University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
2
Ferrous iron is oxidized to ferric iron:
(2) Fe2+ + 1/4 0 + H+ Fe3+ + 1/2 H20
2
Pyrite is oxidized by ferric iron at low pH (< 4) and results in further acid mine
drainage generation:
(3) FeS + 14Fe3+ + 8H20 -M 5Fe2+ + 2S042’ + 16H+
2
(Edwards et al. 1999)
Pyrite dissolution occurs chemically; however, it is a very slow process. Pyrite is
oxidized 3 to 100 times faster by ferric iron (Reaction 3) than by oxygen (Reaction 1)
(Edwards et al. 1999). However, production of ferric iron (Reaction 2) proceeds slowly
in low pH environments, thus representing the rate-limiting step in the oxidation of pyrite
by ferric iron (Edwards et al. 1999). Bacteria catalyze the process and significantly
increase the rate at which it occurs. Chemolithotrophic bacteria, which oxidize inorganic
compounds for energy, use ferrous iron as an electron donor, oxidizing it to ferric iron
enzymatically at a rate up to six orders of magnitude faster than the rate of chemical
oxidation (Newman and Banfield, 2002). These bacteria live in very low pH
environments and are able produce energy using sulfur or ferrous iron as the electron
donor, and oxygen as the terminal electron acceptor (Alexander, 1999).
Thiobacillus ferrooxidans and Leptospirillum ferrooxidans are acidophilic
chemolithotrophs that help catalyze the oxidation of pyrite to ferrous iron. T.
ferrooxidans is generally thought to be the most important species catalyzing metal
sulfide dissolution; however, a study performed at a mine in Iron Mountain, California
demonstrated that T. ferrooxidans did not play an active role in acid generation at that site
(Edwards et al. 1999, Schrenk et al. 1998). Other species, including a new species of
Archaea, Ferroplasma acidarmanus, were found to be present and metabolically active at
the surface of the exposed sulfide minerals at the Iron Mountain site, calling into question
|
Colorado School of Mines
|
3
the presumption that T. ferrooxidans and L. ferrooxidans are the most important acid-
generating microbial species (Edwards et al. 1999).
1.2 Anaerobic Passive Mine Drainage Treatment Systems
Remediation Technologies. Remediation of acid mine drainage requires reduction of the
toxicity, mobility, and bioavailability of metals and increase of alkalinity. One approach
is to stabilize metals as hydroxide, carbonate, or sulfide precipitates. Metal-contaminated
waters have typically been treated by addition of alkaline materials, such as lime
(Ca(OH or CaO), to increase pH and precipitate sulfate as gypsum (CaSO^ and metals
)2
as hydroxides; however, treatment plants utilizing this technique are expensive to build
and operate (Hammack and Edenbom, 1992). Large quantities of sludge resulting from
this process must be dewatered and disposed of as a hazardous waste since the metals
could easily be remobilized if exposed to lower pH (Lyew and Sheppard, 1997).
Traditional site remediation techniques can be very expensive and are not feasible for
remote and abandoned mining-related sites (USEPA, 1995). Cost-effective, low-
maintenance, long-term technologies for removal of metals and acidity are desirable for
remediation of these sites. Passive treatment systems such as permeable reactive barriers
and constructed wetlands have the potential to meet these criteria.
Anaerobic passive treatment systems for remediation of acid mine drainage support
sulfate-reducing bacteria, which generate sulfide. Sulfides are highly reactive with heavy
metal ions, and resulting metal sulfide precipitates typically exhibit low solubility over a
wide pH range (Eccles, 1999, Fortin et al. 1994, Hammack and Edenbom, 1992). Metal
sulfides are often more insoluble than metal carbonates and metal hydroxides (Cocos et
al. 2002, Hammack et al. 1994, Stumm and Morgan, 1996). Thus, an engineered
environment in which the biological and metabolic requirements of sulfate-reducing
bacteria are met is a promising approach for treatment of acid mine drainage.
|
Colorado School of Mines
|
4
Permeable Reactive Barriers. Permeable reactive barriers and constructed wetlands are
designed to provide long-term, in situ treatment of contaminated water. A permeable
reactive barrier is a trench filled with a reactive mixture that is permeable to groundwater.
It functions by intercepting a contaminant plume and treating it as it flows through.
Permeable reactive barriers are used for intercepting shallow contaminant plumes, no
more than 50 to 70 feet deep (USEPA, 2000). They are often anchored into an
impermeable layer, so the contaminant plume will not pass below the reactive media.
They can be placed into aquifers downgradient of acid mine drainage sources. Permeable
reactive barriers do not require pumping, but instead rely on gradient and hydraulic head
to move the contaminant plume through the reactive material. They are designed with
permeability greater than the surrounding aquifer material to ensure that the plume does
not go around the reactive zone. Greater permeability is also necessary so groundwater
flow will not be disrupted as precipitation of metal sulfides begins to occur within the
reactive material (USEPA, 2000).
Anaerobic permeable reactive barriers promoting the activity of sulfate-reducing
bacteria use inexpensive substrate materials and show promise in removing both metals
and acidity from acid mine drainage (Benner et al. 1997). They are a relatively recent
technological development in remediation. The first full-scale commercial permeable
reactive barrier was approved for use in 1994 (USEPA, 1998). A permeable reactive
barrier installed in 1995 at the Nickel Rim mine site in Ontario, Canada was estimated to
be effective for a minimum of 15 years (Benner et al. 1997).
Constructed Wetlands. Constructed wetlands rely on various biogeochemical processes,
generally aerobic, to remove soluble metals from mine drainage (Webb et al. 1998).
Such processes include adsorption, precipitations, organic complexation, and uptake by
plants (August et al. 2002). Constructed wetlands can be engineered to emphasize
processes for removal of specific contaminants (Wildeman and Updegraff, 1997).
Aerobic wetlands emphasize metal precipitation as oxides, hydroxides, or carbonates and
|
Colorado School of Mines
|
5
require plants and algae to provide alkalinity, oxygen, and nutrients (Wildeman and
Updegraff, 1997). Anaerobic wetlands can be engineered so water flows through the
organic substrate, or anaerobic zone, where microbial sulfide production is an important
process leading to metal removal (Wildeman and Updegraff, 1997). Sulfide precipitation
was found to be the primary metal removal process in an anaerobic constructed wetland
engineered to treat acid mine drainage from the Big Five Tunnel in Idaho Springs,
Colorado (Machemer et al. 1993).
Sulfate-Reducing Bacteria. Sulfate-reducing bacteria in passive mine drainage treatment
systems use dissolved sulfate, in acid mine drainage as a terminal electron acceptor
SO 42",
in their respiration process, reducing it to hydrogen sulfide, HjS, in the presence of
organic substrate. The process of sulfate reduction consumes acidity and produces
alkalinity through reduction of the organic carbon substrate (Lyew and Sheppard, 1997).
Dissolved H S then reacts with soluble metals in acid mine drainage to form highly
2
insoluble metal sulfides, which precipitate out of the water:
(4) + 2CH20 + 2H+ -> H2S + 2H20 + 2C0
SO 42" 2
(where “CH O” represents an organic carbon source, and pH > 6.0)
2
(5) Me2+ + H2S -> MeS(S) + 2H+
(Clayton et al. 1998, Blowes et al. 2000).
Sulfate-reducing bacteria grow in the absence of oxygen. In nature, sulfate reducers
are typically found in soils and sediments. Activity of sulfate reducers is severely limited
below pH 5.0 (Wildeman et al. 1997). They require near-neutral pH for optimal activity,
which can be accomplished by adding limestone to the reactive mixture in a passive
treatment system. They can tolerate a variety of heavy metals, as well as dissolved
|
Colorado School of Mines
|
6
sulfide (Chang et al. 2001). They are able to use a wide range of electron donors such as
organic acids, fatty acids, alcohols, and H (Brock and Madigan, 1991).
2
A variety of complex organic materials, including compost, manure, wood chips,
sewage sludge, leaf mulch, sawdust, spent mushroom compost, hay, and alfalfa are used
as substrates to support microbial activity in anaerobic passive treatment systems for acid
mine drainage (Chang et al. 2000, Gilbert et al. 1999, Cocos et al. 2002, Waybrant et al.
1998). Complex carbon sources in a passive treatment system are degraded by other
microbial communities to simple compounds that can later be used by sulfate reducers as
electron donors in their respiration process (Wildeman et al. 1997). Thus, the other
microbial communities within a passive treatment system are critical for long-term
sustainability of passive treatment systems.
1.3 Scope of Research
Anaerobic degradation of complex organic material to simple organic compounds is
carried out by a consortium of microbial species in natural environments (Zehnder,
1988). The same is true in passive treatment systems (Wildeman and Updegraff, 1997).
Sulfate reducers are dependent on other microbial activities to provide the compounds
they require for carbon and energy. This consortium of species, including those that
carry out enzymatic hydrolysis and fermentation, is essential to the long-term
sustainability of passive treatment systems. Conversely, methanogens compete with
sulfate reducers for fermentation products, such as acetate and H2, and can divert the
energy flow to an undesired pathway (Daly et al. 2000). A schematic of key carbon
transformations in anaerobic passive treatment systems, adapted from Gottschalk, in
which complex organic polymers are degraded to simple organic substrates, is presented
in Figure 1.1.
|
Colorado School of Mines
|
A variety of substrates have been used in anaerobic passive treatment systems, with
similar initial sulfate removal rates (Cocos et al. 2002, Wildeman et al. 1997, Wildeman
et al. 1997). However, long-term performance of passive treatment systems remains
variable (Webb et al. 1998). Conditions leading to decline of sulfate-reducing activity
and failure of passive treatment systems are not well understood. Previous research has
focused primarily on the activity of sulfate-reducing bacteria in anaerobic passive
treatment systems (Cocos et al. 2002, Waybrant et al. 1998, Dvorak et al. 1992,
Tsukamoto et al. 1999). Other microbial groups degrade complex organic material to
provide the simple organic compounds required by sulfate reducers and are essential for
long-term sustainability of passive mine drainage treatment systems. Little research has
been focused on understanding the biological processes and carbon flow in passive
treatment systems (Webb et al. 1998).
Several studies have recognized the need for a better understanding of substrate
utilization and carbon flow in passive treatment systems for long-term effectiveness
(Webb et al. 1988, Benner et al. 1999, Wildeman and Updegraff, 1997, Cocos et al.
