical potential of adsorbed substance $j$. We use the Langmuir model [4]; $f(\theta)=\theta /(1-\theta)$ $\theta$ is the fraction of the surface covered by the adsorbate at equilibrium, the fraction $(1- \theta)$ being left bare; note that $\theta$ is an intensive variable. $\text { Then } \mu_{\mathrm{j}}(\mathrm{ad})=\mu_{\mathrm{j}}^{0}(\mathrm{ad})+\mathrm{R} \, \mathrm{T} \, \ln [\theta /(1-\theta)]$ The standard state for adsorbed j molecules corresponds to the situation where $\theta=1 / 2$; i.e $[\theta /(1-\theta)]$ is unity. It is interesting to put some numbers to these variables. We set $\chi=\ln [\theta /(1-\theta)]$. Then for $\theta = 0.1$, $\mathrm{R} \, \mathrm{T} \, \chi=-2.197 \, \mathrm{R} \, \mathrm{T}$. Hence when the surface is less than 50% covered the chemical potential of the adsorbate is less than in the adsorbed standard state. For $\theta =1/2$, $\mathrm{R} \, \mathrm{T} \, \chi=0$; at this stage the chemical potential of substance $j$ in the adsorbed state equals that in the reference (standard) state. For $\theta =0.9$, $\mathrm{R} \, \mathrm{T} \, \chi=2.197 \, \mathrm{R} \, \mathrm{T}$. As the surface occupancy passes 0.5, the chemical potential of the adsorbate increases above that in the reference state. According to equation (i) for a system where solute $–j$ and adsorbate $–j$ have ideal thermodynamic properties, $\mathrm{K}_{\mathrm{ad}}=[\theta /(1-\theta)] \,\left[\mathrm{V} \, \mathrm{c}_{\mathrm{r}} /\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi\right)\right]$ With increase in $\xi$ so $\theta$ increases. Unfortunately we do not know how $\theta$ and $\xi$ are related. One procedure assumes that $\theta$ is proportional to $\xi_{\mathrm{eq}$, the constant of proportionality being $\pi$ [4]. $\mathrm{K}_{\mathrm{ad}}=\left[\pi \, \xi^{\mathrm{eq}} /\left(1-\pi \, \xi^{\mathrm{eq}}\right)\right] \,\left[\mathrm{V} \, \mathrm{c}_{\mathrm{r}} /\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right)\right]$ $\text { Hence, }\left(\mathrm{K}_{\mathrm{ad}} / \mathrm{V}\right) \,\left(1 / \mathrm{c}_{\mathrm{r}}\right) \,\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right)=\pi \, \xi^{\mathrm{eq}} /\left(1-\pi \, \xi^{\mathrm{eq}}\right)$ By definition,$\beta=\left[\mathrm{K}_{\mathrm{ad}} / \mathrm{V} \, \mathrm{c}_{\mathrm{r}}\right]^{-1}$ $\text { Then, } \pi \,\left(\xi^{\mathrm{eq}}\right)^{2}-\xi^{\mathrm{eq}} \,\left[1+\pi \, \mathrm{n}_{\mathrm{j}}^{0}+\beta \, \pi\right]+\mathrm{n}_{\mathrm{j}}^{0}=0$ Equation (t) is a quadratic in the required variable $\xi^{\mathrm{eq}$. With increase in amount of solute $j$ in the system so the extent of adsorption increases. Recognising that $\beta$ is taken as a constant, $\left[\mathrm{d} \xi / \mathrm{dn}_{\mathrm{j}}^{0}\right]=\left[1-\pi \, \xi^{\mathrm{eq}}\right] /\left[\beta \, \pi+\pi \,\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right)+\left(1-\pi \, \xi^{\mathrm{eq}}\right)\right]$ Equation (u) leads to an estimate of the dependence of $\theta$ on the concentration of chemical substance j in the system. This analysis assumes no interactions between adsorbed molecules on the adsorbent. This assumption is probably too drastic. One approach which takes account of such interactions is the Freundlich Adsorption Isotherm [7]. The chemical potentials of the adsorbed chemical substance $j$ is related to $\theta$ using equation (v). $\mu_{\mathrm{j}}(\mathrm{ad})=\mu_{\mathrm{j}}^{0}(\mathrm{ad})+\mathrm{R} \, \mathrm{T} \, \ln [\theta /(1-\theta)]-\mathrm{R} \, \mathrm{T} \, \mathrm{a} \, \theta$ The parameter ‘a’ is an adsorbate-adsorbate interaction parameter. For systems where $\mathrm{a} < 0$ (where $\theta$ is always positive) repulsion between adsorbed molecules raises the chemical potential above that described by the Langmuir model and disfavours adsorption. For system where $\mathrm{a} > 0$, attraction between adsorbed $j$ molecules lowers their chemical potentials below the chemical potentials described by the Langmuir model; i.e. adsorption is enhanced above that required by the ideal model. But as for the Langmuir model, $\theta$ is dependent on the extent of adsorption and thus almost certainly on the geometric properties of guest and host. If the thermodynamic properties of the solute $j$ in solution are assumed to be ideal, equation (r ) is rewritten as follows. $\mathrm{K}_{\mathrm{ad}}=\left[\pi \, \xi^{\mathrm{eq}} /\left(1-\pi \, \xi^{\mathrm{eq}}\right)\right] \, \exp \left(-\mathrm{a} \, \pi \, \xi^{\mathrm{eq}}\right) \,\left[\mathrm{V} \, \mathrm{c}_{\mathrm{r}} /\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi^{\mathrm{eq}}\right)\right]$ Another approach writes $\mathrm{f}(\theta)$ using a general equation having the following form. $\mathrm{f}(\theta)=\left[\frac{\theta}{1-\theta}\right] \,\left[\frac{1}{\mathrm{n}^{\mathrm{n}}}\right] \,\left[\frac{\theta+\mathrm{n} \,(1-\theta)}{1-\theta}\right]^{\mathrm{n}-1}$ For the Langmuir adsorption isotherm, $n$ is unity. Otherwise $n$ is a positive integer; $\mathrm{n}=1,2,3, \ldots$.Thus when $n$ is set at 2 [5], $f(\theta)=\left[\frac{\theta}{1-\theta}\right] \,\left[\frac{1}{4}\right] \,\left[\frac{2-\theta}{1-\theta}\right]$ We comment on terminology. In enzyme chemistry, the term substrate refers to (in relative terms) small molecules which bind to an enzyme, a macromolecule. However in the subject describing adsorption of molecules on a surface, the molecules which are adsorbed are called the adsorbate. The macromolecular host is the adsorbent or substrate. In other words the meaning of the term ‘substrate’ differs in the two subjects. In general terms each enzyme has a unique site at which the adsorbate (substrate) is bound. However the term adsorbent implies that the ‘surface’ has a number of sites at which the adsorbate is adsorbed. The extent to which the sites are specific to a particular adsorbent is often less marked than in the case of enzymes. Nevertheless there is a common theme in which a substance $j$ ‘free’ in solution loses translational freedom by coming in contact with a larger molecular system, being then held by that system. Footnotes [1] I. Langmuir, J.Am.Chem.Soc.,1918,40,1361. [2] D. H. Everett, Pure Appl. Chem.,1986, 58, 967. [3] J. O’M. Bockris and S. U. M. Khan, Surface Electrochemistry, Plenum Press, New York, 1993. [4] B. E. Conway, H. Angerstein-Kozlowska and H. P. Dhar, Electrochim.Acta, 1975, 19,189. [5] M. J. Blandamer, B. Briggs, P. M. Cullis, K. D. Irlam, J. B. F. N. Engberts and J. Kevelam, J. Chem. Soc. Faraday Trans.,1998, 94,259. [6] K. S. Pitzer, Thermodynamics, McGraw-Hill, New York, 3rd edn., 1995,chapter 23. [7] H. Freundlich, Z.Phys.Chem.,1906,57,384.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.01%3A_Adiabatic.txt
We consider the case where in addition to the adsorbent there are two adsorbates, chemical substances $i$ and $j$ in aqueous solution [1]. In the simplest case the thermodynamic properties of the system are ideal. In other words for both solutes and adsorbates there are no $i - i$, $j - j$ and $i - j$ interactions. Analysis of the adsorption using the Langmuir adsorption isotherm leads to two terms describing the surface coverage, $\theta_{i}$ and $\theta_{j}$ plus the total surface coverage, $\theta_{i} + \theta_{j}$. The chemical potentials of the solutes in solution are described in terms of their concentrations (assuming pressure $p$ is close to the standard pressure); \mathrm{c}_{\mathrm{r}}=1 \mathrm{~mol dm}^{-3}. $\mu_{j}(a q)=\mu_{j}^{0}(a q)+R \, T \, \ln \left(c_{j} / c_{r}\right)$ $\mu_{i}(a q)=\mu_{i}^{0}(a q)+R \, T \, \ln \left(c_{i} / c_{r}\right)$ The upper limit of the total surface occupancy is unity and so we expect as $\left(\theta_{i} + \theta_{j} \right)$ approaches unity the sum of the chemical potentials $\mu_{j}(\mathrm{ad})$ and $\mu_{i}(\mathrm{ad})$ approaches $+\infty$, thereby opposing any tendency for further solute to be adsorbed. If we assert that there are no substrate-substrate interactions on the surface, the chemical potential of adsorbate $j$ can be formulated as follows. $\mu_{j}(\mathrm{ad})=\mu_{j}^{0}(\mathrm{ad})+\mathrm{R} \, \mathrm{T} \, \ln \left[\theta_{\mathrm{j}} /\left(1-\theta_{\mathrm{i}}-\theta_{\mathrm{j}}\right)\right]$ The denominator $\left(1-\theta_{i}-\theta_{j}\right)$ takes account of the fact that adsorbate $i$ also occupies the surface. Thus as $\left(\theta_{i} + \theta_{j} \right)$ approaches unity there are no more sites on the surface for adsorbate $j$ (and adsorbate $i$ ) to occupy. $\text { Thus } \operatorname{limit}\left[\left(1-\theta_{\mathrm{i}}-\theta_{\mathrm{j}}\right) \rightarrow 0\right] \mu_{\mathrm{j}}(\mathrm{ad})=+\infty$ The equilibrium between chemical substance $j$ as solute and adsorbate is described as follows. $\text { Solute }-\mathrm{j}+\text { Polymer Surface } \Leftrightarrow \text { Adsorbate } \mathrm{j}$ $\text { Prepared } \mathrm{n}_{\mathrm{j}}^{0} \quad \quad \quad 0 \mathrm{~mol}$ $\text { Equilib. } \quad \mathrm{n}_{\mathrm{j}}^{0}-\xi_{\mathrm{j}} \quad \xi_{\mathrm{j}}+\xi_{\mathrm{i}} \quad \xi_{\mathrm{j}} \mathrm{mol}$ We express the fraction of surface coverage as proportional to the extent of adsorption via a proportionality constant $\pi$ which is a function of the sizes of the solutes and geometric parameters describing the surface. $\text { Then, } \quad \theta_{i}=\pi_{i} \, \xi_{i} \quad \text { and } \theta_{j}=\pi_{j} \, \xi_{j}$ $\text { At equilibrium } \mu_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})=\mu_{\mathrm{j}}^{\mathrm{eq}} \text { (ad) and } \mu_{\mathrm{i}}^{\mathrm{eq}}(\mathrm{aq})=\mu_{\mathrm{i}}^{\mathrm{eq}}(\mathrm{ad})$ Hence for a system having volume $\mathrm{V}$, \begin{aligned} \mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi_{\mathrm{j}}\right) / \mathrm{V} \, \mathrm{c}_{\mathrm{r}}\right] \ &=\mu_{\mathrm{j}}^{0}(\mathrm{ad})+\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\pi_{\mathrm{j}} \, \xi_{\mathrm{j}}}{1-\pi_{\mathrm{j}} \, \xi_{\mathrm{j}}-\pi_{\mathrm{i}} \, \xi_{\mathrm{i}}}\right] \end{aligned} $\text { By definition, } \Delta_{\mathrm{ad}} \mathrm{G}_{\mathrm{j}}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln \mathrm{K}_{\mathrm{j}}=\mu_{\mathrm{j}}^{0}(\mathrm{ad})-\mu_{\mathrm{j}}^{0}(\mathrm{aq})$ $\text { Hence, } \mathrm{K}_{\mathrm{j}}=\left[\frac{\pi_{\mathrm{j}} \, \xi_{\mathrm{j}}}{1-\pi_{\mathrm{j}} \, \xi_{\mathrm{j}}-\pi_{\mathrm{i}} \, \xi_{\mathrm{i}}}\right] \, \frac{\mathrm{V} \, \mathrm{c}_{\mathrm{r}}}{\left(\mathrm{n}_{\mathrm{j}}^{0}-\xi_{\mathrm{j}}\right)}$ A similar equation is obtained for equilibrium constant $\mathrm{K}_{i}$. Both equations are quadratics in the extent of adsorption Footnotes [1] M. J. Blandamer, B. Briggs, P. M. Cullis, K. D. Irlam, J. B. F. N. Engberts and J. Kevelam, J. Chem. Soc. Faraday Trans.,1998, 94, 259. 1.14.04: Apparent Molar Properties- Solutions- Background A given solution, volume $\mathrm{V}$, is prepared using $n_{1}$ moles of solvent (e.g. water) and $n_{j}$ moles of solute, chemical substance $j$. Thus $V(a q)=n_{1} \, V_{1}(a q)+n_{j} \, V_{j}(a q) \label{a}$ Here $\mathrm{V}_{1}(\mathrm{aq})$ is the partial molar volume of the solvent and $\mathrm{V}_{j}(\mathrm{aq})$ is the partial molar volume of the solute-$j$ [1]. Experiment yields the density of this solution at defined $\mathrm{T}$ and $p$. In order to say something about this solution we would like to comment on the two partial molar volumes, $\mathrm{V}_{1}(\mathrm{aq})$ and $\mathrm{V}_{2}(\mathrm{aq})$. But we have only three known variables; the amounts of solvent and solute and the density. If we change the amount of, say, solute then $\mathrm{V}(\mathrm{aq})$ together with the two partial molar volumes change. So we end up with more unknowns than known variables. Hence it would appear that no progress can be made. All is not lost. Equation (a) is rewritten in terms of the molar volume of the solvent, $\mathrm{V}_{1}^{*}(\lambda)$ which is calculated from the density of the pure solvent and its molar mass. At a given $\mathrm{T}$ and $p$, density $\rho_{1}^{*}(\lambda)=\mathrm{M}_{1} / \mathrm{V}_{1}^{*}(\lambda)$. We replace $\mathrm{V}_{j}(\mathrm{aq})$ in Equation \ref{a} by the apparent molar volume, $\phi\left(\mathrm{V}_{j}\right)$; Equation \ref{b}. $\mathrm{V}(\mathrm{aq})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right) \label{b}$ Now we have only one unknown variable. But we anticipate that the apparent molar volume $\phi \left(\mathrm{V}_{j} \right)$ depends on the composition of the solution, the solute, $\mathrm{T}$ and $p$. These comments concerning partial molar volumes establish a pattern which can be carried over to other partial molar properties. The following apparent molar properties of solutes are important; (i) apparent molar enthalpies $\phi\left(\mathrm{H}_{\mathrm{j}}\right)$, (ii) apparent molar isobaric heat capacities $\phi\left(\mathrm{C}_{p j}\right)$, (iii) apparent molar isothermal compressions $\phi\left(\mathrm{K}_{\mathrm{T}_{\mathrm{j}}}\right)$, and (iv) apparent molar isobaric expansions $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$. Apparent molar (defined) isentropic compressions $\phi\left(K_{\mathrm{Sj}} ; \mathrm{def}\right)$, and apparent molar (defined) isentropic expansions $\phi\left(K_{\mathrm{Sj}} ; \mathrm{def}\right)$ are also quoted but new complexities emerge. Lewis and Randall commented [2] that ‘apparent molal quantities have little thermodynamic utility’, a statement repeated in the second [3] but not in the third edition of this classic monograph.[4] Suffice to say, their utility in the analysis of experimental results has been demonstrated by many authors. Apparent molar properties of solutes $\phi\left(\mathrm{E}_{\mathrm{pj}}\right)$, $\phi\left(K_{\mathrm{Sj}} ; \mathrm{def}\right)$, $\phi\left(\mathrm{K}_{\mathrm{T}_{\mathrm{j}}}\right)$, $\phi\left(K_{\mathrm{Sj}} ; \mathrm{def}\right)$ and $\phi\left(\mathrm{C}_{p j}\right)$ are calculated using in turn the extensive properties of solutions, isobaric expansions $\mathrm{E}_{p}$, isentropic expansions $\mathrm{E}_{\mathrm{S}}$, isothermal compressions $\mathrm{K}_{\mathrm{T}}$, isentropic compressions $\mathrm{K}_{\mathrm{S}}$ and isobaric heat capacities $\mathrm{C}_{p}$. Footnotes [1] Equation (a) is interesting . We do not have to add the phrase ‘at constant temperature and pressure’ [2] G. N. Lewis and M. Randall, Thermodynamics and The Free Energy of Chemical Substances, McGraw-Hill, New York, 1923. [The title on the front cover of the monograph is simply ‘Thermodynamics’.] [3] G. N. Lewis and M. Randall, Thermodynamics, revised by K. S. Pitzer and L. Brewer, McGraw-Hill, New York, 2nd. edn. 1961, p. 108. [4] K. S. Pitzer, Thermodynamics, McGraw-Hill, New York, 3d. edn., 1995.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.03%3A_Absorption_Isotherms_-_Two_Absorbates.txt
Consider the general apparent molar property $\phi \left( \mathrm{Q}_j \right)$ and the corresponding extensive property of a solutions, $\mathrm{Q}$; e.g. $\mathrm{C}_{p}$, $\mathrm{E}_{p}$, $\mathrm{E}_{\mathrm{S}}$, $\mathrm{K}_{\mathrm{T}}$ and $\mathrm{K}_{\mathrm{S}}$. The latter are all extensive properties of a given system. The corresponding volume intensive property $q$ is given by the ratio $\mathrm{Q} / \mathrm{V}$; c.f. $q =$ isobaric expansivity $\alpha_{p}$, isentropic expansivity $\alpha_{S}$, isothermal compressibility $\mathrm{K}_{\mathrm{T}}$, isentropic compressibility $\mathrm{K}_{\mathrm{S}}$, and heat capacitance $\sigma$ respectively. The following four general equations [1] show how the volume intensive properties of the solution and solvent, $q$ and ${q_{1}}^{*}$ respectively, form the basis for the calculation of apparent molar property $\phi\left(Q_{j}\right)$ [2]. $\phi\left(Q_{j}\right)=\left(q-q_{1}^{*}\right) \,\left(m_{j} \, \rho_{1}^{*}\right)^{-1}+q \, \phi\left(V_{j}\right)$ $\phi\left(Q_{j}\right)=\left(q-q_{1}^{*}\right) \,\left(c_{j}\right)^{-1}+q_{1}^{*} \, \phi\left(V_{j}\right)$ $\phi\left(Q_{j}\right)=\left(q \, \rho_{1}^{*}-q_{1}^{*} \, \rho\right) \,\left(m_{j} \, \rho \, \rho_{1}^{*}\right)^{-1}+q \, M_{j} \, \rho^{-1}$ $\phi\left(Q_{j}\right)=\left(q \, \rho_{1}^{*}-q_{1}^{*} \, \rho\right) \,\left(c_{j} \, \rho_{1}^{*}\right)^{-1}+q_{1}^{*} \, M_{j} \,\left(\rho_{1}^{*}\right)^{-1}$ In these four equations, $\rho$ is the density of the solution; ${\rho_{1}}^{*}$ is the density of the solvent at the same $\mathrm{T}$ and $p$. The theme running through these equations is the link between the apparent molar property $\phi\left(Q_{j}\right)$ of a given solute and the measured property $q$. Interestingly apparent molar enthalpies break the pattern in that the enthalpy of a solution cannot be measured. Nevertheless apparent molar enthalpies are used in the analysis of calorimetric results. There are no advantages in defining apparent chemical potentials and apparent molar entropies of solutes. Footnotes [1] M. J. Blandamer, M. I. Davis, G.Douheret and J. C. R. Reis, Chem. Soc. Rev.,2001, 30,8. [2] $\phi\left(Q_{j}\right)$ is defined by the following equation with reference to the extensive variable $\mathrm{Q}$ in terms of amounts of solvent and solute $n_{1}$ and $n_{j}$ respectively where ${\mathrm{Q}_{1}}^{*}$ is the molar property of the solvent at the same $\mathrm{T}$ and $p$. $Q=n_{1} \, Q_{1}^{*}+n_{j} \, \phi\left(Q_{j}\right)$ We shift to volume intensive properties $q$ and ${q_{1}}^{*}$. $\mathrm{V} \, \mathrm{q}=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) \, \mathrm{q}_{1}^{*}+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{Q}_{\mathrm{j}}\right)$ We express the volume using the following equation incorporating apparent molar volume $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ and molar volume of the solvent $V_{1}^{*}(\lambda)$ at the same $\mathrm{T}$ and $p$. $\mathrm{V}=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda)+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)$ We solve equation (b) for $\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ using equation (c). $\text{Hence, } \phi\left(\mathrm{Q}_{\mathrm{j}}\right)=\frac{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) \, \mathrm{q}}{\mathrm{n}_{\mathrm{j}}}+\mathrm{q} \, \phi\left(\mathrm{V}_{\mathrm{j}}\right)-\frac{\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) \, \mathrm{q}_{\mathrm{l}}^{*}}{\mathrm{n}_{\mathrm{j}}}$ But [3] $\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) / \mathrm{n}_{\mathrm{j}}=\left(\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}\right)^{-1}$. Equation (e) follows. $\phi\left(Q_{j}\right)=\left(q-q_{1}^{*}\right) \,\left(m_{j} \, \rho_{1}^{*}\right)^{-1}+q \, \phi\left(V_{j}\right)$ Using the latter equation, molalities are converted to concentrations using equation (f). $\left(\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}\right)^{-1}=\left(\mathrm{c}_{\mathrm{j}}\right)^{-1}-\phi\left(\mathrm{V}_{\mathrm{j}}\right)$ $\text { Then } \phi\left(Q_{j}\right)=\left(q-q_{1}^{*}\right) \,\left(c_{j}\right)^{-1}-\phi\left(V_{j}\right) \,\left(q-q_{1}^{*}\right)+q \, \phi\left(V_{j}\right)$ $\text {Or } \phi\left(Q_{j}\right)=\left(q-q_{1}^{*}\right) \,\left(c_{j}\right)^{-1}+q_{1}^{*} \, \phi\left(V_{j}\right)$ We return to equation (b) and express the volume using the following equation. $\mathrm{V}=\left[\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right] / \rho$ $\text { Then [4] } \phi\left(Q_{j}\right)=\frac{n_{1} \, M_{1} \, q}{n_{j} \, \rho}+\frac{M_{j} \, q}{\rho}-\frac{n_{1} \, V_{1}^{*} \, q_{1}^{*}}{n_{j}}$ $\text { Then [3,5] } \phi\left(Q_{j}\right)=\frac{q}{m_{j} \, \rho}+\frac{M_{j} \, q}{\rho}-\frac{q_{1}^{*}}{m_{j} \, \rho_{1}^{*}}$ Or, \begin{aligned} &\phi\left(Q_{j}\right)= \ &\left(q \, \rho_{1}^{*}-q_{1}^{*} \, \rho\right) \,\left(m_{j} \, \rho \, \rho_{1}^{*}\right)^{-1}+q \, M_{j} \, \rho^{-1} \end{aligned} To obtain an equation using concentrations, we use the following equation. $1 / \mathrm{m}_{\mathrm{j}}=\rho / \mathrm{c}_{\mathrm{j}}-\mathrm{M}_{\mathrm{j}}$ Thus [6] $\phi\left(Q_{j}\right)=\frac{q \, \rho_{1}^{*}-q_{1}^{*} \, \rho}{c_{j} \, \rho_{1}^{*}}-\frac{M_{j} \,\left(q \, \rho_{1}^{*}-q_{1}^{*} \, \rho\right)}{\rho_{1}^{*} \, \rho}+\frac{q \, M_{j}}{\rho}$ Or, $\phi\left(Q_{j}\right)=\left(q \, \rho_{1}^{*}-q_{1}^{*} \, \rho\right) \,\left(c_{j} \, \rho_{1}^{*}\right)^{-1}+q_{1}^{*} \, M_{j} \,\left(\rho_{1}^{*}\right)^{-1}$ Because the equations used for converting molalities to concentrations are exact, no approximations are involved. Therefore equations (e), (h),($\lambda$) and m) are rigorously equivalent. [3] \begin{aligned} \mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) / \mathrm{n}_{\mathrm{j}} &=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) \, \rho_{1}^{*}(\lambda) / \mathrm{n}_{\mathrm{j}} \, \rho_{1}^{*}(\lambda) \ &=\mathrm{w}_{1} / \mathrm{n}_{\mathrm{j}} \, \rho_{1}^{*}(\lambda)=\left[\mathrm{m}_{\mathrm{j}} \, \rho_{1}^{*}(\lambda)\right]^{-1} \end{aligned} [4] From equations (b) and (i), $\left(\mathrm{n}_{1} \, \mathrm{M}_{1}+\mathrm{n}_{\mathrm{j}} \, \mathrm{M}_{\mathrm{j}}\right) \, \mathrm{q} / \rho=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\lambda) \, \mathrm{q}_{1}^{*}+\mathrm{n}_{\mathrm{j}} \, \phi\left(\mathrm{Q}_{\mathrm{j}}\right)$ [5] $\mathrm{n}_{1} \, \mathrm{M}_{\mathrm{l}} / \mathrm{n}_{\mathrm{j}}=\left(\mathrm{m}_{\mathrm{j}}\right)^{-1}$ [6] From equation ($\lambda$), $\phi\left(Q_{j}\right)=\left\{\left[q \, \rho_{1}^{*}(\lambda)-q_{1}^{*} \, \rho\right] / \rho \, \rho_{1}^{*}(\lambda)\right\} \,\left[\left(\rho / c_{j}\right)-M_{j}\right]+\left[q \, M_{j} / \rho\right]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.05%3A_Apparent_Molar_Properties-_Solutions-_General.txt
An axiom is a statement of principle which is generally accepted to be true. Axioms cannot be conveniently demonstrated otherwise there would be no need to use axioms. The Laws of Thermodynamics are axioms [1]. However we can test and confirm by experiment many of the consequences following from the Laws of Thermodynamics. We can debate (if we so wish) if these laws were actually discovered in the sense that the laws were always there. Alternatively we might argue that these laws were formulated by brilliant scientists because prior to their formulation these laws did not exist. Footnote [1] M. L. McGlashan, J.Chem.Educ.,1966,43,226. 1.14.07: Boundary The term ‘boundary’ is encountered in several contexts in thermodynamics. Indeed the definition becomes more complicated as we concentrate on the thermodynamic properties of systems. As a starting point, a boundary separates system and surroundings. A boundary is an infinitely thin surface separating system and surroundings such that the properties of system and surroundings change abruptly at the boundary. [1,2] In these terms a reaction vessel is part of the surroundings. We support this careful distinction by observing that if chemical reaction inside the system is exothermic, the liberated heat warms the reaction vessel. In this case, the boundary encloses the system. No molecules can either enter or leave the system. However heat is allowed to cross the boundary. Thus the whole universe is divided into system and surroundings [3], the only role of the boundary is to facilitate communication between system and surroundings. In these terms chemical substances, heat, and electric charge may cross a boundary between a system and surroundings. In some cases the container (e.g. reaction vessel) may be considered part of the system. In many cases [4,5] it is correct to do so and so the boundary is again a hypothetical surface separating ‘reaction solution + reaction vessel’ and the surroundings. In general terms, it is important to define the boundary for a given system. Another term for boundary is ‘envelope’ which indicates something which can be quite dynamic in terms of shape and volume rather than, for example, a glass vessel. Moreover the boundary may be selectively permeable to one or more chemical substances rather like the envelope of unit cells in living systems [6]. The term ‘boundary’ in the context of surface chemistry means a boundary phase (or, capillary phase) [7]. In such a phase there is a concentration gradient of one or more chemical substances across the boundary phase between system and surroundings. Indeed surface chemistry can be described as the chemistry of boundaries. In summary we repeat the point that in a thermodynamic analysis of experimental results, a first requirement is that the system, boundary and surroundings are carefully defined. For the most part we assume that the boundary is an infinitely thin envelope separating system and surroundings. Footnotes [1] K. S. Pitzer, Thermodynamics, McGraw-Hill, New York,1995, 3rd. edn., page 6. [2] G. N. Lewis and M. Randall, Thermodynamics, McGraw-Hill, New York, 1923, page 10. [3] D. H. Everett, Chemical Thermodynamics, Longmans, London, 1959, page 8. [4] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979,page 1. [5] E. B. Smith, Basic Chemical Thermodynamics, Clarendon Press, Oxford, 3rd. edn.,1982, page 2. [6] S. E. Wood and R. Battino, Thermodynamics of Chemical Systems, Cambridge University Press, Cambridge, 1990, page 3. [7] J. N. Bronsted, Physical Chemistry, Heinemann, London, 1937, page 359. 1.14.08: Calculus Consider a variable u defined by the independent variables $x$ and $y$. $\text { We write } u=u[x, y]$ Equation (b) is the general exact differential of equation (a). $\mathrm{du}=\left(\frac{\partial \mathrm{u}}{\partial \mathrm{x}}\right)_{\mathrm{y}} \, \mathrm{dx}+\left(\frac{\partial \mathrm{u}}{\partial \mathrm{y}}\right)_{\mathrm{x}} \, \mathrm{dy}$ In other words the change in u is related to the differential dependence of $\mathrm{u}$ on $x$ at constant $y$ and the differential dependence of $\mathrm{u}$ on $y$ at constant $x$. For the case where u does not change, $\left(\frac{\partial u}{\partial x}\right)_{y}=-\left(\frac{\partial u}{\partial y}\right)_{x} \,\left(\frac{\partial y}{\partial x}\right)_{u}=0 \text { and }\left(\frac{\partial y}{\partial x}\right)_{u}=-\left(\frac{\partial u}{\partial x}\right)_{y} \,\left(\frac{\partial y}{\partial u}\right)_{x}$ A variable $z$ is defined by the independent variables $x$ and $y$. $z=z[x, y]$ Equation (e) is the general differential of equation (d). d z=\left(\frac{\partial z}{\partial x}\right)_{y} \, d x+\left(\frac{\partial z}{\partial y}\right)_{x} \, d y\] We direct attention to the dependence of $z$ on $x$ along a pathway for which $\mathrm{u}$ is constant. $\text { Then }\left(\frac{\partial z}{\partial x}\right)_{u}=\left(\frac{\partial z}{\partial x}\right)_{y}+\left(\frac{\partial z}{\partial y}\right)_{x} \,\left(\frac{\partial y}{\partial x}\right)_{u}$ The latter equation contains the differential dependence of $y$ on $x$ at constant $\mathrm{u}$. The latter dependence can be reformulated using equation (c). Therefore $\left(\frac{\partial z}{\partial x}\right)_{u}=\left(\frac{\partial z}{\partial x}\right)_{y}-\left(\frac{\partial u}{\partial x}\right)_{y} \,\left(\frac{\partial y}{\partial u}\right)_{x} \,\left(\frac{\partial z}{\partial y}\right)_{x}$ The key point to emerge from this exercise centres is the way in which the condition on the partial differential $(\partial z / \partial x)$ can be changed from ‘at constant $y$’ to ‘at constant $\mathrm{u}$’. Another important operation concerns a variable $\mathrm{q}$. $\text { Thus, }\left(\frac{\partial x}{\partial y}\right)_{z}=\left(\frac{\partial x}{\partial q}\right)_{z} \,\left(\frac{\partial q}{\partial y}\right)_{z}$ For composite functions such as $z=z[\mathrm{u}, \mathrm{v}]$, where $z=z[x, y]$, and $\mathrm{u}=\mathrm{u}[\mathrm{x}, \mathrm{y}]$, further important equations are found [1]. $\text { Thus }\left(\frac{\partial z}{\partial u}\right)_{v}=\left(\frac{\partial z}{\partial x}\right)_{y} \,\left(\frac{\partial x}{\partial u}\right)_{v}+\left(\frac{\partial z}{\partial y}\right)_{x} \,\left(\frac{\partial y}{\partial u}\right)_{v}$ Equation (i) is an example of the well-known chain rule, a similar equation holding for $(\partial z / \partial v)_{u}$. This rule allows the total change of independent variables from $z=z[\mathrm{u}, \mathrm{v}]$ to $z=z[x, y]$. $\text { Also }\left(\frac{\partial z}{\partial x}\right)_{y}=\left(\frac{\partial z}{\partial x}\right)_{y, v}+\left(\frac{\partial z}{\partial v}\right)_{y, x} \,\left(\frac{\partial v}{\partial x}\right)_{y}$ The latter equation is useful for introducing an extra constraint on a given differential. Footnote [1] H. B. Callen, Thermodynamics and an Introduction to Thermostatics, Wiley, New York, 2dn. Edn.,1985, Appendix A.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.06%3A_Axioms.txt
A give closed system contains chemical substance j present in both liquid and gas phases. The system is at equilibrium. In terms of the Phase Rule, the following parameters are defined; $\mathrm{P} = 2$, $\mathrm{C} = 1$ and hence $\mathrm{F} = 1$. Hence, if the temperature is fixed by the observer, the equilibrium pressure $\mathrm{p}^{\mathrm{eq}}$ is defined. The equilibrium can be described in terms of an equality of chemical potentials of pure liquid and pure gas. $\mu_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p}) \label{a}$ Both chemical potentials in Equation \ref{a} are functions of both $\mathrm{T}$ and $\mathrm{p}$. In general terms, $\mathrm{d} \mu_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})=\left(\frac{\partial \mu_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})}{\partial \mathrm{T}}\right)_{\mathrm{p}} \, \mathrm{dT}+\left(\frac{\partial \mu_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})}{\partial \mathrm{p}}\right)_{\mathrm{T}} \, \mathrm{dp} \label{b}$ or $\mathrm{d} \mu_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})=-\mathrm{S}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dp} \label{c}$ Similarly $\mathrm{d} \mu_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p})=-\mathrm{S}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dp} \label{d}$ The condition in Equation \ref{a} applies at all $\mathrm{T}$ and $\mathrm{p}$. \begin{aligned} &\text { Then, }-\mathrm{S}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dp} \ &=-\mathrm{S}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p}) \, \mathrm{dp} \end{aligned} or [1], $\left(\frac{\mathrm{dp}}{\mathrm{dT}}\right)^{e q}=\frac{\mathrm{S}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p})-\mathrm{S}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})}{\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p})-\mathrm{V}_{\mathrm{j}}^{*}(\ell ; \mathrm{T} ; \mathrm{p})}$ $\left(\frac{\mathrm{dp}}{\mathrm{dT}}\right)^{\mathrm{eq}}=\frac{\Delta_{\mathrm{vap}} \mathrm{S}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}{\Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}$ But at equilibrium, $\Delta_{\text {vap }} \mathrm{G}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})=\Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})-\mathrm{T} \, \Delta_{\text {vap }} \mathrm{S}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})=0$ $\left(\frac{\mathrm{dp}}{\mathrm{dT}}\right)^{\text {eq }}=\frac{\Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}{\mathrm{T} \, \Delta_{\mathrm{vap}} \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}$ The latter is the Clausius-Clapeyron Equation [2]. In a modern development, equation (i) was exactly integrated [3]. Equation(i) does not have the form of an exact differential in the independent variables $\mathrm{p}$ and $\mathrm{T}$.[3] The corresponding integrating factor is $\mathrm{T}^{-1} \, \Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*}$. Thus $\mathrm{T}^{-1} \, \Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*} \, \mathrm{dp}-\mathrm{T}^{-2} \, \Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*} \, \mathrm{dT}=0$ or, $\mathrm{T}^{-1} \, \Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*} \, \mathrm{dp}+\Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*} \, \mathrm{dT}^{-1}=0$ The latter equation is an exact differential as a consequence of equation ($\ell$) [4]. $\left(\frac{\partial\left(\mathrm{T}^{-1} \, \Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*}\right)}{\partial \mathrm{T}^{-1}}\right)_{\mathrm{p}}=\left(\frac{\partial \Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ A mathematical solution is known for differential equations having the form of equation (k) [3]. A comprehensive set of equations have been derived describing first order transitions for pure substances [5] and hence the phase equilibrium curves For liquid-vapour equilibria, both $\Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p}) \text { and } \Delta_{\text {vap }} \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p}) \text { are }>0$. Therefore the equilibrium vapor pressure of a liquid increases with increase in temperature. A useful approximation assumes that gas $j$ is a perfect gas; i.e. $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{gp})=\mathrm{R} \, \mathrm{T}$ and $\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{gp} ; \mathrm{T} ; \mathrm{p})>>\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{l} ; \mathrm{T} ; \mathrm{p})$. $\left(\frac{\mathrm{d} \ln (\mathrm{p})}{\mathrm{dT}}\right)^{\text {eq }}=\frac{\Delta_{\text {vap }} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}{\mathrm{R} \, \mathrm{T}^{2}}$ $\left(\frac{\mathrm{d} \ln (\mathrm{p})}{\mathrm{d}\left(\mathrm{T}^{-1}\right)}\right)^{\mathrm{eq}}=-\frac{\Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T} ; \mathrm{p})}{\mathrm{R}}$ Within the limits of the approximations outlined above, $\ln \left(p^{e q}\right)$ is a linear function of $\mathrm{T}^{-1}$. Exactly integrated equations have also been established for other first-order transitions ($\mathrm{p}{\mathrm{eq}}$, $\mathrm{T}{\mathrm{eq}}$) curves of pure substances [5]. Footnotes [1] $\frac{\left[\mathrm{N} \mathrm{m}^{-2}\right]}{[\mathrm{K}]}=\frac{\left[\mathrm{J} \mathrm{mol}^{-1} \mathrm{~K}^{-1}\right]}{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]}=\frac{\left[\mathrm{J} \mathrm{m}^{-3}\right]}{[\mathrm{K}]}=\frac{\left[\mathrm{N} \mathrm{m}^{-2}\right]}{[\mathrm{K}]}$ [2] We have derived the equation for vapor-liquid equilibrium which is the generally quoted form. An equivalent form expresses $\left(\frac{\mathrm{dp}}{\mathrm{dT}}\right)^{e q}$ for the equilibrium for chemical substance $j$ in two phases $\alpha$ and $\beta$. [3] C. Mosselman, W. H. van Vugt and H. Vos, J. Chem. Eng. Data 1982,27,246. [4] From $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}$ \begin{aligned} &\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\mathrm{T} \, \left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}}+\mathrm{V} \ &\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=-\mathrm{T} \, \left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{T}}+\mathrm{V} \ &\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\mathrm{T}^{-1} \, \left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}^{-1}}\right)_{\mathrm{T}}+\mathrm{V} \ &\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\left(\frac{\partial\left(\mathrm{T}^{-1} \, \mathrm{V}\right)}{\partial \mathrm{T}^{-1}}\right)_{\mathrm{T}} \end{aligned} [5] L. Q. Lobo and A. G. M. Ferreira, J. Chem. Thermodyn., 2001,33,1597. 1.14.10: Gibbs - Helmholtz Equation The Gibbs energy and enthalpy of a closed system are related; $\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$ The two properties $\mathrm{G}$ and $\mathrm{H}$ are also related by the Gibbs - Helmholtz equation through the dependence of $\mathrm{G}$ on temperature at fixed pressure. We envisage a situation in which a closed system at equilibrium having Gibbs energy $\mathrm{G}$ is displaced to a neighbouring equilibrium state by a change in temperature at constant pressure. We are interested in the partial derivative, $\left[\frac{\partial(\mathrm{G} / \mathrm{T})}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}=0}$. In general terms we consider the isobaric differential dependence of $(\mathrm{G} / \mathrm{T})$ on temperature. $\frac{\mathrm{d}}{\mathrm{dT}}\left(\frac{\mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p}}=\frac{1}{\mathrm{~T}} \,\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{p}-\frac{\mathrm{G}}{\mathrm{T}^{2}}$ $\mathrm{T}^{2} \, \frac{\mathrm{d}}{\mathrm{dT}}\left(\frac{\mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p}}=\mathrm{T} \,\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}}-\mathrm{G}$ But $\mathrm{S}=-\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ For an equilibrium change, equations (b) and (c) yield equation (e). $\mathrm{T}^{2} \, \frac{\mathrm{d}}{\mathrm{dT}}\left(\frac{\mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p}}=-(\mathrm{G}+\mathrm{T} \, \mathrm{S})$ But $\mathrm{H}=\mathrm{G}+\mathrm{T} \, \mathrm{S}$. Then, $\mathrm{H}=-\mathrm{T}^{2} \, \frac{\mathrm{d}}{\mathrm{dT}}\left(\frac{\mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p}}$ For an equilibrium change, $\Delta \mathrm{H}(\mathrm{A}=0)=-\mathrm{T}^{2} \, \frac{\mathrm{d}}{\mathrm{dT}}\left(\frac{\Delta \mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p} ; \mathrm{A}=0}$ or, $\Delta \mathrm{H}(\mathrm{A}=0)=\frac{\mathrm{d}}{\mathrm{dT}^{-1}}\left(\frac{\Delta \mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p} ; \mathrm{A}=0}$ In a similar manner we obtain the Gibbs -Helmholtz equation for a system perturbed at constant composition [1]. $\Delta \mathrm{H}(\text { fixed } \xi)=\frac{\mathrm{d}}{\mathrm{dT}^{-1}}\left(\frac{\Delta \mathrm{G}}{\mathrm{T}}\right)_{\mathrm{p}, \bar{\xi},}$ Equation (f) is the starting point for the development of another important equation. Thus, $\mathrm{H}=-\mathrm{T}^{2} \,\left[-\frac{\mathrm{G}}{\mathrm{T}^{2}}+\frac{1}{\mathrm{~T}} \, \frac{\mathrm{dG}}{\mathrm{dT}}\right]$ Hence, $\mathrm{H}=\mathrm{G}-\mathrm{T} \,\left[\frac{\mathrm{dG}}{\mathrm{dT}}\right]$ Equation (k) is differentiated with respect to temperature at constant pressure and at ‘$\mathrm{A}=0$’. $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0}=\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0}-\mathrm{T} \,\left(\frac{\partial^{2} \mathrm{G}}{\partial \mathrm{T}^{2}}\right)_{p, A=0}-\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0}$ Hence, $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}=0}=-\mathrm{T} \,\left(\frac{\partial^{2} \mathrm{G}}{\partial \mathrm{T}^{2}}\right)_{\mathrm{p}, \mathrm{A}=0}$ But $\left(\frac{\partial^{2} G}{\partial T^{2}}\right)_{p, A=0}=\frac{\partial}{\partial T}\left(\frac{\partial G}{\partial T}\right)=-\left(\frac{\partial S}{\partial T}\right)_{p, A=0}$ Also the equilibrium isobaric heat capacity, $C_{p}(A=0)=\left(\frac{\partial H}{\partial T}\right)_{p, A=0}$ Equations (m), (n) and (o) yield equation (p). $\left(\frac{\partial S}{\partial T}\right)_{p, A=0}=\frac{C_{p}(A=0)}{T}$ Equation (p) relates the isobaric equilibrium dependence of entropy of a closed system on temperature to the isobaric heat capacity. Also starting from, $\mathrm{H}=\mathrm{G}+\mathrm{T} \, \mathrm{S}$, then $(\partial \mathrm{H} / \partial \mathrm{p})_{\mathrm{T}}=(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}}+\mathrm{T} \,(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}$ Using a Maxwell Equation, $(\partial H / \partial p)_{T}=\mathrm{V}-\mathrm{T} \,(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$ Similarly, $(\partial \mathrm{U} / \partial \mathrm{T})_{\mathrm{V}}=\mathrm{C}_{\mathrm{V}}=\mathrm{T} \,(\partial \mathrm{S} / \partial \mathrm{T})_{\mathrm{V}}$ And $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}}=-\mathrm{p}-\mathrm{T} \,(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}} \,(\partial \mathrm{p} / \partial \mathrm{V})_{\mathrm{T}}$ Footnote [1] There are many thermodynamic equations which are of the GibbsHelmholtz type. As a common feature they conform to the following calculus property. Given $\mathrm{f}=\mathrm{f}(\mathrm{x}, \mathrm{y})$ Then $\left(\frac{\partial(f / x)}{\partial(1 / x)}\right)_{y}=-x^{2} \,\left(\frac{\partial(f / x)}{\partial x}\right)_{y}=f-x \,\left(\frac{\partial f}{\partial x}\right)_{y}$ Similarly, $\left(\frac{\partial(f / y)}{\partial(1 / y)}\right)_{x}=-y^{2} \,\left(\frac{\partial(f / x)}{\partial y}\right)_{x}=f-y \,\left(\frac{\partial f}{\partial y}\right)_{x}$ Normally $\mathrm{f}$ stands for a thermodynamic potential and $x$ and $y\ for its natural variables. Thus a total of 8 equations of the Gibbs - Helmholtz type holding for closed systems can be constructed from \(\mathrm{U}=\mathrm{U}(\mathrm{S}, \mathrm{V}), \mathrm{F}=\mathrm{F}(\mathrm{T}, \mathrm{V}), \mathrm{H}=\mathrm{H}(\mathrm{S}, \mathrm{p}) \text { and } \mathrm{G}=\mathrm{G}(\mathrm{T}, \mathrm{p})$.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.09%3A_Clausius_-_Clapeyron_Equation.txt
For binary liquid mixtures at fixed $\mathrm{T}$ and $\mathrm{p}$, an important task is to fit the dependence of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ on $x_{2}$ to an equation in order to calculate the derivative ${\mathrm{dG}_{\mathrm{m}}}^{\mathrm{E}} / \mathrm{dx}_{2}$ at required mole fractions. The Guggenheim-Scatchard [1,2] (commonly called the Redlich-Kister [3] ) equation is one such equation. This equation has the following general form. $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{2} \left(1-\mathrm{x}_{2}\right) \sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{k}} \mathrm{A}_{\mathrm{i}} \left(1-2 \mathrm{x}_{2}\right)^{\mathrm{i}-1} \label{a}$ $\mathrm{A}_{\mathrm{i}}$ are coefficients obtained from a least squares analysis of the dependence of ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ on $x_{2}$. The equation clearly satisfies the condition that ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ is zero at $x_{2} = 0$ and at $x_{2} = 1$. In fact the first term in the $\mathrm{G} - \mathrm{~S}$ equation has the following form. $X_{m}^{E}=x_{2} \left(1-x_{2}\right) A_{1}\label{b}$ According to Equation \ref{b} ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ is an extremum at $x_{2} = 0.5$, the plot being symmetric about the line from ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ to ‘$x_{2} = 0.5$’. In fact for most systems the $\mathrm{A}_{1}$ term is dominant. For the derivative $\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{2}$, we write Equation \ref{a} in the following general form. $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\left(\mathrm{x}_{2}-\mathrm{x}_{2}^{2}\right) \mathrm{Q}\label{c}$ Then $\mathrm{dX}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{x}_{2}=\mathrm{x}_{2} \left(1-\mathrm{x}_{2}\right) \mathrm{dQ} / \mathrm{dx}_{2}+\left(1-2 \mathrm{x}_{2}\right) \mathrm{Q}\label{d}$ where $\mathrm{dQ} / \mathrm{dx}_{2}=-2 \sum_{\mathrm{i}=2}^{\mathrm{i}=\mathrm{k}}(\mathrm{i}-1) \mathrm{A}_{\mathrm{i}} \left(1-2 \mathrm{x}_{2}\right)^{\mathrm{i}-2}\label{e}$ Equation \ref{a} fits the dependence with a set of contributing curves which all pass through points, ${\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}=0$ at $x_{1} = 0$ and $x_{1} =1$. The usual procedure involves fitting the recorded dependence using increasing number of terms in the series, testing the statistical significance of including a further term. Although Equation \ref{a} has been applied to many systems and although the equation is easy to incorporate into computer programs using packaged least square and graphical routines, the equation suffers from the following disadvantage. As one incorporates a further term in the series, (e.g. $\mathrm{A}_{j}$) estimates of all the previously calculated parameters (i.e. $\mathrm{A}_{2}$, $\mathrm{A}_{3}$, ... $\mathrm{A}_{j-1}$) change. For this reason orthogonal polynomials have been increasingly favoured especially where the appropriate computer software is available. The only slight reservation is that derivation of explicit equations for the required derivative ${\mathrm{dX}_{\mathrm{m}}}^{\mathrm{E}}$ is not straightforward. The problem becomes rather more formidable when the second and higher derivatives are required. The derivative $\mathrm{d}^{2}{\mathrm{X}_{\mathrm{m}}}^{\mathrm{E}}$ is sometimes required by calculations concerning the properties of binary liquid mixtures. The derivative $\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{1}$ and ${\mathrm{G}_{\mathrm{m}}}^{\mathrm{E}}$ are combined (see Topic EZ20) to yield an equation for $\ln\left(\mathrm{f}_{1}\right)$. $\ln \left(f_{1}\right)=\frac{G_{m}^{E}}{R T}+\frac{\left(1-x_{1}\right)}{R T} \frac{d G_{m}^{E}}{d x_{1}}\label{f}$ A similar equation leads to estimates of $\ln\left(\mathrm{f}_{2}\right)$. Hence the dependences are obtained of both $\ln\left(\mathrm{f}_{1}\right)$ and $\ln\left(\mathrm{f}_{2}\right)$ on mixture composition. It is of interest to explore the case where the coefficients $\mathrm{A}_{2}, \mathrm{~A}_{3} \ldots$ in Equation \ref{a} are zero. Then $\mathrm{X}_{\mathrm{m}}^{\mathrm{E}}=\mathrm{x}_{2} \left(1-\mathrm{x}_{2}\right) \mathrm{A}_{1}\label{g}$ and $\mathrm{dX}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{dx}_{2}=\left(1-2 \mathrm{x}_{2}\right) \mathrm{A}_{1}\label{h}$ With reference to the Gibbs energies, $\ln \left(\mathrm{f}_{2}\right)=(1 / \mathrm{R} \mathrm{T}) \left[\mathrm{x}_{2} \left(1-\mathrm{x}_{2}\right)+\left(1-\mathrm{x}_{2}\right) \left(1-2 \mathrm{x}_{2}\right)\right] \mathrm{A}_{1}^{\mathrm{G}} \label{i}$ $\ln \left(f_2\right)=\left(A_1^G / R T\right) \left[1-2 x_2 + x_2^{2} \right] \label{j}$ or, $\ln \left(f_{2}\right)=\left(A_{1}^{\mathrm{G}} / \mathrm{R} \mathrm{T}\right) \left[1-\mathrm{x}_{2}\right]^{2}\label{k}$ In fact the equation reported by Jost et al. [4] has this form. Rather than using the Redlich-Kister equation, recently attention has been directed to the Wilson equation [5] written in Equation \ref{l} for a two-component liquid [6]. $\mathrm{G}_{\mathrm{m}}^{\mathrm{E}} / \mathrm{R} \mathrm{T}=-\mathrm{x}_{1} \ln \left(\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}\right)-\mathrm{x}_{2} \ln \left(\mathrm{x}_{2}+\Lambda_{21} \mathrm{x}_{1}\right)\label{l}$ Then , for example [7], $\ln \left(f_{1}\right)=-\ln \left(x_{1}+\Lambda_{12} x_{2}\right)+x_{2} \left(\frac{\Lambda_{12}}{x_{1}+\Lambda_{12} x_{2}}-\frac{\Lambda_{21}}{\Lambda_{21} x_{1}+x_{2}}\right)\label{m}$ The Wilson equation forms the basis for two further developments, described as the NRTL (non-random, two-liquid) equation [8-10] and the UNIQUAC equation [9-10]. Footnotes [1] E. A. Guggenheim, Trans. Faraday Soc.,1937,33,151; equation 4.1. [2] G. Scatchard, Chem. Rev.,1949,44,7;see page 9. [3] O. Redlich and A. Kister, Ind. Eng. Chem.,1948,40,345; equation 8. [4] F. Jost, H. Leiter and M. J. Schwuger, Colloid Polymer Sci., 1988, 266, 554. [5] G. M. Wilson, J. Am. Chem. Soc.,1964,86,127. [6] See also 1. R. V. Orye and J. M. Prausnitz, Ind. Eng. Chem.,1965,57,19. 2. S. Ohe, Vapour-Liquid Equilibrium Data, Elsevier, Amsterdam, 1989. [7] From Equation \ref{l}, \begin{aligned} \frac{1}{\mathrm{R} \mathrm{T}} \frac{\mathrm{dG}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{dx}_{1}}=&-\ln \left(\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}\right)-\frac{\mathrm{x}_{1} \left(1-\Lambda_{12}\right)}{\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}} \ &+\ln \left(\Lambda_{21} \mathrm{x}_{1}+\mathrm{x}_{2}\right)-\frac{\mathrm{x}_{2} \left(\Lambda_{21}-1\right)}{\Lambda_{21} \mathrm{x}_{1}+\mathrm{x}_{2}} \end{aligned} Then using Equation \ref{f} with $1− x_{1} = x_{2}$, \begin{aligned} \ln \left(f_{1}\right)=&-x_{1} \ln \left(x_{1}+\Lambda_{12} x_{2}\right)-x_{2} \ln \left(\Lambda_{21} x_{1}+x_{2}\right) \ &-x_{2} \ln \left(x_{1}+\Lambda_{12} x_{2}\right)-\frac{x_{1} x_{2} \left(1-\Lambda_{12}\right)}{x_{1}+\Lambda_{12} x_{2}} \ &+x_{2} \ln \left(\Lambda_{21} x_{1}+x_{2}\right)+\frac{\left(x_{2}\right)^{2} \left(1-\Lambda_{21}\right)}{\Lambda_{21} x_{1}+x_{2}} \end{aligned} Or, \begin{aligned} \ln \left(f_{1}\right) &=-\left(x_{1}+x_{2}\right) \ln \left(x_{1}+\Lambda_{12} x_{2}\right) \ &+x_{2} \left[\frac{\Lambda_{12} x_{1}-x_{1}}{x_{1}+\Lambda_{12} x_{2}}-\frac{\Lambda_{21} x_{2}-x_{2}}{\Lambda_{21} x_{1}+x_{2}}\right] \end{aligned} But $\Lambda_{12} \mathrm{x}_{1}-\mathrm{x}_{1}=\Lambda_{12} \left(1-\mathrm{x}_{2}\right)-\mathrm{x}_{1}=\Lambda_{12}-\left(\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}\right)$ Hence, \begin{aligned} \ln \left(\mathrm{f}_{1}\right) &=-\ln \left(\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}\right) \ &+\mathrm{x}_{2} \left[\frac{\Lambda_{12}-\left(\mathrm{x}_{1}-\Lambda_{12} \mathrm{x}_{2}\right)}{\mathrm{x}_{1}+\Lambda_{12} \mathrm{x}_{2}}-\frac{\Lambda_{21}-\left(\Lambda_{21} \mathrm{x}_{1}+\mathrm{x}_{2}\right)}{\Lambda_{21} \mathrm{x}_{1}+\mathrm{x}_{2}}\right] \end{aligned} Or, $\ln \left(f_{1}\right)=-\ln \left(x_{1}+\Lambda_{12} x_{2}\right)+x_{2} \left[\frac{\Lambda_{12}}{x_{1}+\Lambda_{12} x_{2}}-\frac{\Lambda_{21}}{\Lambda_{21} x_{1}+x_{2}}\right]$ [8] D. Abrams and J. M. Prausnitz, AIChE J.,1975,21,116. [9] R. C. Reid, J. M. Prausnitz and E. B. Poling, The Properties of Gases and Liquids, McGraw-Hill, New York, 4th edn.,1987, chapter 8. [8] J. M. Prausnitz, R. N. Lichtenthaler and E. G. de Azevedo, Molecular Themodyanamics of Fluid Phase Equilibria, Prentice –Hall, Upper Saddle River, N.J., 3rd edn.,1999,chapter 6. 1.14.12: Legendre Transformations Many important thermodynamic equations are closely related. These relationships are often highlighted by the mathematical technique, Legendre Transformations [1,2]. With reference to thermodynamics, Callen [3] discusses application of Legendre Transformations. The essential features of Legendre Transformations can be understood in the following terms. A primary variable $\mathrm{Q}$ is defined by two dependent variables $x$ and $y$. Thus $Q=Q[x, y]$ Then $\mathrm{dQ}=\left(\frac{\partial \mathrm{Q}}{\partial \mathrm{x}}\right)_{\mathrm{y}} \, \mathrm{dx}+\left(\frac{\partial \mathrm{Q}}{\partial \mathrm{y}}\right)_{\mathrm{x}} \, \mathrm{dy}$ By definition, $\mathrm{u}=\left(\frac{\partial \mathrm{Q}}{\partial \mathrm{x}}\right)_{\mathrm{y}} \quad \text { and } \quad \mathrm{v}=\left(\frac{\partial \mathrm{Q}}{\partial \mathrm{y}}\right)_{\mathrm{x}}$ Then from equation (b), $\mathrm{dQ}=\mathrm{u} \, \mathrm{dx}+\mathrm{v} \, \mathrm{dy}$ A new variable $\mathrm{Z}$ is defined by equation (e). $\mathrm{Z}=\mathrm{Z}[\mathrm{u}, \mathrm{y}] \text { where } \mathrm{Z}=\mathrm{Q}-\mathrm{u} \, \mathrm{x}$ Then, $\mathrm{dZ}=\mathrm{dQ}-\mathrm{x} \, \mathrm{du}-\mathrm{u} \, \mathrm{dx}$ Hence using equation (d), $d Z=u \, d x+v \, d y-x \, d u-u \, d x$ Or, $\mathrm{d} Z=-\mathrm{x} \, \mathrm{du}+\mathrm{v} \, \mathrm{dy}$ Hence, $x=-\left(\frac{\partial Z}{\partial u}\right)_{y} \quad \text { and } \quad v=\left(\frac{\partial Z}{\partial y}\right)_{u}$ Comparison of equations (a) and (e) reveals the transformation, $\mathrm{Q}[\mathrm{x}, \mathrm{y}] \rightarrow \mathrm{Z}[\mathrm{u}, \mathrm{y}]$. We now explore thermodynamic transformations [3]. The following Master Equation relates the change in thermodynamic energy $\mathrm{U}$ with the changes in entropy $\mathrm{S}$ at temperature $\mathrm{T}$, volume $\mathrm{V}$ at pressure $\mathrm{p}$ and composition $\xi$ at affinity $\mathrm{A}$; $\mathrm{U}=\mathrm{U}[\mathrm{S}, \mathrm{V}, \xi]$. $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi$ By definition, enthalpy $\mathrm{H}=\mathrm{U}+\mathrm{p} \, \mathrm{V}$; $\mathrm{dH}=-\mathrm{dU}+\mathrm{p} \, \mathrm{dV}+\mathrm{V} \, \mathrm{dp}$ Using equation (j), $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi+\mathrm{p} \, \mathrm{dV}+\mathrm{V} \, \mathrm{dp}$ Then $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}-\mathrm{A} \, \mathrm{d} \xi$ Or, $\mathrm{H}=\mathrm{H}[\mathrm{S}, \mathrm{p}, \xi]$ The transformation is- $\mathrm{U}[\mathrm{S}, \mathrm{V}, \xi] \rightarrow \mathrm{H}[\mathrm{S}, \mathrm{p}, \xi]$ By definition, Gibbs energy $\mathrm{G}=\mathrm{U}+\mathrm{p} \, \mathrm{V}-\mathrm{T} \, \mathrm{S}$ Or using equation (k), $\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$ Then $\mathrm{dG}=\mathrm{dH}-\mathrm{T} \, \mathrm{dS}-\mathrm{S} \, \mathrm{dT}$ Hence from equation (l) $\mathrm{dG}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}-\mathrm{A} \, \mathrm{d} \xi-\mathrm{T} \, \mathrm{dS}-\mathrm{S} \, \mathrm{dT}$ Or, $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}-\mathrm{A} \, \mathrm{d} \xi$ And, $\mathrm{G}=\mathrm{G}[\mathrm{T}, \mathrm{p}, \xi]$ The transformation is $\mathrm{H}[\mathrm{S}, \mathrm{p}, \xi] \rightarrow \mathrm{G}[\mathrm{T}, \mathrm{p}, \xi]$ Similarly, $\mathrm{U}[\mathrm{S}, \mathrm{V}, \xi] \rightarrow \mathrm{F}[\mathrm{T}, \mathrm{V}, \xi]$ Ledgendre transformations can be examined in the broad context of chemistry and biochemistry [5]. Their importance lies in establishing the general mathematical structure of thermodynamics [6]. Footnotes [1] A. M. Legendre; an eighteenth century mathematician. [2] C. Paus at http://web.mit.edu /8.21/www/ notes/notes/ node7.html [3] B. Callen, Thermodynamics, Wiley, New York,1961. [4] E. Grunwald, Thermodynamics of Molecular Species, Wiley, New York , 1997. [5] R. A. Alberty, Chem. Revs.,1994,94,1457. [6] D. Kondepudi and I. Prigogine, Modern Thermodynamics, Wiley, New York,1998. 1.14.13: Closed System A closed system is effectively a closed reaction vessel. As chemists we are interested in changes in chemical composition of the closed system. The condition “closed” means that while observing the processes taking place inside the system, we do not add more chemical substances to the system from the surroundings or remove chemical substance from the system into the surroundings. Actually the thermodynamic treatment of closed systems is simpler than for other systems. The system and surroundings interact by virtue of, for example, heat passing between system and surroundings.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.11%3A_Guggenheim-Scatchard_Equation_/_Redlich-Kister_Equation.txt