2002). Benner et al. hypothesized that only a fraction of the organic carbon in the
substrates used in passive treatment systems is available to support rapid sulfate reduction
(Benner et al. 1999). A 2002 study by Cocos et al. hypothesized that treatment
performance would decline after labile organic carbon was depleted, but cellulose would
provide a long-term carbon source to sustain the sulfate-reducing population (Cocos et al.
2002).
“There is a rate-limiting step that controls how fast treatment can be accomplished
and determining this rate-limiting step more specifically would greatly help in devising
design modifications that would speed the treatment processes” (Wildeman and
Updegraff, 1997). This research hypothesizes that system performance is limited by one
or more upstream microbial activities that function as rate-limiting steps in generating
substrates for sulfate-reducing bacteria.
|
Colorado School of Mines
|
9
1.4 Research Objectives and Approach
An understanding of relationships between microbial activities and system
performance is fundamental to the design of anaerobic passive treatment systems for
long-term performance. An approach based on microbial ecology that quantifies the
functions of contributing populations, as well as the rate-limiting step(s) that control
sulfate reduction, will lead to better designs and more robust passive treatment systems
(Wildeman and Updegraff, 1997).
Molecular techniques provide information about the presence of bacteria in a system,
but not about their activities (Daly et al. 2000, Raskin et al. 1996, Utgikar et al. 2003).
While many microbial populations may be present in an anaerobic system treating mine
drainage, it is their actual activities that are important when assessing system
performance. Rather than identify specific bacteria, this research investigates an
approach to quantify key microbial degradation activities in an anaerobic column system
treating synthetic mine drainage as the system ages and sulfate removal rates change.
The overall objective of this research was to develop a method to: (1) quantify
important microbial activities that influence sulfate reduction in an anaerobic passive
treatment system, and (2) apply the method to an anaerobic column system treating
synthetic mine drainage to detect differences in activities as the system ages, for the
purpose of determining the rate-limiting step(s) in the degradation of organic material as
they relate to sulfate reduction. Several preliminary experiments were performed leading
up to the final experiment designed to meet the research objectives, and all are described
in the following chapters.
|
Colorado School of Mines
|
10
CHAPTER 2: MICROBIAL PHYSIOLOGY OF ANAEROBIC
PASSIVE MINE DRAINAGE TREATMENT SYSTEMS
2.1 Anaerobic Respiration
Passive treatment systems for remediation of acid mine drainage rely on the activity
of sulfate-reducing bacteria. Sulfate-reducing bacteria are obligate anaerobes (Brock and
Madigan, 1991). Thus, passive treatment systems are designed to emphasize anaerobic
respiration and anaerobic fermentation processes. During anaerobic respiration, an
electron donor is oxidized and the electrons removed are shuttled through a series of
electron transport proteins within the membrane until they reduce a compound known as
the terminal electron acceptor (Brock and Madigan, 1991). Organisms that carry out
anaerobic respiration use compounds other than molecular oxygen, O , as the terminal
2
electron acceptor in their respiration processes. Sulfate-reducing bacteria use sulfate as
the terminal electron acceptor. Other types of anaerobic respirers include denitrifying
bacteria, which use nitrate as the terminal electron acceptor, and iron-reducing bacteria,
which use ferric iron as the terminal electron acceptor.
Many organisms reduce inorganic compounds for nutritional and cell synthesis
purposes (Brock and Madigan, 1991). This process is known as assimilatory metabolism
because the compound is assimilated into the cell. Reduction for the purpose of
generating energy is a different process known as dissimilatory metabolism. During
assimilatory metabolism, just enough of a compound is reduced to meet the nutritional
needs of the cell, while during dissimilatory metabolism the compound is reduced in
much greater amounts and excreted by the cell (Brock and Madigan, 1991). Sulfate-
reducing bacteria carry out dissimilatory metabolism of sulfate. They use sulfate as an
|
Colorado School of Mines
|
11
electron acceptor for energy generation, with the amount of H S excreted much greater
2
than the amount used for cell synthesis (Zehnder, 1988).
2.2 Anaerobic Degradation of Complex Organic Material
Anaerobic degradation of complex organic material, such as the substrates used in
permeable reactive barriers and constructed wetlands, requires a consortium of microbial
species. Cellulose is the principal component of plant material and the most abundant
organic compound on earth (Zehnder, 1988). Thus, a large portion of the carbon in
passive treatment systems is in the form of cellulose. Degradation of cellulose appears to
be dependent on the lignin it is associated with, as the presence of lignin can render much
of the cellulose carbon inaccessible to microbial degradation activities (Pareek et al.
1998). Lignin is a highly branched, complex polymer composed of aromatic subunits
that is found in the cell walls of vascular plants. Lignin is degraded under aerobic
conditions by white rot fungi through non-specific oxidative processes catalyzed by
extracellular enzymes (Pareek et al. 2001). It is generally considered highly recalcitrant
to microbial degradation under anaerobic conditions; however, evidence has shown that
anaerobic degradation of lignin can occur under sulfate-reducing conditions (Pareek et al.
2001).
Cellulose is a polymer of glucose molecules, CôHhOô, connected by /3-1,4-glycosidic
linkages, with cellobiose, C H O , considered the repeating unit (Zehnder, 1988).
12 22 11
Cellulose polymers cannot be used directly for energy in anaerobic processes and must
first be hydrolyzed by cellulolytic enzymes. Hydrolysis of cellulose into cellobiose and
glucose is carried out by extracellular enzymes, called cellulases, which are produced by
fermenters (Zehnder, 1988). Cellulases include three distinct enzymes that may be
associated in a multi-enzyme complex: endoglucanase, which randomly attacks internal
/3-1,4-glycosidic linkages; exoglucanase, which cleaves cellobiose units from the ends of
|
Colorado School of Mines
|
12
cellulose fragments; and beta-glucosidase, which hydrolyzes the resulting cellobiose
subunits to glucose (Zehnder, 1988).
Glucose, CôHuOô, is used as a substrate for anaerobic fermentation. Fermentation is
an energy-yielding process in which a single organic substrate functions as both electron
donor and electron acceptor: part of the molecule is oxidized, while part is reduced
(Brock and Madigan, 1991). Fermenters do not require an external electron acceptor,
which may not be available in many anaerobic environments (Brock and Madigan, 1991).
There are many types of fermentation processes, resulting in a variety of fermentation
products including organic acids and alcohols. Syntrophs are fermenters that use fatty
acids or alcohols as energy sources and produce acetate, CO , and H (Brock and
2 2
Madigan, 1991 Zehnder, 1988). Syntrophs occur in association with F^-consuming
bacteria, such as sulfate reducers or methanogens (Harada et al. 1994). However, sulfate
reducers can also compete with syntrophs for substrates such as ethanol and butyrate
(O’Flaherty et al. 1998, Qatibi et al. 1990). Acetogens are fermenters that use CO as the
2
terminal electron acceptor and produce H and acetate (Brock and Madigan, 1991,
2
O’Flaherty et al. 1998).
Many fermentation products are used by sulfate-reducing bacteria as sources of
carbon and energy (Zehnder, 1988). Sulfate reducers can also compete with fermenters
for substrates such as lactate, which can be fermented further (Zehnder, 1988, Raskin et
al. 1996). A study by Qatibi et al. found that sulfate reducers outcompeted fermenters for
lactate in a mixed bacterial culture (Qatibi et al. 1990). Sulfate-reducing bacteria differ
widely in terms of substrate requirements and morphology, but are grouped together
based on their ability to perform dissimilatory sulfate reduction (Brock and Madigan,
1991).
There are two main metabolic groups of sulfate-reducing bacteria: incomplete
oxidizers and complete oxidizers (Brock and Madigan, 1991). Sulfate reducers that
oxidize lactate and other substrates to acetate as the end product are considered
|
Colorado School of Mines
|
15
that it was due to the fact that acetate-utilizing sulfate reducers do not attach to surfaces
as well as methanogens and may be transported out in the effluent of flow-through
systems (O’Flaherty et al. 1998 and Harada et al. 1994). However, sulfate reducers are
generally favored as long as sulfate is present (Brock and Madigan, 1991, Zehnder, 1988,
Qatibi et al. 1990). A study by Harada et al. using bioreactors containing different levels
of sulfate found a relationship between increased sulfate levels and decreased methane
production (Harada et al. 1994). Sulfate reducers have a stronger affinity for H and can
2
maintain the level of H below the threshold concentration that is required by
2
methanogens for uptake (Raskin et al. 1996, Brock and Madigan, 1991, Zehnder et al.
1988, Qatibi et al. 1990). Methanogens are unable to take up H at concentrations below
2
5 to 10 jitM (Brock and Madigan, 1991). The H2S produced by sulfate reducers can have
inhibitory effects on methanogens and other anaerobic bacteria (Li et al. 1996).
However, in a study by Harada et al. the observed decline in methanogen activity in
anaerobic bioreactors was attributed to substrate competition with sulfate reducers rather
than H2S toxicity (Harada et al. 1994).
This research hypothesized that sulfate reduction in passive treatment systems is
limited by one or more upstream microbial activities that function as rate-limiting steps in
generating substrates for sulfate-reducing bacteria. The purpose of this study was to
investigate key microbial degradation activities upstream of sulfate reduction to reveal
which step ultimately limited performance of laboratory-scale continuous-flow passive
bioreactor systems.
|
Colorado School of Mines
|
16
CHAPTER 3: PILOT BENCH-SCALE COLUMN STUDY
3.1 Experimental Objectives
A bench-scale column study was initiated, based on experiments described in
published studies (Waybrant et al. 2002, Elliott et al. 1998), with the objective of
establishing an active mine drainage treatment system in terms of sulfate reduction, pH
neutralization, and metal removal. The main goal of the pilot column study was to gain
experience assembling and monitoring an experimental column system. The pilot
column study provided valuable information for subsequent experiments in terms of
column setup procedures and effluent sampling and monitoring procedures. It also
provided a data set with which future experimental data could be compared.
3.2 Experimental Methods
3.2.1 Column Specifications.
Three glass columns, each 5 cm in diameter and 30 cm tall with four side sampling
ports, were used for the pilot column experiment. Sampling ports were sealed with
rubber gaskets and screw-on caps. Plastic caps with rubber o-rings inside fit over the
ends of each column and formed a seal. A fine mesh metal screen was placed inside the
caps at both ends of the columns to prevent loss of organic material and clogging.