The molar enthalpy of vaporisation $\Delta_{\text {vap }} \mathrm{H}^{*}$ is the change in enthalpy for one mole of chemical substance $j$ on going from the liquid to the (perfect) gaseous state. The properties of a given liquid-$j$ are determined by $j-j$ intermolecular forces. By definition, there are no intermolecular forces in a perfect gas. Hence $\Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}(\ell)$ offers an insight into the strength of intermolecular forces in the liquid state. We have to be careful not to use the word ‘energy’. By definition enthalpy $\mathrm{H}$ equals ($\mathrm{U} + \mathrm{p} \, \mathrm{V}$). For phase I at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, $\mathrm{U}_{\mathrm{j}}^{*}(\mathrm{I})=\mathrm{H}_{\mathrm{j}}^{*}(\mathrm{I})-\mathrm{p} \, \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{I})$ $\text { Similarly for phase II, } \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{II})=\mathrm{H}_{\mathrm{j}}^{*}(\mathrm{II})-\mathrm{p} \, \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{II})$ $\text { Hence, } \quad U_{j}^{*}(\text { II })-U_{j}^{*}(I)=H_{j}^{*}(\text { II })-H_{j}^{*}(I)-p \,\left[V_{j}^{*}(\text { II })-V_{j}^{*}(I)\right]$ If phase II is a perfect gas and phase I is the corresponding liquid, $\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{II})>>\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{I})$; for one mole of chemical substance $j$, $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}^{*}(\mathrm{II})=\mathrm{R} \, \mathrm{T}$. $\text { Consequently } \Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T})=\Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}(\mathrm{~T})-\mathrm{R} \, \mathrm{T}$ $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T})$ is the molar thermodynamic energy of vaporisation for liquid $j$ at temperature $\mathrm{T}$. Having calculated $\Delta_{\text {vap }} \mathrm{H}_{j}^{*}(\mathrm{~T})$ from experimental data we obtain $\Delta_{\text {vap }} U_{j}^{*}(T)$, a measure of the strength of inter-molecular interactions in the liquid. The differential quantity $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}}$ defines the internal pressure $\pi_{int}(j)$ of chemical substance $j$. For liquid $j$, $\pi_{\mathrm{int}}^{*}(\ell ; \mathrm{j})=\left[\partial \mathrm{U}_{\mathrm{j}}^{*}(\ell) / \partial \mathrm{V}_{\mathrm{j}}^{*}(\ell)\right]_{\mathrm{T}}$ The internal pressure for liquids, of the order $10^{8} \mathrm{~Pa}$, is an indicator of the strength of intermolecular forces [1]. The structure of the terms in equation (e) prompts a slight rewrite using properties that are either readily measured or calculated, namely $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T})$ and the molar volume of the liquid at temperature $\mathrm{T}$, $\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~T})$ [2,3]. The result is the cohesive energy density, c.e.d., a measure of the cohesion within a liquid. $\text { By definition, c.e.d. }=\Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T}) / \mathrm{V}_{\mathrm{j}}^{*}(\ell)$ Intuitively, $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T})$ is a measure of cohesive interactions in the liquid whereas volume is a measure of the repulsive interactions, keeping the molecules in the liquid apart. At constant $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{j}}^{*}(\mathrm{~T})$, c.e.d decreases with increase in molar volume; c.f. repulsion. But at constant $\mathrm{V}_{\mathrm{j}}^{*}(\mathrm{~T})$, c.e.d. increases with increase in $\Delta_{\text {vap }} \mathrm{U}_{j}^{*}(\mathrm{~T})$, the attractive part. $\text { If the vapour is a perfect gas, c.e.d. }=\left[\Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}(\ell)-\mathrm{R} \, \mathrm{T}\right] / \mathrm{V}_{\mathrm{j}}^{*}(\ell)$ If the molar mass of the liquid $j$ equals $\mathrm{M}_{j}$ and the density equals $\rho_{\mathrm{j}}^{*}(\ell)$ $\text { c.e.d. }=\left[\Delta_{\mathrm{vap}} \mathrm{H}_{\mathrm{j}}^{*}(\ell)-\mathrm{R} \, \mathrm{T}\right] \, \rho_{\mathrm{j}}^{*}(\ell) / \mathrm{M}_{\mathrm{j}}$ $\mathrm{M}_{j}$ is expressed in $\mathrm{kg mol}^{-1}$ and $\rho_{\mathrm{j}}^{*}(\ell)$ in $\mathrm{kg m}^{-3}$, consistent with c.e.d. being expressed in ($\mathrm{J mol}^{–1} \mathrm{~m}^{–3}$). At $298.2 \mathrm{~K}$, $\mathrm{R} \, \mathrm{T}=2.48 \mathrm{~kJ} \mathrm{~mol}^{-1}$. The ratio of internal pressure $\pi_{\text {int }}(\mathrm{j})$ to c.e.d. defines a property $\mathrm{n}$ using equation (i). $\mathrm{n}=\left[\partial \mathrm{U}_{\mathrm{j}}^{*}(\ell) / \partial \mathrm{V}_{\mathrm{j}}^{*}(\ell)\right]_{\mathrm{T}} /\left[\Delta_{\mathrm{vap}} \mathrm{U}_{\mathrm{j}}^{*}(\ell) / \mathrm{V}_{\mathrm{j}}^{*}(\ell)\right]$ The dimensionless ratio $\mathrm{n}$ has been used to comment on the strength of intermolecular forces in a liquid [4]. In the context of the properties of liquid mixtures, using the definition of enthalpy $\mathrm{H}(=\mathrm{U}+\mathrm{p} \, \mathrm{V})$ we can write the following equation for a given phase I containing $\mathrm{n}_{1}$ moles of substance 1 and $\mathrm{n}_{2}$ moles of substance 2. $\mathrm{U}\left(\mathrm{I}, \mathrm{n}_{1}+\mathrm{n}_{2}\right)=\mathrm{H}\left(\mathrm{I} ; \mathrm{n}_{1}+\mathrm{n}_{2}\right)-\mathrm{p} \, \mathrm{V}\left(\mathrm{I} ; \mathrm{n}_{1}+\mathrm{n}_{2}\right)$ We assert that phase I is an ideal binary liquid mixture. Then, \begin{aligned} &\mathrm{U}\left(\mathrm{I} ; \mathrm{n}_{1}+\mathrm{n}_{2} ; \mathrm{mix} ; \mathrm{id}\right)= \ &\quad \mathrm{n}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{n}_{2} \, \mathrm{H}_{2}^{*}(\ell)-\mathrm{p} \, \mathrm{V}\left(\mathrm{I} ; \mathrm{n}_{1}+\mathrm{n}_{2} ; \text { mix } ; \mathrm{id}\right) \end{aligned} We assert that phase II is a perfect gas comprising $\mathrm{n}_{1}$ moles of substance 1 and $\mathrm{n}_{2}$ moles of substance 2. Then \begin{aligned} \mathrm{U}\left(\mathrm{II} ; \mathrm{n}_{1}+\right.&\left.\mathrm{n}_{2} ; \mathrm{pfg}\right)=\ & \mathrm{n}_{1} \, \mathrm{H}_{1}^{*}(\mathrm{pfg})+\mathrm{n}_{2} \, \mathrm{H}_{2}^{*}(\mathrm{pfg})-\mathrm{p} \, \mathrm{V}\left(\mathrm{II} ; \mathrm{n}_{1}+\mathrm{n}_{2} ; \mathrm{pfg}\right) \end{aligned} $\text { For a perfect gas, } \mathrm{p} \, \mathrm{V}\left(\mathrm{II} ; \mathrm{n}_{1}+\mathrm{n}_{2} ; \mathrm{pfg}\right)=\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right) \, \mathrm{R} \, \mathrm{T}$ We express the thermodynamic energy of vaporisation for ($\mathrm{n}_{1} + \mathrm{n}_{2}$) moles passing from phase I to phase II. \begin{aligned} &\Delta_{\text {vap }} \mathrm{U}\left(\mathrm{id}, \mathrm{n}_{1}+\mathrm{n}_{2}\right)= \ &\begin{aligned} \mathrm{n}_{1} \, \Delta_{\text {vap }} \mathrm{H}_{1}^{*}+\mathrm{n}_{2} \, \Delta_{\mathrm{vap}} \mathrm{H}_{2}^{*}-\left(\mathrm{n}_{1}\right.&\left.+\mathrm{n}_{2}\right) \, \mathrm{R} \, \mathrm{T} \ &+\mathrm{p} \, \mathrm{V}\left(\mathrm{I} ; \mathrm{n}_{1}+\mathrm{n}_{2} ; \text { mix; id }\right) \end{aligned} \end{aligned} Therefore for one mole, $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{m}}(\mathrm{id})=\mathrm{x}_{1} \, \Delta_{\text {vap }} \mathrm{H}_{1}^{*}+\mathrm{x}_{2} \, \Delta_{\text {vap }} \mathrm{H}_{2}^{*}-\mathrm{R} \, \mathrm{T}+\mathrm{p} \, \mathrm{V}_{\mathrm{m}}(\mathrm{I} ; \mathrm{mix} ; \mathrm{id})$ Suppose however that the thermodynamic properties of the liquid mixture are not ideal. We rewrite equation (k) in the following form (for one mole of mixture) where $\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}$ and $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ are the excess molar enthalpies and excess molar volumes of mixing. \begin{aligned} &\mathrm{U}_{\mathrm{m}}(\mathrm{I}, \operatorname{mix})= \ &\quad\left[\mathrm{x}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{H}_{2}^{*}(\ell)+\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right]-\mathrm{p} \,\left[\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell)+\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}\right] \end{aligned} $\operatorname{Or} \mathrm{U}_{\mathrm{m}}(\mathrm{I}, \operatorname{mix})=\left[\mathrm{x}_{1} \, \mathrm{H}_{1}^{*}(\ell)+\mathrm{x}_{2} \, \mathrm{H}_{2}^{*}(\ell)+\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right]-\mathrm{p} \, \mathrm{V}_{\mathrm{m}}(\operatorname{mix})$ Therefore the molar thermodynamic energy of vaporisation on going from the real mixture to the perfect gas in given by equation (r). $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{m}}=\left[\mathrm{x}_{1} \, \Delta_{\text {vap }} \mathrm{H}_{1}^{*}(\mathrm{~T})+\mathrm{x}_{2} \, \Delta_{\text {vap }} \mathrm{H}_{2}^{*}(\mathrm{~T})-\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right]-\mathrm{R} \, \mathrm{T}+\mathrm{p} \, \mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ The cohesive energy density, c.e.d., for a real binary liquid mixture is given by equation (s). \begin{aligned} \text { c.e.d. }=\left\{\left[\mathrm{x}_{1} \, \Delta_{\mathrm{vap}} \mathrm{H}_{1}^{*}(\mathrm{~T})+\mathrm{x}_{2} \,\right.\right.&\left.\left.\Delta_{\mathrm{vap}} \mathrm{H}_{2}^{*}(\mathrm{~T})-\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}\right] / \mathrm{V}_{\mathrm{m}}(\mathrm{mix})\right\} \ &-\left\{\mathrm{R} \, \mathrm{T} / \mathrm{V}_{\mathrm{m}}(\mathrm{mix})\right\}+\mathrm{p} \end{aligned} The c.e.d. for a given binary mixture is given by the molar enthalpies of vaporisation of the pure components, the excess molar enthalpy of mixing and the molar volume of the mixture. For the corresponding ideal binary mixture, c.e.d.(id) is given by equation (t). \text { c.e.d.(id) } \begin{aligned} =\left\{\left[\mathrm{x}_{1} \, \Delta_{\text {vap }} \mathrm{H}_{1}^{*}(\mathrm{~T})+\mathrm{x}_{2} \,\right.\right.&\left.\left.\Delta_{\text {vap }} \mathrm{H}_{2}^{*}(\mathrm{~T})\right] / \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})\right\} \ &-\left\{\mathrm{R} \, \mathrm{T} / \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})\right\}+\mathrm{p} \end{aligned} The difference between $\Delta_{\text {vap }} \mathrm{U}_{\mathrm{m}} / \mathrm{V}_{\mathrm{m}}(\mathrm{mix})$ and $\Delta_{\mathrm{vap}} \mathrm{U}_{\mathrm{m}} / \mathrm{V}_{\mathrm{m}}(\mathrm{mix} ; \mathrm{id})$ is the excess cohesive energy density, $(\text { c.e.d. })^{\mathrm{E}}$. The sign of $(\text { c.e.d. })^{\mathrm{E}}$ is controlled to a significant extent by the excess molar volume $\mathrm{V}_{\mathrm{m}}^{\mathrm{E}}$ and the excess molar enthalpy $\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}$ [6]. In fact equations (s) and (t) lead to m equation (u). $(\text { c.e.d. })^{\mathrm{E}}=\frac{-\left[\mathrm{x}_{1} \, \Delta_{\mathrm{vap}} \mathrm{H}_{1}^{*}(\mathrm{~T})+\mathrm{x}_{2} \, \Delta_{\mathrm{vap}} \mathrm{H}_{2}^{*}(\mathrm{~T})-\mathrm{R} \, \mathrm{T}\right]}{\mathrm{V}_{\mathrm{m}}(\operatorname{mix}) \, \mathrm{V}_{\mathrm{m}}(\operatorname{mix} ; \mathrm{id})}-\frac{\mathrm{H}_{\mathrm{m}}^{\mathrm{E}}}{\mathrm{V}_{\mathrm{m}}(\mathrm{mix})}$ Recently the cohesive energy density of a liquid has been described as a ‘solvation pressure’ acting on, for example, ethanol in ethanol + water and ethanol + trichloromethane liquid mixtures [7]. Footnotes [1] M. R. J. Dack, Aust. J. Chem.,1976,27,779. [2] $\text { c.e.d. }=\left[\mathrm{J} \mathrm{mol}^{-1}\right] /\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]=\left[\mathrm{J} \mathrm{m}^{-3}\right]=\left[\mathrm{N} \mathrm{m}^{-2}\right]$; the unit of pressure. $\mathrm{R} \, \mathrm{T}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]=\left[\mathrm{J} \mathrm{mol}^{-1}\right]$ [3] A variety of units are used for cohesive energy densities. Despite the fact that there are good grounds for using the unit $\mathrm{J m}^{-3}$, the commonly used unit is calories per $\mathrm{cm}^{3}$, \mathrm{cal cm}^{-3}\). \begin{aligned} &\text { For liquids } 298.15 \mathrm{~K} \text { and ambient pressure. }\ &\begin{array}{lc} \text { Liquid } & \text { c.e.d./ } \mathrm{cal} \mathrm{cm}^{-3} \ \text { water } & 547 \ \text { methanol } & 204 \ \text { benzene } & 85 \ \text { tetrachloromethane } & 74 \end{array} \end{aligned} [4] A. F. M. Barton, J.Chem.Educ.,1971, 48,156. [5] For comments on the role of cohesive energy densities of solvents and rates of disproportionation, see A.P. Stefani, J. Am. Chem. Soc., 1968,90,1694. [6] For comments on cohesive energy densities of binary aqueous mixtures see, 1. D. D. Macdonald and J. B. Hyne, Can. J. Chem.,1971, 49,611,2636. 2. D. D. Macdonald, Can. J. Chem.,1976, 54,3559. [7] N.W. A.van Uden, H. Hubel, D. A. Faux, A. C. Tanczos, B. Howlin and D. J. Dunstan, J. Phys.: Condens. Mater,2003,15,1577.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.14%3A_Cohesive_Energy_Density.txt
A given aqueous solution is prepared using $\mathrm{n}_{1}^{0}$ moles of water and $\mathrm{n}_{\mathrm{A}}^{0}$ moles of a weak acid $\mathrm{HA}$. The composition of the solution at equilibrium (at fixed $\mathrm{T}$ and $\mathrm{p}$) is described as follows. $\mathrm{HA}(\mathrm{aq})$ $\leftrightarrow$ $\mathrm{H}^{+}(\mathrm{aq}) +$ $A^{\prime}(\mathrm{aq})$ At $t=0$ $n_{A}^{0}$   $0$ $0 \mathrm{~mol}$ At equilibirium, $n_{A}^{0}-\xi^{\mathrm{eq}}$   $\xi^{\mathrm{eq}}$ $\xi^{\mathrm{eq}}\mathrm{~mol}$ or, $\mathrm{n}_{\mathrm{A}}^{0} \,\left[1-\frac{\xi^{e q}}{\mathrm{n}_{\mathrm{A}}^{0}}\right]$   $\xi^{\mathrm{eq}}$ $\xi^{\mathrm{eq}} \mathrm{~mol}$ or. $\mathrm{n}_{\mathrm{A}}^{0} \,(1-\alpha)$   $\alpha \, \mathrm{n}_{\mathrm{A}}^{0}$ $\alpha \, \mathrm{n}_{\mathrm{A}}^{0}\mathrm{~mol}$ If the volume of the system is $1 \mathrm{~dm}^{3}$ then, $\mathrm{c}_{\mathrm{A}}^{0} \,(1-\alpha) \quad \alpha \, \mathrm{c}_{\mathrm{A}}^{0} \quad \alpha \, \mathrm{c}_{\mathrm{A}}^{0} \quad \mathrm{~mol} \mathrm{dm}^{-3}$ By definition, the degree of dissociation, $\alpha=\xi^{\mathrm{eq}} / \mathrm{n}_{\mathrm{A}}^{0}$; $\alpha$ is an intensive variable describing the ‘degree’ of dissociation. If the total volume of the solution is $\mathrm{V}$, the concentration $\mathrm{c}_{\mathrm{A}}^{0}=\mathrm{n}_{\mathrm{A}}^{0} / \mathrm{V}$. If the thermodynamic properties of the solution are ideal, the composition of the solution can be described by an equilibrium acid dissociation constant $\mathrm{K}_{\mathrm{A}}$. $\mathrm{K}_{\mathrm{A}}=\alpha^{2} \, \mathrm{c}_{\mathrm{A}}^{0} /(1-\alpha)$ If $1-\alpha \cong 1, \alpha^{2}=\mathrm{K}_{\mathrm{A}} / \mathrm{c}_{\mathrm{A}}^{0}$ If the acid is dibasic, the analysis is a little more complicated. For an acid $\mathrm{H}_{2}\mathrm{A}$, there are two extents of acid dissociation. $\mathrm{H}_{2}\mathrm{A}$ $\rightleftarrows$ $\mathrm{H}^{+} +$ \mathrm{HA}^{-}\) At $t=0$, $\mathrm{n}_{\mathrm{A}}^{0}$   $0$ $0 \mathrm{~mol}$ At equilibrium, $\mathrm{n}_{\mathrm{A}}^{0}-\xi_{1}$   $\xi_{1}+\xi_{2}$ $\xi_{1}-\xi_{2}\mathrm{~mol}$ Or, $n_{A}^{0} \,\left[1-\frac{\zeta_{1}}{n_{A}^{0}}\right]$   $\mathrm{n}_{\mathrm{A}}^{0} \,\left[\frac{\xi_{1}}{\mathrm{n}_{\mathrm{A}}^{0}}+\frac{\xi_{2}}{\mathrm{n}_{\mathrm{A}}^{0}}\right]$ $\mathrm{n}_{\mathrm{A}}^{0} \,\left[\frac{\xi_{1}}{\mathrm{n}_{\mathrm{A}}^{0}}-\frac{\xi_{2}}{\mathrm{n}_{\mathrm{A}}^{0}}\right]\mathrm{~mol}$ By definition $\mathrm{c}_{\mathrm{A}}^{0}=\mathrm{n}_{\mathrm{A}}^{0} / \mathrm{V}$ where $\mathrm{V}$ is the volume of solution expressed in \mathrm{dm}^{3}\). Also by definition $\alpha_{1}=\xi_{1} / \mathrm{n}_{\mathrm{A}}^{0}$ and $\alpha_{2}=\xi_{2} / \mathrm{n}_{\mathrm{A}}^{0}$ Hence from equation (d) $\mathrm{H}_{2}\mathrm{A}$ $\rightleftarrows$ $\mathrm{H}^{+} +$ \mathrm{HA}^{-}\) At equilibrium, $\left(\mathrm{n}_{\mathrm{A}}^{0} / \mathrm{V}\right) \,\left[1-\alpha_{1}\right]$   $\left(\mathrm{n}_{\mathrm{A}}^{0} / \mathrm{V}\right) \,\left[\alpha_{1}+\alpha_{2}\right]$ $\left(\mathrm{n}_{\mathrm{A}}^{0} / \mathrm{V}\right) \,\left[\alpha_{1}-\alpha_{2}\right]\mathrm{~mol}$ Or, $\mathrm{c}_{\mathrm{A}}^{0} \,\left[1-\alpha_{1}\right]$   $\mathrm{c}_{\mathrm{A}}^{0} \,\left[\alpha_{1}+\alpha_{2}\right]$ $\mathrm{c}_{\mathrm{A}}^{0} \,\left[\alpha_{1}-\alpha_{2}\right]\mathrm{~mol}$ $\mathrm{HA}^{-}$ $\rightleftarrows$ $\mathrm{H}^{+} +$ \mathrm{A}^{-2}\) At $t=0$, $0$   $0$ $0 \mathrm{~mol}$ Also At equilibrium, $\xi_{1}-\xi_{2}$   $\xi_{1}+\xi_{2}$ $\xi_{2}\mathrm{~mol}$ Or, $\mathrm{n}_{\mathrm{A}}^{0} \,\left[\frac{\xi_{1}}{\mathrm{n}_{\mathrm{A}}^{0}}-\frac{\xi_{2}}{\mathrm{n}_{\mathrm{A}}^{0}}\right]$   $\mathrm{n}_{\mathrm{A}}^{0} \,\left[\frac{\xi_{1}}{\mathrm{n}_{\mathrm{A}}^{0}}+\frac{\xi_{2}}{\mathrm{n}_{\mathrm{A}}^{0}}\right]$ $n_{A}^{0} \,\left[\frac{\zeta_{2}}{n_{A}^{0}}\right]\mathrm{~mol}$ $c_{\mathrm{A}}^{0} \,\left[\alpha_{1}-\alpha_{2}\right]$   $\mathrm{c}_{\mathrm{A}}^{0} \,\left[\alpha_{1}+\alpha_{2}\right]$ $\mathrm{c}_{\mathrm{A}}^{0} \, \alpha_{2}$ Total amount of $\mathrm{H} in the system $=2 \,\left(\mathrm{n}_{\mathrm{A}}^{0}-\xi_{1}\right)+\xi_{1}+\xi_{2}+\xi_{1}-\xi_{2}=2 \, \mathrm{n}_{\mathrm{A}}^{0}$ Total amount of \(\mathrm{A}$ in the system $=\mathrm{n}_{\mathrm{A}}^{0}-\xi_{1}+\xi_{1}-\xi_{2}+\xi_{2}=\mathrm{n}_{\mathrm{A}}^{0}$ If the thermodynamic properties of the solution are ideal, $\mathrm{K}_{1}=\mathrm{c}_{\mathrm{A}}^{0} \,\left[\alpha_{1}+\alpha_{2}\right] \,\left[\alpha_{1}-\alpha_{2}\right] /\left[1-\alpha_{1}\right]$ If $\mathrm{K}_{2}=0, \alpha_{2}=0, \mathrm{~K}_{1}=\mathrm{c}_{\mathrm{A}}^{0} \, \alpha_{1}^{2} /\left(1-\alpha_{1}\right)$ But $\mathrm{K}_{2}=\left(\alpha_{1}+\alpha_{2}\right) \, \alpha_{2} \, \mathrm{c}_{\mathrm{A}}^{0} /\left(\alpha_{1}-\alpha_{2}\right)$ 1.14.16: Energy and First Law of Thermodynamics A central axiom of chemical thermodynamics is that a given system has a property called energy. In fact the First Law of Thermodynamics centres on the concept of energy. In its broadest terms, the law requires that the energy of the universe is constant [1]. This is a rather overwhelming statement. A more attractive statement is that the thermodynamic energy $\mathrm{U}$ of a typical chemistry laboratory is constant. $\mathrm{U} = \text{ constant} \label{a}$ The latter is the principle of conservation of energy; energy can be neither created nor destroyed. A chemist ‘watches’ energy “move” between system and surroundings. As a consequence of Equation \ref{a} we state that, $\Delta \mathrm{U}(\text { system })=-\Delta \mathrm{U}(\text { surroundings }) \label{b}$ We cannot know the actual energy $\mathrm{U}$ of a closed system although we agree that it is an extensive property of a system. In describing the energy changes we need a convention. We use the acquisitive convention, describing all changes in terms of how the system is affected. Thus $\Delta \mathrm{U} < 0$, means that the energy of the system falls whereas $\Delta \mathrm{U} > 0$ means that the energy increases [2]. In the context of chemistry, chemists agree that the energy of a given closed system can be increased in two ways: 1. heat $\mathrm{q}$ passing from the surroundings into the system, and 2. work $\mathrm{w}$ done by the surroundings on the system. In a wider context the concept of energy is linked with the First Law of Thermodynamics which is based on the following axiom. $\Delta \mathrm{U}=\mathrm{q}+\mathrm{w} \label{c}$ As it stands the symbols $\mathrm{U}$, $\mathrm{q}$ and $\mathrm{w}$ seem rather uninformative. It is the task of chemists to flesh out the meaning of these terms. If only ‘$\mathrm{p}-\mathrm{V}$’ work is involved, $\mathrm{w}=-\mathrm{p} \, \mathrm{dV} \label{d}$ The point of Equation \ref{c} is to separate the work term from the heat term. The significance for chemists is that $\mathrm{q}$ links to the Second Law of Thermodynamics. Thus chemists know that heat flows spontaneously from high to low temperatures. This concept of ‘spontaneous change’ is picked up with enormous impact in the second law. Footnotes [1] Peter Atkins (Galileo’s Finger, Oxford University press, 2003, page 107) speculates that the total energy of the universe ‘may be exactly zero’. [2] In principle it is possible to calculate the total energy of a given system using a scale in conjunction with Einstein’s famous equation, $\mathrm{E}=\mathrm{m} \, \mathrm{c}^{2}$. However the mass corresponding to $1 \mathrm{~kJ}$ is only about $10^{-14} \mathrm{~kg}$.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.15%3A_Degree_of_Dissociation.txt
This Topic takes a rather different approach from the other Topics in this Notebook. Across the galaxy of terms used in thermodynamics, two terms stand out, namely Energy and Entropy. With respect to a given closed system, both terms describe extensive properties, using the letter $\mathrm{U}$ to identify energy and the letter $\mathrm{S}$ to identify entropy. The term ‘energy’ is used quite generally in everyday life. One dictionary [1] describes energy as ‘the power and ability to be physically active’. Perhaps we might not be too happy at the use of the term ‘power’ in this context on the grounds that this term is normally linked to the rate at which energy is supplied [2]. Indeed in every day life we refer to powerful engines in, for example, fast sports cars. Nevertheless in a thermodynamic context the concepts of energy and energy change are part of the language of chemistry; e.g. bond energy, energy of activation, radiant energy… The First Law of Thermodynamics formalises the concept of energy change using the following ostensibly simple equation. $\Delta \mathrm{U}=\mathrm{q}+\mathrm{w}$ Here $\mathrm{U}$ is the thermodynamic energy, a function of state; $\Delta \mathrm{U}$ describes the increase in thermodynamic energy of a closed system when heat $\mathrm{q}$ flows from the surroundings into a given system and work w is done by the surroundings on that system. The distinction between heat and work is crucial. There are many ways in which the surroundings can do work on a system. Caldin [3] lists many examples in which work is given by the product of Intensity and Capacity Factors; e.g. intensity factor pressure, $\mathrm{p}$ and capacity factor volume, $\mathrm{V}$ such that $\mathrm{w}=\mathrm{p} \, \mathrm{dV}$. In the context of energy, chemical thermodynamics quite generally describes quite modest changes in energy. Even in the case of an explosion involving, for example, ignition of a mixture of hydrogen and oxygen gases, the energy change turns out to involve transitions between electronic energy levels in atoms and molecules. Much more dramatic are nuclear reactions which involve the conversion of mass, $\mathrm{m}$ into energy $\mathrm{E}$ as described by Einstein’s famous equation. $\mathrm{E}=\mathrm{m} \, \mathrm{c}^{2}$ Here $\mathrm{c}$ is the speed of light, $3.00 \times 10^{8} \mathrm{~m s}^{-1}$. In nuclear fission the nucleus of an atom breaks into two smaller nuclei of similar mass [4]. Thus $\text{uranium}^{235}$ nuclei bombarded by neutrons split into barium-142 and krypton-92 nuclei. Einstein’s equation shows that $1.0 \mathrm{~g}$ of this uranium isotope undergoes fission with the release of $7.5 \times 10^{10} \mathrm{~J}$ [5], an awesome amount of energy. Here we return to the domain of chemical properties and chemical reactions where nuclei of atoms are not destroyed. Our interest centres on the thermodynamic variable, entropy $\mathrm{S}$, an extensive function of state. However in every day conversation and in articles in newspapers and magazines the term ‘entropy’ is rarely used suggesting that it is not important. This conclusion is incorrect and the message quite misleading. In these Topics we describe the Second Law using an equation based on the formulation given by Clausius as follows. $\mathrm{T} \, \mathrm{dS}=\mathrm{q}+\mathrm{A} \, \mathrm{d} \xi$ Here a positive $\mathrm{q}$ describes heat passing from the surroundings into a closed system; $\mathrm{A}$ is the affinity for spontaneous change, the change being described by the property $\xi$. Following De Donder [6] as discussed by Prigogine and Defay [7] the Second Law is simply stated as follows. $A \, d \xi \geq 0$ For a system moving between equilibrium states (i.e. the system and surroundings are at all stages at equilibrium where $\mathrm{A}$ is zero), $\mathrm{T} \, \mathrm{dS}=\mathrm{q}$ Hence $\mathrm{q}$, measured using a calorimeter, is a direct measure of the change in entropy accompanying a change where the system is always in equilibrium with the surroundings. In fact this statement provides a useful answer to the question ‘what is entropy?’. There is therefore a fundamental link between the two quantities $\mathrm{dS}$ and heat $\mathrm{q}$. Indeed we understand immediately the importance of calorimeters in thermodynamics. At the same time we understand the importance of chemical kinetics because this subject is built around equation (d) which in the basis of the Law of Mass Action. In summary we see how the two foundation stones of thermodynamics, namely energy and entropy, are formalised in two laws for which there are no exceptions. So we can end the Topics here. But chemists do not although there are new hazards. It follows from equation (d) that for a process where $\mathrm{q} < 0$, the entropy of the system decreases. [It is interesting to note that the unit of entropy $\mathm{J K}^{-1}$ is the same as that for heat capacity.] At this point we review the arguments advanced by Lewis and Randall [8]. An important reference system in thermodynamics is the perfect gas. No such gas is actually known in the real world but the concept is very valuable. The properties of a perfect gas conform to the two laws [7b]. We envisage a closed system, volume $\mathrm{V}$ containing n moles of a perfect gas. The first condition states that the thermodynamic energy $\mathrm{U}$ is only a function of temperature. Thus, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=0$ The second condition states that the following equation relates the pressure, volume and temperature of $\mathrm{n}$ moles of a perfect gas. $\mathrm{p} \, \mathrm{V}=\mathrm{n} \, \mathrm{R} \, \mathrm{T}$ Thus for one mole of a perfect gas, having molar volume $\mathrm{V}_{\mathrm{m}}$, $\mathrm{p} \, \mathrm{V}_{\mathrm{m}}=\mathrm{R} \, \mathrm{T}$ A key equation (Topic 2500) relates the change in thermodynamic energy $\mathrm{U}$ to the changes in entropy, volume and composition. Thus $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi$ For an equilibrium transformation, the affinity of spontaneous change is zero. Hence for an equilibrium process, $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}$ For 1 mole of an ideal gas, $\mathrm{dU}_{\mathrm{m}}=\mathrm{T} \, \mathrm{dS}_{\mathrm{m}}-\mathrm{p} \, \mathrm{dV}_{\mathrm{m}}$ Or, $\mathrm{dS}_{\mathrm{m}}=\frac{1}{\mathrm{~T}} \, \mathrm{dU}_{\mathrm{m}}+\frac{\mathrm{P}}{\mathrm{T}} \, \mathrm{dV} \mathrm{m}_{\mathrm{m}}$ Molar isochoric heat capacity $\mathrm{C}_{\mathrm{Vm}}$ is related to $\mathrm{dU}_{\mathrm{m}}$ by equation (m). $\mathrm{C}_{\mathrm{V}_{\mathrm{m}}}=\left(\partial \mathrm{U}_{\mathrm{m}} / \partial \mathrm{T}\right)_{\mathrm{V}(\mathrm{m})}$ Then, $\mathrm{dS}_{\mathrm{m}}=\frac{\mathrm{C}_{\mathrm{Vm}^{\mathrm{m}}}}{\mathrm{T}} \, \mathrm{dT}+\frac{\mathrm{p}}{\mathrm{T}} \, \mathrm{dV} \mathrm{V}_{\mathrm{m}}$ We define the molar entropy of an ideal gas using equation (o). $\mathrm{S}_{\mathrm{m}}=\mathrm{S}_{\mathrm{m}}\left[\mathrm{T}, \mathrm{V}_{\mathrm{m}}\right]$ The total differential of equation (o) takes the following form. $\mathrm{dS}_{\mathrm{m}}=\left(\frac{\partial \mathrm{S}_{\mathrm{m}}}{\partial \mathrm{T}}\right)_{\mathrm{V}(\mathrm{m})} \mathrm{dT}+\left(\frac{\partial \mathrm{S}_{\mathrm{m}}}{\partial \mathrm{V}_{\mathrm{m}}}\right)_{\mathrm{T}} \mathrm{dV} \mathrm{V}_{\mathrm{m}}$ Comparison of equations (n) and (p) reveals the following relation. $\left(\frac{\partial S_{m}}{\partial V_{m}}\right)_{T}=\frac{p}{T}$ Hence using equation h), $\left(\frac{\partial S_{m}}{\partial V_{m}}\right)_{T}=\frac{R}{V_{m}}$ Thus at constant temperature, $\mathrm{dS}_{\mathrm{m}}=\mathrm{T} \, \mathrm{d} \ln \left(\mathrm{V}_{\mathrm{m}}\right)$ Hence the change in entropy for the isothermal expansion of an ideal gas between states where the volumes are $\mathrm{V}_{\mathrm{m}}(\mathrm{B})$ and $\mathrm{V}_{\mathrm{m}}(\mathrm{A})$ is given by equation (t). $S_{m}(B)-S_{m}(A)=R \, \ln \left[\frac{V_{m}(B)}{V_{m}(A)}\right]$ We turn now to a consideration of changes in entropy from a statistical point of view. A given experiment [8] uses two glass flasks of equal volumes connected by a glass tube which includes a tap, all at the same temperature $\mathrm{T}$. The system contains $\mathrm{N}$ gas molecules; e.g. oxygen. The gas molecules pass freely between the two flasks through the open tap. On examining the contents of the two flasks we would not be surprised to discover that there are equal numbers of the gas molecules in the two flasks. The probability of this results from experiment A is expressed by stating that $\mathrm{P}_{\mathrm{y}}^{\mathrm{A}}$ is unity. We return to the two flasks and close the tap. The probability that all the oxygen molecules are to be found in one flask is ${(1/2)}^{\mathrm{N}}$; i.e. a very low probability. If the total system contained only 20 molecules this probability signals a chance of 1 in $2^{20}$. Thus the probability $\mathrm{P}_{\mathrm{y}}^{\mathrm{B}}$ for experiment B is very small; effectively zero. An interesting exercise characterises these probabilities by a property $\sigma$. Then, $\sigma=\frac{R}{N} \, \ln \left(P_{Y}\right)$ Note that the auxiliary property $\sigma$ is generally negative because statistical probabilities vary between zero and unity. For the two experiments, $\sigma_{B}-\sigma_{A}=\frac{R}{N} \, \ln \left(\frac{P_{Y}^{B}}{P_{Y}^{A}}\right)$ Hence, $\sigma_{B}-\sigma_{A}=\frac{R}{N} \, \ln \left(\frac{(1 / 2)^{N}}{1}\right)$ Or, $\sigma_{\mathrm{B}}-\sigma_{\mathrm{A}}=-\mathrm{R} \, \ln (2)$ We can express this result in general terms describing the expansions of one mole of gas from volume $\mathrm{V}_{\mathrm{A}}$ to $\mathrm{V}_{\mathrm{B}}$. Then, $\sigma_{B}-\sigma_{A}=R \, \ln \left[V_{m}(B) / V_{m}(A)\right]$ At this point comparison between equations (t) and (y) is rewarding. Thus we may write the following equation. $\mathrm{S}_{\mathrm{m}}(\mathrm{B})-\mathrm{S}_{\mathrm{m}}(\mathrm{A})=\frac{\mathrm{R}}{\mathrm{N}} \,\left[\ln \left(\mathrm{P}_{\mathrm{Y}}^{\mathrm{B}}\right)-\ln \left(\mathrm{P}_{\mathrm{Y}}^{\wedge}\right)\right]$ In other words the difference between the entropies in the ideal gas state is related to a probability. Thus we might conclude that $\mathrm{S}_{\mathrm{m}}(\mathrm{B})$ is larger that $\mathrm{S}_{\mathrm{m}}(\mathrm{A})$ because there are more ways of arranging molecules in system B than in system A. The state with the more ordered arrangement is the state with the lower entropy. It is a small step (but a very dangerous step) to draw comparison between entropy and (if there is such a word) the muddled-up-ness of a given system. But these are treacherous waters and outside the province of the classic thermodynamics which form the basis of the Topics. Indeed strong feelings are aroused. McGlashan [9,10], for example, takes to task chemists who assume that an increase in entropy implies an increase of disorder or of randomness or of ‘mixed-upness’. We leave the debate here except to note that both authors of these Topics favour the view advanced by McGlashan [9,10] although this view would not a win a popularity contest. But ‘popularity’ is not an acceptable criterion in thermodynamics. Footnotes [1] Cambridge International Dictionary of English, Cambridge University Press, Cambridge, 1995. [2] P. W. Atkins, Concepts in Physical Chemistry, Oxford University Press, Oxford, 1995. [3] E. F. Caldin, An Introduction to Chemical Thermodynamics, Oxford University Press, Oxford, 1958. [4] S. Glasstone, Sourcebook of Atomic Energy, MacMillan, London, 1954. [5] P. W. Atkins and L. Jones, Chemistry: Molecules, Matter and Change, W H Freeman, New York, 3rd edition, 1997, p.875. [6] Th. de Donder, Bull. Acad. Roy. Belg. (Cl.Sc), 1922, 7, 197, 205. [7] I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans Green, London, 1954, (a) chapter 3; (b) chapter 4. [8] G. N. Lewis and M. Randall, Thermodynamics , McGraw-Hill, New York, 1923, chapter VI. [9] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979,pages 112-113; [10] M.L. McGlashan, J. Chem. Educ.,1966,43,226.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.17%3A_Energy_and_Entropy.txt
Electric Current The SI base electrical unit is the AMPERE which is that constant electric current which if maintained in two straight parallel conductors of infinite length and of negligible circular cross section and placed a metre apart in a vacuum would produce between these conductors a force equal to $2 \times 10^{-7}$ newton per metre length. It is interesting to note that definition of the Ampere involves a derived SI unit, the newton. Except in certain specialised applications, electric currents of the order ‘amperes’ are rare. Starter motors in cars require for a short time a current of several amperes. When a current of one ampere passes through a wire about $6.2 \times 10^{18}$ electrons pass a given point in one second [1,2]. The coulomb (symbol $\mathrm{C}$) is the electric charge which passes through an electrical conductor when an electric current of one $\mathrm{A}$ flows for one second. Thus $[\mathrm{C}]=[\mathrm{As}]$ Electric Potential In order to pass an electric current thorough an electrical conductor a difference in electric potential must exist across the electrical conductor. If the energy expended by a flow of one ampere for one second equals one Joule the electric potential difference across the electrical conductor is one volt [3]. Electrical Resistance and Conductance If the electric potential difference across an electrical conductor is one volt when the electrical current is one ampere, the electrical resistance is one ohm, symbol $\Omega$ [4]. The inverse of electrical resistance , the conductance, is measured using the unit siemens, symbol [S]. [4] Ohm’s Law This famous phenomenological law describes the ability of a system to conduct electrical charge. This law describes the relationship between three properties of an electrical conductor; e.g. a salt solution. The three properties are 1. electric current $\mathrm{I}$ described using the SI unit , ampere, symbol $\mathrm{A}$; 2. electric potential difference $\mathrm{V}$ across the electrical conductor using the SI unit , volt, symbol $\mathrm{V}$; and Ohm’s Law is a phenomenological law in that it describes the phenomenon of electrical conductivity. Unlike the laws of thermodynamics, there are exceptions to Ohm’s law for very high voltages and high frequency alternating electric currents. The (electrical) conductance $\mathrm{G}$ of a system is given by the inverse of its resistance $\mathrm{R}$. The conductivity $\kappa$ ($\equiv \sigma$) of a system is given by equation (c) [5]. $\mathrm{j}=\kappa \, \mathrm{E}$ Here $\mathrm{j}$ is the electric current density and $\mathrm{E}$ is the electric field strength. Chemists prefer to think in terms the charge carrying properties of a given system; i.e. the conductance $\mathrm{G}$ ($=1 / \mathrm{R}$) using the unit siemens, symbol S [5]. An interesting contrast often emerges between chemists and physicists, the latter seem to emphasise the property of ‘resistance’ whereas chemists are more interested in how systems transport electrical charge. Certainly the classic subject in chemistry concerns the electrical conductivities of salt solutions where the charge carriers are ions. The subject is complicated by the fact that there are two types of charge carriers in a given solution, cations and anions. Moreover the subject is further complicated by the fact that these charge carriers move in opposite directions with different velocities. In a solution containing a single salt the fraction of electric current carried by the cations and anions are called transport numbers; i.e. $\mathrm{t}_{+}$ and $\mathrm{t}_{-}$ respectively for cations and anions where $\mathrm{t}_{+}+\mathrm{t}_{-}=1$. Ionic Mobility A given aqueous salt solution, volume $\mathrm{V}$, is prepared using $\mathrm{n}_{1}$ moles of water and $\mathrm{n}_{j}$ moles of a salt $j$. A pair of electrodes is placed in the solution, $\mathrm{d}$ metres apart. An electric potential, $\mathrm{V}$ volts, is applied across the solution and an electric current $\mathrm{I}$ is recorded. An electric charge $\mathrm{Q}$ coulombs ($[\mathrm{A s}]$) is passed through the solution. The electric current (unit $[\mathrm{A}]$) is the rate of transport of charge, $\mathrm{dQ} / \mathrm{dt}$. The speed of ion $\mathrm{i}$ through the solution $\mathrm{v}_{\mathrm{i}}$ is given by the ratio of the distance travelled to the time taken. Thus $\mathrm{v}_{\mathrm{i}}$ is a measure of the distance travelled in one second; $\mathrm{v}=(\mathrm{a} / \mathrm{t})\left[\mathrm{ms}^{-1}\right]$. A more interesting property is the velocity of ion $\mathrm{i}$ in an electric field gradient measured using the ratio, volt/metre (Or, $\mathrm{V} / \mathrm{m}$). Thus electric mobility $\mathrm{u}_{\mathrm{i}}$ has the unit, $\left[\mathrm{m} \mathrm{s}^{-1} / \mathrm{V} \mathrm{m} \mathrm{m}^{-1}\right]=\left[\mathrm{v}_{\mathrm{i}} \mathrm{E}^{-1}\right]=\left[\mathrm{m}^{2} \mathrm{~s}^{-1} \mathrm{~V}^{-1}\right]$. The molar conductivity $\Lambda$ of a salt solution is given by the ratio, $\left(\kappa / c_{j}\right)\left\{=\left[S \mathrm{~m}^{2} \mathrm{~mol}^{-1}\right]\right\}$. In fact the majority of research publications describe the dependence of $\Lambda$ on the concentration of salt in a specified solvent at defined $\mathrm{T}$ and $\mathrm{p}$. For salt solutions both cations and anions contribute to $\Lambda$; the transport number of an ion $j$ describes the fraction of current carried by the $j$ ion. Thus $\mathrm{t}_{\mathrm{j}}=\left|\mathrm{z}_{\mathrm{j}}\right| \, \mathrm{c}_{\mathrm{j}} \, \mathrm{v}_{\mathrm{j}} / \sum\left|\mathrm{z}_{\mathrm{i}}\right| \, \mathrm{c}_{\mathrm{i}} \, \mathrm{v}_{\mathrm{i}}$. An electric current (i.e. a flow of electric charge) through a system is impeded by the electrical resistance. The ohm (symbol $\Omega$) is the unit of electrical resistance being the ratio of electric potential (unit = volt) to electric current (unit = ampere) Then, ohm = volt/ampere (d) In other words, Electrical resistance/ohm=[electric potential gradient /volt]/[electric current/ampere] . From equation (c) ohm (symbol $\Omega$) $=\mathrm{V} \mathrm{} \mathrm{A}^{-1}=\mathrm{m}^{2} \mathrm{~kg} \mathrm{~s}^{-3} \mathrm{~A}^{-2}$ The property (electrical) resistance is a measure of the impedance to the flow of electric charge. This somewhat negative outlook is not consistent with the attitude of chemists who are interested in the ‘mechanism’ by which a system conducts electrical charge. Chemists prefer to discuss the property of electric conductance rather than resistance. The conductance is measured using the unit siemen, symbol S [6]. A key component of electrical circuits is the electrical capacitance measured using the unit farad, symbol $\mathrm{F}$ [7]. If the composition of an electrical conductor is uniform, the electrical resistance is directly proportional to its length $\ell$ and inversely proportional to its cross sectional area, $\mathrm{a}$. The material forming the conductor is characterized by its resistivity, $\rho$. Thus[8] resistance, $\mathrm{R}=\rho \, \ell / \mathrm{a}$ Footnotes [1] The electric charge on an electron $=1.602 \times 10^{-19} \mathrm{C}=1.602 \times 10^{-19} \mathrm{~As}$. A given single wire carries a current of 1 ampere. Then in one second, the electric charge carried by that wire $= 1.602 \times 10^{-19} \mathrm{~C}$ For one coulomb to pass a given point, the number of electrons passing $=\frac{1}{1.602 \times 10^{-19}}$ In other words $6.24 \times 10^{18}$ electrons pass by. [2] P.W. Atkins and L. Jones, Chemistry; Molecules, Matter and Change, W. H. Freeman, New York, 1997,p.658. [3] $[\mathrm{V}] \equiv\left[\mathrm{kg} \mathrm{m}{ }^{2} \mathrm{~s}^{-3} \mathrm{~A}^{-1}\right]=\left[\mathrm{J} \mathrm{A}^{-1} \mathrm{~s}^{-1}\right]$ [4] $[\Omega]=\left[\mathrm{VA}^{-1}\right]=\left[\mathrm{S}^{-1}\right]$ [5] \begin{aligned} &\kappa=\left[\mathrm{S} \mathrm{m}^{-1}\right]\ &\left[\mathrm{A} \mathrm{} \mathrm{m}^{-2}\right]=\kappa \,\left[\mathrm{V} \mathrm{} \mathrm{m}^{-1}\right] \end{aligned} [6] $\mathbf{S}=\Omega^{-1}$ [7] $\mathrm{F}=\mathrm{A}^{2} \mathrm{~s}^{4} \mathrm{~kg}^{-1} \mathrm{~m}^{-2}=\mathrm{As} \mathrm{} \mathrm{V}^{-1}=\mathrm{C} \mathrm{V}^{-1}$ [8] $\begin{gathered} {[\mathrm{S}] \equiv\left[\Omega^{-1}\right] \equiv\left[\mathrm{A} \mathrm{V}^{-1}\right] \equiv\left[\mathrm{m}^{-2} \mathrm{~kg}^{-1} \mathrm{~s}^{3} \mathrm{~A}^{2}\right]} \ {[\Omega]=[\Omega \mathrm{m}] \,[\mathrm{m}] \,[\mathrm{m}]^{-2}} \end{gathered}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.18%3A_Electrochemical_Units.txt
In attempting to understand the properties of chemical substances, chemists divide chemistry into two parts. In one part, chemists are interested in understanding intramolecular forces which hold molecules together. For example, using quantum mechanics and associated theories of covalent bonding, chemists describe the cohesive forces holding carbon, hydrogen and nitrogen atoms together in cyanomethane, $\mathrm{CH}_{3}\mathrm{CN}$. At ambient temperature and pressure, cyanomethane is a liquid. In the second part of the sub-division of chemistry, chemists describe the intermolecular forces [1] which hold assemblies of molecules together in, for example, liquid and solid states; e.g. those forces which hold $\mathrm{CH}_{3}\mathrm{CN}$ molecules together in the liquid state. Common experience tells us that intermolecular forces are weaker than intramolecular forces. When we heat $\mathrm{CH}_{3}\mathrm{CN}(\ell)$ at ambient pressure, the liquid boils at a characteristic temperature to form a vapour. The intermolecular separation dramatically increases but the covalent bonds within $\mathrm{CH}_{3}\mathrm{CN}$ do not break. [Of course, these bonds break at very high temperatures - thermolysis.] Here the emphasis centres on intermolecular cohesion. But this cannot be the whole story. If cohesion is the only force operating, molecules would collapse into each other in some nuclear catastrophe. This does not happen. Opposing the forces of cohesion are repulsive forces. In fact everyday experience leads to the idea of "size"; 'size is repulsive'. Basic Physics Molecules contain charged particles; protons (with positive electric charge) and electrons (with negative electric charge-- by convention). Intermolecular forces are understandable in terms of equations describing electrical interactions between electrically charged particles. An SI base unit is the ampere; symbol = $\mathrm{A}$ [2]. The SI unit of electric charge is the coulomb (symbol = $\mathrm{C}) defined as \(\mathrm{A s} [3]. Electric Current An electric current \(\mathrm{I}$ is driven through an electrical resistance $\mathrm{R}$, by an electric potential gradient across the resistance. An ammeter measures the electric current $\mathrm{I}$. The voltmeter records the electric potential gradient, $\Delta \mathrm{E}$ across the resistance. The property called resistance $\mathrm{R}$ is given by Ohm’s Law; $\Delta \mathrm{E}=\mathrm{I} \, \mathrm{R}$ In the IUPAC system the unit of resistance is ohm [symbol $\Omega \equiv \mathrm{VA}^{-1}$ ]. The electric potential difference is measured in volts, symbol $\mathrm{V}$ [4]. Electrical Capacitance In a simple electric circuit, a small battery is connected across a parallel plate capacitance. No current flows in this circuit. The battery produces a set of equal in magnitude but opposite in sign electric charges on the two plates. A capacitance stores electric charge. In practice the extent to which a capacitance stores charge depends on the chemical substance between the two plates. This substance is characterised by its electric permittivity; symbol = $\varepsilon$. Where a vacuum exists between the two plates, the electric permittivity equals $\varepsilon_{0}$[5]. The permittivity of a liquid is measured by comparing capacitance $\mathrm{C}$ when the gap between the plates is filled with this liquid and with capacitance $\mathrm{C}_{0}$ when the gap is "in vacuo". Then $\varepsilon_{\mathrm{r}}=\varepsilon / \varepsilon_{0}=\mathrm{C} / \mathrm{C}_{0}$ or $\mathrm{C}=\varepsilon_{\mathrm{r}} \, \mathrm{C}_{\mathrm{o}}$ For all substances, $\varepsilon_{\mathrm{r}}$ is greater than unity. In other words, with increase in $\varepsilon_{\mathrm{r}}$ so the electrical insulating properties of the system increase. At this stage, we have not offered a molecular explanation of the properly called $\varepsilon_{\mathrm{r}}$ but we have indicated that $\varepsilon_{\mathrm{r}}$ be can measured [6,7]. Intermolecular Forces and Energies Molecule $i$ and molecule $j$ are separated by a distance $\mathrm{r}$; we assert that $\mathrm{r} >>$ molecular radii of molecules $i$ and $j$. Our discussion centres on the assertion that a force (symbol $\mathrm{X}$) exists between the two molecules. Moreover, this force depends on the distance of separation $\mathrm{r}$. Thus $X=f(r)$ Ion-Interactions The force $\mathrm{X}$ between two electric charges $\mathrm{q}_{1}$ and $\mathrm{q}_{2}$ distance $\mathrm{r}$ apart ‘in vacuo’ is given by equation (e); (Couloumb's Law) [8]. $\mathrm{X}=\mathrm{q}_{1} \, \mathrm{q}_{2} / 4 \, \pi \, \varepsilon_{0} \, \mathrm{r}^{2}$ Two ions, $i$ and $j$, have charge numbers $\mathrm{z}_{i}$ and $\mathrm{z}_{j}$ respectively [for $\mathrm{K}^{+}$, $\mathrm{z}_{\mathrm{j}}=+1$; for ${\mathrm{SO}_{4}}^{2-}$, $\mathrm{z}_{j} = -2$]. For two ions ‘in vacuo’, the interionic force is given by equation (f). $\mathrm{F}_{\mathrm{ij}}=\left(\mathrm{z}_{\mathrm{i}} \mathrm{e}\right) \,\left(\mathrm{z}_{\mathrm{j}} \mathrm{e}\right) / 4 \pi \, \varepsilon_{0} \, \mathrm{r}^{2}$ But pairwise potential energy, $\mathrm{U}_{i j}=-\int_{\mathrm{ij}=\infty}^{\mathrm{r}} \mathrm{F}_{\mathrm{ij}} \, \mathrm{dr}$ Hence, $U_{i j}=\left(z_{i} e\right) \,\left(z_{j} e\right) / 4 \pi \, \varepsilon_{0} \, r$ Equation (h) yields the interaction potential energy between a pair of ions [9]. The result is an energy expressed in joules. However, there are often advantages in considering an Avogadro number (i.e. a mole) of such pairwise interactions. $\left.\mathrm{U}_{\mathrm{ij}} / \mathrm{J} \mathrm{mol} \mathrm{mol}_{\mathrm{A}}^{-1}=\mathrm{N}_{\mathrm{i}} \mathrm{e}\right) \,\left(\mathrm{z}_{\mathrm{j}} \mathrm{e}\right) / 4 \pi \, \varepsilon_{0} \, \mathrm{r}$ We consider two classes of ion-ion interactions: 1. Ions $i$ and $j$ have the same sign For cation-cation and anion-anion pairwise interactions the force between the ions is repulsive. The pairwise potential energy increases with decrease in ion-ion separation. To bring two ions having the same charge closer together we have to do work on the system, increasing the pairwise potential energy $\mathrm{U}_{ij}$. 2. Ions $i$ and $j$ have opposite signs For this system, $\left(z_{i} \, Z_{j}\right)<0$. Ion-ion interaction is attractive and the potential energy $\mathrm{U}_{ij}$ decreases with decrease in $\mathrm{r}_{ij}$. We write $\left|z_{i} \, z_{j}\right|$ to indicate the modulus of the product of the charge numbers. $\mathrm{U}_{\mathrm{ij}} / \mathrm{J} \mathrm{mol}^{-1}=-\mathrm{N}_{\mathrm{A}} \,\left|\mathrm{Z}_{\mathrm{i}} \, \mathrm{Z}_{\mathrm{j}}\right| \mathrm{e}^{2} / 4 \pi \, \varepsilon_{0} \, \mathrm{r}$ Hence $\mathrm{U}_{ij}$ has a ($1/\mathrm{r}$) dependence on distance apart. Electric field strength, $\mathrm{E}$ is the force exerted on unit charge at the point in question [10]. At distance $\mathrm{r}$ from charge $\mathrm{q}$, $\mathrm{E}=\mathrm{q} / 4 \pi \, \varepsilon_{0} \, \mathrm{r}^{2}$ Solvent Effects An important topic in Chemistry concerns the effect of solvents on ion-ion interactions. Here we assume that solvents are characterised by their relative permittivities, $\varepsilon_{\mathrm{r}}$. In a solvent the pairwise cation-anion interaction energy is given by equation (l). $\mathrm{U}_{\mathrm{ij}} / \mathrm{J}=-\left|\mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}}\right| \mathrm{e}^{2} / 4 \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}$ $\mathrm{U}_{\mathrm{ij}} / \mathrm{J} \mathrm{mol} \mathrm{m}^{-1}=-\left|\mathrm{z}_{\mathrm{i}} \, \mathrm{Z}_{\mathrm{j}}\right| \, \mathrm{N}_{\mathrm{A}} \, \mathrm{e}^{2} / 4 \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{r}$ As commented above, $\varepsilon_{\mathrm{r}}$ is always greater than unity. Hence for a given system at fixed distance apart $\mathrm{r}$, $\mathrm{U}_{ij}$ increases (becomes less negative) with increase in $\varepsilon_{\mathrm{r}}$. With increase in $\varepsilon_{\mathrm{r}}$, the ions are increasingly insulated and so at given distance $\mathrm{r}$ the stabilisation of the cation-anion pair is less marked. Molecular Dipole Moments A given molecule comprises an assembly of positive and negative charges. Consider a point 0, distance $\mathrm{r}$ from this assembly. We are concerned with the electric field strength at point 0, a short distance from the dipole moment. In the previous section we assumed that this assembly is simply characterised by the electric charge (i.e. $z_{j} \, e$ for ion $j$). However, in those cases where the overall charge is zero, a measurable electric field is detected at point 0. In 1912 Peter Debye showed that this field could be accounted for as a first approximation by characterising a molecule by its dipole moment. In the next approximation the electric field at 0 can also be accounted by an additional contribution from a distribution of charges within a molecule called a quadrupole, and in the next approximation by an additional contribution from a distribution called an octupole [11]. In a homonuclear diatomic molecule such as $\mathrm{H}_{2}$ and $\mathrm{Cl}_{2}$, the positive nuclei are embedded in charge clouds describing the distribution of negatively charged electrons. For such molecules the "centres" of positive charges and negative charges coincide. But for the molecule $\mathrm{HCl}$ the electron distribution favours the more electronegative chlorine atom. Hence the centres of positive and negative charges, magnitude $+\mathrm{q}$ and $-\mathrm{q}$, are separated by a dipole length $\ell$. The molecule has a dipole moment, a characteristic and permanent property of an $\mathrm{HCl}$ molecule. The (molecular) dipole moment $\mu$ is given by the product ‘$\mathrm{q} \, \ell$'. A dipole moment has both magnitude and direction; it is a vector [12]. Footnotes [1] The classic reference in this subject is: J. O. Hirschfelder, C. F. Curtiss and R. B. Bird, Molecular Theory of Gases and Liquids, Wiley, New York 1954; corrected printing ,1964. [2] The ampere is that constant current flowing in two parallel straight conductors, having negligible cross section, one metre apart in vacuo which produces a force between each metre of length equal to $2 \time 10^{-7} \mathrm{~N}$. [3] When a current of one A flows for one second, the total charge passed is one coulomb. In practice, a current of $1 \mathrm{~A}$ is very high and the common unit is milliampere (symbol: $\mathrm{mA}$). The starter motor in a conventional car requires a peak current of around $30 \mathrm{~A}$. Electric charge on a single proton, $\mathrm{e}=1.602 \times 10^{-19} \mathrm{C}$. Faraday, $\mathrm{F}=\mathrm{N}_{\mathrm{A}} \, \mathrm{e}=9.649 \times 10^{4} \mathrm{C} \mathrm{mol}^{-1}$ [4] Just to keep up with the way the units are developing we note: - electric current coulomb $\mathrm{C} = \mathrm{A}$ s electric potential gradient volt $\mathrm{V}=\mathrm{J} \mathrm{A}^{-1} \mathrm{~s}^{-1}=\mathrm{J} \mathrm{C}^{-1}$ ($\mathrm{J =}$ joule). Thus volt expressed as $\mathrm{J C}^{-1}$ is energy per coulomb of electric charge passed. This link between electric potential and energy is crucial. electrical resistance ohm $\Omega=\mathrm{V} \mathrm{A}^{-1}$ Ohm's Law is a phenomenological law. [5] Continuing our concern for units. electrical capacitance: unit = farad $\mathrm{F} \equiv \mathrm{As} \mathrm{} \mathrm{V}^{-1}$ electric permittivity $\varepsilon$; unit = $\mathrm{F m}^{-1}$ electric permittivity of a vacuum, $\varepsilon_{0}=8.854 \times 10^{-12} \mathrm{~F} \mathrm{~m}^{-1}$ relative permittivity $\varepsilon_{\mathrm{r}}=\varepsilon / \varepsilon_{0}$; unit $= 1$ Older literature calls $\epsilon_{\mathrm{r}$, the "dielectric constant". But this property is not a constant for a given substance such as water ($\ell$). Thus $\varepsilon$ and $\varepsilon_{\mathrm{r}}$ depend on both temperature and pressure [6]; $\varepsilon$ and $\varepsilon_{\mathrm{r}}$ for a given liquid depend on electric field strength and frequency of AC current applied to the capacitance. [7] The quantity $\left(4 \, \pi \, \varepsilon_{0} \, 10^{-7}\right)^{-1 / 2}$ equals $2.998 \times 10^{8} \mathrm{m s}^{-1}$ which is the speed of light. [8] We check the units. If $\mathrm{X}$ is a force, the unit for $\mathrm{X}$ is newton (symbol $\mathrm{N}$). Then the right-hand side should simplify to the same unit. Electric charge is expressed in $\mathrm{C}[= \mathrm{A s}]$; $\varepsilon_{0}$ has units of $\mathrm{F} \mathrm{m}{ }^{-1}\left[=\mathrm{As} \mathrm{} \mathrm{V}^{-1} \mathrm{~m}^{-1}\right]$. Then $\mathrm{X}=[\mathrm{C}] \,[\mathrm{C}] /[1] \,[1] \,\left[\mathrm{A} \mathrm{s} \mathrm{} \mathrm{V}^{-1} \mathrm{~m}^{-1}\right] \,[\mathrm{m}]^{2}$ But $[\mathrm{V}]=\left[\mathrm{J} \mathrm{A}^{-1} \mathrm{~s}^{-1}\right]$ Then $\mathrm{X}=\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right] /\left[\mathrm{A} \mathrm{s} \mathrm{J}^{-1} \mathrm{~A} \mathrm{~s} \mathrm{~m}\right]=\left[\mathrm{J} \mathrm{m}^{-1} \mathrm{l}=[\mathrm{N}]\right.$ [9] We check that our units are correct. If $\mathrm{U}_{ij}$ is an energy expressed in joules, the terms on the right-hand side should reduce to joules. $\mathrm{U}_{\mathrm{ij}}=[\mathrm{A} \mathrm{s}] \,[\mathrm{As}] /[1] \,[1] \,\left[\mathrm{As} \mathrm{} \mathrm{V}^{-1} \mathrm{~m}^{-1}\right] \,[\mathrm{m}] =\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right] /\left[\mathrm{As} \mathrm{J}^{-1} \mathrm{As}\right]=[\mathrm{J}]$ [10] \begin{aligned} \mathrm{E} &=[\mathrm{C}] /[1] \,[1] \,\left[\mathrm{As} \mathrm{} \mathrm{V}^{-1} \mathrm{~m}^{-1}\right] \,\left[\mathrm{m}^{2}\right] \ &=[\mathrm{A} \mathrm{s}] /[\mathrm{A} \mathrm{s} \mathrm{V} \mathrm{m}]=\left[\mathrm{V} \mathrm{m}^{-1}\right] \ &=\left[\mathrm{J} \mathrm{A}^{-1} \mathrm{~s}^{-1} \mathrm{~m}^{-1}\right]=\left[\mathrm{J} \mathrm{m}^{-1}\right] /[\mathrm{A} \mathrm{s}]=\left[\mathrm{N} \mathrm{C}^{-1}\right] \end{aligned} Thus electric field strength is expressed in $\mathrm{Vm}^{-1}$ or $\mathrm{N} \mathrm{C}^{-1}$; the latter is clearly a force per unit charge. [11] The classic text is:- P. Debye, Polar Molecules, Chemical Catalog Co., New York 1929 (available as Dover paperback). We do not consider here interactions involving quadrupoles, octupoles, etc. These molecular properties are reviewed by A. D. Buckingham, Quart. Rev., 1959, 13, 183. [12] Dipole moment, $\mu=\mathrm{q} \, \ell=[\mathrm{C}] \,[\mathrm{m}]$ Thus $\mu=[\mathrm{Cm}]$, coulomb metre . Dipole moments are normally quoted using the unit, debye. [The unit is named in honour of Peter Debye.] $1 \mathrm{D}=3.336 \times 10^{-30} \mathrm{Cm}$.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.19%3A_Electrical_Units.txt
At temperature $\mathrm{T}$ and pressure $\mathrm{p}$, the molar conductivity of given salt solution Λ depends on the concentration of salts. This subject has an extensive scientific literature. One of the challenges is to calculate $\Lambda$ for given salt solution knowing the properties of the pure solvent and the salt at specified $\mathrm{T}$ and $\mathrm{p}$. A key quantity is the limiting molar conductivity $\Lambda^{\infty}$ defined for a given salt solution by equation (a). $\operatorname{Lt}\left(c_{j} \rightarrow 0\right) \Lambda=\Lambda^{\infty}$ Moreover as Kohlrausch showed in 1876 ( > 125 years ago) a given $\Lambda^{\infty}$ can be expressed as the sum of limiting molar ionic conductances, $\lambda_{\mathrm{j}}^{\infty}$. Thus $\Lambda^{\infty}=\sum_{j=1}^{j=i} \lambda_{j}^{\infty}$ A difficult theoretical task is to estimate $\lambda_{\mathrm{j}}^{\infty}$ for a given ion at defined $\mathrm{T}$ and $\mathrm{p}$ and specified solvent. Rather more progress has been made in predicting quantitatively the dependence of $\left(\Lambda-\Lambda^{\infty}\right)$ on concentration of salt in a given solvent at defined $\mathrm{T}$ and $\mathrm{p}$ assuming that the ions in solution are characterised by their electric charges and radii. Indeed quantitative treatments of the electrical conductivities of salt solutions have attracted enormous interest and provided a challenge to scientists with good mathematical abilities. Here we summarise briefly the essence of treatments published by Onsager [1-3] and by Fuoss [3,4]. The account given below is based on that set out by N. K. Adam [5]. A relaxation effect and an electrophoretic effect contribute to the magnitude of $\left(\Lambda-\Lambda^{\infty}\right)$ for a real salt solution for which $\Lambda<\Lambda^{\infty}$. In a real solution under the influence of an applied electric field, anions and cations move in opposite direction. The word ‘move’ does not reflect the complexity of the real situation. In a real solution and in the absence of an applied electric field, the ions move in random directions, Brownian motion, as a consequence of the thermal energy of the system. In some sense, ions and solvent molecules are jostling continuously. When an electric potential gradient is applied across the solution, the previously random motion of ions is now biased in a particular direction depending on the ionic charge. If the solution is ‘infinitely dilute’, the velocity of a given ion is characteristic of the ion, solvent, temperature and pressure. In a real solution the mobility decreases with increase in concentration of salt. Two retarding effects are identified, relaxation and electrophoretic effects. The latter emerges from the fact that a given $j$ ion moves against the flow of counter-ions together with associated solvent molecules. In a real solution and in the absence of an applied electric field, a given $j$ ion is at the centre of an ion atmosphere which has an electric charge equal in magnitude but opposite in sign to that of the $j$ ion. Under the impact of an applied electric field the $j$-ion moves away from the centre of the ion atmosphere. The latter pulls the $j$-ion back towards this centre. In other words the $j$-ion is retarded by this relaxation effect. The latter term reflects the fact that the retardation depends on the rate at which the electric charge density in the ion atmosphere grows as the $j$ ion moves through the solution and decays in the wake of the $j$ ion. Ionic Mobility A given salt solution at temperature $\mathrm{T}$ and pressure $p$ contains a sa }\)]. In solution, the motion of ions is quite random, a pattern usually described as Brownian motion. If however an electric field is applied across the solution the movement of ions is biased in a given direction depending on the sign of the charge on the $j$ ion. The electrical mobility $\mathrm{u}_{j}$ of ion $j$ describes the velocity of ion-$j$ in an electric field gradient measured in [$\mathrm{V m}^{-1}$] [6]. In the absence of ion-ion charge-charge interaction , the electrical mobility $\mathbf{u}_{j}^{\infty}$ is characteristic of the $j$ ion, solvent, temperature and pressure. The superscript ‘$\infty$’ identifies that to all intents and purposes the $j$ ion is in an infinitely dilute solution. However in a real solution, concentration $\mathrm{c}_{i}$ in salt $i$, the $j$ ion is surrounded by an ‘ion atmosphere’ which has an electric charge equal in magnitude but opposite in sign to that on the $j$ ion. Electrophoretic Effect The ion atmosphere is modelled as a series of shells, thickness $\mathrm{dr}$ distance $\mathrm{r}$ from the centre of the $j$ ion. The electrical charge $\mathrm{q}_{j}$ on a shell distance $\mathrm{r}$ from the $j$ ion is given by equation (c) [7]. $q_{j}=4 \, \pi \, r^{2} \, \rho \, d r$ Here $\rho$ is the electric charge density, measured in ‘coulombs per cubic metre’. As a result of the electric field gradient operating on the $j$ ion the electric force $\mathrm{F}$ expressed in newtons operating on this shell [8] is given by equation (d). $\mathrm{F}=4 \, \pi \, \mathrm{r}^{2} \, \rho \, \mathrm{E} \, \mathrm{dr}$ According to Stokes Law, a sphere having radius $\mathrm{r}$ and moving with velocity $v$ through a liquid having (shear) viscosity $\eta$ is subject to a viscous resistance $\mathrm{R}_{\eta}$, a force expressed in newtons and given by equation (e) [9]. $\mathrm{R}_{\eta}=6 \, \pi \, \eta \, \mathrm{r} \, \mathrm{V}$ If the speed of the liquid stream increases by $\mathrm{dv}$ when the radius of the shell defining the ion atmosphere increases by $\mathrm{dr}$, the viscous resistance increases by $(6 \, \pi \, \eta \, r \, d v)$. If the motion of the $j$ ion through the solution is steady, the increase in viscous resistance to movement of the $j$ ion equals the electrical force (see equation d). Therefore $6 \, \pi \, \eta \, r \, d v=4 \, \pi \, r^{2} \, \rho_{j} \, E \, d r$ The charge density $\rho_{j}$ is obtained by combining equations (u) and (x) of Topic 680. Thus [10], $\rho_{\mathrm{j}}=-\frac{\left(\mathrm{z}_{\mathrm{j}} \, \mathrm{e}\right) \, \exp \left(\kappa \, \mathrm{a}_{\mathrm{j}}\right)}{\left.\left.4 \, \pi \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \,\right) 1+\kappa \, \mathrm{a}_{\mathrm{j}}\right)} \, \varepsilon_{0} \, \varepsilon_{\mathrm{r}} \, \kappa^{2} \, \frac{\exp (-\kappa \, \mathrm{r})}{\mathrm{r}}$ Then with $\ell^{-1}=\kappa$, $p_{j}=-\frac{z_{j} \, e \, \exp \left(a_{j} / \ell\right) \, \exp (-r / \ell)}{4 \, \pi \, \ell \,\left(a_{j}+\ell\right) \, r}$ From equation (f), $\mathrm{dv}=\frac{2}{3} \, \frac{\rho_{\mathrm{j}}}{\eta} \, \mathrm{E} \, \mathrm{r} \, \mathrm{dr}$ Hence $\mathrm{dv}=-\frac{2 \, \mathrm{z}_{\mathrm{j}} \, \mathrm{e} \, \exp \left(\mathrm{a}_{\mathrm{j}} / \ell\right)}{12 \, \pi \, \eta \, \ell \,\left(\mathrm{a}_{\mathrm{j}}+\ell\right)} \, \mathrm{E} \, \exp (-\mathrm{r} / \ell) \, \mathrm{dr}$ Equation (j) is integrated between limits (i) $r=\sigma$ to $r = \infty$, and (ii) $v = 0$ and $v_{1}$ where $v_{1}$ is the stream velocity of the solution outside the ion atmosphere of the $j$ ion. Then $\mathrm{v}_{1}=\frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e} \, \mathrm{E} \, \exp \left(\mathrm{a}_{\mathrm{j}} / \ell\right)}{6 \, \pi \, \eta \, \ell \,\left(\mathrm{a}_{\mathrm{j}}+\ell\right)} \, \int_{\mathrm{a}_{\mathrm{j}}}^{\infty} \exp (-\mathrm{r} / \ell) \, \mathrm{dv}$ Hence, $\mathrm{v}_{1}=-\frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e} \, \mathrm{E}}{6 \, \pi \, \eta \,\left(\mathrm{a}_{\mathrm{j}}+\ell\right)}$ For dilute solutions, $\mathrm{a}_{\mathrm{j}}<\ll \ell$ such that the stream velocity of the solution outside the ion atmosphere is given by equation (m) Therefore [11] $\mathrm{v}_{1}=-\frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e} \, \mathrm{E}}{6 \, \pi \, \eta \, \ell}$ We shift the reference. The solvent does not physically move when we measure the electrical conductivity of a solution. Therefore the impact of the electrophoretic effect is to retard the $j$-ion in solution. The decrease in electrical mobility of the $j$ ion is given by equation (n) [12]. $-\left(\Delta \mathrm{u}_{\mathrm{j}}\right)_{\mathrm{clectrpbor}}=-\frac{\mathrm{z}_{\mathrm{j}} \, \mathrm{e}}{6 \, \pi \, \eta \, \ell}$ Relaxation Effect In the limit of infinite dilution, a given $j$-ion proceeds through an aqueous solutions at defined $\mathrm{T}$ and $\mathrm{p}$ under the influence of an applied electric field gradient. The impediment to its progress arises from the solvent molecules. However in a real salt solution, the $j$ ion is surrounded by its ion atmosphere which has an electric charge equal in magnitude and opposite on sign. In the absence of an applied electric field the ion atmosphere is spherically symmetric about the $j$ ion. In a real solution, the migrating ion is not at the centre of the ion atmosphere, the latter therefore retarding the migrating ion. This retardation is called the relaxation effect on the grounds that the build-up of the ion atmosphere preceeding the ion and the decay in the wake of the ion is characterised by a relaxation time. The relaxation effect can be understood in terms of irreversible thermodynamics. Thus the flow of cations and anions in opposite directions are coupled. The stronger the coupling the greater is the retardation of the migrating ions. The first treatment of this coupling of flows and forces was developed by Onsager who published a reasonably successful description of the impact of this coupling on ionic mobilities. The analogue of equation (n) describing the relaxation effect takes the following form [13] where $\mathrm{w}$ is a correction factor depending on the type of electrolyte [14]. $-\left(\Delta u_{j}\right)=\frac{e^{3} \, w \, u_{j}^{\infty}}{24 \, \varepsilon^{0} \, \varepsilon_{\mathrm{r}} \, k \, T \, \ell}$ Here $\ell$ is the radius of the ion atmosphere surrounding the $j$ ion ; equation (p) where the concentration of $j$ ions $\mathrm{c}_{j}$ is expressed in $\mathrm{mol dm}^{-3}$. $\ell=\frac{10^{3} \, 4 \, \pi \, \varepsilon^{0} \, \varepsilon_{\mathrm{r}} \, \mathrm{k} \, \mathrm{T}}{8 \, \pi \, \mathrm{e}^{2} \, \mathrm{N}_{\mathrm{A}} \, \mathrm{I}}$ For dilute solutions $I=(0.5) \, \sum_{i=1}^{i=j} c_{i} \, z_{i}^{2}$ Molar Conductivity In summary two retarding effects, electrophoretic and relaxation, mean that the molar conductivity of a given aqueous salt solution is less than the molar conductivity of the corresponding solution at infinite dilution, $\Lambda^{\infty}$. The outcome is the famous Debye-Huckel-Onsager Equation for molar conductivities. For a 1:1 salt (e.g. $\mathrm{KBr}$) in aqueous solution at $298.15 \mathrm{~K}$ and ambient pressure, the molar conductivity $\Lambda$ is given by equation (r) [15,16]. $\Lambda=\Lambda^{\infty}-\left(0.229 \, \Lambda^{\infty}+60.2\right) \,\left(c_{j} / c_{r}\right)^{1 / 2}$ Footnotes [1] L. Onsager, Physik. Z.,1926,27,388. [2] L. Onsager, Trans. Faraday Soc.,1927,23,341. [3] L. Onsager and R. M. Fuoss, J. Phys. Chem.,1932,36,2689. [4] H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions, Reinhold, New York, 1950, 2nd edn. Revised and enlarged [5] N. K. Adam, Physical Chemistry, Oxford, 1956. [6] \begin{aligned} &v_{j}=\left[\mathrm{m} \mathrm{s}^{-1}\right] \ &u_{j}=\left[\mathrm{m} \mathrm{s}^{-1}\right] /[\mathrm{V} \mathrm{m} \end{aligned} [7] $\mathrm{q}_{\mathrm{j}}=[1] \,[1] \,\left[\mathrm{m}^{2}\right] \,\left[\mathrm{Cm}^{-3}\right] \,[\mathrm{m}]=[\mathrm{C}]$ [8] $4 \, \pi \, r^{2} \, \rho \, E \, d r=[1] \,[1] \,\left[\mathrm{m}^{2}\right] \,\left[\mathrm{C} \mathrm{m}^{-3}\right] \,\left[\mathrm{V} \mathrm{m}^{-1}\right] \,[\mathrm{m}] =\left[\mathrm{J} \mathrm{m}^{-1}\right]=[\mathrm{N}]$ [9] $\mathrm{R}_{\eta}=[1] \,[1] \,\left[\mathrm{kg} \mathrm{m}^{-1} \mathrm{~s}^{-1}\right] \,[\mathrm{m}] \,\left[\mathrm{m} \mathrm{s}^{-1}\right]=\left[\mathrm{kg} \mathrm{m} \mathrm{s}^{-22}\right]=[\mathrm{N}]$ [10] $\rho_{j}=\frac{[1] \,[\mathrm{C}] \,[1]}{[1] \,[1] \,\left[\mathrm{Fm}^{-1}\right] \,[1] \,[1]} \, \frac{\left[\mathrm{Fm}^{-1}\right] \,[1] \,[\mathrm{m}]^{2} \,[1]}{[\mathrm{m}]} =\left[\mathrm{C} \mathrm{m}^{-3}\right]$ [11] $\mathrm{v}_{1}=\frac{[\mathrm{l}] \,[\mathrm{C}] \,\left[\mathrm{V} \mathrm{m} \mathrm{m}^{-1}\right]}{[\mathrm{l}] \,[\mathrm{l}] \,\left[\mathrm{kg} \mathrm{m}^{-1} \mathrm{~s}^{-1}\right] \,[\mathrm{m}]} \mathrm{v}_{1}=\frac{[\mathrm{l}] \,[\mathrm{C}] \,\left[\mathrm{V} \mathrm{m} \mathrm{m}^{-1}\right]}{[\mathrm{l}] \,[\mathrm{l}] \,\left[\mathrm{kg} \mathrm{m}^{-1} \mathrm{~s}^{-1}\right] \,[\mathrm{m}]} [12] \(\Delta \mathrm{u}_{\mathrm{j}}=\frac{\left[\mathrm{m} \mathrm{s}^{-1}\right]}{\left[\mathrm{V} \mathrm{} \mathrm{m}^{-1}\right]}=\left[\mathrm{m}^{2} \mathrm{~V}^{-1} \mathrm{~s}^{-1}\right]$ $-\left(\Delta \mathrm{u}_{\mathrm{j}}\right)_{\text {relax }}=\frac{\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right]}{[1] \,\left[\mathrm{Fm}^{-1}\right] \,[1] \,\left[\mathrm{J} \mathrm{K}^{-1}\right] \,[\mathrm{K}] \,[\mathrm{m}]} \, \mathrm{u}_{\mathrm{j}}^{\infty}$ [13] $=\frac{\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right]}{[\mathrm{F}] \,[\mathrm{J}]}=\frac{\left[\mathrm{A}^{2} \mathrm{~s}^{2}\right]}{\left[\mathrm{A}^{2} \mathrm{~s}^{4} \mathrm{~kg}^{-1} \mathrm{~m}^{-2}\right] \,\left[\mathrm{kg} \mathrm{m} \mathrm{m}^{2} \mathrm{~s}^{-2}\right]} \, \mathrm{u}_{\mathrm{j}}^{\infty}$ $=[1] \, \mathrm{u}_{\mathrm{j}}^{\infty}$ [14] For an advanced treatment, see J. O’M. Bockris and A.K.N Reddy, Modern Electrochemistry: Ionics, Plenum Press, New York, 2nd. edn.,1998, chapter 4. [15] P. W. Atkins, Physical Chemistry, Oxford University Press, 1982, 2nd. edn., p.900. [16] $\Lambda=\left[\Omega^{-1} \mathrm{~cm}^{2} \mathrm{~mol}^{-1}\right] ; 0.229=[1] ; 60.2=\left[\Omega^{-1} \mathrm{~cm}^{2} \mathrm{~mol}^{-1}\right]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.20%3A_Electric_Conductivities_of_Salt_Solutions-_Dependence_on_Composition.txt
A given experimental system comprises two compartments, I and II, separated by a membrane. The two compartments contain aqueous solutions at common temperature and pressure. The experimental system is set up by placing in compartment I an aqueous salt solution; e.g. $\mathrm{NaCl}(\mathrm{aq})$ having concentration $\mathrm{c}_{1} \mathrm{~mol dm}^{-3}$. However compartment II contains a salt, $\mathrm{R}^{+}\mathrm{Cl}^{-} (\mathrm{aq})$, concentration $\mathrm{c}_{2} \mathrm{~mol dm}^{-3}$. The membrane is permeable to both $\mathrm{Na}^{+} and \(\mathrm{Cl}^{-}$ ions but not to $\mathrm{R}^{+}$ cations. The sodium and chloride ions spontaneously diffuse across the membrane until the two solutions are in thermodynamic equilibrium. We represent the equilibrium system as follows where | | represents the membrane. [I] $\mathrm{Na}^{+}\left(\mathrm{c}_{1}-\alpha\right)^{\mathrm{cq}} \mathrm{Cl}^{-}\left(\mathrm{c}_{1}-\alpha\right)^{\mathrm{eq}}||[\mathrm{II}] \mathrm{Na}^{+}(\alpha)^{\mathrm{eq}} \mathrm{R}^{+}\left(\mathrm{c}_{2}\right)^{\mathrm{eq}} \mathrm{Cl}^{-}\left(\mathrm{c}_{2}+\alpha\right)^{\mathrm{eq}}$ The solutions on both sides are electrically neutral. A thermodynamic analysis is somewhat complicated if account is taken of the role of ion-ion interactions. However the essential features of the argument are revealed if we identify the activities of the ions as equal to their concentrations. Hence at equilibrium at fixed $\mathrm{T}$ and $\mathrm{p}$, \begin{aligned} &{\left[\mu^{0}\left(\mathrm{Na}^{+} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left\{\left(\mathrm{c}_{1}-\alpha\right)_{\mathrm{Na}}^{\mathrm{eq}} / \mathrm{c}_{\mathrm{r}}\right\}\right]_{\mathrm{I}}} \ &+\left[\mu^{0}\left(\mathrm{Cl}^{-} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left\{\left(\mathrm{c}_{1}-\alpha\right)_{\mathrm{Cl}}^{\mathrm{eq}} / \mathrm{c}_{\mathrm{r}}\right\}\right]_{\mathrm{I}}= \ &{\left[\mu^{0}\left(\mathrm{Na}^{+} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left\{\alpha_{\mathrm{Na+}}^{\mathrm{eq}} / \mathrm{c}_{\mathrm{r}}\right\}\right]_{\mathrm{II}}} \ &+\left[\mu^{0}\left(\mathrm{Cl}^{-} ; \mathrm{aq}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left\{\left(\mathrm{c}_{2}+\alpha\right)_{\mathrm{Cl-}}^{\mathrm{eq}} / \mathrm{c}_{\mathrm{r}}\right\}\right]_{\mathrm{II}} \end{aligned} But $\left(c_{1}-\alpha\right)_{\mathrm{Na+}}^{\mathrm{eq}}=\left(\mathrm{c}_{1}-\alpha\right)_{\mathrm{Cl}-}^{\mathrm{eq}}$ Or, $\left[\left(c_{1}-\alpha\right)_{\mathrm{Na}+}^{\mathrm{eq}}\right]^{2}=\alpha_{\mathrm{Na}+}^{\mathrm{eq}} \,\left(\mathrm{c}_{2}+\alpha\right)^{\mathrm{eq}} \mathrm{CI}^{-}$ Then [1], $\frac{\alpha^{\mathrm{eq}}}{\mathrm{c}_{1}}=\frac{\mathrm{c}_{1}}{2 \, \mathrm{c}_{1}+\mathrm{c}_{2}}$ The latter is Donnan’s Equation. The ratio $\left(\alpha^{\mathrm{eq}} / \mathrm{c}_{1}\right)$ tends to be smaller the larger is $\mathrm{c}_{2}$. This conclusion is confirmed by experiment. We have simplified the algebra by writing $\mathrm{R}^{+} \mathrm{Cl}^{-}$ as the salt in compartment II. In practice the Donnan equilibrium finds major application where salt $\mathrm{RCl}$ is a macromolecule [2-4]. Footnotes [1] $\mathrm{c}_{1}^{2}-2 \, \alpha \, \mathrm{c}_{1}+\alpha^{2}=\alpha \, \mathrm{c}_{2}+\alpha^{2}$ Then, $c_{1}^{2}-2 \, \alpha \, c_{1}=\alpha \, c_{2}$ Or, $\alpha=\frac{\mathrm{c}_{1}^{2}}{2 \, \mathrm{c}_{1}+\mathrm{c}_{2}}$ [2] F. G. Donnan et al, J. Chem. Soc.,1911, 1554; 1914,1941. [3] F. G. Donnan, Chem. Rev.,1924, 1,73. [4] F. G. Donnan and E. A. Guggenheim, Z. Physik Chem. A, 1932,162,346. 1.14.22: Descriptions of Systems Important themes in thermodynamics involve (i) properties (variables) which can be measured (e.g. volumes and/or densities) and (ii) thermodynamic variables which are rigorously defined (e.g. enthalpies). In these terms a measured property (e.g. density) is the reporter of the chemical properties or processes taking place in a system. So it is always important to ask if the “reporter” can be interrogated for the required information. In fact there is often a limit to the amount of information which a given reporter offers to the investigator. These important limitations should be borne in mind. An example makes the point. A system is prepared by placing $\mathrm{n}_{\mathrm{X}}^{0}$ moles of substance in a closed vessel at fixed $\mathrm{T}$ and $\mathrm{p}$. [The superscript ‘0’ implies at zero time.] We explore two possible descriptions of this system. Perhaps two samples were analysed by two independent laboratories. Description A The first laboratory reports that the system is simple and contains the single substance $\mathrm{X}$. Gibbs energy $\mathrm{G}(\mathrm{A})=\mathrm{n}_{\mathrm{X}}^{0} \, \mu_{\mathrm{X}}^{*}(\ell)$ and volume $\mathrm{V}(\mathrm{A})=\mathrm{n}_{\mathrm{X}}^{0} \, \mathrm{V}_{\mathrm{X}}^{*}(\ell)$ Here $\mu_{X}^{*}(\ell)$ and $\mathrm{V}_{\mathrm{x}}^{*}(\ell)$ are the chemical potential and molar volume of the pure chemical substance $\mathrm{X}$. Description B The second laboratory identifies two substances $\mathrm{X}$ and $\mathrm{Y}$ in chemical equilibrium such that the equilibrium amounts of substances $\mathrm{X}$ and $\mathrm{Y}$ are respectively $\mathrm{n}_{\mathrm{X}}^{\mathrm{eq}}$ and $\mathrm{n}_{\mathrm{Y}}^{\mathrm{eq}}$. Gibbs energy $\mathrm{G}(\mathrm{B})=\mathrm{n}_{\mathrm{X}}^{\mathrm{eq}} \, \mu_{\mathrm{X}}^{\mathrm{eq}}+\mathrm{n}_{\mathrm{Y}}^{\mathrm{eq}} \, \mu_{\mathrm{Y}}^{\mathrm{eq}}$ and volume $\mathrm{V}(\mathrm{B})=\mathrm{n}_{\mathrm{X}}^{\mathrm{eq}} \, \mathrm{V}_{\mathrm{X}}^{\mathrm{eq}}+\mathrm{n}_{\mathrm{Y}}^{\mathrm{eq}} \, \mathrm{V}_{\mathrm{Y}}^{\mathrm{eq}}$ Here $\mu_{\mathrm{X}}^{\mathrm{eq}}$ and $\mu_{\mathrm{Y}}^{\mathrm{eq}}$ are the equilibrium chemical potentials; $\mathrm{V}_{\mathrm{X}}^{\mathrm{eq}}$ and $\mathrm{V}_{\mathrm{Y}}^{\mathrm{eq}}$ are the equilibrium partial molar volumes. Description A is “primitive” and Description B is “sophisticated”. Both Gibbs energies and volumes are functions of state so that $\mathrm{V}(\mathrm{A})=\mathrm{V}(\mathrm{B})$ and $\mathrm{G}(\mathrm{A})=\mathrm{G}(\mathrm{B})$. The chemical potential of substance $\mathrm{X}$ describes the change in $\mathrm{G}$ when $\delta n_{X}$ moles of $\mathrm{X}$ are added. This chemical potential is insensitive to the changes taking place in the equilibrium system;$\mu_{X}(\mathrm{~A})=\mu_{X}(\mathrm{~B})$. Consequently, measurement of volume $\mathrm{V}$ [and if it were possible of $\mathrm{G}$] would not distinguish between the two descriptions. Similarly, measurement of $\mathrm{H}$ (if it were possible) would not distinguish between the two descriptions. Footnotes [1] L. P. Hammett, Physical Organic Chemistry, McGraw-Hill, New York, 2nd edn., 1970,p.16. 1.14.23: Enzyme-Substrate Interaction We consider the formation in aqueous solution of an enzyme –substrate complex $\mathrm{ES}$ by an enzyme $\mathrm{E}$ and substrate $\mathrm{S}$. The system is prepared using $\mathrm{n}^{0} (\mathrm{E})$ moles of enzyme and $\mathrm{n}^{0} (\mathrm{S}$) moles of substrate; equation (a) $\mathrm{E}(\mathrm{aq}) +$ $\mathrm{S}(\mathrm{aq}) \Leftrightarrow$ $\mathrm{ES}(\mathrm{aq})$ At $t = 0$ $\mathrm{n}^{0}(\mathrm{E})$ $\mathrm{n}^{0}(\mathrm{S})$ $0 \mathrm{~mol}$ At $t = \infty$ $n^{0}(E)-\xi$ $\mathrm{n}^{0}(\mathrm{~S})-\xi$ $\xi \mathrm{~mol}$ The upper limit of the extent of interaction $\xi$ is controlled by whichever is the smallest amount, either $\mathrm{n}^{0}(\mathrm{E})$ or $\mathrm{n}^{0}(\mathrm{S})$. The latter two variables determine the total amount of $\mathrm{ES}(\mathrm{aq})$ which can be formed in the limit of tight binding. The approach described above can be extended to more complicated schemes involving multip-step reactions. In the following we consider the case where enzyme $\mathrm{E}$ converts substrate $\mathrm{A}$ into product $\mathrm{D}$. The system is prepared using $\mathrm{n}^{0}(\mathrm{E})$ moles of enzyme and $\mathrm{n}^{0}(\mathrm{A})$ moles of substrate $\mathrm{A}$ such that there are two intermediates $\mathrm{EB}(\mathrm{aq})$ and $\mathrm{EC}(\mathrm{aq})$, product $\mathrm{D}$ being liberated from the bound state in the final step. The equilibrium state can be represented by the following scheme. \begin{aligned} &\mathrm{E}(\mathrm{aq}) \quad+\mathrm{A}(\mathrm{aq}) \Leftrightarrow \mathrm{EB}(\mathrm{aq}) \Leftrightarrow \mathrm{EC}(\mathrm{aq}) \Leftrightarrow \mathrm{E}(\mathrm{aq})+\mathrm{D}(\mathrm{aq})\ &\mathrm{n}^{0}(\mathrm{E})-\xi_{1}+\xi_{3} \mathrm{n}^{0}(\mathrm{~A})-\xi_{1} \quad \xi_{1}-\xi_{2} \quad \xi_{2}-\xi_{3} \quad \mathrm{n}^{0}(\mathrm{E})-\xi_{1}+\xi_{3} \xi_{3} \end{aligned} The key point is that at equilibrium the amounts of enzyme $\mathrm{E}(\mathrm{aq})$ identified at both ends of the reaction must be the same. Further a mass balance shows that the total amount of enzyme present equals $\mathrm{n}^{0}(\mathrm{E}) \mathrm{~mol}$. Other features are interesting; $\xi_{3}$ must be zero if $\xi_{2}$ is zero. The method can be applied to more complicated reaction schemes including those where the path from reactant to product involves parallel reactions.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.21%3A_Donnan_Membrane_Equilibria.txt
We take up the challenge of seeking an equation of state for all chemical substances. We confine attention to closed systems containing one chemical substance. We also confine our attention to systems at equilibrium where the affinity for spontaneous change is zero. A calculus operation allows us to relate $\mathrm{p}$, $\mathrm{V}$ and $\mathrm{T}$. $\left(\frac{\partial p}{\partial V}\right)_{T} \,\left(\frac{\partial V}{\partial T}\right)_{p} \,\left(\frac{\partial T}{\partial p}\right)_{V}=-1$ By definition, the equilibrium isobaric expansivity. [1] $\alpha_{p}(A=0)=\frac{1}{V} \,\left(\frac{\partial V}{\partial T}\right)_{p, A=0}$ The equilibrium isothermal compressibility [2], $\kappa_{\mathrm{T}}(\mathrm{A}=0)=-\frac{1}{\mathrm{~V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}=0}$ Hence, $\left(\frac{\partial p}{\partial T}\right)_{V, A=0}=\frac{\alpha_{p}(A=0)}{\kappa_{T}(A=0)}$ $\left(\frac{\partial p}{\partial T}\right)_{V, A=0}$ is the equilibrium isochoric thermal pressure coefficient. [3] Equation (d) shows that this property can be obtained from the experimentally accessible, $\alpha_{p}(A=0)$ and $\kappa_{\mathrm{T}}(\mathrm{A}=0)$. In fact the coefficient can be directly measured, at least for liquids. [4] From the Master Equation where the affinity for spontaneous change is zero, $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}$ At constant $\mathrm{T}$, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{T}}-\mathrm{p}$ Using a Maxwell Equation, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}-\mathrm{p}$ All terms on the right hand side of equation (g) are experimentally accessible. Moreover this equation applies to all systems, solids, liquids and gases. By definition, $\beta_{\mathrm{V}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ Then, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \, \beta_{\mathrm{V}}-\mathrm{p}$ Equation (i) is a Thermodynamic Equation of State. Equation (j) is another Thermodynamic Equation of State. [5] $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\mathrm{V} \,\left(1-\mathrm{T} \, \alpha_{\mathrm{p}}\right)$ For many condensed phases the product $\mathrm{T} \, \boldsymbol{\beta}_{\mathrm{v}}$ is much larger than the pressure $\mathrm{p}$. Hence $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}} \cong \mathrm{T} \, \boldsymbol{\beta}_{\mathrm{V}}$ The partial differential $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}$ is the internal pressure, $\pi_{j}$ for liquid $j$. [6] Originally the term ‘internal pressure’ referred to the product , $\mathrm{T} \, \boldsymbol{\beta}_{\mathrm{v}}$. The closely related ratio of molar thermodynamic energy of vaporisation to molar volume $\left[\frac{\Delta_{\mathrm{vap}} \mathrm{U}^{0}}{\mathrm{~V}^{*}(\ell)}\right]$ is the cohesive energy density, $\text { c.e.d. }=\left[\frac{\Delta_{\mathrm{vap}} \mathrm{U}^{0}}{\mathrm{~V}^{*}(\ell)}\right]$ The rational behind this definition notes that $\Delta_{\text {vap }} \mathrm{U}^{0}$ defines the change in thermodynamic energy when one mole of a given substance passes from the liquid to the vapour state, breaking strong cohesive forces in the liquid. By dividing by the molar volume of the liquid we normalise this change to a fixed volume. [7] Internal pressures are interesting parameters. [8,9] Nevertheless despite their thermodynamic basis, treatments of chemical properties in terms of internal pressures receive only sporadic attention. One feels they should be more informative but it is not always clear how one draws conclusions from analysis of experimental data using these properties. Footnotes [1] $\alpha_{p}=\frac{1}{\left[m^{3}\right]} \, \frac{\left[\mathrm{m}^{3}\right]}{[\mathrm{K}]}=\left[\mathrm{K}^{-1}\right]$ [2] $\kappa_{\mathrm{T}}=\frac{1}{\left[\mathrm{~m}^{3}\right]} \, \frac{\left[\mathrm{m}^{3}\right]}{\left[\mathrm{N} \mathrm{m}^{-3}\right]}=\left[\mathrm{N} \mathrm{m}^{-2}\right]^{-1}=[\mathrm{Pa}]^{-1}$ [3] $\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)=\frac{\left[\mathrm{N} \mathrm{m}^{-2}\right]}{[\mathrm{K}]}=\left[\mathrm{Pa} \mathrm{K}{ }^{-1}\right]$ [4] A small amount of liquid sample is held in a sample cell sealed by a piston; the latter is linked to a device which allows a known pressure to be applied to the sample. The sample cell is held in a thermostat; the temperature of the latter is tightly controlled. The temperature is changed by a small amount; $\Delta \mathrm{T}$. The volume of the liquid in the sample cell (normally) increases. The applied pressure is changed by a small amount $\Delta \mathrm{p}$ in order to recover the original volume. Then for a given liquid at defined $\mathrm{p}$, $\mathrm{V}$ and $\mathrm{T}$, we have the ratio $(\Delta \mathrm{p} / \Delta \mathrm{T})$. [5] J. S. Rowlinson and F. L. Swinton, Liquids and Liquid Mixtures, Butterworths, London, 3rd. edn., 1982, p.12. [6] $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\frac{[\mathrm{J}]}{\left[\mathrm{m}^{3}\right]}=\left[\mathrm{N} \mathrm{m}{ }^{-2}\right]=[\mathrm{Pa}]$ [7] Links between calorimetry and equations of state are discussed by S. L. Randzio, Chem. Soc. Rev., 1995, 24, 359. [8] 1. W. Westwater, H. W. Frantz and J. H. Hildebrand, Phys. Rev., 1928, 31,135. 2. J. H. Hildebrand, Phys. Rev.,1929, 34, 649, 984. 3. J. H. Hildebrand and R. H. Lamoreaux, Ind. Eng. Chem. Fundam., 1974, 13, 110. 4. J. H. Hildebrand, Proc. Natl. Acad.Sci. USA, 1967, 57, 542. [9] $\mathrm{T}/\mathrm{K} = 298.15$ Liquid $\pi_{\mathrm{i} /10^{8} \mathrm{Pa} \quad \mathrm{c.e.d}./10^{8} \mathrm{Pa}$ Methanol $2.930 \quad 8.600$ DMSO $5.166 \quad 7.047$ Water $1.013 \quad 22.98$ 1.14.25: Equation of State- Perfect Gas A closed system contains $\mathrm{n}_{j}$ moles of a gaseous substance $j$. No chemical reaction takes place in the system. The system is at equilibrium where the affinity for spontaneous change is zero. The system is characterised by the thermodynamic energy $\mathrm{U}$. The system is displaced to a neighbouring equilibrium state by a change in entropy $\mathrm{dS}$ and a change in volume $\mathrm{dV}$. The change in thermodynamic energy is given by the Master Equation. $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}$ At equilibrium the isothermal dependence of thermodynamic energy on volume is given by equation (b). $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{T}}-\mathrm{p}$ In attempting to understand from a chemical standpoint the properties of gases, liquid mixtures and solutions, a common approach formulates a set of properties which classify a given system as ideal. The definition of an ideal (or, perfect) system is made with practical chemistry in mind. When examining the properties of gases there is merit in identifying the properties of a perfect gas. [No real gas is perfect!] If the gaseous substance $j$ is a perfect gas, the following conditions [1] are met at all temperatures and pressures. 1. $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}}=0$ 2. $\mathrm{p} \, \mathrm{V}=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ Here $\mathrm{R}$ is the gas constant, $8.314 \mathrm{~J} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}$. Conditions (A) and (B) are equivalent [2]. In most cases condition (B) is quoted because the equation links three practical properties, $\mathrm{p}$, $\mathrm{V}$ and $\mathrm{T}$. Footnotes [1] \begin{aligned} &\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\frac{[\mathrm{J}]}{\left[\mathrm{m}^{3}\right]}=\frac{[\mathrm{N} \mathrm{m}]}{\left[\mathrm{m}^{3}\right]}=\left[\mathrm{N} \mathrm{m}{ }^{2}\right]=[\mathrm{Pa}] \ &\mathrm{p} \, \mathrm{V}=\left[\mathrm{N} \mathrm{m}{ }^{-2}\right] \,\left[\mathrm{m}^{3}\right]=[\mathrm{N} \mathrm{m}]=[\mathrm{J}] \ &\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}=[\mathrm{mol}] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]=[\mathrm{J}] \end{aligned} [2] From definition (A) and equation (b), $p=T \,\left(\frac{\partial S}{\partial V}\right)_{T}$ We use a Maxwell equation; $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ Hence from equations (i) and (ii), $p=T \,\left(\frac{\partial p}{\partial T}\right)_{v}$ From definition (B), $\left(\frac{\partial \mathrm{p}}{\partial T}\right)_{\mathrm{V}}=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} / \mathrm{V}=\mathrm{p} / \mathrm{T}$ Equations (iii) and (iv) are the same. Hence definition (B) is the integrated form of definition $\mathrm{A}$. The gas constant $\mathrm{R}$ is experimentally determined.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.24%3A_Equation_of_State-_General_Thermodynamics.txt