Masterflex Tygon tubing (1/8”) led from the influent storage container to a peristaltic
pump, where tube fittings were used to connect it to 0.89 mm Tygon pump tubing. The
other end of the pump tubing was also connected to / ” tubing, which was then
1 8
connected to 1/4” Masterflex Tygon tubing. The 1/4” tubing was attached to a male luer
lock elbow fitting coupled with a threaded female luer lock fitting fastened to the plastic
|
Colorado School of Mines
|
18
3.2.2 Column Packing.
Columns were loosely packed with a homogenized reactive mixture of 20% dairy
manure, 15% walnut wood shavings, 10% alfalfa pellets, 5% wetland sediment inoculum,
5% pressed limestone pellets, and 45% silica sand mesh) by dry weight on February
(#8
4, 2002. Dairy manure was collected from the College of Agricultural Sciences at
Colorado State University in February 2002 and stored at 4°C. Walnut wood shavings
were obtained from D. Macalady (Colorado School of Mines) in January 2002. Alfalfa
pellets were purchased from Golden Mill in Golden and processed in a Wiley mill
through a 4 mm sieve. Wetland sediment was obtained from the Big Five mine drainage
treatment wetland by T. Wildeman (Colorado School of Mines) in 1998 and stored at
4°C. Silica sand was obtained from T. Illangasekare (Colorado School of Mines) and
limestone pellets were purchased from Pioneer Sand Company in Golden. The materials
were mixed in a large plastic bag until homogenized. Columns were packed wet with DI
water. Each column contained 520 grams of mixed material, 185 grams dry weight.
Column specifications and packing are summarized in Tables 5.1 and 5.2.
3.2.3 Influent Composition.
DI water was used as the influent for the first three days until mine drainage water
was collected. Mine drainage was obtained from the Dinero Tunnel in Leadville,
Colorado on February , 2002 and was used as the column influent beginning on
6
February 7, 2002. It originally consisted of 4.2 mg/L iron, 42.0 mg/L manganese, 11.0
mg/L zinc, and 122 mg/L sulfur, with pH 6.34 (Table 5.3). Over time the concentration
of metals in the solution decreased due to iron oxidation and presumed sorption onto iron
oxides. The influent mine drainage was fed from a common 5-gallon plastic container at
room temperature and was pumped through the columns at a constant flow rate of 165
ml/day with a twelve-channel Ismatec IPC peristaltic pump. Additional mine drainage
water was stored in 5-gallon plastic containers in a constant temperature room at 4°C.
|
Colorado School of Mines
|
19
After 140 days, the influent composition had changed slightly to 0 mg/L iron, 36.6 mg/L
manganese, 9.8 mg/L zinc, and 125 mg/L sulfur, with pH 5.50.
To determine whether substrate supplements would affect sulfate-reducing activity in
the columns, a pulsing experiment was initiated in which the influent for each column
was supplemented with sodium acetate (pH 6.5) or sodium lactate (pH 5.5) for a period
of 24 hours. In the acetate experiment, Column 1 received a 10 mM supplement in the
mine water, Column 2 received a 5 mM supplement, and Column 3 received a 1 mM
supplement. Influent for each column was fed from a separate container during the 24-
hour pulse and fed from the common container with mine drainage following the pulse.
Approximately three weeks after the acetate pulse, all three columns received a 5 mM
lactate supplement. Influent for all three columns was fed from a common container with
mine drainage and sodium lactate during the 24-hour pulse and fed from the container
with only mine drainage following the pulse.
3.2.4 Effluent Collection and Analysis.
Column effluent was collected in plastic bottles kept on ice that were open to the
atmosphere and changed daily. Effluent pH was raised to approximately pH 10 with 1 M
NaOH in an attempt to trap sulfide in the aqueous phase. Samples were stored in the
refrigerator prior to filtration and analysis. Effluent samples and influent samples were
monitored for sulfur and metals (Perkin Elmer ICP-AES) and pH (Orion probe and
meter). Separate subsamples were acidified to pH 2 using trace metals-grade nitric acid
and filtered through 0.45 fim syringe-tip filters for ICP-AES analysis. Addition of 0.5 M
zinc acetate was used to determine qualitatively whether dissolved sulfide was present in
selected samples. An iodometric titration procedure for sulfide was performed with
selected samples to determine whether dissolved sulfide was present (EPA Method
9034).
|
Colorado School of Mines
|
20
3.3 Results and Discussion
Reactive Mixture Selection. The reactive mixture for the pilot columns was chosen based
on a review of the literature. Effective, long-lasting substrate mixtures to support the
activity of sulfate-reducing bacteria in passive treatment systems have been and continue
to be researched. Experiments have been done with inexpensive, readily available
organic substrates including compost, manure, wood chips, sewage sludge, leaf mulch,
sawdust, and alfalfa (Chang et al. 2000, Gilbert et al. 1999, Cocos et al. 2002, Waybrant et
al. 1998). A full-scale permeable reactive barrier installed at the Nickel Rim mine site in
Ontario, Canada used municipal compost, leaf compost, and wood chips as organic
substrate to support sulfate-reducing activity (Benner et al. 1997). Generally, higher
initial sulfate reduction rates have been observed in reactive mixtures containing a variety
of organic sources and sources with high carbon content (Waybrant et al. 1998).
A study performed by Gilbert et al. evaluated several mixtures in batch using mine
water with neutral pH and concluded that a reasonable substrate composition was 50%
limestone, 25% year-old dairy farm manure, 15% aged sawdust, and 10% alfalfa by
weight (Gilbert et al. 1999). Some alfalfa in the mixture was found to be important in
facilitating bacterial growth, most likely due to its high nitrogen content and availability
of low molecular weight organic acids. Aged dairy farm manure was chosen in the study
for its low carbon to nitrogen ratio and high sorptive capacity, as well as its usefulness as
an inoculum source.
The organic component of the reactive mixture for the pilot columns was based on the
mixture recommended by Gilbert et ah, with the following minor differences: (1) Walnut
wood shavings (15%) were used as a long-term source of carbon and energy that would
slowly degrade over time, ( ) limestone pellets were used to increase alkalinity in the
2
columns and silica sand was added to increase hydraulic conductivity, (3) fresh dairy
manure was used as opposed to year-old dairy manure, and (4) wetland sediment was
chosen as an additional inoculum source with the hypothesis that it would contribute
|
Colorado School of Mines
|
21
bacteria more suited to conditions in a subsurface environment, which is an important
consideration when designing a passive treatment system.
The three pilot columns operated for approximately five months. Results of the pilot
column study showed that the system was successful in terms of sulfate reduction, pH
neutralization, and metal removal. The presence of a black deposit in the columns and
tubing was observed, which can be an indication of sulfate reduction (Chang et al. 2000,
Lyew and Sheppard, 2001). Measurable decreases in sulfur were observed in the effluent
as compared to the influent (Figure 3.2). For purposes of measuring sulfate reduction, it
was assumed that all sulfur measured in the effluent was in the form of sulfate (Lyew and
Sheppard, 1997). Dissolved sulfide was below detection (0.5 mg/L) in the effluent
samples from the pilot columns.
Average sulfate reduction rates for the three pilot columns ranged from 0.76 moles
S Vm3/day after two to three weeks of operation to 0.23 moles S Vm3/day after 133
042 042
days of operation, consistent with sulfate reduction rates in the literature. A study by
Wildeman et al. using a pilot reactor treating mine drainage found sulfate reduction rates
ranging from 0.3 to 2 moles S2' produced/m3/day (Wildeman et al. 1997). Active sulfate
reduction was observed to occur within three weeks of column start-up, also consistent
with previous studies in both batch experiments and continuous flow reactors with acid
mine drainage (Cocos et al. 2002, Lyew et al. 1994, Waybrant et al. 1998). In addition,
distinct stages in the sulfate-reducing activity of the columns over time were observed.
Effluent sulfur was observed to decrease as sulfate reduction began after two weeks, to
plateau at a somewhat steady concentration after approximately three weeks, and finally
to increase as the columns began to decline in their ability to reduce sulfate.
Initial concentrations of iron in the column effluent were as high as 25 mg/L, though
influent iron was 4.2 mg/L (Figure 3.3). The source of the additional iron was possibly
the agglomerated limestone pellets used in the column reactive mixture, which were
found to dissolve readily. Influent iron was completely removed by the columns based
on effluent concentrations.
|
Colorado School of Mines
|
23
Zinc was almost completely removed by the columns, although low concentrations
(0.1 to 0.6 mg/L) were observed in the effluent throughout the study, with slightly higher
concentrations (0.1 to 1.4 mg/L) observed following the acetate and lactate pulses after
140 days of operation (Figure 3.4). Based on the Ksp values for metal sulfides, zinc
sulfides were expected to precipitate most readily, followed by iron sulfides and
manganese sulfides (Machemer and Wildeman, 1992). Iron may have been removed
better than zinc because it was present at a lower concentration and iron precipitation in
the influent storage container caused the iron concentration to further decrease before it
even entered the columns.
As expected, manganese was not removed to a high degree by the columns (Figure
3.5). A high pH, near 10, is typically required to rapidly form insoluble manganese
hydroxide, carbonate, or sulfide precipitates (Clayton et al. 1998, Wildeman and
Updegraff, 1997). Manganese removal is likely to be in the form of sorption to organic
material in the system or MnCOs precipitation (Clayton et al. 1998). Loss of manganese
removal has been shown to occur if pH falls below 7 or if sorption sites become filled
(Clayton et al. 1998). Iron has been found to be more strongly sorbed than zinc and
manganese, indicating competition for sorption sites can occur at high metal
concentrations, with manganese less preferentially sorbed (Machemer and Wildeman,
1992). However, iron levels in this system were relatively low compared to manganese
and zinc concentrations.
|
Colorado School of Mines
|
25
Table 3.1. Observations during pilot column study.
Item Observation Response
Column Packing The mixture was packed Subsequent columns were packed
very loosely, with open much more tightly with reactive
pockets and channels mixture.
observed over time.
Walnut Wood The coiled, corkscrew shape The walnut shavings were
of the walnut shavings processed in a Wiley mill for use
hindered column packing, as in subsequent column mixtures.
they did not compress well.
Limestone Limestone pellets were made Limestone used in subsequent
from agglomerated column mixtures was ground from
powdered limestone and limestone rock and classified
dissolved readily in water, so according to size.
they were not likely to be
retained within the columns.
Their dissolution was also
likely to be the cause of the
increased iron initially
observed in the effluent.
Iron Oxidation in Iron precipitated out of the Influent for the next column study
Influent influent solution and was was bubbled constantly with to
likely to precipitate the other remedy the problem of iron
metals as iron oxides, oxidation so the metals would
causing influent composition remain in solution.
to change over time.