The properties of gases pose a formidable challenge for chemists who seek to understand their $\mathrm{p}-\mathrm{V}-\mathrm{T}$ properties. Chemists adopt an approach which starts by defining the properties of a (hypothetical) ideal gas (Topics 1220 and 2588). The fact that the properties of a given real gas are not ideal is understood in terms of intermolecular interactions. In understanding the properties of real liquid mixtures and real solutions, the classic approach identifies the properties of the corresponding systems where the thermodynamic properties are defined as ‘ideal’. In the next stage the reasons why the properties of real liquid mixtures and real solutions are not ideal are discussed in terms of the nature and strength of intermolecular interactions. This general approach mimics the approach used to explain why the properties of real gases are not those of a defined ideal gas (see Topic 2588). In this context the van der Waals equation describing the differences between the properties of a real gas and an ideal gas sets the stage for theories describing the differences between the properties of real and ideal liquid mixtures and the properties of real and ideal solutions. Nevertheless we develop here the argument by considering the properties of a single gas, chemical substance $j$. The starting points are two statements concerning an ideal gas. 1. The actual volume of the molecules making up an ideal gas is negligible compared to the total volume of the system. 2. Neither attractive nor long-range repulsive intermolecular forces are present. In a real gas the molecular volume is not negligible. Also cohesive intermolecular forces mean that the pressure exerted on the containing vessel is less than in the case of an ideal gas. Therefore the equation of state requires that the pressure $\mathrm{p}$ is incremented by a quantity proportional to the density or, by a quantity inversely proportional to the volume. The famous van der Waals [1] equation takes the following form [2] where $\mathrm{V}_{j}$ is the molar volume of gas $j$ at pressure $\mathrm{p}$ and temperature $\mathrm{T}$. $\left(\mathrm{p}+\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{2}}\right) \,\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)=\mathrm{R} \, \mathrm{T}$ The van der Waals equation has played an important role in the development of theories describing fluids; i.e. both liquids and gases. The equation has merit in that it involves just two constants, both characteristic of given chemical substance. Further as McGlashan notes, the equation never leads to physical nonsense and does not predict physically absurd results [3]. Similarly Rowlinson comments that the equation is easy to manipulate and never predicts physically absurd results [4]. Chue comments that despite its simplicity the van der Waals equation ‘comprehends’ both liquid and gaseous states [5]. However other authors are not so enthusiastic. For example, Prigogine and Defay comment that the equation is ‘mainly of qualitative interest’ [6]. Similarly Denbigh states that the a-parameter ‘does not have a sound theoretical basis and interpretation of the a-parameter in terms of intermolecular attraction is ‘intuitive’ [7]. Perhaps the expectation that the $\mathrm{p}-\mathrm{V}-\mathrm{T}$ properties of all gases and liquids can be accounted for using two parameters characteristic of each chemical substance is too optimistic. Nevertheless there is merit in reviewing the van der Waals equation. Equation (a) can be written as an equation for pressure $\mathrm{p}$. $\mathrm{p}=\frac{\mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{j}}-\mathrm{b}}-\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{2}}$ This form of equation (a) highlights the role of the parameter $\mathrm{b}$ in describing the effect of molecular size and the role of parameter a in describing inter-molecular cohesion. A plot of $\mathrm{p}$ as function of $\mathrm{V}_{j}$ at fixed temperature has an extremum at the point defined by equation (c) [8]. $\frac{\mathrm{R} \, \mathrm{T}}{\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)^{2}}=\frac{2 \, \mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{3}}$ Or, $\mathrm{T}=\frac{2 \, \mathrm{a} \,\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)^{2}}{\mathrm{R} \, \mathrm{V}_{\mathrm{j}}^{3}}$ Hence using equations (a) and (d) we obtain [9] an equation relating pressure $\mathrm{p}$ to $\mathrm{V}_{j}$ in terms of the two parameters $\mathrm{p}=\mathrm{a} \,\left[\frac{\mathrm{V}_{\mathrm{j}}-2 \, \mathrm{b}}{\mathrm{V}_{\mathrm{j}}^{3}}\right]$ In a family of curves showing $\mathrm{p}$ as a function of $\mathrm{V}_{j}$, one curve has a point of inflexion with a horizontal tangent where both $\frac{\partial p}{\partial V_{m}} \text { and } \frac{\partial^{2} p}{\partial V_{m}^{2}}$ are zero. This point is the critical point [10]. Interesting features based on the van der Waals equation characterise this point [11]. At the critical point [12,13], $V_{j}^{c}=3 \, b$ $p^{c}=a / 27 \, b^{2}$ and $\mathrm{T}^{\mathrm{c}}=8 \, \mathrm{a} / 27 \, \mathrm{R} \, \mathrm{b}$ A classic plot [10] describes the properties of a fixed amount of carbon dioxide in terms of isotherms showing the dependence of pressure on volume. This plot reported by $\mathrm{T}$. Andrews in 1870 showed that $\mathrm{CO}_{2}(\mathrm{g})$ cannot be liquefied solely by the application of pressure at temperatures above $304.2 \mathrm{~K}$. The latter is the critical temperature $\mathrm{T}^{\mathrm{c}}$ for $\mathrm{CO}_{2}$, the critical pressure $\mathrm{p}_{c}$ being the pressure required to liquefy $\mathrm{CO}_{2}$ at this temperature. The molar volume at pressure $\mathrm{p}^{\mathrm{c}}$ and temperature $\mathrm{T}^{\mathrm{c}}$ is the critical molar volume $\mathrm{V}_{\mathrm{j}}^{\mathrm{c}}$. The critical volume is obtained using the Law of Rectilinear Diameters originally described by L. Caillete and E. Mathias. The law requires that the mean density $\rho_{\mathrm{j}}(\mathrm{T})$ of gas and liquid states of a given chemical substance $j$ at common temperature $\mathrm{T}$ is a linear function of $\mathrm{T}$ Thus, $\rho_{\mathrm{j}}(\mathrm{T})=\rho_{\mathrm{j}}\left(\mathrm{T}^{\mathrm{c}} / \mathrm{K}\right)+\alpha \,(\mathrm{T} / \mathrm{K})$ The parameters $\rho_{\mathrm{j}}\left(\mathrm{T}^{\mathrm{c}} / \mathrm{K}\right)$ and $\alpha$ are characteristic of chemical substance $j$; $\rho_{\mathrm{j}}\left(\mathrm{T}^{\mathrm{c}} / \mathrm{K}\right)$ is the critical density of chemical substance $j$, leading to the critical molar volume $\mathrm{V}_{\mathrm{j}}\left(\mathrm{T}^{\mathrm{c}}\right)$ at critical temperature $\mathrm{T}^{\mathrm{c}}$ and critical pressure $\mathrm{p}^{\mathrm{c}}$. Above the critical temperature the plot of pressure $\mathrm{p}$ and against molar volume at a given temperature $\mathrm{T}$ is a smooth curve. Below $\mathrm{T}^{\mathrm{c}}$ and at low pressures, chemical substance $j$ is a gas. With increase in pressure a stage is reached where a given system comprises two phases, gas and liquid. With further increase in pressure the system comprises a liquid. There is no sharp transition between liquid and gaseous states in contrast to that observed on melting a solid. In other words, gas and liquid states for chemical substance $j$ form a continuity of states. The Law of Corresponding States is an interesting concept, following an observation by J. D. van der Waals in 1881. The pressure, volume and temperature for a given gas $j$ are expressed in terms of the critical pressure $\mathrm{p}_{\mathrm{j}}^{\mathrm{c}}$, volume $\mathrm{V}_{\mathrm{j}}^{\mathrm{c}}$, and temperature $\mathrm{T}_{\mathrm{j}}^{\mathrm{c}}$ using three proportionality constants, $\beta_{1}$, $\beta_{2}$ and $\beta_{3}$ respectively. Thus $\mathrm{p}=\beta_{1} \, \mathrm{p}_{\mathrm{j}}^{\mathrm{c}}$ $\mathrm{V}_{\mathrm{j}}=\beta_{2} \, \mathrm{V}_{\mathrm{j}}^{\mathrm{c}}$ $\mathrm{T}_{\mathrm{j}}=\beta_{3} \, \mathrm{T}_{\mathrm{j}}^{\mathrm{c}}$ Hence from equation (a), $\left(\beta_{1} \, p_{j}^{c}+\frac{a}{\left(\beta_{2} \, V_{j}^{c}\right)^{2}}\right) \,\left(\beta_{2} \, V_{j}^{c}-b\right)=R \, \beta_{3} \, T_{j}^{c}$ Using equations (f), (g) and (h) for $\mathrm{p}_{j}^{c}$, $\mathrm{V}_{j}^{c}$ and $\mathrm{T}_{j}^{c}$ the following equation $j$is obtained from equation (m) [14]. $\left(\beta_{1}+\frac{3}{\beta_{2}^{2}}\right) \,\left(3 \, \beta_{2}-1\right)=8 \, \beta_{3}$ Equation (n) is the van der Waals reduced Equation of State for gas $j$; $\beta_{1}$, $\beta_{2}$ and $\beta_{3}$ being the reduced pressure, volume and temperature respectively. Significantly there are no parameters in equation (n) which can be said to be characteristic of a given chemical substance. In other words the equation has a universal character [15]. The van der Waals equations prompted the development of many equations of state. The van der Waals equation can be modified in two simple ways. In one modification it is assumed that $V_{j} \gg b$. The assumption is that attractive intermolecular processes are dominant Hence, $\left[\mathrm{p}+\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{2}}\right] \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}$ Or, $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}-\left(\mathrm{a} / \mathrm{V}_{\mathrm{j}}\right)$ In another approach it is assumed that repulsive intermolecular forces are dominant. Thus, $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}+\mathrm{p} \, \mathrm{b}$ Clausius Equation One criticism of the van der Waal equation is that no account is taken of the possibility that parameters a and b can depend on temperature. Clausius suggested the following equation in which intermolecular attraction is described as inversely proportional to temperature; $a$, $b$ and $c$ are three constants characteristic of a gas $j$. $\left(\mathrm{p}+\frac{\mathrm{a}}{\mathrm{T} \,\left(\mathrm{V}_{\mathrm{j}}+\mathrm{c}\right)^{2}}\right) \,\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)=\mathrm{R} \, \mathrm{T}$ Nevertheless the advantages gained by recognising that attraction might be dependent on temperature are outweighed by problems associated with using this equation. Bertholot Equation In this approach, the term $\left(\mathrm{V}_{j}+c\right)$ in the Clausius equation is replaced by $\mathrm{V}_{j}$. Then $\left(\mathrm{p}+\frac{\mathrm{a}}{\mathrm{T} \, \mathrm{V}_{\mathrm{j}}^{2}}\right) \,\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)=\mathrm{R} \, \mathrm{T}$ Or, $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}+\mathrm{p} \, \mathrm{b}-\frac{\mathrm{a}}{\mathrm{T} \, \mathrm{V}_{\mathrm{j}}}+\frac{\mathrm{a} \, \mathrm{b}}{\mathrm{T} \, \mathrm{V}_{\mathrm{j}}^{2}}$ The van der Waals, Clausius and Berthelot equations are the forerunners of a large family of cubic equations of state; i.e. equations of state that are cubic polynomials in molar volume. Analysis of experimental data prompted the development of the following equation using critical pressure $\mathrm{p}_{\mathrm{j}}^{\mathrm{c}}$ and temperature $\mathrm{T}_{\mathrm{c}}$. $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T} \,\left[1+\frac{\mathrm{a}}{128} \, \frac{\mathrm{p} \, \mathrm{T}_{\mathrm{j}}^{\mathrm{c}}}{\mathrm{p}_{\mathrm{j}}^{\mathrm{c}} \, \mathrm{T}} \,\left(1-6 \, \frac{\left(\mathrm{T}_{\mathrm{j}}^{\mathrm{c}}\right)^{2}}{\mathrm{~T}^{2}}\right)\right]$ Dieterici Equation This modification suggested in 1899 by C. Dieterici attempts to account for the fact that molecules at the wall of a containing vessel have higher potential energy than molecules in the bulk gas. The following equation was proposed. $\mathrm{p} \,\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)=\mathrm{R} \, \mathrm{T} \, \exp \left(-\frac{\mathrm{a}}{\mathrm{R} \, \mathrm{T} \, \mathrm{V}_{\mathrm{j}}}\right)$ The Virial Equation of State The following virial equation was proposed in 1885 by Thiesen. $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T} \,\left[1+\frac{\mathrm{B}(\mathrm{T})}{\mathrm{V}_{\mathrm{j}}}+\frac{\mathrm{C}(\mathrm{T})}{\mathrm{V}_{\mathrm{j}}^{2}}+\ldots \ldots\right]$ Following a suggestion in 1901 by H. K. Onnes, B(T), C(T),… are called virial coefficients. A modern account of equations of state is given in reference [16]. Boyle Temperature Boyle’s Law requires that the product $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}$ at fixed temperature is independent of pressure. In the case of hydrogen, the product $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}$ at $273 \mathrm{~K}$ increases with increase in pressure. However for many gases (e.g. nitrogen and carbon dioxide) the product $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}$ at fixed $\mathrm{T}$ decreases with increase in pressure, passes through a minimum and then increases. For a given gas the minimum moves to lower pressures with increase in temperature until at high temperatures no minimum is observed. This is the Boyle temperature, which is characteristic of a given gas [11a]. The van der Waals equation offers an explanation of the pattern. Thus from equation (b), $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T} \,\left(\frac{\mathrm{V}_{\mathrm{j}}}{\mathrm{V}_{\mathrm{j}}-\mathrm{b}}\right)-\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}} \label{x}$ Equation \ref{x} is differentiated with respect to pressure at constant temperature. If the plot of $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}$ against $\mathrm{p}$ passes through zero at temperature $\mathrm{T}_{\mathrm{B}}$, then $\mathrm{T}_{\mathrm{B}}$ is given by equation (y) [17]. $\mathrm{T}_{\mathrm{B}}=\mathrm{a} / \mathrm{R} \, \mathrm{b}$ Therefore in terms of Equation (a), a low Boyle temperature is favoured by small $\mathrm{a}$ and large $\mathrm{b}$ parameters. Footnotes [1] For a reproduction of a portrait of Johannes van der Waals (1837-1923) see D. Kondepudi and I. Prigogine, Modern Thermodynamics, Wiley , Colchester, 1998, page 15. [2] \begin{aligned} &V_{j}=\left[m^{3} \mathrm{~mol}^{-1}\right] \quad b=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]\ &\left(\mathrm{p}+\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{2}}\right)=\left(\left[\mathrm{Nm}^{-2}\right]+\frac{\mathrm{a}}{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]^{2}}\right) \end{aligned} Then, \begin{aligned} &\mathrm{a}=\left[\mathrm{N} \mathrm{m} \mathrm{m}^{-2}\right] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]^{2}=\left[\mathrm{N} \mathrm{mol}^{-2} \mathrm{~m}^{4}\right]\ &\mathrm{R} \, T=\left[\mathrm{J} \mathrm{mol} \mathrm{K}^{-1}\right] \,[\mathrm{K}]=\left[\mathrm{J} \mathrm{mol}^{-1}\right] \end{aligned} Then \begin{aligned} &\left(\left[\mathrm{N} \mathrm{m}^{-2}\right]+\frac{\left[\mathrm{N} \mathrm{mol}^{-2} \mathrm{~m}^{4}\right]}{\left[\mathrm{mol}^{-1} \mathrm{~m}^{3}\right]^{2}}\right) \,\left[\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]-\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]\right] \ &=\left(\left[\mathrm{Nm}^{-2}\right]+\left[\mathrm{Nm}^{-2}\right]\right) \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]=\left[\mathrm{N} \mathrm{m} \mathrm{mol}^{-1}\right]=\left[\mathrm{J} \mathrm{mol}^{-1}\right] \end{aligned} [3] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979, page 176. [4] J. S. Rowlinson, Liquids and Liquid Mixtures, Butterworths, London, 2nd edn., 1969, page 66. [5] S. H. Chue, Thermodynamics, Wiley, Chichester, 1977, page 136. [6] I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D.H.Everett, Longmans Green, London, 1954, p. 145. [7] K. Denbigh, The Principles of Chemical Equilibrium, Cambridge University Press. 3rd. edn. 1971, page 119. [8] $\frac{\mathrm{dp}}{\mathrm{dV}}=-\frac{\mathrm{R} \, \mathrm{T}}{\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)^{2}}+\frac{2 \, \mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{3}}$ At an extremum, $\frac{\mathrm{dp}}{\mathrm{dV}_{\mathrm{j}}}=0$; then, $\frac{\mathrm{R} \, \mathrm{T}}{\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)^{2}}=\frac{2 \, \mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{3}}$ [9] \begin{aligned} &\left(p+\frac{a}{V_{j}^{2}}\right) \,\left(V_{j}-b\right)=R \, 2 \, a \, \frac{\left(V_{j}-b\right)^{2}}{V_{j}^{3}} \ &p+\frac{a}{V_{j}^{2}}=2 \, a \, \frac{\left(V_{j}-b\right)}{V_{j}^{3}} \end{aligned} Or, $\mathrm{p}=\mathrm{a} \,\left[\frac{2 \, \mathrm{V}_{\mathrm{j}}-2 \, \mathrm{b}-\mathrm{V}_{\mathrm{j}}}{\mathrm{V}_{\mathrm{j}}^{3}}\right]$ [10] T.Andrews, Phil. Mag.,1870,[4],39,150. [11] See for example, 1. S. Glasstone, Textbook of Physical Chemistry, MacMillan, London, 1948, 2nd. edn., page 435. 2. N. K.Adam, Physical Chemistry, Oxford, 1956, page 83; $\mathrm{CO}_{2}$. 3. J. K. Roberts and A. R. Miller, Heat and Thermodynamics, Blackie, London, 1951, page 110; $\mathrm{CO}_{2}$. 4. P. A . Rock, Chemical Thermodynamics, MacMillan, Toronto, 1969; $\mathrm{H}_{2}\mathrm{O}$. 5. E. F. Caldin, Chemical Thermodynamics, Oxford, 1958, chapter III. 6. J. S. Winn, Physical Chemistry, Harper Collins, New York, 1995, chapter 1. 7. M. L. McGlashan, Chemical Thermodynamics, Academic Press, London, 1979, page 132; xenon. 8. P. Atkins and J. de Paula, Physical Chemistry, Oxford, 2002, 7th edn., Section 1.4. [12] We rewrite equation (e) in the following form. $p=a \,\left[V_{j}-2 \, b\right] \, V_{j}^{-3}$ Or, $\mathrm{p}=\mathrm{a} \,\left\{\left[\mathrm{V}_{\mathrm{j}}-2 \, \mathrm{b}\right] \,\left[\mathrm{V}_{\mathrm{j}}^{-3}\right]\right\}$ Then \begin{aligned} \frac{\mathrm{dp}}{\mathrm{dV}} &=\mathrm{a} \,\left[\mathrm{V}_{\mathrm{j}}^{-3}+\left(\mathrm{V}_{\mathrm{j}}-2 \, \mathrm{b}\right) \,(-3) \, \mathrm{V}_{\mathrm{j}}^{-4}\right] \ &=\mathrm{a} \,\left[\mathrm{V}_{\mathrm{j}}^{-3}-3 \, \mathrm{V}_{\mathrm{j}}^{-3}+6 \, \mathrm{b} \, \mathrm{V}_{\mathrm{j}}^{-4}\right] \ &=\mathrm{a} \,\left[-2 \, \mathrm{V}_{\mathrm{j}}^{-3}+6 \, \mathrm{b} \, \mathrm{V}_{\mathrm{j}}^{-4}\right] \end{aligned} $\frac{d p}{d V_{j}}=0$ where $\left[-2 \, \mathrm{V}_{\mathrm{j}}^{-3}+6 \, \mathrm{b} \, \mathrm{V}_{\mathrm{j}}^{-4}\right]=0$ Or, $\frac{1}{V_{j}^{3}}=\frac{3 \, b}{V_{j}^{4}}$ Or, $V_{j}^{c}=3 \, b$ From equation (e), $p^{c}=\frac{a \, b}{27 \, b^{3}}$ or, $\mathrm{p}^{\mathrm{c}}=\frac{\mathrm{a}}{27 \, \mathrm{b}^{2}}$ Then from equation (d), $T^{c}=\frac{2 \, a \,(3 \, b-b)^{2}}{R \, 27 \, b^{3}}=\frac{8 \, a \, b^{2}}{27 \, R \, b^{3}}=\frac{8 \, a}{27 \, R \, b}$ [13] \begin{aligned} &{\mathrm{V}_{\mathrm{j}}^{\mathrm{c}}=3 \, \mathrm{b}=[1] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]} \ &\mathrm{p}^{c}=\frac{\left[\mathrm{N} \mathrm{m}^{4} \mathrm{~mol}^{-2}\right]}{[1] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]^{2}}=\left[\mathrm{N} \mathrm{m}^{4} \mathrm{~m}^{-6} \mathrm{~mol}^{-2} \mathrm{~mol}^{2}\right]=\left[\mathrm{N} \mathrm{m}^{-2}\right] \ &\mathrm{T}^{\mathrm{c}}=\frac{[1] \,\left[\mathrm{N} \mathrm{m} \mathrm{mol}^{-2}\right]}{[1] \,\left[\mathrm{J} \mathrm{mol} \mathrm{m}^{-1}\right] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]}=\frac{[\mathrm{N} \mathrm{m}]}{[\mathrm{N} \mathrm{m}]} \,[\mathrm{K}]=[\mathrm{K}] \end{aligned} [14] $\left[\beta_{1} \, \frac{\mathrm{a}}{27 \, \mathrm{b}^{2}}+\frac{\mathrm{a}}{\left(\beta_{2} \, 3 \, \mathrm{b}\right)^{2}}\right] \,\left[\beta_{2} \, 3 \mathrm{~b}-\mathrm{b}\right]=\mathrm{R} \, \frac{8 \, \mathrm{a} \, \beta_{3}}{27 \, \mathrm{R} \, \mathrm{b}}$ [15] F. T. Wall, Chemical Thermodynamics, W. H. Freeman, San Francisco, 1965, page 169. [16] S. I. Sandler, H. Orbey and B.-I. Lee , Models for Thermodynamics and Phase Equilibria Calculations , ed. S. I. Sandler, Marcel Dekker, New York, 1994, pp.87-186. [17] \begin{aligned} &\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T} \, \frac{\mathrm{V}_{\mathrm{j}}}{\mathrm{V}_{\mathrm{j}}-\mathrm{b}}-\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}} \ &{\left[\frac{\partial\left(\mathrm{p} \, \mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right]=\left[\mathrm{R} \, \mathrm{T} \,\left(\frac{1}{\left.\mathrm{~V}_{\mathrm{j}}-\mathrm{b}\right)}\right)-\mathrm{R} \, \mathrm{T} \,\left(\frac{\mathrm{V}_{\mathrm{j}}}{\left(\mathrm{V}_{\mathrm{j}}-\mathrm{b}\right)^{2}}\right)+\frac{\mathrm{a}}{\mathrm{V}_{\mathrm{j}}^{2}}\right] \,\left(\frac{\partial \mathrm{V}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}} \end{aligned} Then at $\left[\frac{\partial\left(\mathrm{p} \, \mathrm{V}_{\mathrm{j}}\right)}{\partial \mathrm{p}}\right]=0$, $\left[R \, T \,\left(\frac{V_{j}-b-V_{j}}{\left(V_{j}-b\right)^{2}}\right)+\frac{a}{V_{j}^{2}}\right]=0$ Or, $R \, T=\frac{a}{b} \, \frac{\left(V_{j}-b\right)^{2}}{V_{j}^{2}}$ If the minimum occurs where $\mathrm{p}$ is zero (i.e. where $\mathrm{V}_{j}$ is infinitely large), $\mathrm{R} \, \mathrm{T}_{\mathrm{B}}=\mathrm{a} / \mathrm{b}$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.26%3A_Equation_of_State_-_Real_Gases%2C_van_der_Waals%2C_and_Other_Equations.txt
This theorem emerges from theories concerned with differential equations. The theorem finds many applications in thermodynamics. In particular the theorem concerned with homogeneous functions of the first degree is important. This theorem can be stated as follows [1]. $\mathrm{f}(\mathrm{k} \, \mathrm{x}, \mathrm{k} \, \mathrm{y}, \mathrm{k} \, \mathrm{z})=\mathrm{k} \, \mathrm{f}(\mathrm{x}, \mathrm{y}, \mathrm{z}) \label{a}$ By way of illustration we consider a liquid mixture volume $\mathrm{V}$ prepared using $\mathrm{n}_{1}$ and $\mathrm{n}_{2}$ moles of liquid 1 and 2. If we had used $2 . \mathrm{n}_{1}$ and $2 . \mathrm{n}_{2}$ moles of liquids 1 and 2, then the final volume would have been $2 . \mathrm{V}$. The important theorem allows us to set down the following descriptions. For a system comprising $\mathrm{i}$-chemical substances, it follows that $\mathrm{V}=\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}\label{b}$ where partial molar volume $\mathrm{V}_{\mathrm{j}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \label{c}$ It should be noted that some thermodynamic monographs, when citing Equation \ref{b}, include the phrase ‘at constant temperature and pressure’. Other monographs do not include this phrase on the grounds that the isobaric - isothermal condition is included in the definition of the partial derivative in Equation \ref{c}. In practice nothing is lost by including this phrase simply to indicate that the analysis is concerned with the properties of systems in the $\mathrm{T} – \mathrm{p}$ - composition domain. A similar analysis in the context of Gibbs energies leads to the following two equations and the definition of chemical potentials. $G=\sum_{j=1}^{j=i} n_{j} \, \mu_{j} \label{d}$ where chemical potential $\mu_{j}=\left(\frac{\partial G}{\partial n_{j}}\right)_{T, p, n(i \neq j)} \label{e}$ Footnotes [1] R. J. Tykodi, J. Chem. Educ.,1982,59,557. 1.14.28: First Law of Thermodynamics The first law of thermodynamics centres on the concept of energy. In its broadest sense, the law requires that the energy of the universe is constant. This is a rather overwhelming statement. A more attractive statement requires that the (internal) thermodynamic energy $\mathrm{U}$ of a chemistry laboratory is constant: $\mathrm{U} = \text { constant }$ The latter statement is the principle of conservation of energy; energy can be neither created nor destroyed. All that a chemist can do, during an experiment using a closed reaction vessel, is to watch energy ‘move ‘ between system and surroundings. As a consequence of equation (a), we state that $\Delta \mathrm{U}(\text { system })=-\Delta \mathrm{U}(\text { surroundings })$ We can not know the actual energy of a closed system although we know that it is an extensive property of the system [1]. In describing energy changes we need a convention. So we use the acquisitive convention describing all changes in terms of how a system is affected. Thus the statement $\Delta \mathrm{U} < 0$ means that the energy of a given system decreases during a given process; e.g. chemical reaction. Footnote [1] In principle it is possible to know the total energy of a given system using a scale in conjunction with Einstein’s famous equation, $\mathrm{E} = \mathrm{m} \, \mathrm{c}^{2}$. However, the mass corresponding to $1 \mathrm{~kJ}$ is only about $10^{-14} \mathrm{~kg}$. 1.14.29: Functions of State One Chemical Substance For a system containing one chemical substance we define the volume using equation (a). $\mathrm{V}=\mathrm{V}[\mathrm{T}, \mathrm{p}, \mathrm{n},]$ The variables in the square brackets are the Independent Variables. The term independent' means that, within limits [1], we can change $\mathrm{T}$ independently of the pressure and $\mathrm{n}_{j}$; change $\mathrm{p}$ independently of $\mathrm{T}$ and $\mathrm{n}_{\mathrm{j}}$; and $\mathrm{n}_{\mathrm{i}}$ independently of $\mathrm{T}$ and $\mathrm{p}$. There are some restrictions in our choice of independent variables. At least one variable must define the amount of all substances in the system and one variable must define the hotness' of the system. Actually there is merit in writing equation (a) in terms of three intensive variables which in turn define, for example the, the molar volume of liquid chemical substance 1 at specified temperature and pressure, $\mathrm{V}_{1}^{*}(\ell)$. $\mathrm{V}_{1}^{*}(\ell)=\mathrm{V}(\ell)\left[\mathrm{T}, \mathrm{p}, \mathrm{x}_{1}=1\right]$ Two Chemical Substances If the chemical composition of a given closed system is specified in terms of two chemical substances 1 and 2, four independent variables $\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}\right]$ define the dependent variable $\mathrm{V}$ [2]. Thus $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}\right]$ i - Chemical Substances For a system containing i-chemical substances where the amounts can be independently varied, the dependent variable $\mathrm{V}$ is defined by the following equation. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2} \ldots \mathrm{n}_{\mathrm{i}}\right]$ In a general analysis, we start out with a closed system having Gibbs energy at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, molecular composition (organisation) $\xi$ and affinity for spontaneous change $\mathrm{A}$. We define the Gibbs energy as follows. $\mathrm{G}=\mathrm{G}[\mathrm{T}, \mathrm{p}, \boldsymbol{\xi}]$ In the state defined by equation (e), there is an affinity for spontaneous change (chemical reaction) $\mathrm{A}$. We imagine that starting with the system in the state defined by equation (e), it is possible to change the pressure and perturb the system to a series of neighbouring states for which the affinity for spontaneous change remains constant. The differential dependence of $\mathrm{G}$ on pressure for the original state along the path at constant affinity is given by $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \mathrm{A}}$. Returning to the original state characterised by $\mathrm{T}$, $\mathrm{p}$ and $\xi$, we imagine it is possible to perturb the system by a change in pressure in such a way that the system remains at fixed extent of reaction $\xi$. The differential dependence of $\mathrm{G}$ on pressure for the original state along the path at constant $\xi$ is given by $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \xi}$. We explore these dependences of $\mathrm{G}$ on pressure at fixed temperature and at (i) fixed composition $\xi$ and (ii) fixed affinity for spontaneous change, $\mathrm{A}$. $(\partial G / \partial p)_{\mathrm{T}, \mathrm{A}}$ and $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \xi}$ are related using a standard calculus operation [1]. Thus at fixed temperature, $\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{A}}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right]_{\xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{p}} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{p}}$ This interesting equation shows that the differential dependence of Gibbs energy (at constant temperature) on pressure at constant affinity for change does NOT equal the corresponding dependence at constant extent of chemical reaction. This is inequality is not surprising. But our interest is drawn to the case where the system under discussion is, at fixed temperature and pressure, at thermodynamic equilibrium where $\mathrm{A}$ is zero, $\mathrm{d} \xi / \mathrm{dt}$ is zero, the Gibbs energy is a minimum AND significantly $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero. We conclude that $\mathrm{V}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{T}, \mathrm{A}=0}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{T}, \xi(\mathrm{eq})}$ This rather long winded argument confirms that volume $\mathrm{V}$ is a state variable, the dependence of $\mathrm{G}$ on pressure for differential displacement at constant '$\mathrm{A} =0$' and $\sim^{e} q$ being identical. These comments may seem trivial but the point is made if we go on to consider the volume of a system as a function of temperature at constant pressure. We again use a calculus operation [1] to derive the relationship in equation (h). $\left[\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right]_{\mathrm{A}}=\left[\frac{\partial \mathrm{V}}{\partial \mathrm{p}}\right]_{\xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right]_{\xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{p}} \,\left[\frac{\partial \mathrm{V}}{\partial \xi}\right]_{\mathrm{p}}$ We are not surprised to discover that in general terms the differential dependence of $\mathrm{V}$ on temperature at constant affinity does not equal the differential dependence of $\mathrm{V}$ on temperature at constant composition/organisation. Indeed unlike the simplification we used in connection with equation (e), we cannot assume that the volume of reaction $(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}, \mathrm{p} p}$ is zero at equilibrium. In other words for a closed system at thermodynamic equilibrium at fixed $\mathrm{T}$ and fixed $\mathrm{p}$ {where $\mathrm{A}=0, \xi=\xi^{\mathrm{eq}} \text { and } \mathrm{d} \xi / \mathrm{dt}=0$}, there are two thermal expansions [2]. We consider a closed system in equilibrium state I defined by the set of variables, $\left\{\mathrm{T}[\mathrm{I}], \mathrm{p}, \mathrm{A}=0, \xi^{\mathrm{eq}}[\mathrm{I}]\right\}$. The equilibrium composition is represented by $\xi^{\mathrm{eq}}[\mathrm{I}]$ at zero affinity for spontaneous change. This system is perturbed to nearby state at constant pressure . 1. The state I is displaced to a nearby equilibrium state II defined by the set of variables $\left\{\mathrm{T}[\mathrm{I}]+\delta \mathrm{T}, \mathrm{p}, \mathrm{A}=0, \xi^{\mathrm{eq}}[\mathrm{I}]\right\}$. This equilibrium displacement is characterised by a volume change. $\Delta \mathrm{V}(\mathrm{A}=0)=\mathrm{V}[\mathrm{II}]-\mathrm{V}[\mathrm{I}]$ We record the equilibrium thermal expansion, $\mathrm{E}(\mathrm{A}=0)=\left[\frac{\mathrm{V}(\mathrm{II})-\mathrm{V}(\mathrm{I})}{\Delta \mathrm{T}}\right]$ The equilibrium expansivity, $\alpha(\mathrm{A}=0)=\mathrm{E}) \mathrm{A}=0) / \mathrm{V}$ In order for the system to move from one equilibrium state, I with composition $\xi^{\mathrm{eq}}[\mathrm{I}]$ to another equilibrium state, II with composition $\xi^{\mathrm{eq}}[\mathrm{II}]$, the chemical composition and /or molecular organisation changes. The term expansion' indicates the isobaric dependence of volume on temperature, $\mathrm{E}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}=\left[\mathrm{m}^{3} \mathrm{~K}^{-1}\right]$ $\mathrm{E}$ is an extensive variable. The corresponding volume intensive variable is the expansivity, $\alpha$. $\alpha=\frac{1}{\mathrm{~V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{p}$ where $\alpha=\frac{1}{\left[\mathrm{~m}^{3}\right]} \,\left[\frac{\mathrm{m}^{3}}{\mathrm{~K}}\right]=\left[\mathrm{K}^{-1}\right]$ Hence we define the frozen' expansion, $\mathrm{E}(\sim=\text { fixed })$. An alternative name is the instantaneous expansion because, practically, we would have to change the temperature at such a high rate that there is no change in molecular composition or molecular organisation in the system. Thus, $\mathrm{E}(\xi=\text { fixed })=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \xi}$ Further $\alpha(\xi=\text { fixed })=\frac{1}{\mathrm{~V}} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \xi}$ Similar comments apply to isothermal compressibilities, $\kappa_{\mathrm{T}}$; there are two limiting properties, $\kappa_{\mathrm{T}}(\mathrm{A}=0)$ and $\kappa_{\mathrm{T}}(\xi)$. In order to measure $\kappa_{\mathrm{T}}(\xi)$ we have to change the pressure in a infinitely short time. The entropy $\mathrm{S}$ at fixed composition is given by the partial differential $-\left(\frac{\partial G}{\partial T}\right)_{p, 5}$ and, at constant affinity of spontaneous change by $-\left(\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}}$. At equilibrium where $\mathrm{A} = 0$, the equilibrium entropy, $S(A=0)=-\left(\frac{\partial G}{\partial T}\right)_{p, A=0}$. We carry over the argument described above but now concerned with a system characterised by $\mathrm{T}, \mathrm{p}, \xi$ which is perturbed by a change in temperature. We consider two pathways, at constant $\mathrm{A}$ and at constant $\xi$. $\left[\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ But at equilibrium, $\mathrm{A}$ which equals $-\left[\frac{\partial G}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ is zero and so $\mathrm{S}(\mathrm{A}=0)=\mathrm{S}\left(\xi^{\mathrm{eq}}\right)$. Then just as for volumes, the entropy of a system is not a property concerned with pathways between states; entropy is a function of state. Another important link involving Gibbs energy and temperature is provided by the Gibbs-Helmholtz equation. We explore the relationship between changes in ($\mathrm{G}/\mathrm{T})$ at constant affinity and fixed $\xi$ following a change in temperature. Thus, $\left[\frac{\partial(\mathrm{G} / \mathrm{T})}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}}=\left[\frac{\partial(\mathrm{G} / \mathrm{T})}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi}-\frac{1}{\mathrm{~T}} \,\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ But at equilibrium $\mathrm{A}$ which equals $-(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero. Then $\mathrm{H}(\mathrm{A}=0)=\mathrm{H}\left(\xi^{\mathrm{eq}}\right)$. In other words the variable, enthalpy is a function of state. This is not the case for isobaric heat capacities. Thus, $\left[\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}}=\left[\frac{\partial \mathrm{H}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right]_{\mathrm{p}, \xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{T}, \mathrm{p}} \,\left[\frac{\partial \mathrm{H}}{\partial \xi}\right]_{\mathrm{T}, \mathrm{p}}$ We cannot assume that the triple product term in equation (q) is zero. Hence there are two limiting isobaric heat capacities; the equilibrium isobaric heat capacity $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$ and the frozen' isobaric heat capacity $\mathrm{C}_{\mathrm{p}}\left(\xi^{\mathrm{eq}}\right)$. In other words, an isobaric heat capacity is not a function of state because it is concerned with a pathway between states. Footnotes [1] The phrase independent variable' is important. With reference to the properties of an aqueous solution containing ethanoic acid, the number of components for such a solution is 2, water and ethanoic acid. The actual amounts of ethanoic acid, water, ethanoate and hydrogen ions are determined by an equilibrium constant which is an intrinsic property of this sytem at a given $\mathrm{T}$ and $\mathrm{p}$. From the point of view of the Phase Rule, the number of components equals two. For the same reason when we consider the volume of a system containing $\mathrm{n}$ moles of water we do not take account of evidence that water partly self-dissociates into $\mathrm{H}^{+} (\mathrm{aq})$ and $\mathrm{OH}^{-} (\mathrm{aq})$ ions. [2] In terms of the Phase Rule, we note that for two components (C = 2) and one phase (P = 1) , the number of degrees of freedom F equals equals two. These degrees of freedom refer to intensive variables. Hence for a solution where chemical substance 1 is the solvent and chemical substance 2 is the solute, the system is defined by specifying by the three (intensive) degrees of freedom, \(\mathrm{T}, \mathrm{p} and, for example, solute molality.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.27%3A_Euler%27s_Theorem.txt
Thermodynamics asserts that the energy of a closed system increases if 1. heat $\mathrm{q}$ flows from the surroundings into a system and Separation of the heat term from the work term is extremely important in the context of the Second Law of Thermodynamics. Heat flows spontaneously from high to low temperatures, the word ‘spontaneous’ being absolutely crucial in the context of the Second Law. There are many ways in which the surroundings can do work on a system. At this stage we note the distinction which is drawn between the three variables $\mathrm{U}$, $\mathrm{q}$ and $\mathrm{w}$. [The point is emphasized by use of upper and lower case letters.] The variables $\mathrm{q}$ and $\mathrm{w}$ describe pathways which can result in a change in thermodynamic energy. We make the point by rewriting equation (a) to show the change in thermodynamic energy on going from state I to state II. Thus, $\Delta \mathrm{U}=\mathrm{U}(\mathrm{II})-\mathrm{U}(\mathrm{I})=\mathrm{q}+\mathrm{w}$ If for example $\Delta \mathrm{U} = 100 \mathrm{~J}$, this can be a consequence of many pathways between state I and state II: e.g. 1. $\mathrm{q} = 50 \mathrm{~J}, \mathrm{~w} = 50 \mathrm{~J}$, 2. $\mathrm{q} = 0 \mathrm{~J}, \mathrm{~w} = 100 \mathrm{~J}$ and 3. $\mathrm{q} = 150 \mathrm{~J}, \mathrm{~w} = −50 \mathrm{~J}$. Hence $\mathrm{U}$ is a function of state (or, state variable) although $\mathrm{q}$ and $\mathrm{w}$ are not state variables. This is a triumph of the First Law of Thermodynamics. The task faced by chemists is to identify and describe quantitatively the actual pathway accompanying, for example, a given chemical reaction. Equation (a) signals the energy difference $\Delta \mathrm{U}$ between two states which might involve a comparison of the energies at the start and finish of a chemical reaction in a closed system. In developing our argument there is merit in considering the change in energy of the original system following a small change along the overall reaction pathway. We consider a closed reaction vessel containing ethyl ethanoate (aq; $0.1 \mathrm{~mol}$) and NaOH(aq; excess) . Spontaneous chemical reaction leads to hydrolysis of the ester to form EtOH(aq). The change in thermodynamic energy $\Delta \mathrm{U}$ equals $\mathrm{U}(\mathrm{II}) − \mathrm{~U}(\mathrm{I})$. We subdivide the total chemical reaction into small steps where the change in composition, (i.e. $\mathrm{d}\xi$) is accompanied by a change in thermodynamic energy $\mathrm{dU}$. $\Delta \mathrm{U}=\int_{\text {state } \mathrm{I}}^{\text {state II }} \mathrm{dU}$ If the volume of the system changes by the differential amount $\mathrm{dV}$ such that the pressure within the closed system equals the confining pressure $\mathrm{p}$ [2], $\mathrm{w}=-\mathrm{p} \, \mathrm{dV}$ Then [3], $\mathrm{dU}=\mathrm{q}-\mathrm{p} \, \mathrm{dV}$ We write equation (e) in the following form; $\mathrm{q}=\mathrm{dU}+\mathrm{p} \, \mathrm{dV}$ The right hand side of equation (f) contains the differential changes in two extensive state variables, $\mathrm{U}$ and $\mathrm{V}$. Consequently heat $\mathrm{q}$ is precisely defined by the changes in thermodynamic energy and volume at pressure $\mathrm{p}$. Footnotes [1] The ‘equivalence ‘ of heat and work was first demonstrated in many experiments carried out in the 19th Century by James Joule, the son of a brewer (Salford, England). Joule showed that by doing work on a thermally isolated system the temperature of the latter increases. In other words, doing work on a system is equivalent to passing heat into the system. The SI unit of energy is the joule, symbol $\mathrm{J}$; $\mathrm{J} \equiv \mathrm{kg} \mathrm{m}^{2} \mathrm{~s}^{-2}$ Sometimes one reads that thermodynamics is not concerned with ‘time’. However the concept of energy and the unit of energy involves ’time’. Of course the origins of these concepts are classical mechanics and accompanying discussion of potential and kinetic energies. [2] $\mathrm{p} \, \mathrm{V}=\left[\mathrm{N} \mathrm{m}{ }^{-2}\right] \,\left[\mathrm{m}^{3}\right]=[\mathrm{N} \mathrm{m}]=[\mathrm{J}]$ [3] The fundamental link between heat and work was established by Joule. Interestingly the link between heat and work was apparent previously to A. Haller who suggested that human bodies are heated by the friction between solid particles in the blood passing through the capillaries in the lungs; see comments by 1. P. Epstein, Textbook of Thermodynamics, Wiley, New York, 1937. 2. M. A. Paul, Principles of Chemical Thermodynamics, McGraw-Hill, New York, 1957, p52.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.30%3A_Heat%2C_Work%2C_and_Energy.txt
Gibbs energy is defined with practical chemistry in mind because the definition centres on closed systems held at constant temperature and constant pressure. A similar interest prompt definition of the Helmholtz energy, symbol $\mathrm{F}$. By definition, $\mathrm{F}=\mathrm{U}-\mathrm{T} \, \mathrm{S}$ Physicists use the term ‘Helmholtz Function’. The old term ‘Helmholtz free energy’ is not encouraged [1]. If a closed system is displaced to a neighboring state, the differential change in Helmholtz energy is given by equation (b). $\mathrm{dF}=\mathrm{dU}-\mathrm{T} \, \mathrm{dS}-\mathrm{S} \, \mathrm{dT}$ But the differential change in thermodynamic energy is given by the Master Equation. By incorporating the latter into equation (b), we obtain equation (c), the memorable “all-minus” equation. \begin{aligned} &\mathrm{dF}=-\mathrm{S} \, \mathrm{dT}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi \ &\mathrm{A} \, \mathrm{d} \xi \geq 0 \end{aligned} At fixed temperature and fixed volume (isothermal and isochoric conditions), $\mathrm{dF}=-\mathrm{A} \, \mathrm{d} \xi$ All spontaneous processes at fixed temperature and fixed volume lower the Helmholtz energy of a closed system. In practical terms, for a closed system held at constant volume and temperature, chemical reaction (molecular reorganization) lowers the Helmholtz energy of the system [2]. We presume that the pressure inside the reaction vessel will change, decreasing for some systems and increasing for other systems. As it stands thermodynamics offers no generalization concerning how the pressure changes. In fact if we want to use the Helmholtz energy as an indicator of the direction of spontaneous change we would build the reaction vessel with thick steel walls. This is a practical possibility and so the Helmholtz energy is a practical thermodynamic potential. For equilibrium transformations (i.e. at constant A = zero), $S=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{T}}\right)_{\mathrm{V}, \mathrm{A}=0}$ Similarly [3] $\mathrm{p}=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{V}}\right)_{\mathrm{T}, \mathrm{A}=0}$ Footnotes [1] Since the product $\mathrm{T} \, \mathrm{S}$ is the linked energy, equation (a) shows that $\mathrm{F}$ is the ‘free energy’ of the system. [2] An interesting literature discusses the analysis of experimental data in terms of an isochoric condition. 1. Liquid transport; A. F. M. Barton, Rev. Pure Appl. Chem., 1971, 21, 49. 2. Ionic Conductances in Solution. F. Barreira and G. J. Hills, Trans. Faraday Soc.,1968, 64,1539, and references therein 3. Chemical kinetics 1. M. G. Evans and M. Polanyi, Trans. Faraday Soc.,1935, 31, 873. 2. E. Whalley, Ber. Bunsenges. Phys. Chem., 1966. 70, 958, and references therein. 3. P. G. Wright, J. Chem. Soc. Faraday Trans. 1, 1986, 82, 2557, and 2563. 4. L. M. P. C. Albuquerque and J. C. R. Reis, J. Chem. Soc. Faraday Trans.1, 1989, 85, 207; 1991, 87, 1553. 5. J. B. F. N. Engberts, J. R. Haak, M. J. Blandamer, J. Burgess and H. J. Cowles, J. Chem. Soc .Perkin Trans, 2.1990,1059, and references therein. 4. Chemical Equilibria. M. J. Blandamer, J. Burgess, B. Clarke and J. M. W. Scott, J. Chem. Soc. Faraday 1984, 80, 3359, and references therein.. [3] A similar set of equations describes the differential change in $\mathrm{F}$ at constant molecular composition (molecular organization). Thus $\mathrm{S}=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{T}}\right)_{\mathrm{V}, \xi} \text { and, } \mathrm{p}=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{V}}\right)_{\mathrm{T}, \xi}$ 1.14.32: Hildebrand Solubility Parameter The cohesive energy density (c.e.d.) of a liquid is defined by equation (a). $\text { c.e.d. }=\Delta_{\text {vap }} \mathrm{U}^{0} / \mathrm{V}^{*}(\ell)$ $\Delta_{\text {vap }} \mathrm{U}^{0}$ is the change in thermodynamic energy when one mole of a given chemical substance passes from the liquid to the vapor state. The square root of the c.e.d. for liquid $j$ is the Hildebrand solubility parameter for that liquid. $\delta=(\text { c.e.d. })^{1 / 2}$ $\delta$ can be expressed in many units but following the original definition the customary unit is $\left(\mathrm{cal}^{1 / 2} \mathrm{~cm}^{-3 / 2}\right)$. Property $\delta$ provides an estimate of cohesion within a given liquid. The idea goes a little further in terms of understanding solubilities. A clever idea is based on the following argument. Consider two liquids $\mathrm{A}$ and $\mathrm{B}$. We want to take a small sample of liquid $\mathrm{A}$ (as a solute) and dissolve in liquid $\mathrm{B}$ as the solvent. Within liquid $\mathrm{A}$ the intermolecular interactions $\mathrm{A} \ldots \(\mathrm{A}$ are responsible for the cohesion within this chemical substance. Similarly within liquid $\mathrm{B}$, $\mathrm{B} - \mathrm{~B}$ intermolecular forces are responsible for the cohesion within liquid $\mathrm{B}$. If $\mathrm{B} - \mathrm{~B}$ interactions are much stronger than $\mathrm{A} - \mathrm{~A}$ and $\mathrm{A} - \mathrm{~B}$ intermolecular interactions it is likely that $\mathrm{A}$ will not be soluble in liquid $\mathrm{B}$. Similarly if $\mathrm{A} - \mathrm{~A}$ interactions are stronger than $\mathrm{B} - \mathrm{~B}$ and $\mathrm{A} - \mathrm{~B}$ interactions it is likely that $\mathrm{A}$ will not be soluble in liquid $\mathrm{B}$. If substance $\mathrm{A}$ is to be soluble in liquid $\mathrm{B}$, their cohesive energy densities should be roughly equal.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.31%3A_Helmholtz_Energy.txt