Loss of Fine Fine material was observed Much of the fine material was
Material in Effluent to flow out in the column thought to come from the manure
effluent, especially during in the column mixture, so the next
the first two weeks of set of columns were packed with a
operation, as shown by the mixture containing less manure.
dark brown color of the
effluent.
|
Colorado School of Mines
|
26
After 140 days, when it appeared that sulfate-reducing activity was declining, the
pilot columns were pulsed with acetate (1 mM, 5 mM, and 10 mM) and later lactate (5
mM) to determine if sulfate reduction could be stimulated. An increase in effluent sulfate
concentrations followed the acetate pulses in all three columns, especially visible in
Column 1, which received the largest concentration of acetate (Figure 3.6). Effluent
sulfur levels in Column 1 actually increased above the influent level. Slightly higher
metal concentrations were also observed in the effluent following the acetate pulses.
(The single sulfur data point at day 140 that is higher than the influent level was likely to
be an outlier, while the observed increase beginning at day 144 was clearly represented
by several data points.) Results of the lactate pulses were inconclusive, as no visible
trend was observed (Figure 3.7).
The prominent increase in sulfur in the effluent of the column that received the
greatest concentration of acetate suggests that the observed effect was somehow related
to the acetate concentration. It is possible that the addition of acetate stimulated a
competing acetate-consuming population, such as methanogens, or that sulfate reducers
were inhibited by the acetate. Neither of these effects would account for the observed
increase in sulfur above the influent concentration, however, and would have had to
occur in combination with another effect of the acetate. Another possible explanation is
that suspended metal sulfides flushed from the column were dissolved when the effluent
samples were acidified prior to filtration for ICP-AES analysis. Thus, a mobilization of
sulfur in the effluent samples could have been caused by methods used during the
experiment.
Sulfur in effluent from all three columns exceeded the influent sulfur following the
lactate pulse; however, effluent sulfur was higher than influent sulfur before the lactate
pulse began. Since the columns were not exhibiting sulfate reduction before the lactate
pulses were administered, it may not have been possible to see a response. All three
columns received the same concentration of lactate; however, the effluent sulfur data
following the pulse is very different for each column. Based on results of the pilot
|
Colorado School of Mines
|
29
CHAPTER 4: MINI-COLUMN SUBSTRATE SUPPLEMENT
EXPERIMENT
4.1 Experimental Objectives
A microcosm study was initiated with twelve mini-columns to learn whether a 24-
hour pulse of a substrate supplement through a column would be an effective method to
determine the rate-limiting step(s) in the degradation of organic compounds within the
system as they related to sulfate reduction.
4.2 Experimental Methods
4.2.1 Column Specifications.
Twelve replicate columns were constructed from 40 cm plastic round-bottomed
3
centrifuge tubes. Holes were drilled in the bottom and the cap of each centrifuge tube,
and 1/8” plastic elbow fittings were secured in the holes using epoxy resin. Plastic mesh
screen material was placed in the cap and bottom of each column to prevent loss of
organic material and clogging.
Masterflex Tygon (1/8”) tubing led from the influent container to a peristaltic pump,
where tube fittings connected it to 0.89 mm Tygon pump tubing. The other end of the
pump tubing was also connected to / ” tubing, which attached to the elbow fittings at
1 8
the bottom of each column. Effluent exited each column from the top, where the elbow
fitting in the cap was connected to / ” tubing that led to effluent collection containers.
1 8
|
Colorado School of Mines
|
30
4.2.2 Column Packing.
Columns were packed with a homogenized reactive mixture of 30% walnut wood
shavings, 10% ground brewery waste pellets, 5% dairy manure, 5% wetland sediment
inoculum, 5% crushed limestone rock (between #10 and #20 mesh), and 45% silica sand
mesh), by dry weight. Walnut wood shavings were obtained from D. Macalady
(#8
(Colorado School of Mines) in January 2002. Brewery waste pellets were obtained from
the Coors Brewing Company. Both the walnut wood and the brewery waste pellets were
processed in a Wiley mill through a 4 mm sieve. Dairy manure was collected from the
College of Agricultural Sciences at Colorado State University in February 2002 and
stored at 4°C. Wetland sediment was obtained from the Big Five mine drainage
treatment wetland by T. Wildeman in 1998 and stored at 4°C. Silica sand was obtained
from T. Illangasekare (Colorado School of Mines) and limestone was purchased from
Pioneer Sand Company in Golden. The materials for each column were mixed in
separate plastic bags until homogenized. Columns were packed wet with DI water on
June 21, 2002 with 19 grams of the reactive mixture. Column specifications and packing
are summarized in Table 5.1 and Table 5.2.
4.2.3 Influent Composition.
Influent for the first 20 days consisted of DI water with 540 mg/L sulfate as NazSCL,
pH 5.9. After 20 days, a synthetic mine water was used (pH -6.0, 1400 mg/L sulfate as
Na S and 50 mg/L each of iron as FeSCV 7H 0, manganese as MnSCV H 0, and zinc
2 04 2 2
as ZnSCV 7H 0) with N bubbled into the influent storage container to minimize
2 2
oxidation of iron. Influent composition and flow rates are presented in Table 5.3.
Influent for the columns was fed from a common 5-gallon container at room temperature
and was pumped at a constant flow rate of 24 ml/day throughout the experiment with a
twelve-channel Ismatec IPC peristaltic pump.
Columns were in operation for several weeks to ensure active sulfate reduction and to
allow effluent sulfate to reach a steady concentration. After 33 days, the synthetic mine
|
Colorado School of Mines
|
31
drainage influent for duplicate columns was supplemented with 5 mM of one of the
following for a 24-hour pulse: cellobiose, glucose, sodium lactate (pH 5.9), sodium
acetate (pH 6.5), ammonium nitrate (pH 6.5), or potassium phosphate (pH 6.5). Influent
for each set of duplicate columns was fed from a common -liter container bubbled with
2
N] during the 24-hour pulse. Following the pulse, influent for all twelve columns was
fed from the common 5-gallon container of synthetic mine water.
4.2.4 Effluent Collection and Analysis.
Column effluent was collected in plastic bottles kept on ice that were open to the
atmosphere and changed daily. Effluent pH was raised to approximately pH 10 with 1 M
NaOH in an attempt to trap sulfide in the solution phase. Samples were stored in the
refrigerator prior to filtration and analysis. Influent samples and effluent samples were
collected daily and monitored for sulfur and metals (Perkin Elmer ICP-AES) and pH
(Orion probe and meter). Separate subsamples were acidified to pH 2 using trace metals-
grade nitric acid and filtered through 0.45 fim syringe-tip filters for ICP-AES analysis.
Addition of 0.5 M zinc acetate was used to determine qualitatively whether dissolved
sulfide was present in selected samples. An iodometric titration procedure for sulfide
was performed with selected samples to determine whether dissolved sulfide was present
(EPA Method 9034).
4.3 Results and Discussion
The reactive mixture chosen for the mini-column study differed from the mixture
used for the pilot column study (Table 5.2). During the pilot column study, visual
observations led to the inference that a significant portion of the soluble organic material
in the columns had been flushed out in the effluent during the first several days of
operation. Much of this fine material was thought to have been associated with the dairy
manure component of the reactive mixture. Therefore, the mini-columns were packed
|
Colorado School of Mines
|
32
with a reactive mixture containing less manure and more walnut wood. One of the
experimental objectives was to determine whether the difference in reactive mixture
composition would affect column start-up in terms of sulfate-reducing activity.
The mini-columns were started with influent consisting of 540 mg/L sulfate. After 20
days, when active sulfate reduction had begun, sulfate was increased to 1400 mg/L and
metals were added. Sulfate and metals were increased from the concentrations used for
the pilot columns to resemble more closely concentrations that are common in acid mine
drainage. Sulfate concentration is generally 2-3 orders of magnitude higher than the
metal concentrations in acid mine drainage, in the range of thousands of mg/L (Utgikar et
al. 2003).
The twelve mini-columns were operated for approximately six weeks. The presence
of a black deposit in the columns and tubing was observed, which can be an indication of
sulfate reduction (Chang et al. 2000, Lyew and Sheppard, 2001). Measurable decreases
in sulfur were observed in the effluent as compared to the influent (Figure 4.1). For
purposes of measuring sulfate reduction, it was assumed that all sulfur measured in the
effluent was in the form of sulfate (Lyew and Sheppard, 1997). Dissolved sulfide was
below detection (0.5 mg/L) in the effluent samples from the mini-columns.
The average sulfate reduction rate was approximately 2 moles SCVVmVday during
active sulfate reduction with both influent compositions. This rate is higher than the
sulfate reduction rate of 0.3 moles of S "produced/m3/day that has been used as an
2
estimate for system design purposes; however, that value was based on treatment of mine
water with a pH of 3 to 4, while the water used in this experiment had a pH of
6
(Wildeman et al. 1997). In a study by Wildeman et al., rates of sulfate reduction in a
pilot reactor treating mine drainage were found to reach moles/m3/day during the
2
summer (Wildeman et al. 1997). Effluent sulfur levels began to decline after
approximately 12 days. On day 20, a new influent source was begun with a higher
sulfate concentration and metals. Effluent sulfur exhibited a corresponding increase, but
began decreasing again by day 25.
|
Colorado School of Mines
|
39
Two distinct peaks in the effluent sulfur data were observed immediately following
all but the acetate pulses. It is possible the distinct peaks observed in effluent sulfur data
following the pulses were the result of operational or analytical error. Some of the sulfur
data obtained from the ICP-AES analyses during this time were not reliable, as the sulfur
standards sometimes resulted in values that were elevated by as much as 20 to 30%. The
data plotted were adjusted accordingly, but the samples themselves were never
reanalyzed. Effluent data following the phosphate and glucose pulses exhibited an
increase in sulfur above the influent sulfur value, although it was difficult to discern
whether this apparent mobilization of sulfur species was a real effect of the supplements
or was due to an operational or analytical problem. Operational problems with elevated
pressure from periodic clogs in the fittings as well as from gas produced in the columns
may have contributed to this problem. In addition, the effluent samples were acidified
prior to filtration when preparing them for ICP-AES analysis, which may have released
sulfides from suspended metal sulfide precipitates in the effluent.
As a result of these problems with the mini-columns, systems for gas collection or
release were implemented for subsequent column experiments and effluent samples were
acidified after filtration for ICP-AES analysis to eliminate the possibility of mobilizing
sulfur from metal sulfide precipitates. In addition, it was determined that ICP-AES
analysis for sulfur was not consistently reliable and effluent samples for subsequent
column experiments were instead analyzed for sulfate by ion chromatograph (IC), while
still analyzed for metals by ICP-AES. It was also decided that adding NaOH to the
effluent collection bottles to trap aqueous sulfide in the liquid was no longer necessary, as
dissolved sulfide was below detection in effluent samples from both the pilot column
experiment and the mini-column experiment.
Based on the inconclusive data obtained from the pulsing experiments with the pilot
columns and mini-columns, a modified approach for quantifying microbial activities and
the rate-limiting step(s) was employed for subsequent experiments that involved pilot-
scale columns in correlation with batch analysis of the reactive mixture. It was not
|
Colorado School of Mines
|
40
possible to assess the effects of substrate supplements in column systems, largely due to
their relative complexity. For example, heterogeneities and preferential flow paths could
have resulted in a lack of uniform exposure to the substrate supplements within the
columns and affected the ability to obtain representative samples. It was also difficult to
determine the ideal pulse duration through a column, so as not to significantly change the
microbial community composition within the system.
It was determined that these issues could be minimized by performing batch studies
in serum bottles incubated on a shaker to eliminate variations in exposure to the substrate
supplement by ensuring that all the material was exposed to the same amount of substrate
supplement for the same amount of time. Batch studies would also allow multiple
substrate supplements to be tested at a given time using the same reactive mixture. In
addition, gas analyses were necessary to facilitate the identification of metabolically
active communities within the system at a given time and changes in gas composition
over time could easily be monitored in batch systems.
|
Colorado School of Mines
|
41
CHAPTER 5: BENCH-SCALE COLUMN AND BATCH STUDIES
5.1 Experimental Objectives and Approach
The objectives of the following studies were to (1) develop a method for assessing
microbial activities in anaerobic passive treatment systems, and ( ) apply the method to a
2
column system for the purpose of discerning the rate-limiting step(s) in the degradation
of cellulose-based organic material as they influence sulfate reduction. The following
experimental methods were applied to three different bench-scale studies, referred to as
Column Set 1, Column Set 2, and Column Set 3, each involving a set of replicate
columns in correlation with periodic batch experiments for assessment of microbial
activity by gas analyses. The methods for each column set were generally the same, with
differences noted.
Since gas collection proved difficult with the column systems, changes in gas
composition were measured in batch systems. Gas samples were analyzed throughout the
batch experiments for CO , H S, CH4, and H to quantify the activities stimulated by the
2 2 2
supplements. H S is produced by sulfate reducers, and CH is produced by methanogens,
2 4
while CO is produced throughout the degradation process by fermenters, sulfate
2
reducers, and methanogens and is an indicator of overall activity in the system. H
2
represents a point of competition between sulfate reducers and methanogens (O’Flaherty
et al. 1998, Raskin et al. 1996, Qatibi et al. 1990). Figure 5.1 is a simplified schematic of
carbon flow illustrating the five model compounds used as substrate supplements, along
with the microbial degradation functions targeted by each supplement and the gases
produced.
|
Colorado School of Mines
|
43
5.2 Experimental Methods
5.2.1 Column Specifications.
Eight replicate glass columns, each 5 cm wide and 30 cm tall with four side sampling
ports, were used for all column sets. Sampling ports were sealed with rubber gaskets and
screw-on caps. Plastic caps with rubber o-rings inside fit over each column at the bottom
end and formed a seal. A fine mesh metal screen was placed inside each cap to prevent
loss of organic material and clogging.
Masterflex Tygon tubing (1/8”) led from the influent container to a peristaltic pump,
where tube fittings were used to connect it to 0.89 mm Tygon pump tubing. The other
end of the pump tubing was also connected to / ” tubing, which was then connected to
1 8
1/4” Masterflex Tygon tubing. The 1/4” tubing was attached to a male luer lock elbow
fitting coupled with a threaded female luer lock fitting fastened to the plastic end cap at
the bottom of each column. Rubber stoppers were used in place of plastic caps at the top
of each column. A 1/8” hole was bored into each stopper and a 1/8” glass tube was
inserted. Tygon tubing (1/8”) connected the glass tube to a one-way check valve, which
allowed gas to escape each column, while preventing air from entering.
Effluent exited each column from a side sampling port. A male luer lock tube fitting
was placed into the side port and secured with a rubber o-ring and screw cap. The end of
the tube fitting that protruded from the hole in the screw cap was connected to 1/4”
tubing that led to effluent collection containers. The static water level was maintained
just above the sampling port to allow gas produced in the reactive mixture to exit each
column through the one-way check valve.
Column Set 1. Gas collection bags were connected to the one-way check valves on four
of the columns via 1/8” Tygon tubing. Effluent exited each column from a side sampling
port located approximately cm from the bottom of the column.
22
|
Colorado School of Mines
|
44
Column Set 2. Gas collection bags were not used with any of the columns and gas was
allowed to escape through the one-way check valves. Effluent exited each column from a
side sampling port located approximately cm from the bottom of the column.
22
Column Set 3. Gas collection bags were not used with any of the columns and gas was
allowed to escape through the one-way check valves. Effluent exited each column from a
side sampling port located approximately 15 cm from the bottom of the column. A
schematic of the column setup for Column Set 3 is presented in Figure 5.2 below.
5.2.2 Column Packing.
Columns were packed with 150 grams dry weight of a homogenized reactive mixture
of walnut wood shavings, dairy cattle manure, dried alfalfa, wetland sediment inoculum,
crushed limestone rock (between and mesh), and silica sand mesh).
#10 #20 (#8
Proportions are specified below. Dairy manure was collected from the College of
Agricultural Sciences at Colorado State University and stored at 4°C. Walnut wood
shavings were obtained from D. Macalady (Colorado School of Mines) in January 2002.
Alfalfa pellets were obtained from Golden Mill in Golden. Both the walnut shavings and
the alfalfa pellets were processed in a Wiley mill through a 4 mm sieve. Wetland
sediment was obtained from the Big Five mine drainage treatment wetland by T.
Wildeman (Colorado School of Mines) in 1998 and stored at 4°C. Silica sand was
obtained from T. Illangasekare (Colorado School of Mines). Limestone was purchased
from Pioneer Sand Company in Golden.
The materials for each column were mixed in separate plastic bags until homogenized.
After packing the reactive mixture into the columns, the remaining space in each column
was filled with silica sand. A fine mesh metal screen was used to separate the reactive
mixture and the sand layer in each column and to minimize the loss of organic material in
the flow-through system. Column specifications and packing are summarized in Table
5.1 and Table 5.2.
|
Colorado School of Mines
|
46
Column Set 1. The reactive mixture composition for each column consisted of (dry
weight) 33% walnut wood shavings, 10% dairy cattle manure, 2% alfalfa pellets, 5%
wetland sediment inoculum, 5% crushed limestone rock between #10 and #20 mesh, and
45% mesh silica sand. Dairy manure was collected in February 2002. Columns were
#8
packed wet with DI water on November 13, 2002. The reactive mixture was packed to a
depth approximately 19 cm from the bottom of the column and the remaining space in the
column (approximately cm) was filled with silica sand.
11
Column Set 2. The reactive mixture composition for each column consisted of (dry
weight) 15% walnut wood shavings, 20% dairy cattle manure, 10% dried alfalfa, 5%
wetland sediment inoculum, 5% crushed limestone rock between #10 and #20 mesh, and
45% mesh silica sand. Dairy manure was collected in February 2002. Columns were
#8
packed wet with DI water and 85 ml of 1000 mg/L sodium sulfate solution on January
10, 2003 and allowed to sit for five days. The reactive mixture was packed to a depth
approximately 19 cm from the bottom of the column and the remaining space in the
column (approximately cm) was filled with silica sand.
11
Column Set 3. The reactive mixture composition for each column consisted of (dry
weight) 15% walnut wood shavings, 20% dairy cattle manure, 10% dried alfalfa, 5%
wetland sediment inoculum, 5% crushed limestone rock between #10 and #20 mesh, and
45% mesh silica sand. Fresh dairy manure was collected in February 2003. Columns
#8
were packed wet with DI water on February 11, 2003 and allowed to sit for one day
before starting influent flow. The reactive mixture was packed to a depth approximately
15 cm from the bottom of the column and the remaining space in the column
(approximately 15 cm) was filled with silica sand.
|
Colorado School of Mines
|
47
5.2.3 Influent Composition.
Synthetic mine water was pumped up through the columns at a constant flow rate
with a twelve-channel Ismatec LPC peristaltic pump. Influent composition and flow rates
for Column Sets 1 and 2 are presented in Table 5.3. Influent composition and flow rates
for Column Set 3 are presented in Table 5.4. Synthetic mine drainage was used to ensure
a constant quality of water with each Column Set. Influent was fed from a common 5-
gallon plastic container at room temperature, pumped up through each column bottom,
and removed from a side sampling port.
Column Set 1. Influent flow began on November 14, 2002 with synthetic mine water
(pH -6.0, 1000 mg/L sulfate as Na^SCL and 50 mg/L each of manganese as MnSCV H 0,
2
nickel as NiSCV 6H 0, and zinc as ZnSCV 7H 0) at a constant flow rate of 90 ml/day.
2 2
Influent for each column was removed from a side sampling port located approximately
cm from the bottom of the column, thus traveling through the entire reactive mixture
22
and approximately 3 cm of the sand layer before exiting the columns.
Column Set 2. Influent flow began on January 15, 2003 with synthetic mine water (pH
-6.0, 1000 mg/L sulfate as Na2SCL and 50 mg/L each of manganese as MnSCV H 0,
2
nickel as NiSCV 6H 0, and zinc as ZnSCV 7H 0) at a constant flow rate of 90 ml/day.
2 2
Influent for each column was removed from a side sampling port located approximately
cm from the bottom of the column, thus traveling through the entire reactive mixture
22
and approximately 3 cm of the sand layer before exiting the columns.
Column Set 3. Influent flow began on February 12, 2003 with 1000 mg/L sulfate as
Na2SCV pH -6.0, and was pumped up through the columns at a constant flow rate of 90
ml/day. On day 24, the flow rate was slowed to 30 ml/day and influent for six of the
columns was changed to synthetic mine water (pH -6.0, 1000 mg/L sulfate as Na SO^
2
and 50 mg/L each of manganese as MnSO^ H20 and zinc as ZnSOzr 7H 0), while the
2
|
Colorado School of Mines
|
48
remaining two columns continued to receive the original sulfate influent until they were
sacrificed for batch study on day 27. On day 29, sulfate in the influent for the six
remaining columns was inadvertently increased to 2000 mg/L for a period of days, after
8
which it was decreased again to 1000 mg/L. Influent for each column was removed from
a side sampling port located approximately 15 cm from the bottom of the column, thus
traveling through only the reactive mixture before exiting the columns.