The term ’infinite dilution‘ is often encountered in reviewing the properties of solutions. However some caution has to be exercised when this term is used [1]. There is merit in distinguishing between the properties of aqueous solutions containing simple neutral solutes and those containing salts because the impact of solute - solute interactions plays an important role in the analysis. Further we need to distinguish the properties of solutes and solvents. Neutral Solutes: The Solute The chemical potential of solute $j$, $\mu_{j}(\mathrm{aq})$ is related to the composition of the solution, molality $\mathrm{m}_{j}$, at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ which is assumed to be close to the standard pressure. $\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Or $\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\gamma_{\mathrm{j}}\right)$ For simple solutes in aqueous solutions, $\ln \left(\gamma_{j}\right)$ is a linear function of the molality $\mathrm{m}_{j}$. $\ln \left(\gamma_{\mathrm{j}}\right)=\chi \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Here $\chi$ is a function of temperature and pressure. $\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \chi \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ We note therefore that $\operatorname{limit}\left(\mathrm{m}_{j} \rightarrow 0\right) \mu_{j}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=-\infty$ Hence with increasing dilution of the solution, solute $j$ is increasingly stabilised, $\mu_{j}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ decreasing to ‘$- \infty$’ in an infinite amount of solvent Using the Gibbs-Helmholtz Equation, equation (a) yields equation (f). $-\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) / \mathrm{T}^{2}=-\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) / \mathrm{T}^{2}+\mathrm{R} \,\left(\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right)_{\mathrm{p}}$ $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is the partial molar enthalpy of solute $j$. $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \mathrm{T}^{2} \,\left(\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right)_{\mathrm{p}}$ Using equation (c), $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \mathrm{T}^{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) \,(\partial \chi / \partial \mathrm{T})_{\mathrm{p}}$ Hence, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ As the solution becomes more dilute and approaches infinite dilution so $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ in the limit of infinite dilution approaches the partial molar enthalpy of solute $j$ in the reference solution $\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$.where the partial molar enthalpy is identified as $\mathrm{H}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}). Granted the latter conclusion based on equation (h), this equation offers information concerning the form of the plot of \(\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ against $\mathrm{m}_{j}$. $\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) / \partial \mathrm{m}_{\mathrm{j}}\right]=-\mathrm{R} \, \mathrm{T}^{2} \,\left(\mathrm{m}^{0}\right)^{-1} \,(\partial \chi / \partial \mathrm{T})_{\mathrm{p}}$ In other words the gradient of the plot of $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ against $\mathrm{m}_{j}$ is finite, the gradient being determined by the sign of $(\partial \chi / \partial T)_{p}$. The partial molar isobaric heat capacity of the solute $j$ is given by the differential of equation (h) with respect to temperature. $\mathrm{C}_{\mathrm{pj}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{C}_{\mathrm{pj}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) \,\left[\frac{\partial}{\partial \mathrm{T}}\left(\mathrm{T}^{2} \,(\partial \chi / \partial \mathrm{T})_{\mathrm{p}}\right)\right]$ $\operatorname{Limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{C}_{\mathrm{pj}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is a finite quantity, $\mathrm{C}_{\mathrm{p} j}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$. In other words the limiting partial molar isobaric heat capacity of the solute $\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ equals the standard partial molar isobaric heat capacity, $\mathrm{C}_{\mathrm{pj}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$. $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{C}_{\mathrm{pj}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{C}_{\mathrm{pj}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{C}_{\mathrm{pj}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ A similar conclusion is reached when we turn our attention to partial molar volumes recognizing that for solute $j$, $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right) \,(\partial \chi / \partial \mathrm{p})_{\mathrm{T}}$ Therefore, $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{V}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ The limiting value of $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ is a finite quantity such that the limiting (i.e. infinite dilution) value of $\mathrm{V}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$, namely $\mathrm{V}_{\mathrm{j}}^{\infty}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ equals the standard partial molar volume, $\mathrm{V}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$. The interesting question arises as to why the limiting values of partial molar enthalpies, volumes and isobaric heat capacities are real (and important) properties but limiting chemical potentials are not. We start again with equation (b) recalling that partial molar entropy $\mathrm{S}_{\mathrm{j}}=-\left(\partial \mu_{\mathrm{j}} / \partial \mathrm{T}\right)_{\mathrm{p}}$. \begin{aligned} &\mathrm{S}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})= \ &\quad \mathrm{S}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)-\mathrm{R} \, \ln \left(\gamma_{\mathrm{j}}\right)-\mathrm{R} \, \mathrm{T} \,\left(\partial \ln \left(\gamma_{\mathrm{j}}\right) / \partial \mathrm{T}\right)_{\mathrm{p}} \end{aligned} But $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)=\text { minus infinity }$ Then $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{S}_{\mathrm{j}}=\text { plus infinity }$ With increase in dilution $\mathrm{S}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ tends to the asymptotic limit, plus infinity. For a solution at fixed $\mathrm{T}$ and $\mathrm{p}$ prepared using $1 \mathrm{~kg}$ of water, the Gibbs energy is given by equation (r). $\mathrm{G}\left(\mathrm{T} ; \mathrm{p} ; \mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \, \mu_{1}(\mathrm{aq})+\mathrm{m}_{\mathrm{j}} \, \mu_{\mathrm{j}}(\mathrm{aq})$ Or, \begin{aligned} \mathrm{G}\left(\mathrm{T} ; \mathrm{p} ; \mathrm{aq} ; \mathrm{w}_{1}\right.&=1 \mathrm{~kg})=\left(1 / \mathrm{M}_{1}\right) \,\left[\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right] \ &+\mathrm{m}_{\mathrm{j}} \,\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\gamma_{\mathrm{j}}\right)\right] \end{aligned} $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \mathrm{G}\left(\mathrm{T} ; \mathrm{p} ; \mathrm{aq} ; \mathrm{w}_{1}=1 \mathrm{~kg}\right)=\mu_{1}^{*}(\ell) / \mathrm{M}_{1}$ Salt Solutions; The Salt We consider a dilute 1:1 salt solution, confining the analysis to a consideration of the impact of the Debye - Huckel Limiting Law (DHLL). For a salt solution, molality $\mathrm{m}_{j}$, $\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\gamma_{\pm}\right)$ Or, using the DHLL \begin{aligned} &\mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})= \ &\quad \mu_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)-2 \, \mathrm{R} \, \mathrm{T} \, \mathrm{S}_{\mathrm{\gamma}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2} \end{aligned} From equation (u), $\left[\frac{\partial \mu_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})}{\partial \mathrm{m}_{\mathrm{j}}}\right]=\frac{2 \, \mathrm{R} \, \mathrm{T}}{\mathrm{m}_{\mathrm{j}}}+\mathrm{R} \, \mathrm{T} \,\left[\frac{\partial \ln \left(\gamma_{\pm}\right)}{\partial \mathrm{m}_{\mathrm{j}}}\right]$ As for a non-ionic solute, $\operatorname{limit}\left(m_{j} \rightarrow 0\right)\left[\frac{\partial \mu_{j}(a q ; T ; p)}{\partial m_{j}}\right]=\infty$ From the Gibbs - Helmholtz equation and equation (v), $\mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mathrm{H}_{\mathrm{j}}^{0}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})+2 \, \mathrm{R} \, \mathrm{T}^{2} \, \mathrm{S}_{\mathrm{H}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2}$ where $\mathrm{S}_{\mathrm{H}}=\left(\partial \mathrm{S}_{\gamma} / \partial \mathrm{T}\right)_{\mathrm{p}}$ Further, \begin{aligned} {\left[\partial \mathrm{H}_{\mathrm{j}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) / \partial \mathrm{m}_{\mathrm{j}}\right] } &=\mathrm{R} \, \mathrm{T}^{2} \, \mathrm{S}_{\mathrm{H}} \, /\left(\mathrm{m}_{\mathrm{j}} \, \mathrm{m}^{0}\right)^{1 / 2} \ &+2 \, \mathrm{R} \, \mathrm{T}^{2} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)^{1 / 2} \,\left(\partial \mathrm{S}_{\mathrm{H}} / \partial \mathrm{m}_{\mathrm{j}}\right) \end{aligned} Thus the gradient of a plot of $\mathrm{H}_{\mathrm{j}}(\mathrm{aq})$ against $\mathrm{m}_{j}$ has infinite slope in the $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right)$. A similar pattern emerges in the case of partial molar volumes of the salt [2]. A similar analysis can be undertaken with respect to the partial molar properties of the solvent and apparent molar thermodynamic properties of salts and neutral solutes. Footnote [1] M. Spiro, Educ. Chem.,1966,3,139. [2] J. E. Garrod and T. H. Herrington, J. Chem. Educ.,1969,46,165.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.33%3A_Infinite_Dilution.txt
A given chemical equilibrium exists within a closed system at defined temperature and pressure. The system is perturbed to a neighbouring equilibrium state by a change in pressure at fixed temperature and by change in temperature at fixed pressure. We take these two perturbations in turn, recognising that in both original and perturbed states the affinity for spontaneous chemical reaction is zero. The differential change in composition resulting from a change in pressure is given by Equation \ref{a}. $\left(\frac{\partial \xi}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}}=\frac{(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}, \mathrm{p}}}{(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}} \label{a}$ But at equilibrium $(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}<0$ where $\xi$ is taken as positive in the direction ‘reactants to products’. The property $(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}_{\mathrm{p}}}$ is the volume of reaction $\Delta_{\mathrm{r}}\mathrm{V}$. Thus if $\Delta_{\mathrm{r}}\mathrm{V} > 0$, an increase in pressure produces a shift in the equilibrium position to favour reactants. A shift favouring products results if $\Delta_{\mathrm{r}}\mathrm{V} < 0$. The differential change in composition resulting from a change in temperature is given by Equation \ref{b}. $\left(\frac{\partial \xi}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{A}}=-\frac{\left[\mathrm{A}+(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}}\right]}{\mathrm{T} \,(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}} \label{b}$ At equilibrium $\mathrm{A}$ is zero and $(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}<0$. Then an increase in temperature (at fixed p) for an exothermic reaction (where $\Delta_{\mathrm{r}}\mathrm{H} < 0$) results in a shift in the equilibrium position to favor reactants. An opposite shift results if the reaction is endothermic, i.e. $\Delta_{\mathrm{r}}\mathrm{H} > 0$. The conclusions described above fall under the general heading ‘Theorems of Moderation’. One of the authors (MJB) was taught that the outcome was ‘Nature’s Law of Cussedness’ [= obstinacy]. An exothermic reaction generates heat which might raise the temperature of the system so the system responds, when the temperature is raised by a chemist, by shifting the equilibrium in the direction for which the process is endothermic. This line of argument is not good thermodynamics but makes the point. In the context of ‘obstinacy’, note the switch in sign on going from equations (c) to (d). The Principle of Le Chatelier and Braun is a theorem of moderation [1,2]. Footnotes [1] J. de Heer, J. Chem. Educ. 1957,34,375. [2] J. de Heer, Phenomenological Thermodynamics with Applications to Chemistry, Prentice-Hall, Englewood Cliffs, N.J., 1986, chapter 20. 1.14.34: Internal Pressure According to the Thermodynamic Equation of State, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}-\mathrm{p}$ The partial differential $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}}$ (with units, $\mathrm{N m}^{-2}$) is the internal pressure $\pi_{\mathrm{int}}$. $\pi_{\mathrm{int}}=\mathrm{T} \, \beta_{\mathrm{V}}-\mathrm{p}$ $\pi_{\mathrm{int}$ describes the sensitivity of energy $\mathrm{U}$ to a change in volume. A high $\phi_{\mathrm{int}$ implies strong inter-molecular cohesion [1-8]. For many liquids, $\mathrm{T} \, \boldsymbol{\beta}_{\mathrm{V}}>>\mathrm{p}$ such that $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}} \cong \mathrm{T} \, \beta_{\mathrm{V}}$ $\mathrm{T} \, \boldsymbol{\beta}_{\mathrm{V}}$ is sometimes called the thermal pressure. By definition, for $\mathrm{n}$ moles of a perfect gas, $\mathrm{p} \, \mathrm{V}=\mathrm{n} \, \mathrm{R} \, \mathrm{T}$ Then $\mathrm{V} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\mathrm{n} \, \mathrm{R}$ Or, $\mathrm{T} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\mathrm{n} \, \mathrm{R} \, \mathrm{T} / \mathrm{V}=\mathrm{p}$ From equation (a), for a perfect gas, $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}$ is zero. The internal pressure for water($\ell$) presents an interesting puzzle [9]. From equations (a) and (c), it follows that [1] $\pi_{\mathrm{int}}=\mathrm{T} \,\left(\frac{\alpha_{\mathrm{p}}}{\kappa_{\mathrm{T}}}\right)-\mathrm{p}$ But at the temperature of maximum density (TMD), $\alpha_{p}$ is zero. So near the TMD, $\pi_{\mathrm{int}$ is zero. We understand this pattern if we think about hydrogen bonding. In order to form a strong hydrogen bond between two neighboring water molecules the O-H---O link has to be close to if not actually linear. In other words the molar volume for water($\ell$) is larger than the molar volume of a system comprising close-packed water molecules. Consequently hydrogen bonding has a strong ‘repulsive’ component to intermolecular interaction. However once formed hydrogen bonding has a strong cohesive contribution to intermolecular forces. Hence for water between $273$ and $298 \mathrm{~K}$ cohesive and repulsive components of hydrogen bonding play almost competitive roles. Footnotes [1] Using a calculus operation, $\left(\frac{\partial p}{\partial T}\right)_{V}=-\left(\frac{\partial V}{\partial T}\right)_{p} \,\left(\frac{\partial p}{\partial V}\right)_{T}$ For equilibrium properties, $\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}=\frac{\alpha_{\mathrm{p}}}{\kappa_{\mathrm{T}}}$ [2] Some authors use the term ‘isochoric thermal pressure coefficient’ for the property, $\left(\frac{\partial p}{\partial T}\right)_{V}$ [3] For details of original proposals concerning internal pressures see the following references. 1. W. Westwater, H. W. Frantz and J. H. Hildebrand, Phys.Rev.,1928, 31,135. 2. J. H. Hildebrand, Phys.Rev.,1929,34,649 and 984. 3. See also S. E. Wood, J.Phys.Chem.,1962,66, 600. [4] Internal pressures are quoted in the literature using many units. Here we use $\mathrm{N m}^{-2}$. We list some internal pressures and relative permitivities at $298.15 \mathrm{~K}$. liquid $\varepsilon_{\mathrm{r}}\( \(\pi_{\mathrm{int}} / 10^{5} \mathrm{~N m}^{-2}$ water 78.5 1715 methanol 32.63 2849 ethanol 24.30 2908 propanone 20.7 3368 diethyl ether 4.3 2635 tetrachloromethane 2.24 3447 dioxan 2.2 4991 The above details are taken from M. R. J. Dack, J.Chem.Educ.,1974,51,231;see also 1. Aust. J. Chem.,1976, 29,771 and 779. 2. D. D. MacDonald and J. B. Hyne, Can.J.Chem.,1971, 49,2636 [5] For a discussion of effects of solvents on rates of chemical reactions with reference to internal pressures, see 1. K. Colter and M. L. Clemens, J.Phys.Chem.,1964,68,651. 2. A. P. Stefani, J. Am. Chem.Soc.,1968,90,1694. [6] For comments on solvent polarity and internal pressures see J. E. Gordon, J. Phys. Chem.,1966,70,2413. [7] For comments on internal pressures of binary aqueous mixtures see D. D. Macdonald, Can. J Chem.,1976,54,3559; and references therein. [8] For comments on effect of internal pressure on conformational equilibria see R. J. Ouellette and S. H. Williams, J. Am. Chem.Soc.,1971,93,466. [9] For details concerning the dependence of internal pressure of water and $\mathrm{D}_{2}\mathrm{O}$, see M. J. Blandamer, J. Burgess and A.W.Hakin, J. Chem. Soc. Faraday Trans. 1, 1987, 83, 1783. [10] For comments on the calculation of excess internal pressures for binary liquid mixtures using equation (h) see W. Marczak, Phys.Chem.Chem.Phys.,2002,4,1889. 1.14.35: Internal Pressure: Liquid Mixtures: Excess Property The thermodynamic equation of state takes the form shown in equation (a). $\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\mathrm{T} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}-\mathrm{p}$ The partial differential $(\partial \mathrm{U} / \partial \mathrm{V})_{\mathrm{T}}$ is the internal pressure, $\pi_{\mathrm{int}}$ (with units, $\mathrm{N m}^{-2}$). A calculus operation relates three interesting partial derivatives in the context of $\mathrm{p}-\mathrm{V}-\mathrm{T}$ properties; equation (b). $\left(\frac{\partial p}{\partial T}\right)_{V}=-\left(\frac{\partial V}{\partial T}\right)_{p} \,\left(\frac{\partial p}{\partial V}\right)_{T}$ For a given liquid at defined $\mathrm{T}$ and $\mathrm{p}$, the isobaric (equilibrium) thermal expansion, $\mathrm{E}_{\mathrm{p}}$ equals $(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$. The isothermal (equilibrium) compression $\mathrm{K}_{\mathrm{T}}$ is defined by $-(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{T}}$. According to equations (a) and (b), $\pi_{\mathrm{int}}$ is given by equation (c). $\pi_{\mathrm{int}}=\left(\mathrm{T} \, \mathrm{E}_{\mathrm{p}} / \mathrm{K}_{\mathrm{T}}\right)-\mathrm{p}$ For the purpose of the analysis described here, equation (c) describes the equilibrium molar properties of a given binary liquid mixture at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The internal pressure is a non-Gibbsian property of a liquid. Nevertheless it is interesting to compare internal pressures of real and the corresponding ideal binary liquid mixture [1]. In other words we require an equation for the internal pressure of binary liquid mixture $\pi_{\mathrm{int}}^{\mathrm{id}}$ having thermodynamic properties which are ideal. Marczak[1] uses equation (c) in which the corresponding molar properties of the mixture, mole fraction composition $\mathrm{x}_{2}$, $\mathrm{E}_{\mathrm{pm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)$ and $\mathrm{K}_{\mathrm{Tm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)$ are given by the mole fraction weighted properties of the pure liquids. $\mathrm{E}_{\mathrm{pm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{E}_{\mathrm{pi}}^{*}(\ell)$ $\mathrm{K}_{\mathrm{Tm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)$ Equations (d) and (e) can be generalised to multi-component liquid mixtures. From equation (c) for a binary liquid mixture having thermodynamic properties which are ideal, the internal pressure $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)$ is given by equation (f). $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\mathrm{T} \, \frac{\mathrm{E}_{\mathrm{pm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)}{\mathrm{K}_{\mathrm{Tm}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)}-\mathrm{p}$ Or using equations (d) and (e), $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\frac{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{~T} \, \mathrm{x}_{\mathrm{i}} \, \mathrm{E}_{\mathrm{pi}}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}-\mathrm{p}$ Equation (g) is re-written to establish $\pi_{\mathrm{int,i}}^{*}(\ell)$ as a term on the r.h.s. of the latter equation for $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)$. $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=-\mathrm{p}+\frac{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{~T} \, \mathrm{x}_{\mathrm{i}} \, \mathrm{E}_{\mathrm{pi}}^{*}(\ell) \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell) / \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}$ But according to equation (c) , for the pure liquid–$\mathrm{i}$, $\pi_{\mathrm{im}, \mathrm{i}}^{*}(\ell)+\mathrm{p}=\mathrm{T} \, \mathrm{E}_{\mathrm{pi}}^{*}(\ell) / \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)$ Hence from equation (h), $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=-\mathrm{p}+\frac{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \,\left[\pi_{\mathrm{int,i}}^{*}(\ell)+\mathrm{p}\right] \, \mathrm{K}_{\mathrm{T}_{1}}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}^{*}}(\ell)}$ In other words, \begin{aligned} \pi_{\mathrm{idt}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=&-\mathrm{p}+\frac{\mathrm{x}_{1} \, \pi_{\mathrm{int1} 1}^{*}(\ell) \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}+\frac{\mathrm{x}_{1} \, \mathrm{p} \, \mathrm{K}_{\mathrm{T} 1}^{*}(\ell)}{\sum_{\mathrm{i}=2}^{*} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)} \ &+\frac{\mathrm{x}_{2} \, \pi_{\mathrm{int}, 2}^{*}(\ell) \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}+\frac{\mathrm{x}_{2} \, \mathrm{p} \, \mathrm{K}_{\mathrm{T} 2}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)} \end{aligned} Hence, $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\frac{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \pi_{\mathrm{int}, \mathrm{i}}^{*}(\ell) \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}{\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}$ By definition [1], for liquid component $\mathrm{k}$, $\Psi_{\mathrm{k}}=\frac{\mathrm{x}_{\mathrm{k}} \, \mathrm{K}_{\mathrm{Tk}}^{*}(\ell)}{\sum \mathrm{x}_{\mathrm{i}} \, \mathrm{K}_{\mathrm{Ti}}^{*}(\ell)}$ In other words, $\pi_{\mathrm{int}}^{\mathrm{id}}\left(\mathrm{x}_{2}\right)=\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \psi_{\mathrm{i}} \, \pi_{\mathrm{int,i}}^{*}(\ell)$ The corresponding excess internal pressure at mole fraction $\mathrm{x}_{2}$, $\pi_{\mathrm{int}}^{\mathrm{E}}\left(\mathrm{x}_{2}\right)$ is defined by equation (o). $\pi_{\mathrm{int}}^{\mathrm{E}}\left(\mathrm{x}_{2}\right)=\pi_{\mathrm{int}}\left(\mathrm{x}_{2}\right)-\sum_{\mathrm{i}=1}^{\mathrm{i}=2} \psi_{\mathrm{i}} \, \pi_{\mathrm{int}, \mathrm{i}}^{*}(\ell)$ Marczak [1] reports $\pi_{\text {int }}^{E}\left(X_{2}\right)$ as a function of mole fraction $\mathrm{x}_{2}$ for two binary liquid mixtures at $298.15 \mathrm{~K}$; 1. methanol + propan-1-ol, and 2. tribromomethane + n-octane. Footnotes [1] W. Marczak, Phys. Chem. Chem. Phys.2002,4,1889.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.34%3A_Extent_of_Reaction-_Chemical_Equilibrium-_Dependence_on_Temperature_and_Press.txt
According to the Second Law of Thermodynamics, the change in entropy $\mathrm{dS}$ is related to the affinity for spontaneous change $\mathrm{A}$ using equation (a). $\mathrm{dS}=(\mathrm{q} / \mathrm{T})+(\mathrm{A} / \mathrm{T}) \, \mathrm{d} \xi ; \quad \mathrm{A} \, \mathrm{d} \xi \geq 0$ In these terms chemists usually have in mind a chemical reaction driven by the affinity $\mathrm{A}$ for spontaneous chemical reaction producing extent of reaction $\mathrm{d}\xi$. We generalize the law in the following terms. $\mathrm{dS}=(\mathrm{q} / \mathrm{T})+\mathrm{d}_{\mathrm{i}} \mathrm{S} ; \quad \mathrm{d}_{\mathrm{i}} \mathrm{S} \geq 0$ $\mathrm{d}_{\mathrm{i}\mathrm{S}}$ is the change in entropy of the system by virtue of spontaneous processes in the system. Comparison of equations (a) and (b) yields the following equation. $\mathrm{T} \, \mathrm{d}_{\mathrm{i}} \mathrm{S}=\mathrm{A} \, \mathrm{d} \xi \geq 0$ We introduce two new terms. A quantity $\mathrm{P}[\mathrm{S}]$ describes the rate of entropy production within the system; a quantity $\sigma[\mathrm{S}]$ describes the corresponding rate of entropy production in unit volume of the system. $\mathrm{P}[\mathrm{S}]=\mathrm{d}_{\mathrm{i}} \mathrm{S} / \mathrm{dt}=\int_{\mathrm{V}} \sigma[\mathrm{S}] \, \mathrm{dV} \geq 0$ We combine equations (c) and (d). $\mathrm{P}[\mathrm{S}]=\frac{\mathrm{d}_{\mathrm{i}} \mathrm{S}}{\mathrm{dt}}=\frac{\mathrm{A}}{\mathrm{T}} \, \frac{\mathrm{d} \xi}{\mathrm{dt}} \geq 0$ But if $\mathrm{dn}_{j}$ is the change in amount of chemical substance $j$ in the system, $\mathrm{dn} \mathrm{j}_{\mathrm{j}}=\mathrm{v}_{\mathrm{j}} \, \mathrm{d} \xi$. Then, $P[S]=\frac{d_{i} S}{d t}=\frac{A}{T} \, \frac{1}{v_{j}} \, \frac{d n_{j}}{d t} \geq 0$ We develop this equation into a form which has wider significance. We assume that the system is homogeneous such that for a system volume $\mathrm{V}$, $\sigma[S]=P[S] / V$ Then $\sigma[\mathrm{S}]=\frac{1}{\mathrm{~V}} \, \frac{\mathrm{d}_{\mathrm{i}} \mathrm{S}}{\mathrm{dt}}=\frac{\mathrm{A}}{\mathrm{T}} \, \frac{1}{\mathrm{v}_{\mathrm{j}}} \, \frac{1}{\mathrm{~V}} \, \frac{\mathrm{dn}_{\mathrm{j}}}{\mathrm{dt}} \geq 0$ But the concentration of chemical substance $j$, $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}$. Then, $\mathrm{dc}_{\mathrm{j}}=\mathrm{dn}_{\mathrm{j}} / \mathrm{V}$. Hence from equation (h), $\sigma[S]=\frac{\mathrm{A}}{\mathrm{T}} \, \frac{1}{\mathrm{v}_{\mathrm{j}}} \, \frac{\mathrm{dc}_{\mathrm{j}}}{\mathrm{dt}} \geq 0$ The quantity $\left(1 / v_{j}\right) \, d c_{j} / d t$ describes the change in composition of the system, the flow of the system from reactants to products. In these terms we identify a chemical flow, $\mathrm{J}_{\mathrm{ch}}$. $\mathrm{J}_{\mathrm{ch}}=\left(\mathrm{l} / \mathrm{v}_{\mathrm{j}}\right) \, \mathrm{dc}_{\mathrm{j}} / \mathrm{dt}$ Then, $\sigma[S]=(A / T) \, J_{c h} \geq 0$ Or, $\mathrm{T} \, \sigma[\mathrm{S}]=\mathrm{A} \, \mathrm{J}_{\mathrm{ch}} \geq 0$ The latter equation has an interesting feature; $\mathrm{T} \, \sigma[\mathrm{S}]$ is given by the product of the affinity for spontaneous chemical reaction (the driving force) and the accompanying flow. Indeed $\mathrm{T} \, \sigma[\mathrm{S}]$ is related to the rate of entropy production in the system. Thermodynamics takes us no further. We make an extrathermodynamic leap and suggest that the flow is proportional to the force; i.e. the stronger the driving force the more rapid the chemical flow from reactants to products. In general terms phenomenological equations start out from the basis of a linear model described by a phenomenological law of the general form, $\mathrm{J} = \mathrm{L} \, \(\mathrm{~X}$ where $\mathrm{J}$ is the flow and $\mathrm{X}$ is the conjugate force such that the product $\mathrm{J} \, \mathrm{~X}$ yields the rate of entropy production. These laws are based on experiment. Many such phenomenological laws have been proposed. Some examples are listed below. 1. Phenomenon--- Electrical Conductivity Law: Ohm’s Law (discovered 1826) $\mathrm{I}=\mathrm{L} \, \mathrm{V}$ where $\mathrm{L} =$ conductance; resistance $\mathrm{R} = 1/\mathrm{L}$. 2. Phenomenon--- Diffusion Law: Fick’s Law discovered by Adolf Fick 1855 $\mathrm{J}_{\mathrm{j}}=\mathrm{D} \,\left(-\mathrm{d} \mu_{\mathrm{j}} / \mathrm{dx}\right)$ 3. Phenomenon— Thermal conductivity Law : Fourier’s Law (1822) $\mathrm{J}_{\mathrm{q}}=\lambda \,(-\mathrm{dT} / \mathrm{dx})$ $\lambda =$ thermal conductivity 4. Phenomenon—Chemical reaction If the patterns described above were followed we might write, $J_{c h}=L \, A$ In 1890 Nernst suggested this approach to chemical kinetics. Unfortunately chemists have no method for measuring the affinity; there is no affinity meter. Instead chemists use the Law of Mass Action.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.36%3A_Irreversible_Thermodynamics.txt
The major thrust of the account presented in these Topics concerns reversible processes in which system and surroundings are in thermodynamic equilibrium. When attention turns to non-equilibrium processes, the thermodynamic treatment is necessarily more complicated [1-5]. Here we examine several aspects of irreversible thermodynamics for near equilibrium in open systems[1]. In other words there is a strong ‘communication’ between system and surroundings. With increasing displacement of a given system from equilibrium the thermodynamic analysis becomes more complicated and controversial [2-5]. Here we confine attention to processes in near-equilibrium states [6]. The key assumption is that equations describing relationships between thermodynamic properties are valid for small elemental volumes, the concept of local equilibrium. Hence we can in a description of a given system identify the energy per unit volume and the entropy per unit volume in the context respectively of the first and second laws of thermodynamics. With respect to a small volume $\mathrm{dv}$ of a given system the change in entropy $\mathrm{ds}$ is given by the change in entropy $\mathrm{d}_{\mathrm{i}} \mathrm{s}$ by virtue of processes within a small volume $\mathrm{dv}$ and by virtue of exchange with the rest of the system, des. The rate of change of $\mathrm{d}_{\mathrm{i}} \mathrm{s}$, namely $\mathrm{d}_{\mathrm{i}} \mathrm{s} / \mathrm{dt}$ is the local entropy production and is determined by the following condition. $\sigma \equiv \mathrm{d}_{\mathrm{i}} \mathrm{s} / \mathrm{dt} \geq 0$ Phenomenological Laws For systems close to thermodynamic equilibrium, the entropy production per unit volume $\sigma$ can be expressed as the sum of products of forces $\mathrm{X}_{\mathrm{k}}$ and conjugate flows, $\mathrm{J}_{\mathrm{k}}$. Thus for $\mathrm{k}$ flows and forces, $\sigma=\sum_{\mathrm{k}} \mathrm{X}_{\mathrm{k}} \, \mathrm{J}_{\mathrm{k}}$ The condition ‘conjugate’ is important in the sense that for each flow $\mathrm{J}_{k}$ there is a conjugate force $\mathrm{X}_{k}$. For near equilibrium systems a given flow is a linear function of the conjugate force, $\mathrm{X}_{k}$. Then, $\mathrm{J}_{\mathrm{k}}=\sum_{\mathrm{j}} \mathrm{L}_{\mathrm{kj}} \, \mathrm{X}_{\mathrm{k}}$ The property $\mathrm{L}_{\mathrm{kj}}$ is a phenomenological coefficient describing the dynamic flow and conjugate force. In simple systems there is only one flow and one force such that the flow is directly proportional to the force. A classic example is Ohm’s law which can be written in the following form. $\mathrm{I}=(1 / \mathrm{R}) \, \mathrm{V}$ Thus $\mathrm{I}$ is the electric current, the rate of flow of electric charge for a system where the driving force is the electric potential gradient $\mathrm{V}$. The relevant property of the system under consideration is the resistance $\mathrm{R}$ or, preferably, its conductance $\mathrm{L} (= 1/\(\mathrm{R}$). A similar phenomenological law is Fick’s Law of diffusion relating the rate of diffusion of chemical substance $j$, $\mathrm{J}_{j}$ to the concentration gradient $\mathrm{dc}_{j}/\mathrm{dx}$ where $\mathrm{D}_{j}$ describes the property of diffusion. Thus $\mathrm{J}_{\mathrm{j}}=\mathrm{D}_{\mathrm{j}} \,\left(\mathrm{dc}_{\mathrm{j}} / \mathrm{dx}\right)$ The Law of Mass Action is a similar phenomenological law. In other words throughout chemistry (and indeed all sciences) there are phenomenological laws which do not, for example, follow from the first and second laws of thermodynamics. Onsager Equations Following on a proposal by Lord Rayleigh relating to mechanical properties, in 1931 Onsager [7] extended the ideas discussed above to include all forces and flows. For a system involving two flows and forces we may write the following two equations to describe near –equilibrium systems. $\mathrm{J}_{1}=\mathrm{L}_{11} \, \mathrm{X}_{1}+\mathrm{L}_{12} \, \mathrm{X}_{2}$ $\mathrm{J}_{2}=\mathrm{L}_{21} \, \mathrm{X}_{1}+\mathrm{L}_{22} \, \mathrm{X}_{2}$ This formulation recognises that force $\mathrm{X}_{2}$ may also produce a coupled flow $\mathrm{J}_{1}$. In each case the products $\mathrm{L}_{11} \, \mathrm{X}_{1}, \mathrm{~L}_{12} \, \mathrm{X}_{2}, \mathrm{~L}_{21} \, \mathrm{X}_{1}$ and $\mathrm{L}_{22} \, \mathrm{X}_{2}$ involve conjugate flows and forces such that the product, $\mathrm{J}_{\mathrm{i}} \, \mathrm{X}_{\mathrm{i}}$ has the dimension of the rate of entropy production. The cross terms $\mathrm{L}_{12}$ and $\mathrm{L}_{21}$ are the coupling coefficients such that for example, force X2 produces flow J1. Onsager’s Law The key theoretical advance made by Onsager was to show that for near-equilibrium states the matrix of coefficients is symmetric. Then, for example,[8] $\mathrm{L}_{12}=\mathrm{L}_{21}$ The point can be developed by considering a system involving two flows and two forces. According to equation (b) $\sigma=\mathrm{J}_{1} \, \mathrm{X}_{1}+\mathrm{J}_{2} \, \mathrm{X}_{2}$ Hence from equations (f) and (g) $\sigma=\mathrm{L}_{11} \, \mathrm{X}_{1}^{2}+\left(\mathrm{L}_{12}+\mathrm{L}_{21}\right) \, \mathrm{X}_{1} \, \mathrm{X}_{2}+\mathrm{L}_{22} \, \mathrm{X}_{2}^{2}>0$ It also follows that [8] $\mathrm{L}_{11} \, \mathrm{X}_{1}^{2} \geq 0 \quad ; \quad \mathrm{L}_{22} \, \mathrm{X}_{2}^{2} \geq 0$ And, $\mathrm{L}_{11} \, \mathrm{L}_{22} \geq \mathrm{L}_{12}^{2}$ Electrokinetic Phenomena[1] These phenomena illustrate the application of the equations discussed above. A membrane separates two salt solutions; an electric potential E and a pressure gradient are applied across the membrane. There are two flows; 1. solution flows through the membrane, described as a volume flow; 2. an electric current. The dynamics of the system are described by the dissipation function $\phi$ given by equation (m), the sum of products of flows and forces. $\phi=\mathrm{J}_{\mathrm{V}} \, \Delta \mathrm{p}+\mathrm{I} \, \mathrm{E}$ The dynamics of the system are described by two dynamic equations, $\mathrm{J}_{\mathrm{V}}=\mathrm{L}_{11} \, \Delta \mathrm{p}+\mathrm{L}_{12} \, \mathrm{E}$ $\mathrm{I}=\mathrm{L}_{21} \, \Delta \mathrm{p}+\mathrm{L}_{22} \, \mathrm{E}$ Onsager’s law requires that, $\mathrm{L}_{12}=\mathrm{L}_{21}$ In an experiment we set $\mathrm{E}$ at zero. Then $\mathrm{L}_{11}=\left(\frac{\mathrm{J}_{\mathrm{v}}}{\Delta \mathrm{p}}\right)_{\mathrm{E}=0}$ However the electric current is not zero. According to equation (o), $\mathrm{I}=\mathrm{L}_{21} \, \Delta \mathrm{p}$ In other words, there is a coupled flow of ions. Katchalsky and Curran [1] discuss numerous experiments which illustrate this type of coupling of flows and forces. Footnotes [1] A. Katchalsky and P. F. Curran, Non-Equilibrium Thermodynamics in Biophysics, Harvard University Press, 1965. [2] P. Glandsorff and I. Prigogine, Thermodynamics of Structure Stability and Fluctuations, Wiley-Interscience, London,1971. [3] G. Nicolis and I. Prigogine, Self-Organization in Nonequilibrium Systems, Wiley, New York, 1977. [4] P Gray and S. K. Scott, Chemical Oscillations and Instabilities, Oxford,1990. [5] B. Lavenda, Thermodynamics of Irreversible Processes, MacMillan Press, London, 1978. [6] D. Kondepudi and I. Prigogine, Modern Thermodynamics, Wiley, New York, 1998. [7] L. Onsager, Phys. Rev.,1931,38,2265. [8] D. G. Miller, Chem. Rev.,1960,60,15.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.37%3A_Irreversible_Thermodynamics%3A_Onsager_Phenomenological_Equations.txt
An important property of a given gas is its Joule-Thomson coefficient [1-3]. These coefficients are important from two standpoints; 1. intermolecular interaction, and 2. liquefaction of gases. A given closed system contains one mole of gaseous chemical substance $\mathrm{j}$ at temperature $\mathrm{T}$ and pressure $\mathrm{p}$. The molar enthalpy of the gas $\mathrm{H}_{\mathrm{j}}$ describes its molar enthalpy defined by equation (a). $\mathrm{H}_{\mathrm{j}}=\mathrm{H}_{\mathrm{j}}[\mathrm{T}, \mathrm{p}]$ Then, $\mathrm{dH}_{\mathrm{j}}=\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \, \mathrm{dT}+\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \, \mathrm{dp}$ Hence at constant enthalpy, $\mathrm{H}$, $\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{p}} \, \mathrm{dT}=-\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}} \, \mathrm{dp}$ Or, $\left(\frac{\partial T}{\partial p}\right)_{H}=-\left(\frac{\partial H_{\mathrm{j}}}{\partial p}\right)_{T} \,\left(\frac{\partial T}{\partial H_{j}}\right)_{p}$ The Joule-Thomson coefficient for gas $\mathrm{j}$, $\mu_[\mathrm{j}}$ is defined by equation (e). $\mu_{\mathrm{j}}=\left(\frac{\partial T}{\partial p}\right)_{\mathrm{H}(\mathrm{j})}$ For all gases (except helium and hydrogen) at $298 \mathrm{~K}$ and moderate pressures $\mu_{\mathrm{j} > 0$. At room temperature and ambient pressure, $\mu_{\mathrm{j}}$ is $0.002 \mathrm{~K Pa}^{-1}$ for nitrogen and $0.025 \mathrm{~K Pa}^{-1}$ for 2,2-dimethylpropane [3]. Further the isobaric heat capacity for chemical substance $\mathrm{j}$ is defined by equation (f). $\mathrm{C}_{\mathrm{pj}}=\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ Hence from equations (d), (e) and (f), $\mu_{\mathrm{j}}=-\frac{(\partial \mathrm{H} / \partial \mathrm{p})_{\mathrm{T}}}{\mathrm{C}_{\mathrm{pj}}}$ Then, $\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=-\mu_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}$ Equation (h) marks an important stage in the analysis. For example, $\mathrm{C}_{\mathrm{pj}} > 0$. From the definition of enthalpy $\mathrm{H}_{\mathrm{j}}$, $\mathrm{U}_{\mathrm{j}}=\mathrm{H}_{\mathrm{j}}-\mathrm{p} \, \mathrm{V}_{\mathrm{j}}$ Equation (i) is differentiated with respect to $\mathrm{V}_{\mathrm{j}} at fixed \(\mathrm{T}$. Thus, $\[\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}-\mathrm{V}_{\mathrm{j}} \,\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}-\mathrm{p}$ Or, $\left(\frac{\partial U_{j}}{\partial V_{j}}\right)_{T}=\left(\frac{\partial p}{\partial V_{j}}\right)_{T} \,\left[\left(\frac{\partial H_{j}}{\partial V_{j}}\right)_{T} \,\left(\frac{\partial V_{j}}{\partial p}\right)_{T}-V_{j}\right]-p$ Then, $\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{T} \,\left[\left(\frac{\partial \mathrm{H}_{\mathrm{j}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}-\mathrm{V}_{\mathrm{j}}\right]-\mathrm{p}$ Using equation (h), $\left(\frac{\partial U_{j}}{\partial V_{j}}\right)_{T}=-\left(\frac{\partial p}{\partial V_{j}}\right)_{T} \,\left[\mu_{j} \, C_{p j}+V_{j}\right]-p$ An important application of equation (m) concerns the case where chemical substance $\mathrm{j}$ is a perfect gas. In this case, $\mathrm{p} \, \mathrm{V}_{\mathrm{j}}=\mathrm{R} \, \mathrm{T}$ Or, $\mathrm{p}=\mathrm{R} \, \mathrm{T} \, \frac{1}{\mathrm{~V}_{\mathrm{j}}}$ Hence, $\left(\frac{\partial p}{\partial V_{j}}\right)_{T}=-R \, T \, \frac{1}{V_{j}^{2}}=-\frac{p}{V_{j}}$ Then from equation (m), $\left(\frac{\partial U_{j}}{\partial V_{j}}\right)_{T}=\frac{p}{V_{j}} \,\left[\mu_{j} \, C_{p j}+V_{j}\right]-p$ Or, $\left(\frac{\partial U_{j}}{\partial V_{j}}\right)_{T}=\frac{p \, \mu_{j} \, C_{p j}}{V_{j}}$ But $\operatorname{limit}(\mathrm{p} \rightarrow 0) \frac{\mathrm{p} \, \mu_{\mathrm{j}} \, \mathrm{C}_{\mathrm{pj}}}{\mathrm{V}_{\mathrm{j}}}=0$ Then $\operatorname{limit}(\mathrm{p} \rightarrow 0)\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}=0$ A definition of a perfect gas is that $\left(\frac{\partial \mathrm{U}_{\mathrm{j}}}{\partial \mathrm{V}_{\mathrm{j}}}\right)_{\mathrm{T}}$ is zero. Then all real gases are perfect in the $\operatorname{limit}(\mathrm{p} \rightarrow 0)$. Footnotes [1] James Prescott Joule(12818-1889) William Thomson (1824-1907); Later Lord Kelvin Some authors refer to the Joule-Thomson coefficient; e.g. E. B. Smith, Basic Chemical Thermodynamics, Clarendon Press, Oxford, 1982, 3rd. edn., page 119. Other authors refer to the Joule –Kelvin Effect; e.g. E. F. Caldin, Chemical Thermodynamics, Clarendon Press, Oxford, 1958,page 81. Other authors refer to either the Joule-Thomson or Joule-Kelvin Effect; e.g. M. H. Everdell, Introduction to Chemical Thermodynamics, English Universities Press, London 1965, page 57. [2] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London 1979, page 94. [3] Benjamin Thompson (1753-1814); later Count von Rumford, married Lavoisier’s widow. 1.14.39: Kinetic Salt Effects The chemical potential of a given solute $\mathrm{j}$ in an aqueous solution is related to the concentration $\mathrm{c}_{\mathrm{j}}$ using equation (a) where $\mathrm{c}_{\mathrm{r}}$ is a reference concentration, $1 \mathrm{~mol dm}^{-3}$, and yj is the solute activity coefficient. $\mu_{\mathrm{j}}(\mathrm{aq})=\mu_{\mathrm{j}}^{0}\left(\mathrm{c}_{\mathrm{j}}=1 \mathrm{~mol} \mathrm{dm}{ }^{-3} ; \mathrm{aq} ; \mathrm{id}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{c}_{\mathrm{j}} \, \mathrm{y}_{\mathrm{j}} / \mathrm{c}_{\mathrm{r}}\right)$ By definition, at all $\mathrm{T}$ and $\mathrm{p}$, $\operatorname{limit}\left(\mathrm{c}_{\mathrm{j}} \rightarrow 0\right) \mathrm{y}_{\mathrm{j}}=1.0$ In the application of equation (a) to the rates of chemical reactions in solution, transition state theory [1] is used. In the case of a second order bimolecular reaction involving solutes $\mathrm{X}(\mathrm{aq})$ and $\mathrm{Y}(\mathrm{aq})$, the reaction proceeds as described by equation (c). $\mathrm{X}(\mathrm{aq})+\mathrm{Y}(\mathrm{aq}) \Leftarrow \Rightarrow \mathrm{TS}^{\neq} \rightarrow \text { products }$ An equilibrium between reactants and transition state, $\mathrm{TS}^{\neq}$ is described by an equilibrium constant $\mathrm{K}^{\neq}$. Hence, $\Delta^{\neq} \mathrm{G}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{K}^{\neq}\right)=\mu_{\neq}^{0}(\mathrm{aq})-\mu_{\mathrm{X}}^{0}(\mathrm{aq})-\mu_{\mathrm{Y}}^{0}(\mathrm{aq})$ At equilibrium, $\mu^{\mathrm{eq}}(\mathrm{X} ; \mathrm{aq})+\mu^{\mathrm{eq}}(\mathrm{Y} ; \mathrm{aq})=\mu^{\mathrm{eq}}(\mathrm{TS} ; \mathrm{aq})$ Using equation (a), $\mathrm{K}^{\neq}=\frac{\mathrm{c}^{\neq}(\mathrm{aq}) \, \mathrm{y}^{\neq}(\mathrm{aq}) \, \mathrm{c}_{\mathrm{r}}}{\mathrm{c}_{\mathrm{x}}^{\mathrm{eq}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{X}}(\mathrm{aq}) \, \mathrm{c}_{\mathrm{Y}}^{\mathrm{eq}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{Y}}(\mathrm{aq})}$ According to $\mathrm{TS}$ theory [1] rate constant $\mathrm{k}$ is related to $\mathrm{K}^{\neq}$ using equation (g) where $\kappa$ is a transmission coefficient, customarily set to unity. Then, $\mathrm{k}=\mathrm{K} \,(\mathrm{k} \, \mathrm{T} / \mathrm{h}) \, \mathrm{K}^{\neq} \, \mathrm{y}_{\mathrm{X}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{Y}}(\mathrm{aq}) / \mathrm{y}_{\neq}(\mathrm{aq})$ In the event that the thermodynamic properties of the aqueous solution are ideal, equation (g) simplifies to equation (h). $\mathrm{k}(\mathrm{id})=\kappa \,(\mathrm{k} \, \mathrm{T} / \mathrm{h}) \, \mathrm{K}^{\neq}$ For a real system, $\mathrm{k}=\mathrm{k}(\mathrm{id}) \, \mathrm{y}_{\mathrm{X}}(\mathrm{aq}) \, \mathrm{y}_{\mathrm{Y}}(\mathrm{aq}) / \mathrm{y}_{\neq}(\mathrm{aq})$ The Bronsted-Bjerrum analysis concerns rates of chemical reaction between ions having electric charges, $\mathrm{z}_{\mathrm{x}} \, \mathrm{e}$ and $\mathrm{z}_{\mathrm{y}} \, \mathrm{e}$ where the transition state has charge z ⋅ e ≠ ( z e z e) X Y = ⋅ + ⋅ . In most applications, the activity coefficients are related to the ionic strength of the solution using the Debye - Huckel Limiting Law. For reactant $\mathrm{j}$, $\ln \left(\mathrm{y}_{\mathrm{j}}\right)=-\mathrm{S}_{\mathrm{Y}} \, \mathrm{z}_{\mathrm{j}}^{2} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2}$ Then, $\ln (\mathrm{k})=\ln (\mathrm{k}(\mathrm{id}))+\ln \left(\mathrm{y}_{\mathrm{X}}\right)+\ln \left(\mathrm{y}_{\mathrm{Y}}\right)-\ln \left(\mathrm{y}_{z}\right)$ $\ln (\mathrm{k})=\ln (\mathrm{k}(\mathrm{id}))-\mathrm{S}_{\mathrm{y}} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2} \,\left[\mathrm{z}_{\mathrm{X}}^{2}+\mathrm{z}_{\mathrm{Y}}^{2}-\left(\mathrm{z}_{\mathrm{X}}+\mathrm{z}_{\mathrm{Y}}\right)^{2}\right]$ Or, $\ln (\mathrm{k})-\ln (\mathrm{k}(\mathrm{id}))=\mathrm{S}_{\mathrm{y}} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2} \,\left[2 \, \mathrm{z}_{\mathrm{X}} \, \mathrm{Z}_{\mathrm{Y}}\right]$ Equation (m) forms the basis of the classic and oft-quoted plot of $[\ln (\mathrm{k})-\ln (\mathrm{k}(\mathrm{id}))]$ against $\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2}$ in which the slope is determined by the product of charge numbers, $\mathrm{z}_{\mathrm{x}} \, \mathrm{z}_{\mathrm{y}}$; [1;see Footnote (1), page 429]. An interesting feature was noted by Rosseinsky [2]. Equation (m) can be written in a quite general form for a reaction involving $\mathrm{n}$ ions. Then, $\ln (\mathrm{k})-\ln (\mathrm{k}(\mathrm{id}))=\mathrm{S}_{\mathrm{y}} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2} \, \sum_{\mathrm{i}}^{\mathrm{n}} \sum_{\mathrm{j}}^{\mathrm{n}} \mathrm{z}_{\mathrm{i}} \, \mathrm{z}_{\mathrm{j}} \quad(\mathrm{i} \neq \mathrm{j})$ For chemical reaction involving cations and anions , cases can arise where the double sum in equation(n) is zero. Hence the rate constant will be independent of ionic strength. Rosseinsky cites the following reaction as a case in point [3]. $2 \mathrm{Mn}^{2+}(\mathrm{aq})+\mathrm{MnO}_{4}^{-} \text {(aq) } \rightarrow \mathrm{Mn}_{3} \mathrm{O}_{4}^{3+}$ Footnotes [1] S. A. Glasstone, K. J. Laidler and H. Eyring, The Theory of Rate Processes, McGraw-Hill, New York, 1941, pp. 427-429. [2] D. R. Rosseinsky, J. Chem. Phys.,1968,48, 4806. [3] D. R. Rosseinsky and M. J. Nicol, Trans. Faraday Soc.,1965,61, 2718.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.38%3A_Joule-Thomson_Coefficient.txt
A remarkable feature of the subject called thermodynamics is the extent to which it is founded on four laws: Zeroeth, First, Second and Third. These laws summarize elegantly the results of experiments. Actually these are not laws in the sense of being laid down by government or by religious doctrine. Rather the laws are axioms. As McGlashan notes [1] each axiom is a ‘rule of the game’. These axioms refer to state variables such as temperature, pressure, energy and entropy. At this level the laws are not of immediate interest to chemists. However chemists have discovered how to ‘tell’ these axioms about chemical substances and chemical reactions. The First Law invokes the concepts of energy and energy change. The law states that the energy of the universe is constant. In a realistic sense, at least for chemists, the law states that the energy of a chemical laboratory is constant. Then if the energy of system held in a reaction vessel increases, an equivalent amount of energy is lost from the rest of the laboratory. Then $\Delta \mathrm{U}(\text { system })+\Delta \mathrm{U}(\text { surroundings })=0$ The Second Law of thermodynamics invokes the concepts of entropy and entropy change. In summary the law states that heat cannot flow spontaneously from low to high temperatures. The elegant studies carried out by James Prescott Joule (1818 -1889) were crucial to the development of thermodynamics [2]. Footnotes [1] M. L. McGlashan, Chemical Thermodynamics, Academic Press, London 1979. [2] L. Woodcock and L. Lue, Chem. Britain, 2001, August, p. 38. 1.14.41: Lewisian Variables A given liquid mixture is prepared using $\mathrm{n}_{1}$ moles of liquid 1 and $\mathrm{n}_{2}$ moles of liquid 2. If the thermodynamic properties of the liquid mixture are ideal the volume of the mixture is given by the sum of products of amounts and molar volumes (at the same $\mathrm{T}$ and $\mathrm{p}$); equation (a). $\mathrm{V}(\operatorname{mix} ; \mathrm{id})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ If the thermodynamic properties of the mixture are not ideal, the volume of the (real) mixture is given by equation (b). $V(\operatorname{mix})=\mathrm{n}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{n}_{2} \, \mathrm{V}_{2}(\mathrm{mix})$ $\mathrm{V}_{1}(\operatorname{mix})$ and $\mathrm{V}_{2}(\operatorname{mix})$ are the partial molar volumes of chemical substances 1 and 2 defined by equations (c) and (d). $\mathrm{V}_{1}(\operatorname{mix})=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{n}_{\mathrm{l}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2)}$ $\mathrm{V}_{2}(\operatorname{mix})=\left(\frac{\partial V}{\partial n_{2}}\right)_{T_{, p, n(1)}}$ The similarities between equations (a) and (b) are obvious and indicate an important method for describing the extensive properties of a given system. This was the aim of G. N. Lewis who sought equations of the form show in equation (b). In general terms we identify an extensive property $\mathrm{X}$ of a given system such that the variable can be written in the general form shown in equation (e). $\mathrm{X}=\mathrm{n}_{1} \, \mathrm{X}_{1}+\mathrm{n}_{2} \, \mathrm{X}_{2}$ where $\mathrm{X}_{1}=\left(\frac{\partial \mathrm{X}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2)}$ $\mathrm{X}_{2}(\operatorname{mix})=\left(\frac{\partial \mathrm{X}}{\partial \mathrm{n}_{2}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(1)}$ Other than the composition variables, the conditions on the partial differentials in equations (f) and (g) are intensive properties; 1. mechanical variable, pressure, and 2. thermal variable, temperature. Lewisian partial molar variables can be used to describe the thermodynamic energy $\mathrm{U}$, entropy $\mathrm{S}$ and volume $\mathrm{V}$ together with their Legendre transforms, Helmholtz energy, enthalpy and Gibbs energy. With respect to other thermodynamic properties of a closed system, the case for identifying similar Lewisian partial molar properties has to be established. It turns out that partial molar expansions [e.g. $\mathrm{E}_{\mathrm{p} j}(\mathrm{T}, \mathrm{p})$] and partial molar compressions [e.g. $\mathrm{K}_{\mathrm{T} j}(\mathrm{T}, \mathrm{p})$] for chemical substance $j$ in a closed single phase system are Lewisian but partial molar isentropic compressions are not . 1.14.42: L'Hospital's Rule In several important cases, analysis of thermodynamic properties of solutions (and liquid mixtures) requires consideration of a term having the general form $x \, \ln (x)$ where $x$ is an intensive composition variable; e.g. molality, concentration or mole fraction. The accompanying analysis requires an answer to the question --- what value does the product $x \, \ln (x)$ take in the limit that $x$ tends to zero. But $\operatorname{limit}(x \rightarrow 0) \ln (x)=-\infty$. The thermodynamic analysis has to take account of the answer to this question. In fact most accounts assume that the answer to the above question is ‘zero’. Confirmation that the latter statement is correct emerges from application of L’Hospital’s Rule (G. F. A. de l’Hospital, 1661-1704, marquis de Saint-Mesme). This rule allows the evaluation of terms having indeterminate forms. Most applications of this method usually involve the ratio of two terms each being a function of $x$. If $\mathrm{f}(\mathrm{x}) / \mathrm{F}(\mathrm{x})$ approaches either [0/0] or $[\infty / \infty]$ when $x$ approaches a, and $\mathrm{f}^{\prime}(\mathrm{x}) / \mathrm{F}^{\prime}(\mathrm{x})$ [where $\mathrm{f}^{\prime}(\mathrm{x})$ and $\mathrm{F}^{\prime}(\mathrm{x})$ are first derivatives of $\mathrm{f}(\mathrm{x})$ and $\mathrm{F}(\mathrm{x})$] approaches a limit as $x$ approaches a, then $\mathrm{f}(\mathrm{x}) / \mathrm{F}(\mathrm{x})$ approaches the same limit. Example $1$ If $f(x)=x^{2}-1$ and $F(x)=x-1$ then $\frac{f(x)}{F(x)}=\frac{x^{2}-1}{x-1}$ and $\frac{f^{\prime}(x)}{F^{\prime}(x)}=\frac{2 \, x^{2}}{1}$ then $\operatorname{limit}(x \rightarrow 1) \frac{f^{\prime}(x)}{F^{\prime}(x)}=2$ Hence, $\operatorname{limit}(x \rightarrow 1) \frac{f(x)}{F(x)}=2$ This rule can be proved using three assumptions. 1. In the neighborhood of $x = a$, $F(x) \neq 0 \text { if } x \neq \mathrm{a}$. 2. $f(x)$ and $F(x)$ are continuous in the neighbourhood of $x = \mathrm{a}$ except perhaps at $\mathrm{a}$. 3. $\mathrm{f}^{\prime}(x)$ and $\mathrm{F}^{\prime}(x)$ exist is some neighborhood of $x = \mathrm{a}$ (except perhaps at $x = \mathrm{a}$) and do not vanish simultaneously for $x \neq \mathrm{a}$. In the present context the terms under consideration have a different form. With reference to the term, $x \, \ln (x)$, $f(x)=\ln (x) \text { and } F(x)=1 / x$ Then $\mathrm{f}^{\prime}(\mathrm{x})=1 / \mathrm{x}$ and $\mathrm{F}^{\prime}(\mathrm{x})=-1 / \mathrm{x}^{2}$. Hence, $\mathrm{f}^{\prime}(\mathrm{x}) / \mathrm{F}^{\prime}(\mathrm{x}) = -\mathrm{x}$. Thus $\operatorname{limit}(x \rightarrow 0) \mathrm{f}^{\prime}(x) / \mathrm{F}^{\prime}(\mathrm{x})=0$ Hence, $\operatorname{limit}(x \rightarrow 0) x \, \ln (x)=0$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.40%3A_Laws_of_Thermodynamics.txt