5.2.4 Hydraulic Residence Time.
A ninth replicate column was used to determine the hydraulic residence time of the
columns in Column Set 3 after the flow rate was slowed to 30 ml/day. DI water was
pumped up through the column at the same flow rate as the other columns and effluent
conductivity was monitored and recorded every 30 minutes. When the conductivity had
reached a steady level, salt solution was delivered to the column and effluent conductivity
was recorded every 30 minutes until an increase was observed. The increase in effluent
conductivity occurred after 4.13 days.
5.2.5 Effluent Collection and Analysis.
Effluent from the columns was collected in plastic bottles open to the atmosphere and
was monitored for sulfate (Dionex ICS-90, AS14A column, CO /HCO eluent with flow
3 3
rate 1 ml/min), sulfur and metals (Perkin Elmer ICP-AES), pH (Orion probe and meter),
alkalinity (HACH digital titration, method #8203), and conductivity (YSI 35 probe and
meter). Separate subsamples were filtered through 0.2 [xm syringe-tip filters for IC and
ICP-AES analyses. Samples for ICP-AES analysis were acidified to pH 2 using trace
metals-grade nitric acid. Filtered effluent samples were frozen in 1.7 ml microcentrifuge
tubes for possible future analyses. Modifications to the above suite of analyses are noted
below for specific column setups.
|
Colorado School of Mines
|
49
Column Set 1. Effluent samples from Column Set 1 were not analyzed for sulfur or
metals. Conductivity was monitored over the first month of column operation. Filtered
effluent samples were frozen for possible future analyses, including sulfur and metals.
Column 7 exhibited problems with clogging and leaking and was terminated after
approximately two weeks of operation.
Column Set 2. Effluent samples from Column Set 2 were not analyzed for sulfur, metals,
or alkalinity. Conductivity and pH were monitored over the first three weeks of column
operation. Filtered effluent samples were frozen for possible future analyses, including
sulfur and metals.
5.2.6 Batch Experiments.
An initial batch study was performed with a portion of the reactive mixture used to
pack the columns in each column set, as well as with the reactive mixture from sacrificed
columns in Column Sets 2 and 3.
Methods used for the batch studies were adapted from previously published studies.
Serum bottle sizes, proportions of solid to liquid, and sampling protocols were adapted
from studies that investigated dechlorinating systems (Fennell et al. 2001, He et al. 2002).
The volume of gas and liquid samples removed during these studies was determined by
the requirements of the analytical instruments, such as the GC used for these batch
studies, which required a sample size of 1 to 1.5 ml of gas for analysis.
Column Set 1. An initial batch study was performed with a portion of the reactive
mixture used in the columns. In an anaerobic glovebox, 10 g of the reactive mixture (dry
weight) was placed into each of ten 160 ml serum bottles, along with 85 ml of liquid
(1000 mg/L sulfate as Na S and a 10 mM substrate supplement, with pH ranging from
2 04
6.4 and 7.1, no metals added) previously purged with N2. The serum bottles were sealed
with thick rubber stoppers and aluminum crimp tops. Negative controls did not receive
|
Colorado School of Mines
|
50
substrate supplements, while experimental serum bottles received 10 mM of one of the
following: cellulose (Supelco), glucose (Aldrich), sodium lactate (Sigma), or sodium
acetate (Sigma). Each treatment was duplicated for a total of 10 serum bottles.
The serum bottles were incubated at room temperature on a shaker for 48 hours.
Liquid and gas samples were taken at 0, , 12, 24, 30, 36, and 48 hours. Sterile 23-gauge
6
needles and 3 ml syringes were used for sampling. Approximately 1.5 ml of gas and 3 ml
of liquid were removed from each bottle when sampled. Gas samples were analyzed for
CO], CEL, H]S, H], and O] (Agilent P200 Micro GC with helium carrier gas, Column A:
molesieve 5A 100m, Column B: ppu m). Liquid samples were frozen in 1.7 ml
8
microcentrifuge tubes for possible future analyses.
Column Set 2. An initial batch study was performed with a portion of the reactive
mixture used in the columns. A second batch study was performed beginning on
February 18, 2003 with the reactive mixtures from two columns sacrificed after operating
for 38 days. For all batch studies with mixtures from sacrificed columns, the reactive
mixtures taken from the two columns were combined and homogenized, and
approximately g (dry weight) of the homogenized mixture was placed into each serum
10
bottle for batch study. Beginning with the second batch study of Column Set 2 and
continuing for all subsequent batch studies, the anaerobic glovebox was no longer used to
prepare the serum bottles. After adding the organic material and liquid to the serum
bottles, the liquid was sparged with N] for approximately minutes, and the headspace
20
for approximately 7 minutes, before sealing the bottles with rubber stoppers and
aluminum crimp tops.
Batch studies of Column Set 2 and all subsequent batch studies included 10 mM
cellobiose (Sigma) as an additional substrate supplement, for a total of 12 serum bottles.
The serum bottles were incubated at room temperature on a shaker for 60 hours. Liquid
and gas samples were taken at 0, , 12, 24, 30, 36, 48, 54, and 60 hours. Needles and
6
syringes were flushed with N] before sampling to ensure that air was not introduced into
|
Colorado School of Mines
|
51
the serum bottles. Luer-lock mininert syringe valves were attached to the syringes used
for gas sampling and all gas samples were analyzed immediately. Liquid samples were
frozen in 1.7 ml microcentrifuge tubes for possible future analyses.
Column Set 3. An initial batch study was performed with a portion of the reactive
mixture used in the columns. A second batch study was performed beginning on 3-10-03
with the reactive mixtures from two sacrificed columns after operating for 27 days. A
third batch study was performed beginning on 3-24-03 with the reactive mixtures from
two sacrificed columns after operating for 41 days. A fourth batch study was performed
beginning on 5-21-03 with the reactive mixtures from two sacrificed columns after
operating for 99 days.
Each liquid sample taken during batch studies of Column Set 3 was split between two
1.7 ml microcentrifuge tubes, with 1.5 ml added to each tube. One of the two tubes
contained 100 /ft of 0.1 M zinc acetate to trap any dissolved sulfide in the sample (Spear,
1999). The liquid samples were frozen for future analyses of sulfate (Dionex ICS-90)
and sulfur and metals (Perkin Elmer ICP-AES). Liquid samples taken at 0, 36, and 60
hours during the first three batch studies were thawed and spun in a microcentrifuge at
maximum speed for three minutes. The supernatant from each sample was filtered
through a 0.2 fim filter into a clean microcentrifuge tube. The samples that received zinc
acetate were analyzed for sulfate (Dionex ICS-90), while the samples without zinc
acetate were analyzed for sulfur and metals (Perkin Elmer ICP-AES). Samples for ICP-
AES analysis were acidified to pH 2 using trace metals-grade nitric acid.
5.2.7 Column Set Termination.
Column Set 1 was terminated after 55 days due to its failure to achieve active sulfate
reduction. Column Set 2 was terminated after 38 days for the same reason. The final
batch study of Column Set 3 took place with two columns sacrificed after 99 days of
operation.
|
Colorado School of Mines
|
61
5.3.2 Batch Analysis of Activity.
Five substrate supplements of central importance in cellulose degradation and sulfate
reduction, based on the schematic of key carbon transformations presented in Figure 1.1
and illustrated in Figure 5.1, were added in batch experiments to the reactive mixture
from sacrificed columns as a way to probe important microbial activities. The substrate
supplements chosen, cellulose, cellobiose, glucose, lactate, and acetate, ranged from a
complex polymer to simple organic acids and each targeted a distinct microbial function
at a specific step in the degradation process of cellulose, which is typically a predominant
component of mixtures used in passive treatment systems.
As described in 5.1 Experimental Objectives and Approach, changes in activities
following the addition of supplements were quantified by gas analyses for CO , H S,
2 2
CH4, and H2 to provide information about the rate-limiting step(s) in the degradation of
cellulose-based organic material as they relate to sulfate reduction. If a specific substrate
supplement was found to cause a significant decrease in sulfate and/or increase in
headspace H S, concentrations of that particular component would be present in limited
2
amounts within the system. Thus, supplements that resulted in the highest response
would be downstream of the rate-limiting step, while supplements that resulted in the
lowest response would be either at or upstream of the rate-limiting step.
H was below detection in all samples, as the GC was unable to detect visible H
2 2
peaks in the samples. The lowest standard measured on the GC was 0.01% H by mole.
2
The failure to detect H was likely to have been due to rapid consumption by sulfate
2
reducers and methanogens, which compete for H as an electron donor (Chapter 2,
2
Section 2.3). H is immediately consumed by sulfate reducers and methanogens, only
2
accumulating when sulfate reducers or methanogens are inhibited (Brock and Madigan,
1991). A small amount of O was present in the headspace gas initially, but decreased
2
between the 0 h sampling point and the h sampling point. Initial concentrations of O
6 2
during the batch studies averaged 3%, dropping to approximately 1% after h and 0.7%
6
|
Colorado School of Mines
|
62
after 12 h. The remainder of the headspace gas at the 0 h sampling point was composed
of N , which was used to sparge the samples before the serum bottles were sealed.
2
Column Set 1. An initial batch study for Column Set 1 was performed with a portion of
the reactive mixture used to construct the columns. The initial batch experiment showed
that glucose stimulated increased CO production (Figure 5.11). The CO production in
2 2
the samples that received cellulose, lactate, and acetate was very similar to production in
the samples that received no substrate supplement, indicating that they had no effect on
CO production. By 48 h the samples from batches that received glucose measured
2
19.3% CO , while the others ranged from . % to 8.3% CO . H S results were difficult
2 6 8 2 2
to interpret and showed no discernible trends, except for possible stimulation by glucose
(Figure 5.12). No methane production was observed for the initial mixture under any of
the conditions tested (Figure 5.13).
—■— No Supplement 1
- No Supplement 2
—■—Cellulose 1
—A—Cellulose 2
- —■—Glucose 1
A Glucose 2
—m— Lactate 1
_ —A— Lactate 2
—m— Acetate 1
—A— Acetate 2
70
Time (h)
Figure 5.11. Headspace CO in the initial batch study of the reactive mixture for
2
Column Set 1.
|
Colorado School of Mines
|
64
Column Set 2. Column Set 2 batch studies were performed with a portion of the reactive
mixture used to construct the columns and later with the reactive mixture from a pair of
columns after 38 days of operation when Column Set 2 was terminated. Overall activity
in the initial mixture, as indicated by CO production, was stimulated by cellobiose and
2
possibly glucose and lactate as compared to the samples that received no supplement
(Figure 5.14). The initial batch study showed very little stimulation of sulfate-reducing
activity, with the exception of one of the samples that received acetate (Figure 5.15).
production with no supplement was higher than production with no supplement
CH4 H 2S
(Figure 5.16). It appears that cellulose, cellobiose, glucose, and lactate may have had
inhibitory effects on methanogens because they resulted in less CFL* production than the
controls. Acetate was the only supplement that increased and production above
CH4 H 2S
the controls, but this only occurred in one of the replicates.
70
No Supplement 1
60 No Supplement 2
Cellulose 1
50 Cellulose 2
Cellobiose 1
<N
40 Cellobiose 2
O
O Glucose 1
55 30 Glucose 2
Lactate 1
20 Lactate 2
Acetate 1
10
Acetate 2
0
10 20 30 40 50 60 70
Time (h)
Figure 5.14. Headspace CO in the initial batch study of the reactive mixture for
2
Column Set 2.
|
Colorado School of Mines
|
66
The Column Set 2 batch study performed with the reactive mixture after 38 days
showed decreased overall activity, as measured by CO production, compared to the
2
initial batch study (Figure 5.17). In this second batch study, CO production was not
2
aftected by cellulose, but was stimulated greatly by the substrates immediately
downstream of cellulose hydrolysis, namely cellobiose and glucose, and to a lesser extent
by lactate. The reactive mixture taken from the sacrificed columns did not produce any
H S in batch experiments without the addition of a substrate supplement, but lactate,
2
cellobiose, and glucose did stimulate sulfate-reducing activity (Figure 5.18).
Methanogens were stimulated by lactate and acetate and inhibited by cellobiose and
glucose (Figure 5.19). Cellulose again had no effect.
70
No Supplement 1
60 No Supplement 2
Cellulose 1
Cellulose 2
Cellobiose 1
8 Cellobiose 2
" Glucose 1
5? 30 Glucose2
Lactate 1
20
Lactate 2
Acetate 1
10
Acetate 2
10 20 30 40 50 60 70
Time (h)
Figure 5.17. Headspace CO in the batch study of Column Set 2 after 38 days.
2
|
Colorado School of Mines
|
68
Column Set 3. Column Set 3 batch studies were performed using a portion of the initial
reactive mixture used to construct the columns and later using mixtures from sacrificed
columns. The first pair of columns was sacrificed for batch study after 27 days of
operation when they were actively reducing sulfate. The second pair of columns was
sacrificed for batch study after 41 days of operation when active sulfate reduction was
still occurring at the same rate as day 27. The third pair of columns was sacrificed for
batch study after 99 days when sulfate reduction in the columns had declined. Although
four columns were still operating. Columns 3 and 4 were sacrificed because they had
both declined in their ability to reduce sulfate compared to Columns 5 and (Figure 5.5).
6
Column Set 3 data is presented below in Figures 5.20 through 5.31.
Cellobiose and glucose again stimulated the greatest CO production in the initial
2
mixture for Column Set 3, which was stimulated to a lesser degree by lactate and
cellulose (Figure 5.20). Very little sulfate-reducing activity and no methane production
were exhibited in the initial mixture (Figure 5.21 and Figure 5.22).
No Supplement 1
No Supplement 2
Cellulose 1
Cellulose 2
Cellobiose 1
Cellobiose 2
Glucose 1
Glucose 2
Lactate 1
Lactate 2
Acetate 1
Acetate 2
30 40
Time (h)
Figure 5.20. Headspace CO in the initial batch study of the reactive mixture for
2
Column Set 3.
|
Colorado School of Mines
|
76
5.4 Discussion
5.4.1 Column Performance.
Sulfate Reduction. The relatively low sulfate-reducing activity observed in Column Set 1
was thought to be attributable to the reactive mixture composition, which contained %
2
alfalfa. Alfalfa has been found to aid in system start-up (Gilbert et al. 1999). As a result,
the reactive mixture for Column Set 2 was modified to consist of the same mixture
composition as that used for the pilot columns, which contained 10% alfalfa. While
higher sulfate-reducing activity was observed with Column Set 2, it still failed to exhibit
an initial sulfate reduction rate as high as that found in the pilot and mini-columns. It was
determined that the influent composition may have been the problem. Both Column Set
1 and Column Set 2 were fed influent that included 50 mg/L nickel. Iron, zinc and
manganese were used at concentrations of 50 mg/L each during the mini-column
experiment with no inhibitory effects observed; however, this was the first experiment in
which nickel was used in the influent composition. The nickel may have caused
inhibition of sulfate-reducing activity directly, or inhibited a group upstream of sulfate
reducers that function to degrade complex compounds to simple ones used by sulfate
reducers.
Soluble heavy metals can inhibit bacteria by deactivating enzymes, denaturing
proteins, or competing with essential trace cations, resulting in reduced activity or
mortality (Utgikar et al. 2003). The production of sulfide protects sulfate-reducing
bacteria from the effects of soluble heavy metals because it readily binds with most
metals to form metal sulfides (Lyew and Sheppard, 1997). If the metal load to the system
exceeds production of sulfide by sulfate-reducing bacteria, the bacteria can be adversely
affected (Utgikar et al. 2003).
A 1994 study performed by Hao et al. found that nickel was toxic to a mixed culture
of sulfate-reducing bacteria at concentrations between 10 and 20 mg/L (Utgikar et al.
|
Colorado School of Mines
|
77
2003). However, a study performed by Hammack and Edenbom in 1992 found
significant nickel removal rates in compost-based sulfate-reducing columns operated with
influent nickel concentrations of 50 to 1000 mg/L and the maximum sulfate reduction
rates were found to be higher with increasing nickel concentration up to 500 mg/L
(Hammack and Edenbom, 1992). In a study by Waybrant et al., batch reactors were
spiked with 480 mg/L nickel and exhibited almost complete removal with no effect on
sulfate-reducing activity (Waybrant et al. 1998). Thus, nickel did not appear to be
inhibitory to sulfate-reducing bacteria in these studies. A study performed by Fortin et al.
found bacteria isolated from a mixed bacterial culture originating from a metal refining
plant produced a nickel protein, induced by high nickel concentrations, which functioned
to complex the nickel and increase their tolerance to it (Fortin et al. 1994). However,
nickel-resistant bacteria are uncommon in environments where nickel is absent in
significant amounts (Brock and Madigan, 1991).
Two variables differed between Column Set 2, with low initial sulfate reduction, and
Column Set 3, with active initial sulfate reduction: (a) the manure used in the reactive
mixture; and (b) the presence or absence of nickel in the influent water. Column Set 3
was assembled using the same reactive mixture as used for Column Set 2, except that
fresh manure was used in place of the dairy manure collected in February 2002, which
had been stored at 4°C for one year. No metals were added in the influent until sulfate
reduction was occurring, and nickel was no longer used as a component of the influent.
A comparison of the initial batch studies for each of these column sets shows that overall
activity was higher in the mixture using fresh manure (Figure 5.19), but that the stored
manure was still active (Figure 5.13), as evidenced by its use in both the pilot columns
and mini-columns with no start-up problems. This suggests that nickel, or the
combination of nickel with manganese and zinc, was responsible for adversely affecting
sulfate-reducing activity in Column Sets 1 and 2.
|
Colorado School of Mines
|
78
Metal Removal. Zinc was completely removed throughout the lifetime of Column Set 3.
A column study by Lyew and Sheppard using synthetic mine drainage influent with
aluminum, copper, iron, manganese, and zinc found that the process of sulfate reduction
appeared to be most effective in removal of zinc (Lyew and Sheppard, 1997).
Manganese was removed from 50 mg/L in the influent to 25 mg/L in effluent during the
column study. This removal was greater than observed with the pilot and mini-columns
and may have been related to the higher effluent pH (up to 8.5) achieved with Column
Set 3. A study by Hammack et al. found up to 92% initial manganese removal when
effluent pH was , decreasing to 32% removal when pH was below 5, and concluded that
8
the high removal observed at pH was due to precipitation in the form of MnCOg
8
(Hammack et al. 1994). Anaerobic manganese removal in this study would have been
most likely to occur by sorption to column material or MnCOg precipitation, as formation
of stable manganese sulfides typically requires pH near 10 (Clayton et al. 1998).
Decreased efficiency in manganese removal has been shown to occur, for example, if pH
falls below 7 or if sorption sites become filled (Clayton et al. 1998).
Alkalinity and pH. Based on the stoichiometries presented in Chapter 1, alkalinity would
have been expected to increase as effluent sulfate decreased. A study by Wildeman et al.
(Wildeman et al. 1997) found that sulfate reduction could be tracked by changes in
sulfate, alkalinity, or sulfide in the water, with mole of sulfate reduced generating
1
approximately 2 moles of alkalinity and 1 mole of sulfide. The observed effluent
alkalinities for Column Sets 1 and 3 were consistent with alkalinity produced by sulfate
reduction.
Average alkalinity generated in Column Set 1 from days 11 through 46 was 270 mg/L
(2.7 mM), while average sulfate reduction was 112 mg/L (1.16 mM). 1.16 mM of sulfate
reduced would be expected to result in 2.3 mM of alkalinity generated. Thus, 17% more
alkalinity was generated by the columns than could be accounted for based on sulfate
reduction alone. The additional alkalinity could have been generated by the column
|
Colorado School of Mines
|
79
mixture, which included limestone. Alkalinity was not monitored for Column Set 2;
however, effluent pH was observed to increase slightly. Column Set 3 exhibited the
highest effluent pH, increasing up to pH 8.5.
Alkalinity for Column Set 3 was also much higher than alkalinity generated by
Column Set 1, never decreasing below 600 mg CaCOg/L. Column Set 3 alkalinity began
to increase after approximately days, which correlated with increased sulfate
21
reduction. In addition, Column 5, which exhibited the greatest sulfate-reducing activity,
also generated the greatest alkalinity. Average alkalinity generated in Column Set 3 from
days 13 to 76 was mg/L (8.65 mM), while average sulfate reduction for days 13
866
through 30 and days 57 through 76 was 506 mg/L (5.27 mM). The actual sulfate
concentrations in the columns from days 31 through 56 were unknown. 5.27 mM of
sulfate reduced would be expected to result in 10.5 mM of alkalinity generated. Thus, the
average alkalinity observed accounts for 82% of the alkalinity that could have been
produced according to the sulfate reduction observed. The remaining alkalinity could
have been consumed during formation of carbonate precipitates in the columns. In
addition, analytical error in the sulfate analyses with the IC could have been up to 10%,
while the alkalinity titrations relied on a color endpoint, which could also have
represented a source of analytical error.