Two important laws of thermodynamics describe spontaneous change in a closed system. First Law $\mathrm{dU}=\mathrm{q}-\mathrm{p} \, \mathrm{dV}$ Second Law $\mathrm{T} \, \mathrm{dS}=\mathrm{q}+\mathrm{A} \, \mathrm{d} \xi ; \mathrm{A} \, \mathrm{d} \xi \geq 0$ Heat $\mathrm{q}$ is common to these equations which we combine. The result is a very important equation. \begin{aligned} &\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi \ &\mathrm{A} \, \mathrm{d} \xi \geq 0 \end{aligned} We use the description ‘Master Equation’. A case can be made for the statement that chemical thermodynamics is based on this Master Equation. The Master Equation describes the differential change in the thermodynamic energy of a closed system. $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}-\mathrm{A} \, \mathrm{d} \xi$ where $\mathrm{T}=\left(\frac{\partial \mathrm{U}}{\partial \mathrm{S}}\right)_{\mathrm{V}, \xi}$ and $\mathrm{p}=-\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{s}_{, \xi}}$ Symbol $\xi$ represents the chemical composition of the system (and quite generally molecular organization). The thermodynamic energy of a closed system containing $\mathrm{k}$ chemical substances is defined by the independent variables $\mathrm{S}$, $\mathrm{V}$ and amounts of each chemical substance. $\mathrm{U}=\mathrm{U}\left[\mathrm{S}, \mathrm{V}, \mathrm{n}_{1}, \mathrm{n}_{2} \ldots \mathrm{n}_{\mathrm{k}}\right]$ We assert that we can independently add $\delta \mathrm{n}_{j}$ moles of any one of the k chemical substances in the system and that the entropy $\mathrm{S}$ and $\mathrm{V}$ can change independently. Based on equation (f), the following (often called the Gibbs equation) is a key relationship. $\mathrm{dU}=\left(\frac{\partial \mathrm{U}}{\partial \mathrm{S}}\right)_{\mathrm{V}, \mathrm{n}(\mathrm{i})} \, \mathrm{dS}+\left(\frac{\partial \mathrm{U}}{\partial \mathrm{V}}\right)_{\mathrm{S}, \mathrm{n}(\mathrm{i})} \, \mathrm{dV}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}}\left(\frac{\partial \mathrm{U}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{s}, \mathrm{V}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \, \mathrm{dn}_{\mathrm{j}}$ Here $\mathrm{n}(\mathrm{i})$ represents the amounts of each of the $\mathrm{k}$ chemical substances in the system. Hence from equation (e), $\mathrm{dU}=\mathrm{T} \, \mathrm{dS}-\mathrm{p} \, \mathrm{dV}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}}\left(\frac{\partial \mathrm{U}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{s}, \mathrm{V}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ The importance of these equations is indicated by imagining a closed system held at constant volume ($\mathrm{dV}=0$) and entropy ($\mathrm{dS}=0$). Under these constraints $\mathrm{dU}(\mathrm{S} \text { and } \mathrm{V} =$ constant) equals - $\mathrm{A} \, \mathrm{d} \xi$. But according to equation(c), the product $\mathrm{A} \, \mathrm{d} \xi$ is always positive for spontaneous reactions. Hence $\mathrm{dU}(\mathrm{S} \text { and } \mathrm{V} =$ constant) is negative. In other words, all spontaneous chemical reactions in a closed system at constant $\mathrm{S}$ and constant $\mathrm{V}$ proceed in a direction which lowers the thermodynamic energy $\mathrm{U}$ of the system. This conclusion is universal, independent of the type of chemical reaction and of the mechanism of chemical reaction. For this reason the thermodynamic energy is the thermodynamic potential function for processes in closed systems at constant $\mathrm{S}$ and constant $\mathrm{V}$ [1]. There is however a problem in terms of practical chemistry. We can envisage designing a reaction vessel which has constant volume. In fact we would probably use heavy steel walls because the conclusions reached above tell us nothing about a possible change in pressure as we face the challenge of holding the volume constant. But it is not obvious what we have to do to hold the entropy constant. Clearly the line of argument is important. Indeed a similar analysis based on the definitions of enthalpy $\mathrm{H}$, Helmholtz energy $\mathrm{F}$ and Gibbs energy $\mathrm{G}$ leads to the following three key equations for changes in enthalpy, Helmholtz energy and Gibbs energy respectively. $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}}\left(\frac{\partial \mathrm{H}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{S}, \mathrm{p}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \, d n_{j}$ $\mathrm{dF}=-\mathrm{S} \, \mathrm{dT}-\mathrm{p} \, \mathrm{dV}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}}\left(\frac{\partial F}{\partial n_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{V}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \, \mathrm{dn}_{\mathrm{j}}$ $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}}\left(\frac{\partial \mathrm{G}}{\partial n_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \, \mathrm{dn} \mathrm{j}_{\mathrm{j}}$ The four partial derivatives with respect to $\mathrm{n}_{j}$ in the four equations define the chemical potential, $\mu_{j}$. \begin{aligned} \mu_{\mathrm{j}}=\left(\frac{\partial \mathrm{U}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{s}, \mathrm{v}, \mathrm{n}(i \neq \mathrm{j})} &=\left(\frac{\partial \mathrm{H}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{S}, \mathrm{p}, \mathrm{n}(i \neq \mathrm{j})} \ &=\left(\frac{\partial \mathrm{F}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{V}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})}=\left(\frac{\partial \mathrm{G}}{\partial \mathrm{n}_{\mathrm{j}}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(\mathrm{i} \neq \mathrm{j})} \end{aligned} For example, $\mathrm{dG}=-\mathrm{S} \, \mathrm{dT}+\mathrm{V} \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{k}} \mu_{\mathrm{j}} \, \mathrm{dn}_{\mathrm{j}}$ In context of chemistry, the latter equation is very important. Footnote [1] An analogy is drawn with electric potential. In an electrical circuit, electric charge flows spontaneously from high to low electric potential.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.43%3A_Master_Equation.txt
Important relationships[1] in thermodynamics are based on Maxwell Equations [2-4]. Consider the state variable G for a given closed system characterized by the two independent variables, $\mathrm{T}$ and $\mathrm{p}$. Hence, $\partial^{2} \mathrm{G} / \partial \mathrm{T} \, \partial \mathrm{p}=\partial^{2} \mathrm{G} / \partial \mathrm{p} \, \partial \mathrm{T}$ or, $\left(\frac{\partial[\partial \mathrm{G} / \partial \mathrm{T}]_{\mathrm{p}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\left(\frac{\partial[\partial \mathrm{G} / \partial \mathrm{p}]_{\mathrm{T}}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ But at both fixed composition $\xi$ and at equilibrium, $\mathrm{A} = 0$, $\mathrm{V}=[\partial \mathrm{G} / \partial \mathrm{p}]_{\mathrm{T}}$ and $\mathrm{S}=-[\partial \mathrm{G} / \partial \mathrm{T}]_{\mathrm{p}}$ Then $\mathrm{E}_{\mathrm{p}}=-(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}=(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$ For the most part we use this relationship in the context of an equilibrium displacement; i.e. at $\mathrm{A} = 0$. Equation (c) shows that at equilibrium the isothermal dependence of entropy on pressure equals, with opposite signs, the isobaric dependence of volume on temperature. $\mathrm{E}_{\mathrm{p}}$ is the isobaric expansion. This equation has practical importance. Suppose we require for either practical or theoretical reasons the dependence of the molar entropy of water($\ell$), $\mathrm{S}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ on pressure at a given temperature. This has all the signs of being a difficult project. However the Maxwell Equation (c) shows that the information is obtained by measuring the dependence of molar volume $\mathrm{V}^{*}\left(\mathrm{H}_{2} \mathrm{O} ; \ell\right)$ on temperature at constant pressure, a simpler approach to the problem. Equation (c) finds several important applications. One application concerns the isothermal dependence of enthalpy on pressure. We start with the equation, $\mathrm{H}=\mathrm{G}-\mathrm{T} \, \mathrm{S}$. We are interested in the dependence of the properties of a given system on pressure at, for example, equilibrium, $\mathrm{A} = 0$ and constant temperature. Then, $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right)_{\mathrm{T}}+\mathrm{T} \,\left(\frac{\partial \mathrm{S}}{\partial \mathrm{p}}\right)_{\mathrm{T}}$ But at $\mathrm{A} = 0, \mathrm{V}=(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}}$. Using equation (c), $\left(\frac{\partial \mathrm{H}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\mathrm{V}-\mathrm{T} \,\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ We see that the isothermal dependence of enthalpy on pressure is readily obtained knowing the volume of a system and its isobaric dependence on temperature. This is another interesting way in which Maxwell equations often simplify tasks facing chemists when probing the properties of systems. In fact equation (e) is fascinating bearing in mind that we can never know the enthalpy $\mathrm{H}$ of a system but we can calculate in a straightforward manner using volumetric properties the isothermal dependence of enthalpy on pressure. In fact the integrated form of equation (e) is also useful. For a system at constant temperature [and at either constant composition $\xi$ or at equilibrium, $\mathrm{A} = 0$], $\mathrm{H}\left(\mathrm{T}, \mathrm{p}_{2}\right)-\mathrm{H}\left(\mathrm{T}, \mathrm{p}_{1}\right)=\int_{\mathrm{p}_{1}}^{\mathrm{p}_{2}}\left[\mathrm{~V}-\mathrm{T} \,(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}\right] \, \mathrm{dp}$ Another important Maxwell Equation is based on the Helmholtz energy, $\mathrm{F}$, of a closed system. $\mathrm{F}=\mathrm{F}[\mathrm{V}, \mathrm{T}, \xi]$ For a closed system at fixed composition $\xi$ (or at equilibrium when $\mathrm{A} = 0$) $\left(\frac{\partial[\partial \mathrm{F} / \partial \mathrm{T}]_{\mathrm{V}}}{\partial \mathrm{p}}\right)_{\mathrm{T}}=\left(\frac{\partial[\partial \mathrm{F} / \partial \mathrm{V}]_{\mathrm{T}}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ But, $\mathrm{S}=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{T}}\right)_{\mathrm{p}}$ and $\mathrm{p}=-\left(\frac{\partial \mathrm{F}}{\partial \mathrm{V}}\right)_{\mathrm{T}}$. Hence, $\left(\frac{\partial \mathrm{S}}{\partial \mathrm{V}}\right)_{\mathrm{T}}=\left(\frac{\partial \mathrm{p}}{\partial \mathrm{T}}\right)_{\mathrm{V}}$ The right-hand-side of equation (i) involves the three practical properties, $\mathrm{p}$, $\mathrm{V}$ and $\mathrm{T}$. In summary, the isochoric dependence of pressure on temperature equals the isothermal dependence of entropy on volume. Two interesting Maxwell Equations develop from the Gibbs energy $\mathrm{G}$. For a system at fixed pressure, $\frac{\partial}{\partial T}\left(\frac{\partial G}{\partial \xi}\right)_{T}=\frac{\partial}{\partial \xi}\left(\frac{\partial G}{\partial T}\right)_{\xi}$ But $\mathrm{A}=-\left(\frac{\partial \mathrm{G}}{\partial \dot{\xi}}\right)_{\mathrm{T}, \mathrm{p}}$, and $S=-\left(\frac{\partial G}{\partial T}\right)_{p, \xi}$, Then, $\left(\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \xi}=\left(\frac{\partial \mathrm{S}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}$ This interesting equation concerns the temperature dependence of the affinity for spontaneous reaction at fixed pressure and composition. In fact this dependence equals the isothermal-isobaric entropy of reaction, $(\partial \mathrm{S} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$. Also with respect to the Gibbs energy we explore the properties of a closed system at fixed temperature. Thus, $\frac{\partial}{\partial p}\left(\frac{\partial G}{\partial \xi}\right)=\frac{\partial}{\partial \xi}\left(\frac{\partial G}{\partial p}\right)$ But, $\left(\frac{\partial \mathrm{G}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}=-\mathrm{A}$, and $\left(\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \xi}=\mathrm{V}$. Then, $-\left(\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right)_{\mathrm{T}, \xi}=\left(\frac{\partial \mathrm{V}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}$ In other words at constant composition the isothermal dependence of the affinity for spontaneous change on pressure equals (minus) the volume of reaction, $(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$. Maxwell Equations are used in the analysis of parameters describing chemical equilibria. In general terms the limiting enthalpy of reaction, $\Delta_{\mathrm{r}} \mathrm{H}^{\infty}$ depends on pressure and the limiting volume of reaction. $\Delta_{\mathrm{r}} \mathrm{V}^{\infty}$ depends on temperature. Further the entropy of reaction at temperature $\mathrm{T}$, $\Delta_{\mathrm{r}} \mathrm{S}^{\#}$ depends on pressure. These complexities signal more complexities in data analysis. Fortunately two Maxwell Equations assist the analysis. [Here ∆rS# refers to the difference between partial molar entropies of reactants and products in solution reference states at a pressure significantly different from the standard pressure.] The isothermal dependence of entropy of reaction on pressure is related to the isobaric dependence of limiting volume of reactions on temperature. $\left[\frac{\partial \Delta_{\mathrm{r}} \mathrm{S}^{\#}}{\partial \mathrm{p}}\right]_{\mathrm{T}}=-\left[\frac{\partial \Delta_{\mathrm{r}} \mathrm{V}^{\infty}}{\partial \mathrm{T}}\right]_{\mathrm{p}}$ Further the isothermal pressure dependence of the limiting enthalpy of reaction is related to the limiting volume of reaction and its isobaric temperature dependence. Thus, $\left[\frac{\partial \Delta_{\mathrm{r}} \mathrm{H}^{\infty}}{\partial \mathrm{p}}\right]_{\mathrm{T}}=\Delta_{\mathrm{r}} \mathrm{V}^{\infty}-\mathrm{T} \,\left[\frac{\partial \Delta_{\mathrm{r}} \mathrm{V}^{\infty}}{\partial \mathrm{T}}\right]_{\mathrm{p}}$ The relationships offer a check of derived quantities and the numerical analysis when equilibrium constants are reported as functions of temperature and pressure. The beauty of thermodynamics is appreciated when one realizes that these relationships are precise. Discovery that a set of data and associated analyses do not conform to these equations does not disprove these Maxwell Equations. Rather one must conclude that analysis of the original experimental results is flawed. In fact Maxwell Equations offer an interesting exercise in units of derived and measured parameters. The isentropic expansion $\mathrm{E}_{\mathrm{S}}$ is related to the isochoric dependence of entropy on pressure [5,6]. From $\mathrm{U}=\mathrm{U}[\mathrm{S}, \mathrm{V}]$, $\partial^{2} U / \partial S \, \partial V=\partial^{2} U / \partial V \, \partial S$ Then, $(\partial \mathrm{T} / \partial \mathrm{V})_{\mathrm{S}}=-(\partial \mathrm{p} / \partial \mathrm{S})_{\mathrm{V}}$ We invert the latter equation. Hence, $\mathrm{E}_{\mathrm{S}}=(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{s}}=-(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{v}}$ Two other Maxwell Equations are worthy of note. From, $\mathrm{dH}=\mathrm{T} \, \mathrm{dS}+\mathrm{V} \, \mathrm{dp}-\mathrm{A} \, \mathrm{d} \xi$ At equilibrium and fixed composition, $\left[\partial(\partial H / \partial S)_{p} / \partial p\right]_{\mathrm{S}}=(\partial T / d p)_{\mathrm{S}}$ and $\left[\partial(\partial H / \partial \mathrm{p})_{\mathrm{s}} / \partial \mathrm{T}\right]_{\mathrm{p}}=(\partial \mathrm{V} / \mathrm{dS})_{\mathrm{p}}$ Then, $(\partial \mathrm{T} / \partial \mathrm{p})_{\mathrm{s}}=(\partial \mathrm{V} / \partial \mathrm{S})_{\mathrm{p}}$ From $(\partial \mathrm{S} / \partial \mathrm{p})_{\mathrm{T}}=-(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$. Then, $(\partial \mathrm{S} / \partial \mathrm{V})_{\mathrm{T}} \,(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{T}}=-(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$ Hence, $(\partial \mathrm{S} / \partial \mathrm{V})_{\mathrm{T}}=-(\partial \mathrm{p} / \partial \mathrm{V})_{\mathrm{T}} \,(\partial \mathrm{V} / \partial \mathrm{T})_{\mathrm{p}}$ Then, $(\partial \mathrm{p} / \partial \mathrm{T})_{\mathrm{V}}=(\partial \mathrm{S} / \partial \mathrm{V})_{\mathrm{T}}$ Footnotes [1] The extent of information available from thermodynamic partial derivatives is explored by: 1. R. Gilmore, J. Chem. Phys., 1981,75, 5964; 1982,77, 5853. 2. M. Ishara, Bull. Chem. Soc. Jpn., 1986,59, 5853. 3. E. Grunwald, J. Am. Chem. Soc., 1984,106, 5414. [2] E. F. Caldin comments on 1010 possible relationships; Chemical Thermodynamics, Oxford, 1958 (page 158). [3] H. Margenau and G. M. Murphy, The Mathematics of Physics and Chemistry, van Nostrand 1943. [4] Extending the observation made by A. B. Pippard [The Elements of Classical Thermodynamics, Cambridge, 1957, p. 46], Maxwell Equations are dimensionally homogeneous in that cross-multiplication yields the following pairs of variables; 1. $\mathrm{p} - \mathrm{~V}$, 2. $\mathrm{T} - \mathrm{~S}$ and 3. $\mathrm{A} - \xi$. The product of each pair is energy, with unit ‘Joule’. $\begin{array}{r} \mathrm{T} \, \mathrm{S}=[\mathrm{K}] \,\left[\mathrm{J} \mathrm{K}^{-1}\right]=[\mathrm{J}] \ \mathrm{p} \, \mathrm{V}=\left[\mathrm{Nm}^{-2}\right] \,\left[\mathrm{m}^{3}\right]=[\mathrm{N} \mathrm{m}]=[\mathrm{J}] \ \mathrm{A} \, \xi=\left[\mathrm{J} \mathrm{mol}^{-1}\right] \,[\mathrm{mol}]=[\mathrm{J}] \end{array}$ [5] With reference to equation (o), \begin{aligned} {\left[\frac{\partial \Delta_{\mathrm{r}} \mathrm{H}^{\infty}}{\partial \mathrm{p}}\right]_{\mathrm{T}}=\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right]+[\mathrm{K}] \,\left[\frac{\mathrm{m}^{3} \mathrm{~mol}^{-1}}{[\mathrm{~K}]}\right] } \ &=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]}{\left[\mathrm{N} \mathrm{m}^{-2}\right]}=\frac{\left[\mathrm{J} \mathrm{mol}^{-1}\right]}{\left[\mathrm{N} \mathrm{m}^{-2}\right]} \end{aligned} [6] S. D. Hamann, Aust. J. Chem.,1984,37,867.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.44%3A_Maxwell_Equations.txt
For a closed system, the dependence of chemical composition $\xi$ on temperature $\mathrm{T}$ at affinity $\mathrm{A}$ and constant pressure is given by equation (a). $\left(\frac{\partial \xi}{\partial T}\right)_{\mathrm{p}, \mathrm{A}}=-\left[\frac{\mathrm{A}+(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}}}{\mathrm{T} \,(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}}\right]$ Similarly for a closed system, the dependence of chemical composition $\xi$ on pressure at fixed temperature is given by equation (b). $\left(\frac{\partial \xi}{\partial \mathrm{p}}\right)_{\mathrm{T}, \mathrm{A}}=\left[\frac{(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}, \mathrm{p}}}{(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}}\right]$ These two equations form the basis of ‘Laws of Moderation’ for closed systems at chemical equilibrium. These equations yield the sign for the two quantities $\left(\frac{\partial \xi}{\partial T}\right)_{p, A=0}$ and $\left(\frac{\partial \zeta}{\partial p}\right)_{T, A=0}$ which describe the change in composition when a system at equilibrium is perturbed to a neighboring equilibrium state. We recall that by definition $\xi$ is positive for displacement in composition from reactants to products; $\left(\frac{\partial V}{\partial \xi}\right)_{T, A=0}$ is the volume of reaction. If $\left(\frac{\partial V}{\partial \xi}\right)_{T, A=0}$ is positive, $\left(\frac{\partial \xi}{\partial p}\right)_{T, A=0}$ is negative because $\left(\frac{\partial \mathrm{A}}{\partial \xi}\right)_{\mathrm{T}, \mathrm{p}}<0$. According to equation (b), an increase in pressure favors a swing in the equilibrium position towards more reactants [1]. Similarly it follows from equation (a) that an increase in temperature favors a swing in the equilibrium position towards more reactants for an exothermic reaction [2]. Moderation is a striking example of the Second Law of Thermodynamics in action with reference to the direction of spontaneous changes in a closed system following changes in either $\mathrm{T}$ or $\mathrm{p}$. Here the stress on the word ‘closed’ reminds us that these laws of moderation do not apply to open system although the point is not always stressed. Therefore controversy often surrounds what is often called Le Chatelier’s Principle. Consider a closed system in which the following chemical equilibrium is established at defined $\mathrm{T}$ and $\mathrm{p}$. $x X+y Y \Leftrightarrow z Z$ As often argued, if $\delta \mathrm{n}_{\mathrm{Y}}$ moles of chemical substance $\mathrm{Y}$ are added to the system, then the equilibrium amount of chemical substance $\mathrm{Z}$ increases. In fact such moderation of composition only occurs if $\sum_{j=1}^{j=i} v_{j}$ is zero for a chemical equilibrium involving i chemical substances. An interesting case concerns the Haber Synthesis. $\mathrm{N}_{2}(\mathrm{g})+3 \mathrm{H}_{2}(\mathrm{g}) \Leftrightarrow 2 \mathrm{NH}_{3}(\mathrm{g})$ If in the equilibrium system mole fraction $\mathrm{x}\left(\mathrm{N}_{2}\right) < 0.5$, addition of a small amount of $\mathrm{N}_{2}(\mathrm{g})$ leads to an increase in the amount of ammonia. However if $\mathrm{x}\left(\mathrm{N}_{2}\right) > 0.5$ addition of a small amount of $\mathrm{N}_{2}(\mathrm{g})$ leads to dissociation of ammonia to form more $\mathrm{N}_{2}(\mathrm{g})$ and $\mathrm{H}_{2}(\mathrm{g})$ [3]. Footnotes [1] This conclusion is called a Theorem of Moderation. Co-author MJB was taught that the outcome was “Nature’s Law of Cussedness” ($\equiv$ Obstinacy). An exothermic reaction operates to generate heat so the system responds when the temperature is raised in the direction for which the process is endothermic. This line of argument is not good thermodynamics but does make the point. [2] Another example of Nature’s Obstinacy; see [1]. Note the switch in sign on the r.h.s of equations (a) and (b). [3] I. Prigogine and R. Defay, Chemical Thermodynamics, transl. D. H. Everett, Longmans - Green, London, 1953, page 268. 1.14.46: Molality and Mole Fraction Molality For a solution prepared using $\mathrm{n}_{j}$ moles of solute and $\mathrm{w}_{\mathrm{s}}$ kg of solvent , molality $m_{j}=n_{j} / w_{s}$ Molality $\mathrm{m}_{j}$ expressed in ‘$\mathrm{mol kg}^{-1}$’ is independent of temperature and pressure being defined by the masses of solvent and solute. The solvent may comprise a mixture of liquids, the composition of the solvent being described using mole fractions, weight-per-cent or volume-per-cent. Mole Fraction For a closed system comprising $\mathrm{n}_{1}, \mathrm{~n}_{2}, \mathrm{~n}_{3} \ldots \mathrm{~n}_{i}$ moles of each $\mathrm{k}$ chemical substance, the mole fraction of chemical substance $\mathrm{j}$, $\mathrm{x}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \sum_{\mathrm{k}=1}^{\mathrm{k}=\mathrm{i}} \mathrm{n}_{\mathrm{k}}$ where $\sum_{k=1}^{k=1} x_{k}=1$ Mole fraction xj is independent of temperature and pressure (in the absence of chemical reaction between the chemical substances in the system).
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.45%3A_Moderation.txt
The Newton-Laplace Equation is the starting point for the determination of isentropic compressibilities of solutions [1,2] using the speed of sound $\mathrm{u}$ and density $\rho$; equation (a) [3]. $\mathrm{u}^{2}=\left(\kappa_{\mathrm{s}} \, \rho\right)^{-1}$ Densities of liquids and speeds of sound at low frequencies can be precisely measured [4,5] .The isentropic condition means that as the sound wave passes through a liquid the pressure and temperature fluctuate within each microscopic volume but the entropy remains constant. The condition ‘at low frequencies‘ is important because at high frequencies ( e.g. $> 100 \mathrm{~MHz}$) there is a velocity dispersion and absorption of sound as the sound wave couples with molecular processes within the liquid [2,6,7]. Several points emerge from a consideration of equation (a). For example one might ask --- is it just assumed that the correct term is $\kappa_{\mathrm{S}}$ and not $\kappa_{\mathrm{T}}$? The point is that in their examination of the properties of aqueous solutions and aqueous mixtures authors often write something along the following lines -- ‘we used the Newton-Laplace equation to calculate $\kappa_{\mathrm{S}}$ from measured speeds of sound’. One might then ask-- can one prove equation (a) and is the proof thermodynamic? Rowlinson states that the speed of sound defined by equation (a) is, and we quote, ‘of course a purely thermodynamic quantity’ [1]. This comment raises the issue as to whether or not the defined quantity equals the measured speed of sound. Intuitively the task of measuring the isothermal property $\mathrm{K}_{\mathrm{T}}\left[=-(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{T}}\right]$ might seem less problematic than measuring the isentropic property, $\mathrm{K}_{\mathrm{s}}\left[=-(\partial \mathrm{V} / \partial \mathrm{p})_{\mathrm{s}}\right]$. $\mathrm{K}_{\mathrm{T}}$ would be obtained by measuring the change in volume following an increase in pressure. However as Tyrer warned [8] in 1914, the isothermal condition is difficult to satisfy and estimated compressions and compressibilities reported up to that time and in the majority of cases were certainly between isentropic and isothermal values. Tyrer did in fact measure $\kappa_{\mathrm{T}}$ and calculated $\kappa_{\mathrm{S}}$ using equation (b). $\kappa_{\mathrm{S}}=\kappa_{\mathrm{T}}-\mathrm{T} \,\left[\alpha_{\mathrm{p}}\right]^{2} / \sigma$ Other authors [9,10] have measured $\kappa_{\mathrm{T}}$ directly by, for example, the volume increase on sudden decompression of a liquid from high to ambient pressure. Nevertheless the conventional approach uses the Newton-Laplace equation. Historically this subject has its origins in the attempts initiated in the 17th Century to measure the speed of sound in air [11]. A sound wave traveling through a fluid produces a series of compressions and rarefactions. Consequently planes of molecules perpendicular to the direction of the sound waves are displaced. The displacement $\varepsilon$ depends on both position $\mathrm{x}$ and time $t$. Thus $\varepsilon=\varepsilon[\mathrm{x}, \mathrm{t}]$ The speed of the sound wave $\mathrm{u}$ is related to the displacement ε using equation (d), the wave equation. $\left(\partial^{2} \varepsilon / \partial x^{2}\right)=\left(1 / u^{2}\right) \,\left(\partial^{2} \varepsilon / \partial t^{2}\right)$ A classic analysis [12] in terms of equation (d) and stress-strain relationships for an isotropic phase using Hooke’s Law yields equation (a). At this stage we could consider both the isothermal compressibility $\kappa_{\mathrm{T}}$ and the isentropic compressibility $\kappa_{\mathrm{S}}$. If equation (a) is correct then, either (a) the speed of sound can be calculated knowing $\kappa_{\mathrm{S}}$ and $\rho$, or (b) $\kappa_{\mathrm{S}}$ can be calculated by measuring speed of sound $\mathrm{u}$ and density $\rho$. Another line of argument states that equation (a) defines the speed of sound in terms of $\kappa_{\mathrm{S}}$ and density $\rho$. The question arises -- is the speed of sound calculated using equation (a) equal to the measured speed of sound? The analysis up to and including equation (d) was familiar to Newton (I. Newton 1642-1727) [13]. Newton using Boyle's Law assumed that the fluid is an ideal gas and that the compressions and rarefactions are isothermal (and in a thermodynamic sense, reversible); Hence $\mathrm{u}^{2}= \mathrm{p} / \rho$ Equation (e) was particularly important to Newton because the three quantities in equation (e) can independently determined for (dry) air. Using the density ρ for air at pressure p one can calculate the speed of sound in air. The agreement between observed and calculated speeds was, somewhat disappointingly, only fair but encouraging. The disagreement was an underestimate by 20% as was noted by Newton. The argument is interesting in the sense of testing if the analysis yields the measured speed of sound. Clearly the equations do not. An important contribution was made by Laplace [14] who assumed that the compressions and rarefactions are perfect and isentropic; i.e. $\mathrm{p} \, \mathrm{V}^{\gamma}=$ constant where $\gamma$ is the ratio of isobaric and isochoric heat capacities. This is the assertion made by Laplace. The overall condition is isentropic for a gas at temperature T. The condition refers to macroscopic properties. Within each microscopic volume both temperature and pressure fluctuate but the entropy remains constant. [The equilibrium and isentropic conditions mean that there is no loss of heat on compression and no gain of heat on rarefaction when the sound wave passes through the system; everything is in phase.] Assuming that $\gamma$ is independent of $\mathrm{p}$, $\mathrm{u}^{2}=\gamma \, \mathrm{p} / \rho$ The point is that Laplace knew $\gamma$ for (dry) air at $273 \mathrm{~K}$ and standard pressure equals 1.4. With this information Laplace obtained good agreement between theory and experiment for the speed of sound in air. In other words Laplace confirmed his assertion that for air (a fluid with low density, $1.29 \times 10^{-3} \mathrm{~g cm}^{-3}$) compressions and rarefactions are isentropic and not isothermal. Hence the fame of the Newton-Laplace equation which is based on an assertion. Laplace did not prove that the processes are isentropic but having shown agreement between theory and experiment one must conclude that the assertion is correct for air. Equation (a) is the Newton-Laplace Equation. The key point is that the equation emerges from an Equation of State for isentropic compressions of a particular gas, air. Indeed the success achieved by the Newton-Laplace equation in term of predicting the speed of sound in a gas is noteworthy. However we need to comment on the link between $\kappa_{\mathrm{S}}$ measured directly and obtained from measurements of $\kappa_{\mathrm{T}}$, \alpha_{\mathrm{p}}\) and $\mathrm{C}_{\mathrm{p}}$ using equation (b) [15]. We direct attention to a given closed system containing liquid water. From a practical standpoint, the difference between isothermal and isentropic compressibilities (cf. equation (b)) written here for the pure liquid water, $\frac{\left[\alpha_{\mathrm{pl}}^{*}(\ell)\right]^{2} \, \mathrm{V}_{1}^{*}(\ell) \, \mathrm{T}}{\mathrm{C}_{\mathrm{pl}}^{*}(\ell)}$. is reasonably accessible. The molar volume $\mathrm{V}_{1}^{*}(\ell)$ is obtained from the density $\rho_{1}^{*}(\ell); \alpha_{p 1}^{*}(\ell)$ is obtained from the dependence of density on temperature at fixed pressure. The molar isobaric heat capacity $\mathrm{C}_{p 1}^{*}(\ell)$ is also experimentally accessible. The most frequently cited data set for $\mathrm{V}_{1}^{*}(\ell)$ and $\alpha_{p 1}^{*}(\ell)$ was published by Kell and Whalley in 1965 [16]; see also reference [17]. The isothermal compressibility is less accessible . In 1967 Kell summarized [18] the results obtained by Kell and Whalley [16] and quoted that at 25 Celsius, $\alpha_{\mathrm{p} 1}^{*}(\ell)=257.05 \times 10^{-6} \mathrm{~K}^{-1}$ and $\kappa_{\mathrm{T} 1}^{*}(\ell)=45.24 \mathrm{Mbar}^{-1}$. In 1969 Millero and co-workers[19] directly measured isothermal compressions of water($\ell$) drawing comparisons with the estimates made by Kell and Whalley [16,17]. They reported that for water($\ell$) at 25 Celsius, $\kappa_{\mathrm{Tl}}^{*}(\ell)=(45.94 \pm 0.06) \mathrm{Matm}^{-1}$. Millero et al. comment [19] on the excellent agreement. In 1970, Kell addressed the issue which is of interest here [18]. Equation (b) is the key to the debate because we obtain an estimate of $\kappa_{\mathrm{S} 1}^{*}(\ell)$ from measured $\kappa_{\mathrm{2} 1}^{*}(\ell), \alpha_{\mathrm{pl}}^{*}(\ell), \mathrm{~V}_{1}^{*}(\ell) \text { and } \mathrm{C}_{\mathrm{pl}}^{*}(\ell)$; i.e. $\kappa_{\mathrm{S}}^{*}(\ell ; \text { density })$. Alternatively we obtain $\kappa_{\mathrm{S} 1}^{*}(\ell)$ using equation (a); i.e. speed of sound and density yielding $\kappa_{\mathrm{S}}$ (acoustic) . The key question is --- are $\kappa_{\mathrm{S}}^{*}(\ell ; \text { density })$ and $\kappa_{\mathrm{S}}$ (acoustic) equal? How confident are we that they are equal? There are no assumptions underlying the calculation of $\kappa_{\mathrm{S}}^{*}(\ell ; \text { density })$. In the case of $\kappa_{\mathrm{S}}$ (acoustic), the sound wave perturbs the system isentropically; cf. Laplace analysis. Kell comments [20] that speeds of sound can be precisely measured and also a precise estimate of the defined $\kappa_{\mathrm{S}}$ (acoustic) is obtained. Granted the validity of equation (a) one can re-express equation (b) as an equation for $\kappa_{\mathrm{T} 1}^{*}(\ell)$ in terms of measured $\kappa_{\mathrm{S} 1}^{*}(\ell), \alpha_{\mathrm{p} 1}^{*}(\ell), \mathrm{~V}_{1}^{*}(\ell) \text { and } \mathrm{C}_{\mathrm{p} 1}^{*}(\ell)$. Examination of various sets of data showed that $\kappa_{\mathrm{S}}$ (acoustic) has less systematic errors than $\kappa_{\mathrm{S}}^{*}(\ell ; \text { density })$ but that they are effectively the same, a point confirmed by Fine and Millero [21]. Footnotes [1] J. S. Rowlinson and F. L. Swinton, Liquid and Liquid Mixtures, Butterworths, London , 3rd. edn., 1982, pp. 16-17. [2] J. O. Hirschfelder, C. F. Curtis and R. B. Bird, Molecular Theory of Gases and Liquids, Wiley, New York, corrected printing 1964, chapters 5 and 11. [3] $\mathrm{u}^{2}=\frac{1}{\kappa_{\mathrm{s}}} \, \frac{1}{\rho}=\left[\mathrm{N} \mathrm{m}^{-2}\right] \, \frac{1}{\left[\mathrm{~kg} \mathrm{~m}^{-3}\right]}=\frac{\left[\mathrm{kg} \mathrm{m} \mathrm{s}^{-2} \mathrm{~m}^{-2}\right]}{\left[\mathrm{kg} \mathrm{m}^{-3}\right]}=\left[\mathrm{m}^{2} \mathrm{~s}^{-2}\right]$ $\mathrm{u}=\left[\mathrm{m} \mathrm{s}^{-1}\right]$ [4] A.T. J. Hayward, Brit. J. Appl. Phys.,1967,18,965, [5] A. T. J. Hayward, J. Phys. D: Appl. Physics, 1971.4,938. [6] A.T. J .Hayward, Nature, 1969,221.1047 [7] M. J. Blandamer, Introduction to Chemical Ultrasonics, Academic Press,1973. [8] D. Tyrer, J. Chem. Soc.,1914,105,2534. [9] D. Harrison and E. A. Moelwyn-Hughes, Proc. R. Soc. London, Ser.A, 1957, 239. 230. [10] L. A. K. Staveley, W. I. Tupman and K. Hart, Trans. Faraday, Soc.,1955,51,323. [11] P. Costabel and L.Auger, in Science in the Nineteenth Century; ed. R. Taton, transl. A. J. Pomerans, Basic Books, New York,1961, p. 170. [12] S. G. Starling and A. J. Woodhall, Physics, Longmans, London, 2nd edn.,1957. [13] I. Newton, Philosophicae Naturalis Principia Mathematica, Vol.II, Sect VII, Prop.46, London,1687. [14] S. Laplace, Ann. Chim. Phvs., 1816,3,328. [15] D.-P. Wang and F. J. Millero, J. Geophys. Res.,1973,78,7122. [16] G. S. Kell and E. Whalley, Philos. Trans. R. Soc. London, Ser.A.,1965,258,565. [17] G. S. Kell, G. E. McLaurin and E. Whalley, Proc. R. Soc. London, Ser,.A, 1978, 360, 389. [18] G. S. Kell, J. Chem. Eng. Data, 1967,12,66;1970,l5, 119. [19] F. J. Millero, R.W. Curry and W. Drost-Hansen, J. Chem. Eng. Data,1969,14,422. [20] G. S. Kell, in Water A Comprehensive Treatise, ed. F. Franks, Plenum Press, New York,1973,volume I, chapter 10. [21] R. A. Fine and F. J. Millero, J Chem. Phys.,1973,59,5529;1975,63,89.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.47%3A_Newton-Laplace_Equation.txt
The chemical thermodynamics of open systems [1-2] is more complicated than that of closed systems because chemical substances exchange between system and surroundings, crossing the boundary of the system. Footnote [1] D. Kondepudi and I. Prigogine, Modern Thermodynamics; From Heat Engines to Dissipative Structures, Wiley, New York, 1998. [2] Clearly a treatment of the chemical thermodynamics of the human body has to take account of the fact that such systems are open. Farmers are very practical chemical thermodynamic experts because in feeding their livestock they judge if the animals they are feeding will 1. produce milk for sale, 2. meat for food, 3. skin for the manufacture of leather and/or 4. be used for breeding. Farmers do not leave these options to chance as they cope in very practical way with such open systems. 1.14.49: Osmotic Coefficient There is possible disadvantage in an approach using the mole fraction scale to express the composition of a solution. Granted 1. that our interest is often in the properties of solutes in aqueous solutions, 2. that the amount of solvent greatly exceeds the amount of solute in a solution, and 3. that the sensitivity of equipment developed by chemists is sufficient to probe the properties of quite dilute solutions, the mole fraction scale for the solvent is not the most convenient method for expressing the composition of a given solution [1-3]. Hence another equation relating $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ to the composition of a solution finds favor. By definition, for a solution containing a single solute, chemical substance $\mathrm{j}$ [4], $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ Or, in terms of the standard chemical potential for water at temperature $\mathrm{T}$ and standard pressure $\mathrm{p}^{0}$, $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{1}^{*}\left(\ell ; \mathrm{T} ; \mathrm{p}^{0}\right)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}+\int_{\mathrm{p}^{0}}^{\mathrm{p}} \mathrm{V}_{1}^{*}(\ell ; \mathrm{T}) \, \mathrm{dp}$ $\mathrm{M}_{1}$ is the molar mass of water; $\phi$ is the practical osmotic coefficient which is characteristic of the solute, molality mj, temperature and pressure. By definition $\phi$ is unity for ideal solutions at all temperatures and pressures. $\operatorname{limit}\left(\mathrm{m}_{\mathrm{j}} \rightarrow 0\right) \phi=1.0 \text { at all T and } \mathrm{p}$ Further for ideal solutions, the partial differentials $(\partial \phi / \partial T)_{p}, \left(\partial^{2} \phi / \partial T^{2}\right)_{p} \text { and } (\partial \phi / \partial \mathrm{p})_{\mathrm{T}}$ are zero. For an ideal solution [5], $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p} ; \mathrm{id})=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ We rewrite equation (d) in the following form: $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p} ; \mathrm{id})-\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})=-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ Hence with an increase in molality of solute in an ideal aqueous solution, the solvent is stabilized, being at a lower chemical potential than that for pure water. We contrast the chemical potentials of the solvent in real and ideal solutions using an excess chemical potential, $\mu_{1}^{E}(a q ; T ; p)$; \begin{aligned} \mu_{1}^{\mathrm{E}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}) &=\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})-\mu_{1}(\mathrm{aq} ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}) \ &=(1-\phi) \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}} \end{aligned} The term $(1 - \phi)$ is often encountered because it expresses succinctly the impact of the solute on the properties of the solvent. At a given molality (and fixed temperature and pressure), $\phi$ is characteristic of the solute. In the case of a salt $\mathrm{j}$ which on complete dissociation forms ν ions the analogue of equation (a) takes the following form. $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})-\mathrm{v} \, \phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ Footnotes [1] N. Bjerrum, Z. Electrochem., 1907, 24,259. [2] G. N. Lewis and M. Randall, Thermodynamics, revised by K. S. Pitzer and L. Brewer, McGraw-Hill, New York, 2nd edn., 1961, chapter 22. [3] Mole fractions of solvent $\mathrm{x}_{1}$ for aqueous solutions having gradually increasing molality of solute $\mathrm{m}_{j}. (A) \(\mathrm{m}_{\mathrm{j}} / \mathrm{mol} \mathrm{kg}^{-1}=10^{-3}$; $\mathrm{x}_{1}=0.999982$ $x_{j}=1.8 \times 10^{-5}$ (B) $\mathrm{m}_{\mathrm{j}} / \mathrm{mol} \mathrm{kg}=10^{-2}$; $\mathrm{x}_{1}=0.99982$ $x_{\mathrm{j}}=1.8 \times 10^{-4}$ (C) $\mathrm{m}_{\mathrm{j}} / \mathrm{mol} \mathrm{kg}=10^{-1}$; $\mathrm{x}_{1}=0.9982$ $x_{j}=1.8 \times 10^{-3}$ (D) $\mathrm{m}_{\mathrm{j}} / \mathrm{mol} \mathrm{kg}{ }^{-1}=0.5$; $\mathrm{x}_{1}=0.9911$ $\mathrm{x}_{\mathrm{j}}=8.9 \times 10^{-3}$ (E) $\mathrm{m}_{\mathrm{j}} / \mathrm{mol} \mathrm{kg}=1.0$ $\mathrm{x}_{1}=0.9823$ $\mathrm{x}_{\mathrm{j}}=1.77 \times 10^{-2}$ [4] $\left[\mathrm{J} \mathrm{mol}^{-1}\right]=\left[\mathrm{J} \mathrm{mol}^{-1}\right]-[1] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}] \,\left[\mathrm{kg} \mathrm{mol}^{-1}\right] \,\left[\mathrm{mol} \mathrm{kg}^{-1}\right]$ The definitions of ideal solutions expressed in equations (i) and (ii) are not in conflict. $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p} ; \mathrm{id})=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}$ $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p} ; \mathrm{id})=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)$ Thus for an ideal solution these equations require that, $v-\mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}=\ln \left(\mathrm{x}_{1}\right)$ But $\ln \left(\mathrm{x}_{1}\right)=\ln \left[\mathrm{M}_{1}^{-1} /\left(\mathrm{M}_{1}^{-1}+\mathrm{m}_{\mathrm{j}}\right)\right]=-\ln \left(1.0+\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}\right)$ Bearing in mind that $\mathrm{M}_{1} = 0.018 \mathrm{~kg mol}^{-1}$, then for dilute solutions $\ln \left(1.0+\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}\right)=\mathrm{m}_{\mathrm{j}} \, \mathrm{M}_{1}$.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.48%3A_Open_System.txt
A semi-permeable membrane [1] separates an aqueous solution (where the mole fraction of water equals $\mathrm{x}_{1}$) and pure solvent at temperature $\mathrm{T}$ and ambient pressure. Solvent water flows spontaneously across the membrane thereby diluting the solution. This flow is a consequence of the chemical potential of the solvent in the solution being lower than the chemical potential of pure solvent at the same $\mathrm{T}$ and $\mathrm{p}$. If a pressure ($\mathrm{p} + \pi$) is applied to the solution, the spontaneous process stops because the solution at pressure ($\mathrm{p} + \pi$) and the solvent at pressure $\mathrm{p}$ are in thermodynamic equilibrium; $\pi$ is the osmotic pressure. Thus at equilibrium, $\mu_{1}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}+\pi)=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})$ Under this equilibrium condition solvent flows in both directions across the semi-permeable membrane but the net flow is zero. In the analysis presented here we take account of the fact, writing $\mathrm{p}^{\prime}$ for ($\mathrm{p} + \pi$), the chemical potential of water in the aqueous solution is given by equation (b). \begin{aligned} &\mu_{1}^{\mathrm{eq}}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{\prime}\right)= \ &\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0)+\mathrm{p}^{\prime} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \mathrm{K}_{\mathrm{Tl}}^{*}(\ell) \, \mathrm{p}^{\prime}\right] \ & \ &+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right) \end{aligned} Here $\mathrm{f}_{1}$ is the activity coefficient expressing the extent to which the thermodynamic properties of water in the aqueous solution are not ideal. For the pure solvent water at pressure $\mathrm{p}$ (i.e. on the other side the of the semi-permeable membrane), \begin{aligned} &\mu_{1}^{\mathrm{eq}}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})= \ &\quad \mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0)+\mathrm{p}^{\prime} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{Tl}}^{*}(\ell) \, \mathrm{p}\right] \end{aligned} But osmosis experiments explore an equilibrium characterized by equation (d). $\mu_{1}\left(\mathrm{aq} ; \mathrm{T} ; \mathrm{p}^{\prime}\right)=\mu_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})$ Therefore using equations (b) and (c), $\begin{array}{r} \mathrm{p}^{\prime} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}^{\prime}\right]+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)= \ \mathrm{p} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}\right] \end{array}$ Or, \begin{aligned} &\mathrm{p} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}^{\prime}\right] \ &-\mathrm{p} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}\right]=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right) \end{aligned} The terms $\mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}^{\prime}\right]$ and $\mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p}=0) \,\left[1-(1 / 2) \, \kappa_{\mathrm{T} 1}^{*}(\ell) \, \mathrm{p}\right]$ describe the molar volumes of water at pressures $\mathrm{p}$ and $\mathrm{p}^{\prime}$; i.e. at pressure $\mathrm{p}$ and ($\mathrm{p}+\pi$). We assume that both terms can be replaced by the molar volumes at average pressure $[(2 \, \mathrm{p}+\pi) / 2]$; namely $\mathrm{V}_{1}^{*}[\ell ; \mathrm{T} ;(2 \, \mathrm{p}+\pi) / 2]$. Therefore $\pi \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ;(2 \, \mathrm{p}+\pi) / 2]=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ In the event that $\pi<<2 \, p$, $\pi \, \mathrm{V}_{1}^{*}[\ell ; \mathrm{T} ; \mathrm{p}]=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1} \, \mathrm{f}_{1}\right)$ If the thermodynamic properties of the solutions are ideal, $\mathrm{f}_{1}$ equals unity. Then $\pi^{\mathrm{id}} \, \mathrm{V}_{1}^{*}[\ell ; \mathrm{T} ; \mathrm{p}]=-\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{x}_{1}\right)$ In the latter two equations $\mathrm{V}_{1}^{*}(\ell ; \mathrm{T} ; \mathrm{p})$ is treated as a constant, independent of the thermodynamic properties of the solution. A further interesting development of equation (i) is possible for a solution prepared using $\mathrm{n}_{1}$ moles of solvent water and $\mathrm{n}_{j}$ moles of solute. Thus $-\ln \left(\mathrm{x}_{1}\right)=\ln \left(\frac{1}{\mathrm{x}_{1}}\right)=\ln \left(\frac{\mathrm{n}_{1}+\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}\right)=\ln \left(1+\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}\right)$ But . $\left(\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{1}\right)<<1$. We expand the last term in equation (j). $\ln \left(1+\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}\right)=\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}-\frac{1}{2} \,\left(\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}\right)^{2}+\frac{1}{3} \,\left(\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}}\right)^{3}-\ldots . .$ If we retain only the first term; $\pi^{\mathrm{id}} \, \mathrm{V}_{1}^{*}[\ell ; \mathrm{T} ; \mathrm{p}]=\mathrm{R} \, \mathrm{T} \,\left(\mathrm{n}_{\mathrm{j}} / \mathrm{n}_{1}\right)$ But for a dilute solution, the volume of the solution $\mathrm{V}$ is given by $\mathbf{n}_{1} \, \mathrm{V}_{1}^{*}[\ell ; \mathrm{T} ; \mathrm{p}]$. Or [2], $\pi^{\mathrm{id}} \, \mathrm{V}(\mathrm{aq})=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ But concentration $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}$ Then [3] $\pi^{\mathrm{id}}=\mathrm{c}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ Agreement between $\pi(\mathrm{obs})$ and $\pi^{\mathrm{id}}$ for aqueous solutions containing neutral solutes (e.g. sucrose) confirms the validity of the thermodynamic analysis. Footnotes [1] The term ‘semi-permeable’ in the present context means that the membrane is only permeable to the solvent. Perhaps the optimum semipermeable membrane is the vapor phase. [2] Historically, equation(o) owes much to the equation of state for an ideal gas; i.e. $\mathrm{p} \, \mathrm{V}=\mathrm{n} \, \mathrm{R} \, \mathrm{T}$. From an experimentally found proportionality between $\pi$ and $\mathrm{c}_{j}$, van’t Hoff showed that the proportionality constant can be approximated by $\mathrm{R} \, \mathrm{T}$. [3] $\pi=\left[\mathrm{mol} \mathrm{m}^{-3}\right] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]=\left[\mathrm{J} \mathrm{m}^{-3}\right]=\left[\mathrm{N} \mathrm{m} \mathrm{m}^{-3}\right]=\left[\mathrm{N} \mathrm{m}^{-2}\right]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.50%3A_Osmotic_Pressure.txt
The laws of thermodynamics and the associated treatment of the thermodynamic properties of closed systems concentrate attention on macroscopic properties. Although we may define the composition of a closed system in terms of the amounts of each chemical substance in a system, general thermodynamic treatments direct our attention to macroscopic properties such as, volume $\mathrm{V}$, Gibbs energy $\mathrm{G}$, enthalpy $\mathrm{H}$ and entropy $\mathrm{S}$. We need to ‘tell’ these thermodynamic properties that a given system probably comprises different chemical substances. In this development the analysis is reasonably straightforward if we define the system under consideration by the ‘Gibbsian ‘set of independent variables; i.e. $\mathrm{T}, \mathrm{~p}$ and amounts of each chemical substance. Thus, $\mathrm{G}=\mathrm{G}\left[\mathrm{T}, \mathrm{p}, \mathrm{n}_{1}, \mathrm{n}_{2}, \ldots \mathrm{n}_{\mathrm{i}}\right]$ In equation (a), the Gibbs energy is defined by intensive variables $\mathrm{T}$ and $\mathrm{p}$ together with extensive composition variables. In many cases the task of a chemist is to assay a system to determine the number and amounts of each chemical substance in the system. The analysis leads to the definition of the chemical potential for each substance $\mathrm{j}$, $\mu_{\mathrm{j}}$ in a closed system. Consider a solution comprising $\mathrm{n}_{1}$ moles of solvent, liquid chemical substance 1, and $\mathrm{n}_{\mathrm{j}}$ moles of solute, chemical substance $\mathrm{j}$. We ask—what contributions are made by the solvent and by the solute to the volume of the solution at defined $\mathrm{T}$ and $\mathrm{p}$? In fact we can only guess at these contributions [1]. This is disappointing. The best we can do is to probe the sensitivity of the volume of a given solution to the addition of small amounts of either solute or solvent. This approach leads to a set of properties called partial molar volumes. Here we explore the definition of these properties. The starting point is the Gibbs energy of a solution. We develop the argument in a way which places the Gibbs energy at the centre from which all other thermodynamic variables develop. For a closed system containing i-chemical substances, $\mathrm{G}=\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{n}_{\mathrm{j}} \, \mu_{\mathrm{j}}$ The later equation signals that the total Gibbs energy is given by the sum of products of amounts and chemical potentials of each chemical substance in the system. For an aqueous solution containing $\mathrm{n}_{\mathrm{j}}$ moles of solute $\mathrm{j}$ and $\mathrm{n}_{1}$ moles of solvent 1 ( water), $G(a q)=n_{1} \, \mu_{1}(a q)+n_{j} \, \mu_{j}(a q)$ We do not have to attach to equation (c) the condition ‘at fixed $\mathrm{T} and \mathrm{p}$’. Similarly the volume of the solution is given by equation (d). $V(a q)=n_{1} \, V_{1}(a q)+n_{j} \, V_{j}(a q)$ The same argument applies in the case of a system prepared using $\mathrm{n}_{1}$ moles of water, $\mathrm{n}_{\mathrm{X}}$ moles of solute $\mathrm{X}$ and $\mathrm{n}_{\mathrm{Y}}$ moles of solute $\mathrm{Y}$. $\mathrm{G}(\mathrm{aq})=\mathrm{n}_{1} \, \mu_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{X}} \, \mu_{\mathrm{X}}(\mathrm{aq})+\mathrm{n}_{\mathrm{Y}} \, \mu_{\mathrm{Y}}(\mathrm{aq})$ Complications emerge however if solute $\mathrm{X}$ and $\mathrm{Y}$ are in chemical equilibrium; e.g. $X(a q) \Leftrightarrow Y(a q)$. Then account must be taken of the fact that $\mathrm{n}_{\mathrm{x}}^{\mathrm{eq}}$ and $\mathrm{n}_{\mathrm{x}}^{\mathrm{eq}}$ depend on $\mathrm{T}$ and $\mathrm{p}$. Footnote [1] I have on my desk a flask containing water ($\ell ; 100 \mathrm{~cm}^{3}$) and 20 small round steel balls, each having a volume of $0.1 \mathrm{~cm}^{3}$. I add a steel ball to the flask and the volume of the system, water + steel ball, is $100.1 \mathrm{~cm}^{3}$. I add one more steel ball and the volume of the system increases by $0.1 \mathrm{~cm}^{3}$. So in this simple case I can equate directly the volume of the pure steel balls $\mathrm{V}^{*}$(balls) with the partial molar volume of the balls in the system, water + balls. I have on my desk an empty egg carton designed to hold six eggs. The volume of the carton is represented as $\mathrm{V}(\mathrm{c})$ as judged by the volume occupied in a food store.. The volume of one egg is $\mathrm{V}^{*}$(egg), the superscript * indicating that we are discussing the property of pure eggs. I now ‘add’ one egg to the egg carton which does not change its volume ----again as judged by the volume occupied in a food store. In other words the partial molar volume of eggs in the egg carton $\mathrm{V}(\mathrm{egg})$ is zero; $V(\text { egg })=\left(\frac{\partial V(\text { system })}{\partial n(\text { egg })}\right)=\text { zero. Or, } V(\text { egg })-V^{*}(\text { egg })<0$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.51%3A_Partial_Molar_Properties%3A_General.txt
A given liquid mixture is prepared using $\mathrm{n}_{1}$ moles of liquid 1 and $\mathrm{n}_{2}$ moles of liquid 2. If the thermodynamic properties of the liquid mixture are ideal the volume of the mixture is given by the sum of products of amounts and molar volumes (at the same $\mathrm{T}$ and $\mathrm{p}$); equation (a). $\mathrm{V}(\operatorname{mix} ; \mathrm{id})=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)+\mathrm{n}_{2} \, \mathrm{V}_{2}^{*}(\ell)$ If the thermodynamic properties of the mixture are not ideal, the volume of the (real) mixture is given by equation (b). $\mathrm{V}(\operatorname{mix})=\mathrm{n}_{1} \, \mathrm{V}_{1}(\operatorname{mix})+\mathrm{n}_{2} \, \mathrm{V}_{2}(\operatorname{mix})$ $\mathrm{V}_{1}(\operatorname{mix})$ and $\mathrm{V}_{1}(\operatorname{mix})$ are the partial molar volumes of chemical substances 1 and 2 defined by equations (c) and (d). $V_{1}(m i x)=\left(\frac{\partial V}{\partial n_{1}}\right)_{T, p, n(2)}$ $V_{2}(\operatorname{mix})=\left(\frac{\partial V}{\partial n_{2}}\right)_{T, p, n(1)}$ The similarities between equations (a) and (b) are obvious and indicate an important method for describing the extensive properties of a given system. This was the aim of G. N. Lewis [1] who sought equations of the form shown in equation (b). In general terms, we identify an extensive property $\mathrm{X}$ of a given system such that the variable can be written in the general form shown in equation (e). $\mathrm{X}=\mathrm{n}_{1} \, \mathrm{X}_{1}+\mathrm{n}_{2} \, \mathrm{X}_{2}$ where $\mathrm{X}_{1}=\left(\frac{\partial \mathrm{X}}{\partial \mathrm{n}_{1}}\right)_{\mathrm{T}, \mathrm{p}, \mathrm{n}(2)}$ $X_{2}(\operatorname{mix})=\left(\frac{\partial X}{\partial n_{2}}\right)_{T_{, p, n(1)}}$ Other than the composition variables, the conditions on the partial differentials in equations (f) and (g) are intensive properties; 1. mechanical variable, pressure, and 2. thermal variable, temperature. Partial molar properties can also be defined for different pairs of intensive thermal and non-thermal variables, other than $\mathrm{T}$ and $\mathrm{p}$ [2]. The concept of a partial property was extended to intensive properties such as isothermal and isentropic compressibilities [3]. A further distinction between Lewisian and non-Lewisian partial molar properties has been proposed [2,4]. Footnotes [1] G. N. Lewis, Proc. Am. Acad. Arts Sci.,1907,43,259. [2] J. C. R. Reis, J. Chem. Soc Faraday Trans.,2,1982,78,1575. [3] J. C. R. Reis, J. Chem. Soc Faraday Trans.,1998,94,2385. [4] J. C. R. Reis, M. J. Blandamer, M. I. Davis and G. Douheret, Phys. Chem.Chem.Phys.,2001,3,1465. 1.14.53: Phase Rule According to the Gibbs-Duhem Equation, the properties of a single phase at equilibrium containing i chemical substances are related; we divide the Gibbs-Duhem Equation by the total amount in the system such that $\mathrm{x}_{j}(\alpha)$ is the mole fraction of substance $j$ in the $\alpha$ phase. The Gibbs-Duhem Equation. requires that $0=\mathrm{S}_{\mathrm{m}}(\alpha) \, \mathrm{dT}-\mathrm{V}_{\mathrm{m}}(\alpha) \, \mathrm{dp}+\sum_{\mathrm{j}=1}^{\mathrm{j}=\mathrm{i}} \mathrm{x}_{\mathrm{j}}(\alpha) \, \mathrm{d} \mu_{\mathrm{j}}(\alpha)$ Within this phase, the definition of mole fraction means that over all $\mathrm{i}$-chemical substances, $\sum_{j=1}^{j=i} x_{j}(\alpha)=1$ The number of independent intensive variables is $[\mathrm{P} \,(\mathrm{C}-1)+2]$ where $\mathrm{C}$ is the number of independent chemical substances in phase $\alpha$. The additional two variables refer to the intensive temperature and pressure. We consider the case where the closed system contains $\mathrm{P}$ phases. Therefore we can set down $\mathrm{P}$ equations of the form shown in equation (a). With reference to the chemical potential of substance $j$, the overall equilibrium condition requires that the chemical potentials of this substance over all phases ( i.e. $\alpha_{1}, \alpha_{2}, \alpha_{3}, \ldots \ldots \alpha_{p}$) are equal. $\mu_{j}\left(\alpha_{1}\right)=\mu_{j}\left(\alpha_{2}\right)=\mu_{j}\left(\alpha_{3}\right)=\ldots \ldots \ldots . .=\mu_{j}\left(\alpha_{p}\right)$ Hence with reference to the intensive chemical potentials there are ($\mathrm{P} - 1$) constraints. Therefore the number of independent intensive variables for this system comprising $\mathrm{i}$ chemical substances distributed through $\mathrm{P}$ phases, namely $\mathrm{F}$, equals $(\mathrm{C} −1) + 2 − (\mathrm{P} −1)$. Therefore $\mathrm{P}+\mathrm{F}=\mathrm{C}+2$ The latter is the Phase Rule. This equation is possibly the most elegant and practical equation in chemistry.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.52%3A_Partial_Molar_Properties%3A_Definitions.txt
A given closed system comprises chemical substance j in two homogeneous subsystems which are separated by an appropriate semipermeable diaphragm and which are at the same temperature but different pressures. The subsystems I and II are in thermodynamic equilibrium. Thus (cf. Topic 690), $\mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)=\mu_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right)$ For subsystem I, $\mathrm{d} \mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)=\left(\frac{\partial \mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)}{\partial \mathrm{T}}\right)_{\mathrm{p}} \, \mathrm{dT}+\left(\frac{\partial \mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)}{\partial \mathrm{p}}\right)_{\mathrm{T}} \, \mathrm{dp}_{1}$ Or, $\mathrm{d} \mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)=-\mathrm{S}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right) \, \mathrm{dp}_{1}$ Here $\mathrm{S}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)$ and $\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{\mathrm{l}}\right)$ are molar properties of chemical substance $j$. Similarly, $\mathrm{d} \mu_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right)=-\mathrm{S}_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right) \, \mathrm{dT}+\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right) \, \mathrm{dp}_{2}$ The equality expressed in equation (a) is valid at all $\mathrm{T}$ and $\mathrm{p}$. Clearly this condition can only be satisfied if the following equation is satisfied. $\mathrm{d} \mu_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)=\mathrm{d} \mu_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right)$ Then at constant temperature, $V_{j}^{*}\left(I, T, p_{1}\right) \, d p_{1}=V_{j}^{*}\left(I I, T, p_{2}\right) \, d_{2}$ Hence, $\frac{\mathrm{dp}_{1}}{\mathrm{dp}_{2}}=\frac{\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}_{1}, \mathrm{p}_{2}\right)}{\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)}$ The latter is the Poynting Equation [1]. An interesting application of this equation concerns the case where system II is the vapor phase and system I is the liquid phase. The vapor phase is described as an ideal gas using equation (h) for one mole of chemical substance $j$. $\mathrm{p}_{2} \, \mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{II}, \mathrm{T}, \mathrm{p}_{2}\right)=\mathrm{R} \, \mathrm{T}$ The liquid phase comprises one mole of liquid j for which $\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)$ is the molar volume which is assumed to be a constant, independent of pressure. Hence from equations (g) and (h), $\frac{\mathrm{dp}_{1}}{\mathrm{dp}_{2}}=\frac{1}{\mathrm{~V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)} \, \frac{\mathrm{R} \, \mathrm{T}}{\mathrm{p}_{2}}$ Or, $\mathrm{R} \, \mathrm{T} \, \mathrm{d} \ln \left(\mathrm{p}_{2}\right)=\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right) \, d \mathrm{p}_{1}$ The assumption is made that, phase I being a liquid, $\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right)$ is independent of pressure. Then equation (j) is integrated between pressure limits $\mathrm{p}_{2}$ and ${\mathrm{p}_{2}}^{\prime}$ and between $\mathrm{p}_{1}$ and ${\mathrm{p}_{1}}^{\prime}$. Hence, $\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{p}_{2}^{\prime} / \mathrm{p}_{2}\right)=\mathrm{V}_{\mathrm{j}}^{*}\left(\mathrm{I}, \mathrm{T}, \mathrm{p}_{1}\right) \,\left[\mathrm{p}_{1}^{\prime}-\mathrm{p}_{1}\right]$ An interesting application of equation (k) concerns the impact of an increase in pressure from $\mathrm{p}_{1}$ and ${\mathrm{p}_{1}}^{\prime}$ on liquid $j$. This increase might be produced for example by an increase in confining pressure of an inert gas insoluble in liquid $j$. Equation (k) describes the increase in vapor pressure from $\mathrm{p}_{2}$ to ${\mathrm{p}_{2}}^{\prime}$ of liquid $j$. This pattern might seem intuitively somewhat unexpected. Footnotes [1] J. J. Vanderslice, H. W. Schamp Jr and E. A. Mason, Thermodynamics, Prentice Hall,Englewood Cliffs, N.J., 1966, page 106. [2] Poynting, Phil. Mag.,1881,[4],12,32. 1.14.55: Process In order to document the thermodynamics of processes a convention has been agreed. In general, the thermodynamic variable takes the following form. $\Delta_{\text {proc }} \mathrm{X}^{0}$ Here 1. $\Delta$ signals a change is the thermodynamic extensive variable $\mathrm{X}$; 2. the subscript ‘proc’ signals the process; e.g. • $\mathrm{f} =$ formation • $\mathrm{c} =$ combustion • $\mathrm{vap} =$ vaporisation • $\mathrm{r} =$ chemical reaction • $\mathrm{aln} =$ = solution In recognition of the long tradition of using a ‘double-dagger’, a superscript $\neq$ indicates activation as in the formation of ‘transition state from reactants. 1. the superscript ‘0’ means under standard conditions which should be defined. Example. • $\Delta_{\mathrm{c}} \mathrm{H}^{0}=$ standard enthalpy of combustion • $\Delta^{\neq} \mathrm{V}^{0}=$ standard volume of activation
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.54%3A_Poynting_Relation.txt
A given closed system having Gibbs energy $\mathrm{G}$ at temperature $\mathrm{T}$, pressure $\mathrm{p}$, molecular composition (organization $\xi$) and affinity for spontaneous change $\mathrm{A}$ is described by equation (a). $\mathrm{G}=\mathrm{G}[\mathrm{T}, \mathrm{p}, \xi]$ In the state defined by equation (a), there is an affinity for spontaneous chemical reaction $\mathrm{A}$. Starting with the system in the state defined by equation (a) it is possible to change the pressure and perturb the system to a series of neighboring states for which affinity remains constant. The differential dependence of $\mathrm{G}$ on pressure for the original state along the path at constant $\mathrm{A}$ is given by $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \mathrm{A}}$. Returning to the original state characterized by $\mathrm{T}$, $\mathrm{p}$ and $\xi$, we imagine that it is possible to perturb the system by a change in pressure in such a way that the system remains at fixed extent of reaction, $\xi$. The differential dependence of $\mathrm{G}$ on pressure for the original state along the path at constant $\xi$ is given by $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \xi}$. We explore these dependences of $\mathrm{G}$ on pressure at fixed temperature and at 1. fixed composition, $\xi$ and 2. fixed affinity for spontaneous change, $\mathrm{A}$. The procedure for relating $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \mathrm{A}}$, and $(\partial \mathrm{G} / \partial \mathrm{p})_{\mathrm{T}, \xi}$ is a standard calculus operation. At fixed temperature, $\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{A}}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\xi}-\left[\frac{\partial \mathrm{A}}{\partial \mathrm{p}}\right]_{\xi} \,\left[\frac{\partial \xi}{\partial \mathrm{A}}\right]_{\mathrm{p}} \,\left[\frac{\partial \mathrm{G}}{\partial \xi}\right]_{\mathrm{p}}$ This interesting equation shows that the differential dependence of Gibbs energy (at constant temperature) on pressure at constant affinity for spontaneous change does NOT equal the corresponding dependence at constant extent of chemical reaction. This inequality is not surprising. But our interest is drawn to the case where the system under discussion is, at fixed temperature and pressure, at thermodynamic equilibrium where $\mathrm{A}$ is zero, $\mathrm{d} \xi / \mathrm{dt}$ is zero, Gibbs energy is a minimum AND, significantly, $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero. Hence $\mathrm{V}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{T}, \mathrm{A}=0}=\left[\frac{\partial \mathrm{G}}{\partial \mathrm{p}}\right]_{\mathrm{T}, \xi^{\mathrm{eq}}}$ The dependence of $\mathrm{G}$ on pressure for differential displacements at constant ‘$\mathrm{A} = 0$’ and $\xi^{\mathrm{eq}}$ are identical. We confirm that the volume $\mathrm{V}$ of a system is a ‘strong’ state variable. These comments seem trivial but the point is made if we go on to consider the volume of a system as a function of temperature at constant pressure. We use a calculus operation to derive equation (d). $\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{A}}=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\xi}-\left(\frac{\partial \mathrm{A}}{\partial \mathrm{T}}\right)_{\xi} \,\left(\frac{\partial \xi}{\partial \mathrm{A}}\right)_{\mathrm{T}} \,\left(\frac{\partial \mathrm{V}}{\partial \xi}\right)_{\mathrm{T}}$ Again we are not surprised to discover that in general terms the differential dependence of $\mathrm{V}$ on temperature at constant affinity does not equal the differential dependence of $\mathrm{V}$ on temperature at constant composition/organization. Indeed, unlike the simplification we could use in connection with equation (b), {namely that at equilibrium $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero} we cannot assume that the volume of reaction, $(\partial \mathrm{V} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero at equilibrium. In other words for a closed system at thermodynamic equilibrium at fixed $\mathrm{T}$ and fixed $\mathrm{p}$ {when $\mathrm{A}=0$, $\xi = \xi^{\mathrm{eq}}$ and $\mathrm{d} \xi^{\mathrm{eq}} / \mathrm{dt}=0$}, there are two thermal expansions, at constant $\mathrm{A}$ and at constant $\xi$ξ. We consider a closed system in equilibrium state I defined by the set of variables,$\left\{\mathrm{T}[\mathrm{I}], \mathrm{p}, \mathrm{A}=0, \xi^{\mathrm{eq}}[\mathrm{I}]\right\}$. The equilibrium composition is $\xi^{\mathrm{eq}}[\mathrm{I}]$ at zero affinity for spontaneous change. This system is perturbed to two nearby states at constant pressure. 1. State I is displaced to a nearby equilibrium state II defined by the set of variables, $\left\{\mathrm{T}[\mathrm{I}]+\delta \mathrm{T}, \mathrm{p}, \mathrm{A}=0, \xi^{\mathrm{eq}}[\mathrm{II}]\right\}$. This equilibrium displacement is characterized by a volume change; $\Delta \mathrm{V}(\mathrm{A}=0)=\mathrm{V}[\mathrm{II}]-\mathrm{V}[\mathrm{I}]$ At constant pressure we record the equilibrium thermal expansion; $\mathrm{E}_{\mathrm{p}}(\mathrm{A}=0)=\left[\frac{\mathrm{V}[\mathrm{II}]-\mathrm{V}[\mathrm{I}]}{\Delta \mathrm{T}}\right]_{\mathrm{p}, \mathrm{A}=0}$ The equilibrium isobaric expansibility, $\alpha_{\mathrm{p}}(\mathrm{A}=0)=\mathrm{E}_{\mathrm{p}}(\mathrm{A}=0) / \mathrm{V}$ In order for the system to move from one equilibrium state, I with composition $\xi^{\mathrm{eq}}[\mathrm{I}]$ to another equilibrium state, II with composition $\xi^{\mathrm{eq}}[\mathrm{II}]$, the system changes by a change in chemical composition and/or molecular organization. Hence we define the ‘frozen’ isobaric expansion, $\mathrm{E}_{\mathrm{p}} (\xi = \text { fixed})$. An alternative name is the instantaneous expansion because, practically, we would have to change the temperature at a such a high rate that there is no change in molecular composition or organisation in the system. $\mathrm{E}_{\mathrm{p}}(\xi=\text { fixed })=\left(\frac{\partial \mathrm{V}}{\partial \mathrm{T}}\right)_{\mathrm{p}, \xi}$ Further $\alpha_{p}(\xi=\text { fixed })=\frac{1}{V} \,\left(\frac{\partial V}{\partial T}\right)_{p, \xi}$ Similar comments apply to isothermal compressibilities, $\mathrm{K}_{\mathrm{T}}$; there are two limiting quantities $\kappa_{\mathrm{T}}(\mathrm{A}=0)$ and $\kappa_{\mathrm{T}}(\xi)$. In order to measure $\kappa_{\mathrm{T}}(\xi)$ we have to change the pressure also in an infinitely short time. The entropy $\mathrm{S}$ is given by the partial differential, $-(\partial \mathrm{G} / \partial \mathrm{T})_{\mathrm{p}, \xi}$. At equilibrium where $\mathrm{A}=0, \mathrm{~S}=-(\partial \mathrm{G} / \partial \mathrm{T})_{\mathrm{p}, \mathrm{A}=0}$. We carry over the argument described in the previous section but now concerned with a change in temperature. We consider the two pathways, constant $\mathrm{A}$ and constant $\xi$. \begin{aligned} &(\partial \mathrm{G} / \partial \mathrm{T})_{\mathrm{p}, \mathrm{A}}= \ &\quad(\partial \mathrm{G} / \partial \mathrm{T})_{\mathrm{p}, \xi}-(\partial \xi / \partial \mathrm{A})_{\mathrm{T}, \mathrm{p}} \,(\partial \mathrm{A} / \partial \mathrm{T})_{\mathrm{p}, \bar{\xi}} \,(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}} \end{aligned} But at equilibrium, $\mathrm{A}$ which equals $-(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero, and so $\mathrm{S}(\mathrm{A}=0)$ equals $\mathrm{S}\left(\xi^{\mathrm{eq}}\right)$. Then just as for volumes, the entropy of a system is not a property concerned with pathways between states; entropy is a strong function of state. Another important link involving Gibbs energy and temperature is provided by the Gibbs-Helmholtz equation. We explore the relationship between changes in $(\mathrm{G} / \mathrm{T})$ at constant affinity $\mathrm{A}$ and at fixed $\xi$, following perturbation by a change in temperature. \begin{aligned} &{[\partial(\mathrm{G} / \mathrm{T}) / \partial \mathrm{T}]_{\mathrm{p}, \mathrm{A}}=} \ &{[\partial(\mathrm{G} / \mathrm{T}) / \partial \mathrm{T}]_{\mathrm{p}, \xi}-(1 / \mathrm{T}) \,(\partial \xi / \partial \mathrm{A})_{\mathrm{T}, \mathrm{p}} \,(\partial \mathrm{A} / \partial \mathrm{T})_{\mathrm{p}, \xi} \,(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}} \end{aligned} But at equilibrium, $\mathrm{A}$ which equals $-(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero. Then $\mathrm{H}(\mathrm{A}=0)=\mathrm{H}\left(\xi^{\mathrm{eq}}\right)$. In other words, the variable enthalpy is another strong function of state. This is not the case for isobaric heat capacities. \begin{aligned} &(\partial \mathrm{H} / \partial \mathrm{T})_{\mathrm{p}, \mathrm{A}}= \ &(\partial \mathrm{H} / \partial \mathrm{T})_{\mathrm{p}, \xi}-(\partial \xi / \partial \mathrm{A})_{\mathrm{T}, \mathrm{p}} \,(\partial \mathrm{A} / \partial \mathrm{T})_{\mathrm{p}, \xi} \,(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}} \end{aligned} We cannot assume that the triple product term in the latter equation is zero. Hence, there are two limiting isobaric heat capacities; the equilibrium isobaric heat capacity $C_{p}(A=0)$ and the frozen isobaric heat capacity $\mathrm{C}_{\mathrm{p}}(\xi \mathrm{eq})$. In other words, an isobaric heat capacity is not a strong function of state because it is concerned with a pathway between states. Unless otherwise stated, we use the symbol $\mathrm{C}_{\mathrm{p}}$ to indicate an equilibrium transformation, $\mathrm{C}_{\mathrm{p}}(\mathrm{A}=0)$.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.56%3A_Properties%3A_Equilibrium_and_Frozen.txt
In thermodynamics the term 'reversible' means that in such a system the affinity for spontaneous change $\mathrm{A}$ is zero; we can in fact characterize the composition of the system by the symbol $\xi^{\mathrm{eq}}$, indicating a time independent extent of chemical reaction. The composition of the system does not change because the affinity for spontaneous change is zero. For a reversible change the affinity for spontaneous change is zero at all stages. The composition is represented by $\xi^{\mathrm{eq}}$, and the rate of change $\mathrm{d} \xi^{\mathrm{eq}} / \mathrm{dt}$ is zero, at defined $\mathrm{T}$ and $\mathrm{p}$. We represent the volume of the system using following equation. $\mathrm{V}=\mathrm{V}\left[\mathrm{T}, \mathrm{p}, \xi^{\mathrm{eq}}, \mathrm{A}=0\right]$ This equation means that the volume, a dependent variable, is unambiguously defined by the set of variables in the square brackets, [... ]. The pressure is changed from $\mathrm{p}$ to $\mathrm{p} + \Delta \mathrm{p}$, such that the new equilibrium composition is $\xi + \Delta \xi$ where the affinity for spontaneous change is zero. $\mathrm{V}=\mathrm{V}\left[\mathrm{T},(\mathrm{p}+\Delta \mathrm{p}), \xi^{\mathrm{eq}}(\mathrm{p}+\Delta \mathrm{p}), \mathrm{A}=0\right]$ Under these circumstances the change from $\mathrm{V}(\mathrm{p})$ to $\mathrm{V}(\mathrm{p} + \Delta \mathrm{p})$ is from one equilibrium state where $\mathrm{A} = 0$ to another equilibrium state where $\mathrm{A}$ is also zero. Such an equilibrium transformation is, in thermodynamic terms, reversible. All changes under the constraint that $\mathrm{A}$ remains at zero are reversible. 1.14.58: Reversible Chemical Reactions Two important themes in thermodynamics concern the description of chemical equilibria and the kinetics of chemical reactions in closed systems at fixed temperature and pressure. These two themes are often linked in descriptions of chemical reactions. We comment on this link. A given aqueous solution (at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, which is close to the standard pressure, $\mathrm{p}^{0}$) is prepared using chemical substance $\mathrm{X}$. Spontaneous chemical reaction forms chemical substance $\mathrm{Z}$ in the following chemical reaction. $\mathrm{X}(\mathrm{aq}) \rightarrow \mathrm{Z}(\mathrm{aq})$ Experiment confirms that the extent of chemical reaction is given by equation (b). $\frac{1}{\mathrm{~V}} \, \frac{\mathrm{d} \xi}{\mathrm{dt}}=-\mathrm{k}_{1} \,[\mathrm{X}]$ In this system, $\mathrm{c}_{\mathrm{x}}\{=[\mathrm{X}]\}=\mathrm{n}_{\mathrm{x}} / \mathrm{V}$. The common assumption is that for dilute solutions both $\mathrm{k}_{1}$ and volume $\mathrm{V}$ are independent of time. $\mathrm{dc}_{\mathrm{x}} / \mathrm{dt}=-\mathrm{k}_{1} \, \mathrm{c}_{\mathrm{x}}$ Rate constant $\mathrm{k}_{1}$ is expressed using the unit, $\mathrm{s}^{-1}$. We consider a system prepared using chemical substance $\mathrm{Z}$ which undergoes spontaneous chemical reaction to form chemical substance $\mathrm{X}$. The analogue of equation (c) takes the following form.; $\mathrm{dc}_{\mathrm{Z}} / \mathrm{dt}=-\mathrm{k}_{2} \, \mathrm{c}_{\mathrm{Z}}$ We assert that the chemical reaction described by equations (c) and (d) proceed until the properties of the system (at fixed $\mathrm{T}$ and $\mathrm{p}$) are independent of time. In other words the system is in thermodynamic equilibrium with the surroundings with $\mathbf{c}_{X}=\mathbf{c}_{X}^{e q}$ and $\mathbf{c}_{Z}=\mathbf{c}_{Z}^{e q}$ where, macroscopically, $\mathrm{dc}_{\mathrm{X}} / \mathrm{dt}$ and $\mathrm{dc}_{\mathrm{Z}} / \mathrm{dt}$ are zero. Also $\operatorname{limit}\left(\mathrm{m}_{\mathrm{z}} \rightarrow 0 ; \mathrm{m}_{\mathrm{x}} \rightarrow 0\right) \gamma_{\mathrm{z}}=1$ and $\operatorname{limit}\left(\mathrm{m}_{\mathrm{x}} \rightarrow 0\right.$; and $\left.\mathrm{m}_{\mathrm{z}}=0\right) \gamma_{\mathrm{x}}=1$ Thus $\mathrm{k}_{1} \, \mathrm{c}_{\mathrm{x}}^{\mathrm{eq}}=\mathrm{k}_{2} \, \mathrm{c}_{\mathrm{Z}}^{\mathrm{eq}}$ From a thermodynamic point of view, at equilibrium (at fixed $\mathrm{T}$ and $\mathrm{p}$) the affinity for spontaneous change $\mathrm{A}$ is zero, and the system is at a minimum in Gibbs energy $\mathrm{G}$. If the molalities of substances $\mathrm{X}$ and $\mathrm{Z}$ are $\mathrm{m}_{\mathrm{X}}^{\mathrm{eq}}$ and $\mathrm{m}_{\mathrm{Z}}^{\mathrm{eq}}$ respectively, the standard increase in Gibbs energy $\Delta_{\mathrm{r}} \mathrm{G}^{0}(\mathrm{~T}, \mathrm{p}, \mathrm{aq})$ is related to a (dimensionless) thermodynamic equilibrium constant $\mathrm{K}(\mathrm{aq} ; \mathrm{T} ; \mathrm{p})$ using equation (f). $\Delta_{\mathrm{r}} \mathrm{G}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln (\mathrm{K})=\mu_{\mathrm{Z}}^{0}(\mathrm{aq})-\mu_{\mathrm{x}}^{0}(\mathrm{aq})$ where, $\mathrm{K}=\left(\mathrm{m}_{\mathrm{z}} \, \gamma_{\mathrm{z}} / \mathrm{m}^{0}\right)^{\mathrm{eq}} /\left(\mathrm{m}_{\mathrm{x}} \, \gamma_{\mathrm{x}} / \mathrm{m}^{0}\right)^{\mathrm{eq}}$ Here $\gamma_{\mathrm{X}}$ and $\gamma_{\mathrm{Z}}$ are the activity coefficients for solutes $\mathrm{X}$ and $\mathrm{Z}$. If the aqueous solution is quite dilute, we can assume that the thermodynamic properties of the solution are ideal. Moreover the ratio $\left(m_{Z} / m_{X}\right)^{e q}$ is, based on the same approximation, equal to $\left(c_{Z} / c_{X}\right)^{e q}$. In other words equations (e) and (g) can be written in the following forms. Law of Mass Action $\left(c_{Z} / c_{X}\right)^{e q}=\left(k_{1} / k_{2}\right)^{e q}$ Thermodynamics $\left(\mathrm{c}_{\mathrm{Z}} / \mathrm{c}_{\mathrm{X}}\right)^{\mathrm{eq}}=\mathrm{K}$ Within the context of the assumptions outlined above , we obtain by comparing equations (h) and (i) the following classic equation. $\mathrm{K}=\mathrm{k}_{1} / \mathrm{k}_{2}$ Equation (j) is fascinating because the two sides of the equation have different origins, Law of Mass Action and the Laws of Thermodynamics. Indeed equation (j) is often used in an introduction to the concept of chemical equilibrium, the latter emerging as a ‘balance of rates of reaction’. In a wider context equation (j) is used in treatments of fast chemical reactions where a given closed system is only marginally displaced from equilibrium by transient changes in electric field, magnetic field, pressure or temperature [1-3]. Footnotes [1] E. Caldin, Fast Reactions in Solution, Blackwell Scientific Publications, Oxford, 1964. [2] M. J. Blandamer, Introduction to Chemical Ultrasonics, Academic Press, London, 1973. [3] The analysis takes a similar form in cases where the reaction stoichiometry is more complicated. Consider the case of an association reaction in aqueous solution. $\mathrm{X}(\mathrm{aq})+\mathrm{Y}(\mathrm{aq}) \rightarrow \mathrm{Z}(\mathrm{aq})$ Law of Mass Action For the forward reaction $\mathrm{dc}_{\mathrm{X}} / \mathrm{dt}=-\mathrm{k}_{1} \, \mathrm{c}_{\mathrm{X}} \, \mathrm{c}_{\mathrm{Y}}$ For the reverse reaction $\mathrm{Z}(\mathrm{aq}) \rightarrow \mathrm{X}(\mathrm{aq})+\mathrm{Y}(\mathrm{aq})$ $\mathrm{dc}_{\mathrm{Z}} / \mathrm{dt}=-\mathrm{k}_{2} \, \mathrm{c}_{\mathrm{z}}$ For a system where, macroscopically, $\mathrm{dc}_{\mathrm{X}} / \mathrm{dt}=\mathrm{dc}_{\mathrm{Y}} / \mathrm{dt}=\mathrm{dc}_{\mathrm{Z}} / \mathrm{dt}=0$, $\mathrm{k}_{1} \,\left(\mathrm{c}_{\mathrm{X}} \, \mathrm{c}_{\mathrm{Y}}\right)^{\mathrm{cq}}=\mathrm{k}_{2} \,\left(\mathrm{c}_{\mathrm{Z}}\right)^{\mathrm{eq}}$ Or, $\mathrm{k}_{1} / \mathrm{k}_{2}=\left(\mathrm{c}_{\mathrm{Z}}\right)^{\mathrm{eq}} /\left(\mathrm{c}_{\mathrm{X}} \, \mathrm{c}_{\mathrm{Y}}\right)^{\mathrm{eq}}$ From a thermodynamic viewpoint, at equilibrium (at fixed $\mathrm{T}$ and $\mathrm{p}$), $\Delta_{\mathrm{r}} \mathrm{G}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln (\mathrm{K})=\mu_{\mathrm{Z}}^{0}(\mathrm{aq})-\mu_{\mathrm{X}}^{0}(\mathrm{aq})-\mu_{\mathrm{Y}}^{0}(\mathrm{aq})$ where $\mathrm{K}=\left(\mathrm{m}_{\mathrm{Z}} \, \gamma_{\mathrm{Z}} / \mathrm{m}^{0}\right)^{\mathrm{eq}} /\left(\mathrm{m}_{\mathrm{X}} \, \gamma_{\mathrm{X}} / \mathrm{m}^{0}\right)^{\mathrm{eq}} \,\left(\mathrm{m}_{\mathrm{Y}} \, \gamma_{\mathrm{Y}} / \mathrm{m}^{0}\right)^{\mathrm{eq}}$ In the limit that the solution is dilute, $\left(\gamma_{\mathrm{Z}}\right)^{e q}=\left(\gamma_{\mathrm{x}}\right)^{\mathrm{eq}}=\left(\gamma_{\mathrm{Y}}\right)^{\mathrm{eq}}=1$. Then $\mathrm{K}=\left(\mathrm{c}_{\mathrm{Z}} / \mathrm{c}_{\mathrm{r}}\right)^{\mathrm{eq}} /\left(\mathrm{c}_{\mathrm{X}} / \mathrm{c}_{\mathrm{r}}\right)^{\mathrm{eq}} \,\left(\mathrm{c}_{\mathrm{Y}} / \mathrm{c}_{\mathrm{r}}\right)^{\mathrm{eq}}$ Comparison of equations (e) and (h) allows identification of the ratio $\mathrm{k}_{1} / \mathrm{k}_{2}$ with the equilibrium constant $\mathrm{K}$. Comment If at equilibrium $\mathrm{dc}_{\mathrm{X}} / \mathrm{dt}=0$ and $\mathbf{c}_{\mathrm{x}}^{\mathrm{cq}} \neq 0$ then according to equation (c), $\mathrm{k}_{1}$ must be zero. The same problem arises from equation (d). To circumvent this objection, the Principle of Microscopic Reversibility states that at Equilibrium, the amount of chemical substance $\mathrm{X}$ consumed by equation (a) and described by equation (c) equals the amount of chemical substance $\mathrm{X}$ produced by the reverse reaction described by equation (d).
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.57%3A_Reversible_Change.txt
A classic subject concerning the properties of salt solutions has centred for more than a century on the effects of an added salt on the solubility of an apolar (volatile) solute. In terms of the Phase Rule, a given closed system contains two phases, gas and liquid. The liquid phase is an aqueous salt solution. A volatile chemical substance is distributed between the vapour and liquid phases. Hence the number of phases $\mathrm{P}$ equals 2; the number of components $\mathrm{C}$ equals 3; i.e. water + salt + volatile chemical substance. Hence the number of degrees of freedom $\mathrm{F}$ equals 3. If therefore we define the temperature, pressure and concentration of salt in the aqueous salt solution, the thermodynamic equilibrium is completely defined. Similarly if the closed system contains pure liquid $j$ and an aqueous salt solution which also contains solute $j$, the number of degrees of freedom is again 3. Then the equilibrium state is completely defined by specifying $\mathrm{T}, \mathrm{~p}$ and the concentration (molality) of salt in solution. In this case the equilibrium at defined $\mathrm{T}$ and $\mathrm{p}$ is defined by equation \ref{a}. $\mu_{\mathrm{j}}^{*}(\ell)=\mu_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})\label{a}$ In the absence of salt, treating substance $j$ as a solute in aqueous solution, equation \ref{b} describes this equilibrium in terms of the equilibrium composition of the solution assuming ambient pressure is close to the standard pressure. $\mu_{\mathrm{j}}^{*}(\ell)=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)_{\mathrm{aq}}^{\mathrm{cq}}\label{b}$ A similar equilibrium is established but this time the aqueous solution contains a salt, molality $\mathrm{m}_{\mathrm{s}}$. Equation \ref{b} takes the following form. $\mu_{\mathrm{j}}^{*}(\ell)=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)_{\mathrm{s}}^{\mathrm{\alpha q}}\label{c}$ The subscript on the last term in equation \ref{c} indicates that the aqueous solution contains salt $\mathrm{S}$ as well as apolar solute $j$. According to equations \ref{b} and \ref{c}, the two solubilities of substance $j$ are related. $\left(m_{j} \, \gamma_{j}\right)_{a q}^{e q}=\left(m_{j} \, \gamma_{j}\right)_{s}^{e q}\label{d}$ Equation \ref{d} is thermodynamically correct. The change in solubility of chemical substance $j$ on adding salt $\mathrm{S}$, molality ms is compensated by a change in the activity coefficient of solute $j$. The corresponding equation on the concentration scale has the form shown in equation \ref{e}. $\left(c_{j} \, y_{j}\right)_{a q}^{e q}=\left(c_{j} \, y_{j}\right)_{s}^{e q}\label{e}$ The latter is the usual form of the equation. The analysis is readily repeated for the case where the chemical substance $j$ is a volatile gas. At this stage a number of extra-thermodynamic assumptions are built into the analysis. To reduce the clutter of symbols we drop the designation ‘$\mathrm{eq}$’ taking this condition as implicit in all that follows. Further we assume that substance $j$ is sparingly soluble so that in the aqueous solution $j-j$ solute-solute interactions are unimportant. Therefore the properties of solute $j$ are ideal; $\left(\mathrm{y}_{\mathrm{j}}\right)_{\mathrm{aq}}=1$. Hence from equation \ref{e}, $\ln \left[\left(c_{j}\right)_{\mathrm{aq}} /\left(\mathrm{c}_{\mathrm{j}}\right)_{\mathrm{s}}\right]=\ln \left[\left(\mathrm{y}_{\mathrm{j}}\right)_{\mathrm{s}}\right]\label{f}$ In these terms $\left(\mathrm{y}_{j}\right)_{\mathrm{s}}$ is the activity coefficient of solute $j$ in the aqueous salt solutions where the concentration of salt is represented by $\mathrm{c}_{\mathrm{s}}$. For dilute salt solutions the assumption is made that $\ln \left[\left(\mathrm{y}_{\mathrm{j}}\right)_{\mathrm{s}}\right]$ is a linear function of $\mathrm{c}_{\mathrm{s}}$. $\ln \left[\left(\mathrm{y}_{\mathrm{j}}\right)_{\mathrm{s}}\right]=\mathrm{k} \, \mathrm{c}_{\mathrm{s}}\label{g}$ Combination of equations \ref{f} and \ref{g} yields the following equation. $\ln \left[\left(c_{j}\right)_{a q} /\left(c_{j}\right)_{s}\right]=k \, c_{s}\label{h}$ Equation \ref{h} is one form of the Setchenow equation in which constant $\mathrm{k}$ (at fixed $\mathrm{T}$ and $\mathrm{p}$) is characteristic of salt $\mathrm{S}$ and solute $j$. An alternative form starts by expressing $\left(c_{j}\right)_{s}$ as $\left(c_{j}\right)_{a q}-\delta c_{j}$ implying a reduction in the solubility of solute $j$ when a salt is added; i.e. a salting-out.[2] Hence, $\delta \mathrm{c}_{\mathrm{j}} /\left(\mathrm{c}_{\mathrm{j}}\right)_{\mathrm{aq}}=\mathrm{k} \, \mathrm{c}_{\mathrm{s}}\label{i}$ This Setchenow Equation requires that $\delta \mathrm{c}_{\mathrm{j}} /\left(\mathrm{c}_{\mathrm{j}}\right)_{\mathrm{aq}}$ is a linear function of $\mathrm{c}_{\mathrm{s}}$. A positive $\mathrm{k}$ describes a salting-out; a negative $\mathrm{k}$ describes a salting-in. The phenomenon by which solubilities of gases in aqueous solutions are changed by adding a salt attracts enormous interest, both from practical and theoretical standpoints [3,4]. Conway reviewed theoretical models which attempt to account quantitatively for the phenomenon [5]. Considerable attention has been given to theories based on the relationship between the impact of the non-polar solute on the dielectric properties of the solvent and hence the chemical potential of the salt in solution [6,7]. For the most part salting-out is the commonly observed pattern [8]. Nevertheless there are some interesting cases where apolar solutes are salted-in by tetra-alkylammonium salts; benzene[9,10] , methane[11] and helium[12] in $\mathrm{Bu}_{4}\mathrm{N}^{+} \mathrm{~Br}^{-} (\mathrm{aq})$. It would appear that an added apolar solute is stabilized by interaction with the apolar alkyl groups of the cations [13]. Footnotes [1] J.Setchenow, Z. Phys. Chem.,1889,4,117. [2] \begin{aligned} &\ln \left[\frac{\left(c_{j}\right)_{a q}}{\left(c_{j}\right)_{s}}\right]=\ln \left[\frac{\left(c_{j}\right)_{a q}}{\left(c_{j}\right)_{a q}-\delta c_{j}}\right]=-\ln \left[\frac{\left(c_{j}\right)_{\mathrm{aq}}-\delta c_{j}}{\left(c_{j}\right)_{\mathrm{aq}}}\right] \ &\quad-\ln \left[1-\frac{\delta c_{j}}{\left(c_{j}\right)_{\mathrm{aq}}}\right] \equiv \frac{\delta c_{j}}{\left(c_{j}\right)_{\mathrm{aq}}} \end{aligned} [3] F. A. Long and W. F. McDevit, Chem. Rev.,1952,51,119. [4] For a review of the definitions of units used in this subject area see H. L. Clever, J. Chem. Eng. Data, 1983,28,340. [5] B. E. Conway, Pure Appl. Chem.,1985,57,263. [6] P. Debye and J. MacAulay, Z. Phys. Chem.,1925,26,22. [7] B. E. Conway, J. E. Desnoyers and A. C. Smith, Philos. Trans. R. Soc.,A 1964, 256A, 389. [8] 1. $\mathrm{N}_{2}$ and $\mathrm{CH}_{4}$ in NaCl(aq); T. D. O’Sullivan and N. O. Smith, J. Phys. Chem., 1970, 70 , 1460. 2. Phenolic salts in salt solutions; B. Das and R. Ghosh, J. Chem. Eng. Data, 1984, 29,137. 3. Oxygen in salt solutions; W. Lang and R. Zander, Ind. Eng. Chem. Fundam., 1986,25,775. 4. Hg in NaCl(aq); D. N. Glew and D. A. Hames, Can.J.Chem.,1972,50,3124. 5. Aromatic hydrocarbons in salt solutions; I. Sanemaa, S. Arakawa,M. Araki and T. Deguchi, Bull. Chem. Soc. Jpn.,1984,57,1539. 6. n-Butane in NaCl(aq); P. A. Rice, R. P. Gale and A. J. Barduhn, J. Chem. Eng. Data, 1976,21,204. 7. Ar in MX(aq); L. Clever and C. J. Holland, J. Chem. Eng. Data, 1968,13,411. 8. Benzene in NaCl(aq); D.F. Keeley, M.A. Hoffpaulr and J. Meriwether, J. Chem. Eng. Data, 1988,33,87. 9. B. E. Conway, D. M. Novak and L. H. Laliberte, J. Solution Chem.,1974,3,683; Ar in $\mathrm{R}_{4}\mathrm{NX}(\mathrm{aq})$. 10. R. Aveyard and R. Heselden, J. Chem. Soc. Faraday Trans. 1, 1974,70,1953. 11. Ar in benzene in $\mathrm{R}_{4}\mathrm{NX}(\mathrm{aq})$; A. Ben-Naim, J. Phys. Chem.,1967,71,1137. 12. A. Ben-Naim and M. Egel-Thal, J. Phys. Chem.,1965,69,3250; Ar in MX(aq). 13. $\mathrm{Et}_{3}\mathrm{N}$ in $\mathrm{R}_{4}\mathrm{NCl}(\mathrm{aq})$; A. F. S. S. Mendonca, D. T. R. Formingo and I. M. S. Lampreia, J Solution Chem.,2002,31,653. 14. $\mathrm{Et}_{3}\mathrm{N}$ in $\mathrm{CaCl}_{2}(\mathrm{aq})$; A. F. S. S. Mendonca, D. T. R. Formingo and I. M. S. Lampreia, J Solution Chem.,2003,32,1033. [9] Benzene in $\mathrm{R}_{4}\mathrm{N}^{+} \mathrm{~Br}^{-} (\mathrm{aq})$;J. E. Desnoyers, G. E. Pelletier and C. Jolicoeur, Can. J. Chem.,1965,43,3232. [10] Benzene in R4NBr(aq); H. E. Wirth and A. LoSurdo, J. Phys. Chem.,1968,72,751. [11] RH in $\mathrm{R}_{4}\mathrm{NBr}(\mathrm{aq})$;W.-Y. Wen and J. H. Hung, J.Phys.Chem.,1970,74,170. [12] A. Feillolay and M. Lucas, J. Phys. Chem.,1972,76,3068. [13] C. Treiner and A. K. Chattopadhyay, J. Chem. Soc Faraday Trans. 1, 1983,79,2915. 1.14.60: Second Law of Thermodynamics The Second Law introduces an extensive function of state, a property of a given system, called the entropy, symbol $\mathrm{S}$. Spontaneous chemical reaction in a closed system is driven by the affinity for spontaneous change $\mathrm{A}$ producing a change in chemical composition $\xi$. The change in entropy $\mathrm{dS}$ at temperature $\mathrm{T}$ is given by Equation \ref{a}. $\mathrm{T} \, \mathrm{dS}=\mathrm{q}+\mathrm{A} \, \mathrm{d} \xi \label{a}$ where $\mathrm{A} \, \mathrm{d} \xi>0 \label{b}$ The latter inequality is the LAW. This inequality is the key to chemistry. In effect the law states that if there is an affinity for a given chemical reaction ( i.e. a driving ‘force’ for reaction) the chemical reaction will spontaneously proceed in that direction. This is the thermodynamic selection rule for which there are no exceptions. In the limit that a system undergoes a ‘reversible ‘ change, $\mathrm{A}$ is zero; the system is at equilibrium with the surroundings. For a reversible change $\mathrm{T} \, \mathrm{d} \mathrm{S}=\mathrm{q} \label{c}$ Often texts seek to answer the question ‘what is entropy?’ This is a fruitless task unless one draws attention to Equation \ref{c} which reminds us that the product $\mathrm{T} \, \mathrm{dS}$ is in fact a thermal energy. Chemists are familiar with spontaneous chemical reactions and Equations \ref{a} and \ref{b} present no conceptual problems [1]. Footnotes [1] Robert Park, Voodoo Science, Oxford,2000. From page 7; ‘The first law says you can’t win; the second law says you can’t even break even’. This comment is with respect to fraudulent claims of discoveries of perpetual motion machines. 1.14.61: Solubility Products A given closed system at temperature $\mathrm{T}$ and pressure $\mathrm{p}$ (which is close to ambient) contains an aqueous solution of a sparingly soluble salt $\mathrm{MX}$; e.g. $\mathrm{AgCl}$. The system also contains solid salt $\mathrm{MX}$. When a soluble salt (e.g. $\mathrm{KNO}_{3}$) is added the solubility of salt $\mathrm{MX}$ increases. This remarkable observation is readily accounted for. The equilibrium involving the sparingly soluble salt is represented as follows. $\mathrm{MX}(\mathrm{s})$ $\Leftrightarrow$ $\mathrm{M}^{+}\mathrm{~X}^{-}(\mathrm{aq})$ solid   solution We represent the salt $\mathrm{MX}$ by the symbol $j$. At equilibrium, $\mu_{\mathrm{j}}^{\prime \prime}(\mathrm{s})=\mu_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})$ In terms of the solubility $\mathrm{S}_{j}$ of the salt $\mathrm{MX}$, a 1:1 salt, $\mu_{\mathrm{j}}^{*}(\mathrm{~s})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+2 \, \mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{S}_{\mathrm{j}} \, \gamma_{\pm} / \mathrm{m}^{\mathrm{o}}\right)$ By definition $\Delta_{\text {sol }} G^{0}=-R \, T \, \ln K_{s}=\mu_{j}^{0}(a q)-\mu_{j}^{*}(s)$ $\mathrm{K}_{\mathrm{S}}$ is the solubility product, a characteristic property of salt $\mathrm{MX}$ (at defined $\mathrm{T}$ and $\mathrm{p}$). $\mathrm{K}_{\mathrm{S}}=\left[\mathrm{S}_{\mathrm{j}} \, \gamma_{\pm} / \mathrm{m}^{0}\right]^{2}$ Or, $\ln \left(\mathrm{S}_{\mathrm{j}} / \mathrm{m}^{0}\right)=(1 / 2) \, \ln \left(\mathrm{K}_{\mathrm{s}}\right)-\ln \left(\gamma_{\pm}\right)$ According to the DHLL, $\ln \left(\gamma_{\pm}\right)=-S_{\gamma} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2}$ $\mathrm{I}$ is the ionic strength of the solution which can be changed by adding a soluble salt. From equations (e) and (f), $\ln \left(\mathrm{S}_{\mathrm{j}} / \mathrm{m}^{0}\right)=(1 / 2) \, \ln \left(\mathrm{K}_{\mathrm{s}}\right)+\mathrm{S}_{\gamma} \,\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2}$ The key point to note is the positive sign in equation (g) showing that the theory accounts for the observed salting–in of the sparingly soluble salt. Further a plot of $\ln \left(S_{j} / m^{0}\right)$ against $\left(\mathrm{I} / \mathrm{m}^{0}\right)^{1 / 2}$ is linear yielding an estimate for $\mathrm{K}_{\mathrm{S}}$ from the intercept.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.59%3A_Salting-In_and_Salting-Out.txt
Comparison of the solubilities of volatile chemical substance $j$ in liquids $\ell_{1}$ and $\ell_{2}$ yields an estimate of the difference in reference chemical potentials, $\Delta\left(\ell_{1} \rightarrow \ell_{2}\right) \mu_{\mathrm{j}}^{0}(\mathrm{~T})$.This is a classic subject [1-10] with two consequences. 1. A vast amount of information has been published, not all, unfortunately, of high quality. 2. Many terms and definitions have been developed. Determination of thermodynamic parameters characterizing gaseous solubilities is not straightforward. Account has to be taken of the fact that the properties of real gases are not perfect. A closed system contains two phases, liquid and gaseous, at temperature $\mathrm{T}$. The liquid is water; a sparingly soluble chemical substance $j$ exists in both gas and liquid phases. A phase equilibrium is established for substance $j$ in the two phases. In terms of the Phase Rule, there are two phases and two components. Hence there are two degrees of freedom. If the temperature and pressure are defined, the compositions of the two phases are fixed. In terms of chemical potentials with reference to substance $j$ the following condition holds. $\mu_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{aq} ; \mathrm{x}_{\mathrm{j}}, \mathrm{p}, \mathrm{T}\right)=\mu_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{g} ; \mathrm{p}_{\mathrm{j}}, \mathrm{T}\right)$ Here $\mathrm{p}_{j}$ is the equilibrium partial pressure of substance $j$ in the gas phase at pressure $\mathrm{p}$ where pressure $\mathrm{p}$ equals $\left(\mathrm{p}_{j} + \mathrm{p}_{1}\right)^{\mathrm{eq}}$ where $\mathrm{p}_{1}$ is the equilibrium partial pressure of water in the vapor phase. Equation (a) establishes the thermodynamic basis of the phenomenon discussed here. However historical and practical developments resulted in quite different approaches to the description of the solubilities of gases in liquids. The thermodynamic treatment is not straightforward if we recognize that the thermodynamic properties of the vapor (i.e. gas phase) and the solution are not ideal. When both the solubility and the partial pressure of the ‘solute’ in the gas phase are low, the assumption is often made that the thermodynamic properties of gas and solution are ideal. Then the analysis of solubility is reasonably straightforward [1-3]. In a sophisticated analysis, account must be taken of the intermolecular interactions in the vapor phase and solute-solute interactions in solution [4-10]. We review the basis of analyses where the thermodynamic properties of gas and liquid phases are ideal. Nevertheless equation (a) is the common starting point for the analysis. Bunsen Coefficient, $\alpha$ By definition, the Bunsen Coefficient $\alpha$ is the volume of a gas at $273.15 \mathrm{~K}$ and standard pressure $\mathrm{p}^{0}$ which dissolves in unit volume of a solvent when the partial pressure of the gas equals $\mathrm{p}^{0}$. Experiment yields the volume $\mathrm{V}_{j}(\mathrm{g})$ of gas $j$ at temperature $\mathrm{T}$ and partial pressure $\mathrm{p}_{j}$ absorbed by volume $\mathrm{V}_{\mathrm{s}}$ of solvent at temperature $\mathrm{T}$. The volume of gas at $273.15 \mathrm{~K}$ and standard pressure $\mathrm{p}^{0}$, $\mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{p}^{0} ; 273.15 \mathrm{~K}\right)$ is given by equation (b) assuming that gas $j$ has the properties of a perfect gas. $\mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{p}^{0} ; 273.15 \mathrm{~K}\right)=\left[\mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{T} ; \mathrm{p}_{\mathrm{j}}\right) \, 273.15 \mathrm{~K} / \mathrm{p}^{0} \, \mathrm{T}\right]$ Hence experiment yields the ratio, $\left[\mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{T} ; \mathrm{p}_{\mathrm{j}}\right) / \mathrm{V}_{\mathrm{s}}(\ell ; \mathrm{T} ; \mathrm{p})\right]$. By simple proportion we obtain the volume, $\mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{p}^{0} ; 273.15 \mathrm{~K}\right)$ in the event that the gas $j$ was at pressure $\mathrm{p}^{0}$ above the liquid phase. Bunsen coefficient, $\alpha=\frac{\mathrm{V}_{\mathrm{j}}\left(\mathrm{g} ; \mathrm{T} ; \mathrm{p}_{\mathrm{j}}\right)}{\mathrm{V}_{\mathrm{s}}(\ell ; \mathrm{T} ; \mathrm{p})} \,\left(\frac{\mathrm{p}_{\mathrm{j}}}{\mathrm{p}^{0}}\right) \, \frac{273.15 \mathrm{~K}}{\mathrm{~T}}$ At ambient pressure, if the partial pressure of the solvent is negligibly small, $\alpha=\frac{\mathrm{V}_{\mathrm{j}}(\mathrm{g} ; \mathrm{T} ; \mathrm{p})}{\mathrm{V}_{\mathrm{s}}(\ell ; \mathrm{T} ; \mathrm{p})} \,\left(\frac{273.15 \mathrm{~K}}{\mathrm{~T}}\right)$ The assumption that substance $j$ is a perfect gas can be debated but the correction is often less that 1%. Oswald Coefficient A given closed system comprises gaseous and liquid phases, at temperature $\mathrm{T}$. The system is at equilibrium such that equation (a) holds. We assume that the thermodynamic properties of the solution and the gas phase are ideal and that an equilibrium exists for chemical substance $j$ between the two phases. The solution is at pressure $\mathrm{p}^{0}$, standard pressure which is close to ambient pressure. \begin{aligned} \mu_{\mathrm{j}}^{0}\left(\mathrm{~g} ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, & \ln \left(\mathrm{p}_{\mathrm{e}}^{\mathrm{eq}} / \mathrm{p}^{0}\right) \ &=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{c}_{\mathrm{j}} / \mathrm{c}_{\mathrm{r}}\right) \end{aligned} For a perfect gas, $\mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{g})=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ For a solution, $\mathrm{c}_{\mathrm{j}}=\mathrm{n}_{\mathrm{j}} / \mathrm{V}(\mathrm{aq})$ In these terms $\mathrm{V}(\mathrm{aq})$ is the volume of solution which dissolves $\mathrm{n}_{j}$ moles of chemical substance $j$ from the gas phase. \begin{aligned} \mu_{\mathrm{j}}^{0}\left(\mathrm{~g} ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{j}}(\mathrm{g})} \, \frac{1}{\mathrm{p}^{0}}\right) \ &=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)+\mathrm{R} \, \mathrm{T} \, \ln \left(\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{V}(\mathrm{aq})} \, \frac{1}{\mathrm{c}_{\mathrm{r}}}\right) \end{aligned} The Ostwald Coefficient $\mathrm{L}$ is defined in terms of reference chemical potentials of substance $j$ in solution and gas phase. $\Delta_{s \ln } \mathrm{G}^{0}\left(\mathrm{~T}, \mathrm{p}^{0}\right)=-\mathrm{R} \, \mathrm{T} \, \ln (\mathrm{L})=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)-\mu_{\mathrm{j}}^{0}\left(\mathrm{~g} ; \mathrm{id} ; \mathrm{T} ; \mathrm{p}^{0}\right)$ From equation (h), $\Delta_{\mathrm{s} \ln } \mathrm{G}^{0}\left(\mathrm{~T}, \mathrm{p}^{0}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{j}}(\mathrm{g})} \, \frac{\mathrm{V}(\mathrm{aq}) \, \mathrm{c}_{\mathrm{r}}}{\mathrm{n}_{\mathrm{j}} \, \mathrm{p}^{0}}\right]$ $\Delta_{s \ln } \mathrm{G}^{0}\left(\mathrm{~T}, \mathrm{p}^{0}\right)=\mathrm{R} \, \mathrm{T} \, \ln \left[\frac{\mathrm{V}(\mathrm{aq})}{\mathrm{V}_{\mathrm{j}}(\mathrm{g})} \, \frac{\mathrm{R} \, \mathrm{T} \, \mathrm{c}_{\mathrm{r}}}{\mathrm{p}^{0}}\right]$ Because $\mathrm{V}(\mathrm{aq})$ and $\mathrm{V}_{j}(\mathrm{g})$ are expressed in the same units, the Oswald coefficient is dimensionless. The key assumption is that the thermodynamic properties of gas and solution phases are ideal. Ostwald coefficients can be defined in several ways [4]. In the analysis set out above we refer to the volume of the solution containing solvent and solute $j$. Another definition refers to the volume of pure liquid which dissolves a volume of gas $\mathrm{V}_{j}$. $\mathrm{L}^{0}=\left[\mathrm{V}_{\mathrm{g}} / \mathrm{V}^{*}(\ell)\right]^{\mathrm{eq}}$ A third definition refers to ratio of concentrations of substances $j$ in liquid and gas phases. $\mathrm{L}_{\mathrm{c}}=\left[\mathrm{c}_{\mathrm{j}}^{\mathrm{L}} / \mathrm{c}_{\mathrm{j}}^{\mathrm{V}}\right]^{\mathrm{eq}}$ In effect an Oswald coefficient describes an equilibrium distribution for a volatile solute between gas phase and solution. The Oswald coefficient is related to the (equilibrium) mole fraction of dissolved gas, $x_{j}$ using equation (n) where $\mathrm{p}_{j}$ is the partial pressure of chemical substance $j$ and $\mathrm{V}_{1}^{*}(\ell)$ is the molar volume of the solvent [11]. $\mathrm{x}_{2}=\left\{\left[\mathrm{R} \, \mathrm{T} / \mathrm{L} \, \mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{1}^{*}(\ell)\right]+1\right\}^{-1}$ Henry’s Law Constant In a given closed system at temperature $\mathrm{T}$, gas and solution phases are in equilibrium. The thermodynamic properties of both phases are ideal. Then according to Henry’s Law, the partial pressure of volatile solute $j$ is a linear function of the concentration $\mathrm{c}_{j}$ at fixed temperature. $\mathrm{p}_{\mathrm{j}}(\operatorname{vap})=\mathrm{K}_{\mathrm{c}} \, \mathrm{c}_{\mathrm{j}} / \mathrm{c}_{\mathrm{r}}$ $\mathrm{K}_{\mathrm{c}}$ is the Henry’s Law constant (on the concentration scale), characteristic of solvent, solute and temperature. Similarly on the mole fraction scale, $\mathrm{p}_{\mathrm{j}}(\operatorname{vap})=\mathrm{K}_{\mathrm{x}} \, \mathrm{x}_{\mathrm{j}}$ This subject, gas solubilities, is enormously important. We draw attention to some interesting reports concerning solubilities with particular reference to aqueous solutions and the environment [12]. Footnote [1] R. Battino and H. L. Clever, Chem.Rev.,1966,66,395. [2] E. Wilhelm and R. Battino, Chem.Rev.,1973,73,1. [3] E. Wilhelm, R. Battino and R. J. Wilcock, Chem.Rev.,1977,77, 219. [4] R. Battino, Fluid Phase Equilib.,1984,15,231. [5] E. Wilhelm, Pure Appl.Chem.,1985,57,303. [6] E. Wilhelm, Fluid Phase Equilib.,1986,27,233 [7] E. Wilhelm, Thermochim. Acta,1990,162,43. [8] R.Fernandez-Prini and R. Crovetto, J. Phys. Chem. Ref.Data,1989,18,1231. [9] R. Battino, T. R. Rettich and T. Tominaga, J. Phys. Chem. Ref. Data, 1984,13,563. [10] T. R. Rettich, Y. P. Handa, R. Battino and E. Wilhelm, J. Phys. Chem.,1981,85,3230. [11] \begin{aligned} &\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}}{\mathrm{V}_{\mathrm{j}}(\mathrm{g})} \, \frac{\mathrm{V}(\mathrm{aq}) \, \mathrm{c}_{\mathrm{r}}}{\mathrm{n}_{\mathrm{j}} \, \mathrm{p}^{0}}=\frac{[\mathrm{mol}] \,\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]}{\left[\mathrm{m}^{3}\right]} \, \frac{\left[\mathrm{m}^{3}\right] \,\left[\mathrm{mol} \mathrm{m}^{-3}\right]}{[\mathrm{mol}] \,\left[\mathrm{N} \mathrm{m}^{-2}\right]} \ &=\frac{[\mathrm{N} \mathrm{m}]}{\left[\mathrm{m}^{3}\right]} \, \frac{1}{\left[\mathrm{~N} \mathrm{\textrm {m } ^ { - 2 } ]}\right.}=[1] \end{aligned} For a perfect gas $j$, $\mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{\mathrm{j}}(\mathrm{g})=\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}$ $V_{j}(g)=n_{j} \, R \, T / p_{j}$ For $\mathrm{n}_{1}$ moles of liquid 1, density $\rho_{1}^{*}(\ell)$, $\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell)=\mathrm{n}_{1} \, \mathrm{M}_{1} / \rho_{1}^{*}(\ell)$ By definition, at temperature $\mathrm{T}$, $\mathrm{L}=\mathrm{V}_{\mathrm{j}}(\mathrm{g}) / \mathrm{V}_{1}^{*}(\ell)$ $\mathrm{L}=\frac{\mathrm{n}_{\mathrm{j}} \, \mathrm{R} \, \mathrm{T}}{\mathrm{p}_{\mathrm{j}}} \, \frac{\rho_{\mathrm{l}}^{*}(\ell)}{\mathrm{n}_{1} \, \mathrm{M}_{\mathrm{l}}}$ Hence, $\frac{\mathrm{n}_{1}}{\mathrm{n}_{\mathrm{j}}}=\frac{\mathrm{R} \, \mathrm{T}}{\mathrm{L} \, \mathrm{p}_{\mathrm{j}}} \, \frac{\rho_{1}^{*}(\ell)}{\mathrm{M}_{1}}=\frac{\mathrm{R} \, \mathrm{T}}{\mathrm{L} \, \mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{1}^{*}(\ell)}$ But mole fraction of solute $j$ in solution, $\mathrm{x}_{\mathrm{j}}=\frac{\mathrm{n}_{\mathrm{j}}}{\mathrm{n}_{1}+\mathrm{n}_{\mathrm{j}}}=\frac{1}{\left(\mathrm{n}_{\mathrm{l}} / \mathrm{n}_{\mathrm{j}}\right)+1}$ From equations (f) and (g), $\mathrm{x}_{\mathrm{j}}=\left[\left\{\mathrm{R} \, \mathrm{T} / \mathrm{L} \, \mathrm{p}_{\mathrm{j}} \, \mathrm{V}_{1}^{*}(\ell)\right\}+1\right]^{-1}$ [12] 1. $\mathrm{O}_{2}(\mathrm{aq})$; E. Douglas, J. Phys.Chem.,1964,68,169. 2. $\mathrm{H}_{\mathrm{g}}(\mathrm{aq})$: E. Onat, J. Inorg. Nucl. Chem..,1974,36,2029. 3. $\mathrm{H}_{2}\mathrm{S}(\mathrm{aq})$; E .C .Clarke and David N. Glew, Can. J. Chem., 1961,49,691. 4. $\mathrm{N}_{2}(\mathrm{aq})$; T. R. Rettich, R. Battino and E. Wilhelm, J. Solution Chem.,1984,13,335. 5. Benzene(aq): D. S. Arnold, C. A. Plank, E. E. Erikson and F. P. Pike, Ind. Eng. Chem.,1958,3,253. 6. Polychlorinated Biphenyls(aq): W. Y. Shiu and D. Mackay, J. Phys. Chem. Ref. Data, 1986,15,911. 7. Cummenes; D. N. Glew and R. E. Robertson, J. Phys. Chem.,1956,60,332. 8. Fluorocarbons(aq); W.-Y. Wen and J. A. Muccitelli, J. Solution Chem.,1979,8,225. 9. Hydrocarbons; C. McAuliffe, J. Phys. Chem.,1966,70,1267. 10. $\mathrm{O}_{2}(\mathrm{g})$ in water + alcohol mixtures; R.W.Cargill, J. Chem. Soc., Faraday Trans.,1996,72,2296. 11. Hydrocarbons in alcohol + water mixtures; R. W. Cargill and D. E Macphee, J. Chem. Res.(S),1986,2301. 12. $\mathrm{N}_{2}(\mathrm{aq}); \mathrm{~H}_{2}(\mathrm{aq}); 298 - 640 \mathrm{~K}$; J. Alvarez, R. Crovetto and R. Fernandez-Prini, Ber. Bunsenges. Phys. Chem.,1988,92,935. 13. A(aq); R. W. Cargill and T.J. Morrison, J. Chem. Soc. Faraday Trans.1,1975, 71,620. 14. Gases in ethylene glycol; R. Fernandez-Prini, R. Crovetto and N. Gentili, J. Chem. Thermodyn.,1987,19,1293. 15. A(aq; dixoan); A. Ben-Naim and G. Moran, Trans. Faraday Soc., 1965, 61,821. 16. T. Park, T. R. Rettich, R. Battino, D. Peterson and E. Wilhelm, J. Chem. Eng. Data, 1982,27,324. 17. RH(aq); W.E. May, S.P. Wasik, M. E. Miller, Y.B. Tewari, J.M. Brown-Thomas and R. N. Goldberg, J. Chem. Eng. Data, 1983,28,197. 18. $\mathrm{C}_{2}\mathrm{H}_{4}(\mathrm{aq} + \mathrm{~amine})$; E. Sada, H. Kumazawa and M. A. Butt, J. Chem. Eng. Data, 1977, 22,277. 19. A(aq); A. Ben-Naim, J. Phys. Chem.,1965,69,3245;1968,72,2998. 20. Viny chloride(aq); W. Hayduk and H. Laudie, J. Chem. Eng. Data, 1974,19,253. 21. Halogenated hydrocarbons(aq); A.L.Horvath, J. Chem. Documentation, 1972,12,163. 22. $\mathrm{N}_{2}(\mathrm{aq}), \mathrm{~A}(\mathrm{aq}) \text{ and } \mathrm{Xe}(\mathrm{aq})$; R.P. Pennan and G.L.Pollock, J. Chem. Phys.,1990,93,2724. 23. $\mathrm{CO}_{2}(\mathrm{aq} + \mathrm{~ROH})$;R. W. Cargill, J.Chem. Res(S),1982,230.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.62%3A_Solubilities_of_Gases_in_Liquids.txt
This very large subject can be divided into two groups. The first group concerns the solubility of a given solid substance $j$ in a given solvent, liquid $\ell_{1}$. The second group involves comparison of the solubilities of a given solid in two liquids, $\ell_{1}$ and $\ell_{2}$. A closed system (at defined $\mathrm{T}$ and $\mathrm{p}$, the latter being close to the standard pressure) contains solid substance $j$ in equilibrium with an aqueous solution containing solute $j$. The system is characterized by the (equilibrium) solubility, $\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})$. At equilibrium, $\mu_{\mathrm{j}}^{*}(\mathrm{~s})=\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) \, \gamma_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) / \mathrm{m}^{0}\right]$ Then $\Delta \mu_{\mathrm{j}}^{0}=\mu_{\mathrm{j}}^{0}(\mathrm{aq})-\mu_{\mathrm{j}}^{*}(\mathrm{~s})=-\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) \, \gamma_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) / \mathrm{m}^{0}\right]$ If the aqueous solution is dilute and the solubility is low, it can often be assumed that the properties of the solution are ideal. Hence, $\Delta \mu_{\mathrm{j}}^{0}=\mu_{\mathrm{j}}^{0}(\mathrm{aq})-\mu_{\mathrm{j}}^{\mathrm{*}}(\mathrm{s})=-\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) / \mathrm{m}^{0}\right]$ It should be noted that the sign of $\Delta \mu_{j}^{0}$ depends on whether or not $m_{j}^{e q}(a q)$ is larger or less than unity. We illustrate the second approach by considering a combination of the experiment described above and an experiment where the solvent is a binary aqueous mixture, mole fraction composition $\mathrm{x}_{2}$. At equilibrium, $\mu_{j}^{*}(s)=\mu_{j}^{0}\left(s \ln ; x_{2}\right)+R \, T \, \ln \left[m_{j}^{\mathrm{eq}}\left(s \ln ; x_{2}\right) \, \gamma_{j}^{\mathrm{eq}}\left(s \ln ; x_{2}\right) / m^{0}\right]$ \begin{aligned} \Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}=\mu_{\mathrm{j}}^{0}\left(\mathrm{~s} \ln ; \mathrm{x}_{2}\right)-\mu_{\mathrm{j}}^{0}(\mathrm{aq}) \ =-\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) \, \gamma_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq}) \, \gamma_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})\right] \end{aligned} If both solutions are dilute in substance $j$, the ratio, $\gamma_{j}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \gamma_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})$ can be assumed to be close to unity. In fact this is a better approximation than assuming both activity coefficients are unity. Then $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{0}=-\mathrm{R} \, \mathrm{T} \, \ln \left[\mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}\left(\mathrm{s} \ln ; \mathrm{x}_{2}\right) / \mathrm{m}_{\mathrm{j}}^{\mathrm{eq}}(\mathrm{aq})\right]$ In other words if the solubility of substance $j$ increases on adding solvent component 2 then $\Delta\left(\mathrm{aq} \rightarrow \mathrm{x}_{2}\right) \mu_{\mathrm{j}}^{\mathrm{c}}$ is negative. This stabilization is a consequence of a difference in solute-solvent interactions. 1.14.64: Solutions: Solute and Solvent A given solution (at temperature $\mathrm{T}$ and pressure $\mathrm{p}$, where the latter is close to the standard pressure) is prepared using $1 \mathrm{~kg}$ of water($\ell$) and $\mathrm{m}_{j}$ moles of a simple solute. The Gibbs energy $\mathrm{G}\left(\mathrm{w}_{1}=1 \mathrm{~kg}\right)$ is given by equation (a). \begin{aligned} \mathrm{G}\left(\mathrm{w}_{1}=1 \mathrm{~kg}\right)=\left(1 / \mathrm{M}_{1}\right) \,\left[\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right] \ &+\mathrm{m}_{\mathrm{j}} \,\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right)\right] \end{aligned} In the event that the thermodynamic properties of the solution are ideal, \begin{aligned} \mathrm{G}\left(\mathrm{w}_{1}=1 \mathrm{~kg} ; \mathrm{id}\right)=&\left(1 / \mathrm{M}_{1}\right) \,\left[\mu_{1}^{*}(\ell)-\mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right] \ &+\mathrm{m}_{\mathrm{j}} \,\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)\right] \end{aligned} The excess Gibbs energy $\mathrm{G}^{\mathrm{E}}$ for the solution prepared using $1 \mathrm{~kg}$ of water($\ell$) is given by equation (c). $\mathrm{G}^{\mathrm{E}}=\mathrm{G}\left(\mathrm{w}_{1}=1 \mathrm{~kg}\right)-\mathrm{G}\left(\mathrm{w}_{1}=1 \mathrm{~kg} ; \mathrm{id}\right)$ Therefore [1] $\mathrm{G}^{\mathrm{E}}=\mathrm{R} \, \mathrm{T} \, \mathrm{m}_{\mathrm{j}} \,\left[1-\phi+\ln \left(\gamma_{\mathrm{j}}\right)\right]$ Hence at fixed $\mathrm{T}$ and $\mathrm{p}$, the dependence of $\mathrm{G}^{\mathrm{E}}$ on $\mathrm{m}_{j}$ is given by equation (e). $(1 / \mathrm{R} \, \mathrm{T}) \, \mathrm{dG}^{\mathrm{E}} / \mathrm{dm}_{\mathrm{j}}=1-\phi+\ln \left(\gamma_{\mathrm{j}}\right)-\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi / \mathrm{dm}_{\mathrm{j}}+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm} \mathrm{j}_{\mathrm{j}}$ According to the Gibbs-Duhem equation, the chemicals potentials of solvent $\mu_{1}(\mathrm{aq})$ and solute $\mu_{j}(\mathrm{aq})$ are linked. At fixed $\mathrm{T}$ and $\mathrm{p}$, $\mathrm{n}_{1} \, \mathrm{d} \mu_{1}(\mathrm{aq})+\mathrm{n}_{\mathrm{j}} \, \mathrm{d} \mu_{\mathrm{j}}(\mathrm{aq})=0$ Then for a solution prepared using $1 \mathrm{~kg}$ of water($\ell$), $\left(1 / M_{1}\right) \, d \mu_{1}(a q)+m_{j} \, d \mu_{j}(a q)=0$ In terms of the impact of adding $\mathrm{dm}_{j}$ moles of solute, $\left(1 / M_{1}\right) \, d \mu_{1}(a q) / d m_{j}+m_{j} \, d \mu_{j}(a q) / d m_{j}=0$ The Gibbs-Duhem relation describes moderation of the effects of added $\mathrm{dm}_{j}$ moles of solute $j$ on the changes in chemical potentials of solute and solute. We use the equation which relates these chemical potentials to the composition of the solution. For simple solutes (e.g. urea) at ambient pressure, equation (g) takes the following form. \begin{aligned} &{\left[1 / \mathrm{M}_{1}\right] \, \mathrm{d}\left[\mu_{1}^{*}(\ell)-\phi \, \mathrm{R} \, \mathrm{T} \, \mathrm{M}_{1} \, \mathrm{m}_{\mathrm{j}}\right]} \ &\quad+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\mu_{\mathrm{j}}^{0}(\mathrm{aq})+\mathrm{R} \, \mathrm{T} \, \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right]=0\right. \end{aligned Hence, $\mathrm{d}\left[-\phi \, \mathrm{m}_{\mathrm{j}}\right]+\mathrm{m}_{\mathrm{j}} \,\left[\mathrm{d} \ln \left(\mathrm{m}_{\mathrm{j}} \, \gamma_{\mathrm{j}} / \mathrm{m}^{0}\right]=0\right.$ The simple differential equation (j) is important in developing links between the thermodynamic properties of solutions, solvent and solute. The integrated form of this equation is important. From equation (j), $\mathrm{d}\left[-\phi \, \mathrm{m}_{\mathrm{j}}\right]+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\mathrm{m}_{\mathrm{j}}\right)+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right)=0$ Therefore, $-\phi \, d m_{j}-m_{j} \, d \phi+d m_{j}+m_{j} \, d \ln \left(\gamma_{j}\right)=0$ Or, $\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right)=(\phi-1) \, \mathrm{dm} \mathrm{m}_{\mathrm{j}}+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi$ Or, with a slight re-arrangement, $d \ln \left(\gamma_{\mathrm{j}}\right)=\mathrm{d} \phi+(\phi-1) \, \mathrm{d} \ln \left(\mathrm{m}_{\mathrm{j}}\right)$ Hence we obtain an equation for $\ln \left(\gamma_{j}\right)$ in terms of the dependence of $(\phi - 1)$ on molality of solute bearing in mind that $\ln \left(\gamma_{j}\right)$ equals zero and $\phi$ equals 1 at $\mathrm{m}_{j} = 0$. $\ln \left(\gamma_{\mathrm{j}}\right)=(\phi-1)+\int_{\mathrm{o}}^{\mathrm{m}_{\mathrm{j}}}(\phi-1) \, \mathrm{d} \ln \left(\mathrm{m}_{\mathrm{j}}\right)$ In another approach we start again with equation (j). $\mathrm{d}\left[\phi \, \mathrm{m}_{\mathrm{j}}\right]=\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)\right]+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\gamma_{\mathrm{j}}\right)\right]$ Or, $\mathrm{d}\left[\phi \, \mathrm{m}_{\mathrm{j}}\right]=\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\mathrm{m}_{\mathrm{j}}\right)-\ln \left(\mathrm{m}^{0}\right)\right]+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\gamma_{\mathrm{j}}\right)\right]$ Or, $\mathrm{d}\left[\phi \, \mathrm{m}_{\mathrm{j}}\right]=\mathrm{dm}_{\mathrm{j}} \,+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\gamma_{\mathrm{j}}\right)\right]$ Following integration from ‘$\mathrm{m}_{j} =0$’ to $\mathrm{m}_{j}$, $\phi \, \mathrm{m}_{\mathrm{j}}=\mathrm{m}_{\mathrm{j}}+\int_{0}^{\mathrm{m}_{\mathrm{j}}} \mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right)$ $\phi=1+\left(1 / \mathrm{m}_{\mathrm{j}}\right) \, \int_{0}^{\mathrm{m}_{\mathrm{j}}} \mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right)$ $\phi-1=\left(1 / m_{j}\right) \, \int_{0}^{m_{j}} m_{j} \, d \ln \left(\gamma_{j}\right)$ In other words $(\phi - 1)$ is related to the integral of $m_{j} \, d \ln \left(\gamma_{j}\right)$ between the limits ‘$\mathrm{m}_{j} = 0$’ and $\mathrm{m}_{j}$. Equation (e) can be re-expressed as an equation of $\ln \left(\gamma_{\mathrm{j}}\right)$. $\ln \left(\gamma_{\mathrm{j}}\right)=-(1-\phi)+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi / \mathrm{dm}_{\mathrm{j}}-\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm}_{\mathrm{j}}+(1 / \mathrm{R} \, \mathrm{T}) \, \mathrm{dG}{ }^{\mathrm{E}} / \mathrm{dm} \mathrm{j}_{\mathrm{j}}$ Hence from equation (r), $\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi+\phi \, \mathrm{dm} \mathrm{j}_{\mathrm{j}}=\mathrm{dm} \mathrm{m}_{\mathrm{j}}+\mathrm{m}_{\mathrm{j}} \, \mathrm{d}\left[\ln \left(\gamma_{\mathrm{j}}\right)\right]$ Or, $\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi / \mathrm{dm} \mathrm{m}_{\mathrm{j}}+\phi-1-\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm} \mathrm{m}_{\mathrm{j}}=0$ Or, $-(1-\phi)+\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \phi / \mathrm{dm}_{\mathrm{j}}-\mathrm{m}_{\mathrm{j}} \, \mathrm{d} \ln \left(\gamma_{\mathrm{j}}\right) / \mathrm{dm} \mathrm{m}_{\mathrm{j}}=0$ Then with reference to equation (v), [2] $\ln \left(\gamma_{\mathrm{j}}\right)=(1 / \mathrm{R} \, \mathrm{T}) \, \mathrm{dG}^{\mathrm{E}} / \mathrm{dm}_{\mathrm{j}}$ Combination of equations (z) and (d) yields an equation for $(1 - \phi)$ in terms of $\mathrm{G}^{\mathrm{E}}$. Thus $\mathrm{G}^{\mathrm{E}} / \mathrm{R} \, \mathrm{T}=\mathrm{m}_{\mathrm{j}} \,(1-\phi)+\left(\mathrm{m}_{\mathrm{j}} / \mathrm{R} \, \mathrm{T}\right) \, \mathrm{dG}^{\mathrm{E}} / \mathrm{dm}_{\mathrm{j}}$ Or [3], $(1-\phi)=(1 / \mathrm{R} \, \mathrm{T}) \,\left[\mathrm{G}^{\mathrm{E}} / \mathrm{m}_{\mathrm{j}}-\mathrm{dG}^{\mathrm{E}} / \mathrm{dm}_{\mathrm{j}}\right]$ Or, $(1-\phi)=-\left(\mathrm{m}_{\mathrm{j}} / \mathrm{R} \, \mathrm{T}\right) \,\left\{\mathrm{d}\left[\mathrm{G}^{\mathrm{E}} / \mathrm{m}_{\mathrm{j}}\right] / \mathrm{dm}_{\mathrm{j}}\right\}$ Footnotes [1] $\mathrm{G}^{\mathrm{E}}=\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}] \,\left[\mathrm{mol} \mathrm{kg}{ }^{-1}\right] \,[1]=\left[\mathrm{J} \mathrm{kg}^{-1}\right]$ [2] $\ln \left(\gamma_{\mathrm{j}}\right)=\frac{1}{\left[\mathrm{~J} \mathrm{~K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]} \, \frac{\left[\mathrm{J} \mathrm{kg}^{-1}\right]}{\left[\mathrm{mol} \mathrm{kg}^{-1}\right]}=[1]$ [3] $(1-\phi)=\left[\frac{[1]}{\left[\mathrm{J} \mathrm{K}^{-1} \mathrm{~mol}^{-1}\right] \,[\mathrm{K}]}\right] \,\left[\frac{\left[\mathrm{J} \mathrm{kg}^{-1}\right]}{\left[\mathrm{mol} \mathrm{kg}^{-1}\right]}\right]=[1]$
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.63%3A_Solubilities_of_Solids_in_Liquids.txt
By definition, $\mathrm{G}=\mathrm{H}-\mathrm{T} \, \mathrm{S}$ $\mathrm{G}, \mathrm{~H} \text { and } \mathrm{S}$ are extensive functions of state. At fixed $\mathrm{T}$ and $\mathrm{p}$, the dependences of these variables on extent of reaction, $\xi$ are related. $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}=(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}}-\mathrm{T} \,(\partial \mathrm{S} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ For a spontaneous change $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}<0$ where the affinity for spontaneous change $\mathrm{A}\left[=-(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}\right]$ is positive. This can arise under two limiting circumstances. 1. The spontaneous process is exothermic; [i.e.$\left[\text { i.e. }(\partial \mathrm{H} / \partial \xi)_{\mathrm{T}, \mathrm{p}}<0\right]$ such that $\left.\left|\mathrm{T} \,(\partial \mathrm{S} / \partial \xi)_{\mathrm{T}, \mathrm{p}}\right|<\mid \partial \mathrm{H} / \partial \xi\right)_{\mathrm{T}, \mathrm{p}} \mid$. The decrease in $\mathrm{G}$ is enthalpy driven. 2. The spontaneous process is endothermic; $\left[\text { i.e. }(\partial H / \partial \xi)_{\mathrm{T}, \mathrm{p}}>0\right]$ such that $\left.\left|\mathrm{T} \,(\partial \mathrm{S} / \partial \xi)_{\mathrm{T}, \mathrm{p}}\right|>\mid \partial \mathrm{H} / \partial \xi\right)_{\mathrm{T}, \mathrm{p}} \mid$. The decrease in $\mathrm{G}$ is entropy driven. If for a given possible process, $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}>0$, then the process is not spontaneous; there is no affinity for spontaneous change. If $(\partial \mathrm{G} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is zero at defined $\mathrm{T}$ and $\mathrm{p}$, the system is at equilibrium with the surroundings; the affinity for spontaneous change is zero. The chemical equilibrium is stable if $(\partial \mathrm{A} / \partial \xi)_{\mathrm{T}, \mathrm{p}}$ is negative. 1.14.66: Spontaneous Chemical Reaction A closed reaction vessel at $298.2 \mathrm{~K}$ and $101325 \mathrm{~N m}^{-2}$ is filled with a solution having the initial composition, water ($1.2 \mathrm{~mol}$), $\mathrm{NaOH}(\mathrm{aq}, 0.5 \mathrm{~mol})$, $\mathrm{CH}_{3}.\mathrm{COOC}_{2}\mathrm{H}_{5}(\mathrm{aq}, 0.2 \mathrm{~mol})$. Experiment shows that the system spontaneously changes composition. We write the overall chemical reaction as: $\mathrm{CH}_{3} \mathrm{COOC}_{2} \mathrm{H}_{5}+\mathrm{OH}^{-} \rightarrow \mathrm{CH}_{3} \mathrm{COO}^{-}+\mathrm{C}_{2} \mathrm{H}_{5} \mathrm{OH}$ In this connection we state that the reaction is driven by the affinity for spontaneous reaction leading to a change in chemical composition characterized by the extent of reaction, $\xi$. 1.14.67: Standard States: Reference States: Processes • Pressure: • [1] standard pressure $10^{5} \mathrm{~N m}^{-2} • States: • Solids: • pure solids • Liquids: • pure liquids • Gases: • ideal gas at standard pressure • Solutions: • Simple Solution • Solvent: pure liquid • Solute: ideal solution at unit molality; \(1 \mathrm{~mol kg}^{-1}$ • Process/Change [2,3,4] $\Delta_{\text {proc }} X^{0}$ • Subscript ‘proc’ indicates the process. Examples include • $\mathrm{f} =$ formation • $\mathrm{c} =$ combustion • $\mathrm{vap} =$ vaporisation • $\mathrm{r} =$ chemical reaction • $\mathrm{soln} =$ solution] • $\neq =$ activation; an attempt to represent the classic double dagger symbol suggested by Eyring . The symbol is written as a superscript; e.g. $\Delta^{\neq} V^{0}=$ standard volume of activation. • Example; $\Delta_{\mathrm{c}} \mathrm{H}^{0}=$ standard enthalpy of combustion. Footnotes [1] R. D. Freeman, Bull. Chem.Thermodyn.,1982,25,523. [2] Pure Appl. Chem.,1982,54,1239. [3] M. L. McGlashan, Physico-Chemical Quantities and Units, RIC, London, Number 15, 1971. [4] Quantities, Units, Symbols in Physical Chemistry, IUPAC, Blackwell, Oxford, 1988. 1.14.68: Surroundings and System System The word ‘system’ describes that part of the universe which we have identified for the purpose of studying its chemical properties. The term "universe" in the latter sentence is somewhat pretentious (= implying ‘of enormous importance’ in a way that is doubtful). From the practical point of view, a chemist identifies the system as the contents of the reaction vessel (flask) under investigation. Surroundings The rest of the universe comprises the surroundings. We as observers of the properties plus all our measuring equipment including spectrophotometers and calorimeters are part of the surroundings. As far as chemists are concerned the surroundings mean the laboratory (+ chemist!) surrounding the system.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.65%3A_Spontaneous_Change%3A_Isothermal_and_Isobaric.txt
At ambient pressure, the molar volume of water ($\ell$) is a minimum near $277 \mathrm{~K}$, the temperature of maximum density, $\mathrm{TMD}$. The $\mathrm{TMD}$ is sensitive to the concentration and nature of added solute. Generally added salts lower the $\mathrm{TMD}$, the extent of lowering being often written $\Delta \theta$. For dilute salt solutions, $\mathrm{TMD}$ is a linear function of the molality of salt, $\mathrm{m}_{j}$,$\left(\partial \Delta \theta / \partial m_{\mathrm{j}}\right)$ being negative; Despretz Law. However considerable interest is generated by the observation that some organic solutes at low mole fractions raise the $\mathrm{TMD}$; i.e. $\Delta \theta > 0$; e.g. 2-methylpropan-2-ol. Although the phenomenon of a shift in $\mathrm{TMD}$ is straightforward from an experimental standpoint, explanations distinguish between possible contributions to the shift in $\mathrm{TMD}$. Most treatments identify two contributions to the shift in $\mathrm{TMD}$, an ‘ideal’ shift and a contribution which reflects the fact that the thermodynamic properties of the aqueous system are not ideal [1,2]. At the outset we assume that the molar volume of water at ambient pressure in the region of the $\mathrm{TMD}$ is a quadratic function of the difference ($\mathrm{T}-\mathrm{TMD}^{*}$) where $\mathrm{TMD}^{*}$ is the temperature of maximum density of water ($\ell$) [2]. At temperature $\mathrm{T}$ the molar volume $\mathrm{V}_{1}^{*}(\ell, \mathrm{T})$ is given by equation (a) where $\chi_{1}$ is a dimensionless property of water ($\ell$) [3]. $\mathrm{V}_{1}^{*}(\ell, \mathrm{T})=\mathrm{V}_{1}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \,\left\{1+\chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right)^{2} /[\mathrm{K}]^{2}\right\}$ We consider briefly three types of systems; 1. binary aqueous mixtures, 2. aqueous solutions, and 3. aqueous salt solutions. (i) Binary Aqueous Mixtures For the non-aqueous component, the dependence of molar volume $\mathrm{V}_{2}^{*}(\ell, \mathrm{T})$ on temperature is given by equation (b) where $\chi_{2}$ is a dimensionless property of the non-aqueous component. $\mathrm{V}_{2}^{*}(\ell, \mathrm{T})=\mathrm{V}_{2}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \,\left\{1+\chi_{2} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) /[\mathrm{K}]\right\}$ $\mathrm{V}_{2}^{*}\left(\ell, \mathrm{TMD}^{*}\right)$ is the molar volume of non-aqueous component at the temperature $\mathrm{TMD}^{*}$. The volume of a mixture prepared using $\mathrm{n}_{1}$ and $\mathrm{n}_{2}$ moles of the two liquids at temperature $\theta$ under the no-mix condition is given by equation (c). $\mathrm{V}(\mathrm{no}-\operatorname{mix} ; \theta)=\mathrm{n}_{1} \, \mathrm{V}_{1}^{*}(\ell ; \theta)+\mathrm{n}_{2} \, \mathrm{V}_{2}^{*}(\ell ; \theta)$ Hence, \begin{aligned} \mathrm{V}(\mathrm{no}-\operatorname{mix} ; \theta) &=\mathrm{n}_{1} \,\left[\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left\{1+\chi_{1} \,\left(\theta-\mathrm{TMD}^{*}\right)^{2} /[\mathrm{K}]^{2}\right\}\right] \ &+\mathrm{n}_{2} \,\left[\mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left\{1+\chi_{2} \,\left(\theta-\mathrm{TMD}^{*}\right) /[\mathrm{K}]\right\}\right] \end{aligned} At temperature $\theta$, the volume of the real mixture $\mathrm{V}(\operatorname{mix} ; \theta)$ is given by equation (e) where $\mathrm{V}_{1}(\operatorname{mix} ; \theta)$ and $\mathrm{V}_{2}(\operatorname{mix} ; \theta)$ are the partial molar volumes of the two components in the mixture at temperature $\theta$. $\mathrm{V}(\operatorname{mix} ; \theta)=\mathrm{n}_{1} \, \mathrm{V}_{1}(\operatorname{mix} ; \theta)+\mathrm{n}_{2} \, \mathrm{V}_{2}(\operatorname{mix} ; \theta)$ The volume of mixing at temperature $\theta$ is given by equation (f). By definition, $\Delta_{\text {mix }} V(\theta)=V(\operatorname{mix} ; \theta)-V(\text { no }-\operatorname{mix} ; \theta)$ Hence, \begin{aligned} \Delta_{\text {mix }} \mathrm{V}(\theta)=& \mathrm{n}_{1} \,\left\{\mathrm{V}_{1}(\operatorname{mix} ; \theta)-\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{1} \,\left(\theta-\mathrm{TMD}^{*}\right)^{2} /[\mathrm{K}]^{2}\right]\right\} \ &+\mathrm{n}_{2} \,\left\{\mathrm{V}_{2}(\operatorname{mix} ; \theta)-\mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{2} \,\left(\theta-\mathrm{TMD}^{*}\right) /[\mathrm{K}]\right]\right\} \end{aligned} But the molar volume of mixing at temperature $\theta$, $\Delta_{\text {mix }} \mathrm{V}_{\mathrm{m}}(\theta)=\Delta_{\text {mix }} \mathrm{V}(\theta) /\left(\mathrm{n}_{1}+\mathrm{n}_{2}\right)$ Hence, \begin{aligned} \Delta_{\operatorname{mix}} \mathrm{V}_{\mathrm{m}}(\theta) &=\mathrm{x}_{1} \, \mathrm{V}_{1}(\operatorname{mix} ; \theta)+\mathrm{x}_{2} \, \mathrm{V}_{2}(\operatorname{mix} ; \theta) \ -\left\{\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD}) \,\left[1+\chi_{1} \,\left(\theta-\mathrm{TMD}^{*}\right)^{2} /[\mathrm{K}]^{2}\right]\right\} \ &-\left\{\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{2} \,\left(\theta-\mathrm{TMD}^{*}\right) /[\mathrm{K}]\right]\right\} \end{aligned} Then the differential of $\Delta_{\mathrm{mix}} \mathrm{~V}_{\mathrm{m}}$ is given by equation (j). \begin{aligned} \mathrm{d} \Delta_{\operatorname{mix}} \mathrm{V}_{\mathrm{m}}(\mathrm{T})=& \mathrm{x}_{1} \, \mathrm{dV}(\operatorname{mix} ; \mathrm{T})+\mathrm{V}_{1}(\operatorname{mix} ; \mathrm{T}) \, \mathrm{dx}{ }_{1} \ &+\mathrm{x}_{2} \, \mathrm{dV}_{2}(\operatorname{mix} ; \mathrm{T})+\mathrm{V}_{2}(\mathrm{mix} ; \mathrm{T}) \, \mathrm{dx} \mathrm{x}_{2} \ &-\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{1} \,(\mathrm{T}-\mathrm{TMD})^{2} \,[\mathrm{K}]^{-2}\right] \, \mathrm{dx} \ &\left.-\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD}) \, 2 \, \chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) \,[\mathrm{K}]^{-2}\right] \, \mathrm{dT} \ &-\mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{1} \,(\mathrm{T}-\mathrm{TMD})^{2} \,[\mathrm{K}]^{-2}\right] \, \mathrm{dx}_{2} \ &-\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell ; \mathrm{TMD}) \, \chi_{2} \,[\mathrm{K}]^{-1} \, \mathrm{dT} \end{aligned} But according to the Gibbs-Duhem Equation, at fixed pressure $x_{1} \, d V_{1}(\operatorname{mix} ; T)+x_{2} \, d V_{2}(\operatorname{mix} ; T)=E_{p m}(\operatorname{mix} ; T) \, d T$ In equation (k), $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{T})$ is the molar isobaric expansion of the mixture. Equation (j) can therefore be reorganized into equation (l). \begin{aligned} &\mathrm{d} \Delta_{\operatorname{mix}} \mathrm{V}_{\mathrm{m}}(\mathrm{T})=\ &\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{T}) \, \mathrm{dT}\ &+\left\{\mathrm{V}_{1}(\operatorname{mix} ; \mathrm{T})-\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right)^{2} \,[\mathrm{K}]^{-2}\right\} \, \mathrm{dx} \mathrm{x}_{1}\right.\ &+\left\{\mathrm{V}_{2}(\mathrm{mix} ; \mathrm{T})-\mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \,\left[1+\chi_{2} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) \,[\mathrm{K}]^{-1}\right] \, \mathrm{dx}\right.\ &\left.-\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, 2 \, \chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) \,[\mathrm{K}]^{-2}\right] \, \mathrm{dT}\ &-\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, \chi_{2} \,[\mathrm{K}]^{-1} \, \mathrm{dT} \end{aligned} \ We note that $\mathrm{dx}_{1} = −\mathrm{dx}_{2}$ and that the coefficients of $\mathrm{dx}_{1}$ and $\mathrm{dx}_{2}$ are in fact excess partial molar volumes at temperature $\mathrm{T}$. Hence, \begin{aligned} \mathrm{d} \Delta_{\operatorname{mix}} \mathrm{V}_{\mathrm{m}}(\mathrm{T})=& \ \mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \mathrm{T}) \, & \mathrm{dT}+\left[\mathrm{V}_{2}^{\mathrm{E}}(\mathrm{T})-\mathrm{V}_{1}^{\mathrm{E}}(\mathrm{T})\right] \, \mathrm{dx} \mathrm{x}_{2} \ &\left.-\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD}) \, 2 \, \chi_{1} \,(\mathrm{T}-\mathrm{TMD}) \,[\mathrm{K}]^{-2}\right] \, \mathrm{dT} \ &-\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}(\ell ; \mathrm{TMD}) \, \chi_{2} \,[\mathrm{K}]^{-1} \, \mathrm{dT} \end{aligned} Or, \begin{aligned} \frac{\mathrm{d} \Delta_{\text {mix }} \mathrm{V}_{\mathrm{m}}(\mathrm{T})}{\mathrm{dT}} &=\mathrm{E}_{\mathrm{pm}}(\mathrm{mix} ; \mathrm{T})+\left[\mathrm{V}_{2}^{\mathrm{E}}(\mathrm{T})-\mathrm{V}_{1}^{\mathrm{E}}(\mathrm{T})\right] \, \frac{\mathrm{dx}}{\mathrm{dT}} \ &\left.-\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, 2 \, \chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) \,[\mathrm{K}]^{-2}\right] \ &-\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, \chi_{2} \,[\mathrm{K}]^{-1} \end{aligned} We use equation (n) at temperature ‘$\mathrm{T} = \theta$’ and at fixed composition; i.e. $\mathrm{dx}_{2} = 0$. Moreover, by definition at the $\mathrm{TMD}$, $\mathrm{E}_{\mathrm{pm}}(\operatorname{mix} ; \theta)$ is zero. Hence the dependence of $\Delta_{\text {mix }} \mathrm{V}_{\mathrm{m}}(\theta)$ on temperature at fixed pressure and composition is given by equation (o). \begin{aligned} \left(\frac{\partial \Delta_{\text {mix }} V_{\mathrm{m}}(\mathrm{T})}{\partial \mathrm{T}}\right)_{\mathrm{p}, \mathrm{x}(2)}=&\left.-\mathrm{x}_{1} \, \mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, 2 \, \chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right) \,[\mathrm{K}]^{-2}\right] \ &-\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, \chi_{2} \,[\mathrm{K}]^{-1} \end{aligned} We identify temperature $\theta$ with the recorded $\mathrm{TMD}$. Hence from equation (o) with $\Delta \theta=\theta-\mathrm{TMD}^{*}$ [4], $\begin{array}{r} \Delta \theta=-\frac{\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{2} \,[\mathrm{K}]^{-1}}{2 \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{V}_{1}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{1} \,[\mathrm{K}]^{-2}} \ -\frac{\left[\partial \Delta_{\text {mix }} \mathrm{V}_{\mathrm{m}}(\theta) / \partial \mathrm{T}\right]_{\mathrm{p}, \mathrm{x}(2)}}{2 \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{V}_{1}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{1} \,[\mathrm{K}]^{-2}} \end{array}$ In other words, the shift in the $\mathrm{TMD}$, $\Delta \theta$, is made up of two contributions. For binary system having thermodynamic properties which are ideal, the second term on the r.h.s. of equation (p) is zero. The first term on the r.h.s. side of equation (p) predicts that $\Delta \theta$ is negative in agreement with the Despretz rule. In summary therefore equation (p) can be written in the following simple form. $\Delta \theta=\Delta \theta(\text { ideal })+\Delta \theta(\text { struct })$ The sign of $\Delta \theta (\text{struct})$ is determined by the sign of $\left[\partial \Delta_{\operatorname{mix}} V_{\mathrm{m}}(\theta) / \partial \mathrm{T}\right]$. If the latter term is negative, $\Delta \theta (\text{struct})$ is positive and for some systems can be the dominant term. As noted above, this is the case at low mole fractions $\mathrm{x}_{2}$ for 2-methylpropan-2-ol, a trend attributed to enhancement of water-water hydrogen bonding by the non-aqueous component. (ii) Aqueous Solutions The volume of a solution at temperature $\mathrm{TMD}$, prepared using $1 \mathrm{~kg}$ of solvent water and mj moles of a simple neutral solute is given by equation (r). $\mathrm{V}(\mathrm{aq} ; \mathrm{TMD})=\left(\mathrm{l} / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD})+\mathrm{m}_{\mathrm{j}} \, \phi\left(\mathrm{V}_{\mathrm{j}} ; \mathrm{TMD}\right)$ But at the $\mathrm{TMD}$, $\left(1 / \mathrm{M}_{1}\right) \,\left[\partial \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD}) / \partial \mathrm{T}\right]=-\mathrm{m}_{\mathrm{j}} \,\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}} ; \mathrm{TMD}\right) / \partial \mathrm{T}\right]$ We use equation (a) to relate $\mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD})$ to $\mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right)$ with $\Delta \mathrm{T}$ representing ($\mathrm{TMD} - \mathrm{~TMD}^{*}$). $\left(1 / \mathrm{M}_{1}\right) \, \mathrm{V}_{1}^{*}(\ell ; \mathrm{TMD}) \, \chi_{1} \, 2 \, \Delta \mathrm{T} /[\mathrm{K}]^{2}=-\mathrm{m}_{\mathrm{j}} \,\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}} ; \mathrm{TMD}\right) / \partial \mathrm{T}\right]$ We assume that for dilute real solutions $\phi\left(\mathrm{V}_{j}\right)$ is a linear function of the molality of solute $j$ and that the proportionality term is the pairwise volumetric interaction parameter $\mathrm{v}_{jj}$. Thus, $\phi\left(\mathrm{V}_{\mathrm{j}}\right)=\phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty}+\mathrm{v}_{\mathrm{jj}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Then $\mathrm{d} \phi\left(\mathrm{V}_{\mathrm{j}}\right) / \mathrm{dT}=\mathrm{d} \phi\left(\mathrm{V}_{\mathrm{j}}\right)^{\infty} / \mathrm{dT}+\left[\partial \mathrm{v}_{\mathrm{ij}} / \partial \mathrm{T}\right]_{\mathrm{p}} \,\left(\mathrm{m}_{\mathrm{j}} / \mathrm{m}^{0}\right)$ Hence, \begin{aligned} &\left\{2 \, \mathrm{V}_{1}^{*}\left(\ell ; \mathrm{TMD}^{*}\right) \, \chi_{1} / \mathrm{M}_{1} \,[\mathrm{K}]^{2}\right\} \, \Delta \mathrm{T}= \ &\quad-\mathrm{m}_{\mathrm{j}} \,\left[\partial \phi\left(\mathrm{V}_{\mathrm{j}}\right)=\partial \mathrm{T}\right]-\left[\partial \mathrm{v}_{\mathrm{ij}} / \partial \mathrm{T}\right] \,\left(\mathrm{m}^{0}\right)^{-1} \,\left(\mathrm{m}_{\mathrm{j}}\right)^{2} \end{aligned} We rewrite equation (w) as an equation for the change in $\mathrm{TMD}$, $\Delta \mathrm{T}$ using a quadratic in $\mathrm{m}_{j}$ [5]. Thus, $\Delta \mathrm{T}=\mathrm{q}_{1} \, \mathrm{m}_{\mathrm{j}}+\mathrm{q}_{2} \, \mathrm{m}_{\mathrm{j}}^{2}$ Consequently a plot of $\Delta \mathrm{T}$ $\mathrm{m}_{j}$ is linear having intercept $\mathrm{q}_{1}$ and slope $\mathrm{q}_{2}$ [1]. If the solution is ideal [i.e. $\mathrm{v}_{jj}$ is zero] then $\mathrm{q}_{2}$ in zero and $\left[\Delta \mathrm{T} / \mathrm{m}_{\mathrm{j}}\right]$ is constant independent of $\mathrm{m}_{j}$ [6]. (iii) Aqueous Salt Solutions The above analysis forms the basis for an analysis of the effects of salts on $\mathrm{TMD}$ except that the dependence of $\phi\left(\mathrm{V}_{j}\right)$ is expressed using the following equation where $\mathrm{S}_{\mathrm{V}}$ is the Debye-Huckel Limiting Law volumetric parameter [7-12]. $\phi\left(V_{j}\right)=\phi\left(V_{j}\right)^{\infty}+S_{v} \,\left(m_{j} / m^{0}\right)^{1 / 2}+b \,\left(m_{j} / m^{0}\right)$ The foregoing analysis has been extended to include consideration of isobaric expansions [13] and limiting partial molar expansions [14-17]. Footnotes [1] C. Wada and S.Umeda, Bull. Chem. Soc. Jpn,1962,35,646,1797. [2] F. Franks and B. Watson, Trans. Faraday Soc.,1969,65,2339. [3] The symbol [K] indicates the unit of temperature, kelvin. The term $\left\{1+\chi_{1} \,\left(\mathrm{T}-\mathrm{TMD}^{*}\right)^{2} /[\mathrm{K}]^{2}\right\}$ is dimensionless as required by equation (a). [4] As noted $\chi_{1}$ and $\chi_{2}$ are dimensionless and characteristic properties of the two components. \begin{aligned} &\frac{\mathrm{x}_{2} \, \mathrm{V}_{2}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{2} \,[\mathrm{K}]^{-1}}{2 \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{V}_{1}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{1} \,[\mathrm{K}]^{-2}} \ &=\frac{[1] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,[1] \,[\mathrm{K}]^{-1}}{[1] \,[1] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,[1] \,[\mathrm{K}]^{-2}}=[\mathrm{K}] \ &\frac{\left[\partial \Delta_{\text {mix }} \mathrm{V}_{\mathrm{m}}(\theta) / \partial \theta\right]}{2 \,\left(1-\mathrm{x}_{2}\right) \, \mathrm{V}_{1}^{*}\left(\ell, \mathrm{TMD}^{*}\right) \, \chi_{1} \,[\mathrm{K}]^{-2}}=\frac{\left[\mathrm{m}^{3} \mathrm{~mol}^{-1} \mathrm{~K}^{-1}\right]}{[1] \,[1] \,\left[\mathrm{m}^{3} \mathrm{~mol}^{-1}\right] \,[1] \,[\mathrm{K}]^{-2}}=[\mathrm{K}] \end{aligned} [5] M. V. Kaulgud, J. Chem. Soc. Faraday Trans.1,1979,75,2246; 1990,86,911. [6] J. R. Kuppers, J.Phys.Chem.,1974,78,1041. [7] T. H. Lilley and S. Murphy, J.Chem. Thermodyn., 1973,5,467. [8] T. Wakabayashi and K. Takazuimi, Bull. Chem. Soc. Jpn., 1982,55,2239. [9] T. Wakabayashi and K. Takazuimi, Bull. Chem. Soc. Jpn., 1982,55,3073. [10] For comments on salts in D2O, see A. J. Darnell and J. Greyson, J. Phys. Chem.,1968, 73,3032. [11] G. Wada and M. Miura, Bull. Chem. Soc. Jpn., 1969,42,2498. [12] J. R. Kuppers, J. Phys. Chem.,1975,79,2105. [13] D. A. Armitage, M. J. Blandamer, K. W. Morcom and N. C. Treloar, Nature, 1968,219,718. [14] J. E. Garrod and T. M. Herrington, J. Phys.Chem.,1970,74,363. [15] T. M. Herrington and E. L. Mole, J. Chem. Soc. Faraday Trans.1,1982,78,213. [16] D. D. Macdonald and J. B. Hyne, Can. J. Chem.,1976,54,3073. [17] D. D. Macdonald, B. Dolan and J. B. Hyne, J. Solution Chem.,1976,5,405.
textbooks/chem/Physical_and_Theoretical_Chemistry_Textbook_Maps/Topics_in_Thermodynamics_of_Solutions_and_Liquid_Mixtures/1.24%3A_Misc/1.14.69%3A_Temperature_of_Maximum_Density%3A_Aqueous_Solutions.txt