Conductivity. The high initial conductivities observed in the effluent of both column sets
corresponded to the large amounts of soluble material that were observed to wash out of
the columns during the first two weeks of operation. In addition, Column Set 3 displayed
an increase in effluent conductivity that corresponded to the increase in effluent sulfate at
day 29.
|
Colorado School of Mines
|
80
5.4.2 Batch Analysis of Activity.
Batch Experiments in the Context of Previous Studies. Previous research has focused
primarily on the activity of sulfate-reducing bacteria in anaerobic passive treatment
systems (Cocos et al. 2002, Waybrant et al. 1998, Dvorak et al. 1992, Tsukamoto et al.
1999). Many studies have focused on solution phase measurements of sulfate and
dissolved sulfide to determine the activity of sulfate reducers and have not monitored the
activities of other microbial groups or measured other gases produced (Waybrant et al
2002, Wildeman et al, 1997, Waybrant et al. 1998). For example, a batch study
performed by Reynolds et al. used substrate from a treatment wetland with sodium lactate
and hay extract as nutrient amendments and monitored aqueous sulfate concentration and
sulfide production (Reynolds et al. 1991). A batch study of reactive mixtures by
Waybrant et al. vented the reaction flasks to allow gas to escape and monitored the liquid
for sulfate-reducing activity (Waybrant et al. 1998).
The approach of using substrate supplements and analyzing the gases produced to
probe degradation activities has been used in several fields, including experiments for
treatment of sewage sludge and wastewater, as well as landfill applications, generally to
understand the interactions between sulfate reducers and methanogens (Harada et al.
1994, Li et al. 1996, Qatibi et al. 1990). A 1985 study of sulfate-reducing activity in salt
marsh sediments investigated the effects of substrate supplements on sulfate reduction
and H] production and consumption using numerous supplements, including the ones
used in these experiments (Dicker and Smith, 1985).
Column Set 1. The failure of cellulose in stimulating CO production, compared to the
2
increased production observed with glucose, which is produced from cellulose
degradation, indicates that the rate-limiting step(s) in the degradation process were
upstream of glucose. Based on this result, cellobiose was included as one of the substrate
supplements in subsequent experiments to further probe the rate-limiting step, since
|
Colorado School of Mines
|
81
cellobiose is between cellulose and glucose in the degradation process. It was important
to determine which supplements resulted in the most rapid rates of gas production, as the
rate-limiting step in degradation would have to be upstream of the degradation processes
stimulated by those supplements. Gas samples collected at night during this batch study
were kept until the next morning when they could be analyzed on the GC. This may have
affected the H S data, which exhibited a great degree of variability and were very
2
difficult to interpret.
Column Set 2. Stimulation of overall activity in the initial batch study was seen with
cellobiose and glucose, while no effect was seen with the addition of cellulose. This
suggests that the rate-limiting step(s) lay between cellulose and cellobiose in the
degradation of cellulose. The inhibition of methanogens by cellulose, cellobiose,
glucose, and lactate could have been explained by possible competition for nutrients or
electron donors by another anaerobic population that was stimulated by those substrates.
Acetate was the only supplement that increased and H S production above the
CH4 2
control. Since this only occurred in one of the duplicate samples that received acetate, it
is difficult to determine whether the observed increases were actual effects of the acetate.
It is possible that something happened to one of the two acetate batch samples to cause
the replicates to have behaved so differently with acetate addition. Perhaps the bottle that
exhibited the stimulatory effects with acetate received a portion of reactive mixture that
was not representative of the overall column mixture received in the other bottles.
The substantial decline in overall activity after 38 days of column operation may have
indicated that much of the easily consumed, labile carbon in the reactive mixture had
been degraded or had been transported out of the column in the effluent. However, the
composition of the microbial community present in the mixture initially would have
changed after adjusting to anaerobic column conditions with sulfate and metals, and this
change in community composition was likely to have been responsible for the observed
|
Colorado School of Mines
|
82
difference in activities between the initial mixture and the mixture taken from sacrificed
columns.
A methanogen population was active in the column mixture after 38 days. Again, the
apparent inhibition of methanogens by cellobiose and glucose could be explained by
possible competition for nutrients or electron donors by another anaerobic population that
was stimulated by cellobiose and glucose. Sulfate-reducing activity was only observed
when substrate supplements were added. As Column Set 2 did not exhibit active initial
sulfate reduction while operating, this result was not surprising. The fact that sulfate
reduction could be stimulated in the reactive mixture from Column Set 2 signifies that the
population was present; however, it was apparently not receiving the necessary carbon
source to be active. Lactate stimulated sulfate reducers directly, while cellobiose and
glucose stimulated sulfate reducers indirectly. Cellobiose and glucose are degradation
products of cellulose that cannot be directly used by sulfate reducers for carbon and
energy; however, their stimulatory effect on sulfate reduction suggests that degradation to
simpler substrates used by sulfate reducers proceeds rapidly enough to show a
stimulatory effect in the TLS measurements.
Stimulation of sulfate reduction may also have been due to the fact that metals were
not included in the solutions used in the batch studies and that the nickel that had been
added to the column mixture in the influent may have been diluted to levels that were no
longer toxic. However, the fact that the control samples did not exhibit sulfate-reducing
activity indicates that dilution of nickel was probably not the reason for the observed
activity.
Column Set 3. The greatest stimulation of overall activity in the initial batch study was
seen with cellobiose and glucose. Sulfate reduction was very low initially and no
methane production was observed during the initial batch study.
The batch study conducted on day 27 showed that overall activity in the reactive
mixture had decreased significantly, further supporting the hypothesis that labile organic
|
Colorado School of Mines
|
83
carbon may have diminished greatly. The effectiveness of passive mine drainage
treatment systems declines as easily degraded carbon sources are rapidly consumed
(Tsukamoto et al. 1999). However, as discussed above, the microbial community
composition in the mixture initially would have differed from the community
composition after adjusting to anaerobic column conditions with sulfate and metals, and
was likely to have had an important role in the observed difference in activities between
the initial mixture and the mixture taken from sacrificed columns. CO production,
2
reflecting overall carbon-degrading activity, was stimulated most by addition of glucose
and cellobiose. Sulfate-reducing activity had increased after 27 days, consistent with the
active sulfate reduction observed in the columns, and was most stimulated by lactate,
glucose, and cellobiose. Lactate is known to be a preferred substrate for several groups
of sulfate-reducing bacteria and has frequently been used to stimulate the activity of
sulfate reducers in treatment systems (Tsukamoto and Miller, 1999, Dvorak et al. 1992).
The fact that cellobiose and glucose stimulated sulfate reduction, while cellulose had no
effect, indicates that a rate-limiting step existed in the hydrolysis of cellulose to
cellobiose, suggesting that degradation to simpler substrates from cellobiose proceeds
rapidly enough to show a stimulatory effect in the H S measurements. Methanogen
2
activity was observed after 27 days, with no significant differences in CH production
4
between the supplements.
The columns sacrificed after 41 days were actively reducing sulfate and had been
exposed to metals for approximately two weeks, while the columns sacrificed at day 27
had not yet received metals in the influent. CO production and H S production remained
2 2
very similar to the production observed in the columns sacrificed after 27 days,
suggesting that the addition of metals had not affected activity. A clear rate-limiting step
was again observed between cellulose and cellobiose. After 41 days, methanogen
activity had increased slightly and was stimulated most by acetate.
The columns sacrificed after 99 days of operation did not reduce sulfate as actively as
previous columns at the time of their batch studies. This was reflected in the batch study
|
Colorado School of Mines
|
84
data, as CO production had decreased and H S production with no supplement had
2 2
decreased. Again, a very clear rate-limiting step was visible between cellulose and
cellobiose. Addition of lactate, glucose, and cellobiose stimulated sulfate-reducing
activity to nearly the same levels observed in the previous two batch studies, suggesting
that the population had the potential to be active, but was lacking the proper substrates.
The fact that CH production was stimulated by acetate, while H S production was not,
4 2
indicates that the populations were probably not in competition for organic substrates and
were more likely to have been competing for H , which has been well documented in
2
previous studies and is described in Chapter 2.
Results of the batch studies were very similar to results reported in a 1985 study,
which investigated sulfate-reducing activity in salt marsh sediments. The study examined
several substrate supplements and reported that cellobiose and glucose caused the
greatest increases in sulfate-reducing activity, while no effect was seen with cellulose or
acetate (Dicker and Smith, 1985). To further test the rate of degradation of cellulose in
the study, the amount of cellulose added to the systems was increased, as well as the
incubation time, yet neither of these variables affected cellulose degradation, leading to
the conclusion that anaerobic degradation of cellulose was a very slow process in that
environment (Dicker and Smith, 1985). In an actual passive treatment system the rate-
limiting step is likely to be more pronounced. The cellulose used in these experiments
was in a pure form, but as described in Chapter 2, cellulose is usually associated with
lignin, which causes the cellulose to be less available for degradation.
Comparison of Initial Mixture Batch Studies. A comparison of the initial mixtures gives
insight into the importance of reactive mixture composition for system performance. The
mixture used in Column Set 2 exhibited higher overall microbial activity, as measured by
CO production when no supplement was added, than the mixture used in Column Set 1.
2
The reactive mixture for Column Set 2 contained a larger percentage of manure, which
may have had a higher concentration of microorganisms, and a larger percentage of
|
Colorado School of Mines
|
85
alfalfa, which has a high percentage of easily degraded soluble organic material and
nitrogen. Column Set 3 showed even higher overall activity in controls than Column Set
2. Though the mixture composition was the same for both column sets, fresh dairy
manure was used in Column Set 3, as opposed to year-old dairy manure used in Column
Set 2. The fresh manure may have contained a higher concentration of active
microorganisms and/or a higher percentage of easily degraded soluble organic material
than the older manure.
Sulfate-reducing activity was relatively low in the initial mixtures of all column sets,
which was not surprising, as the mixtures had not been exposed to sulfate until the batch
study. The initial mixture from Column Set 2 exhibited the presence of a potentially
competing methanogen population, while no methane production was observed for the
initial mixtures from Column Sets lor 3. It is possible that a methanogen population
could have been established in the manure and grown more active between November,
2002, when Column Set 1 was started and January, 2003, when Column Set 2 was
started.
Comparison of Batch Studies of First Sacrificed Columns (Column Sets 2 and 3). CO
2
production, and thus overall activity, was significantly reduced compared to initial batch
studies in both Column Set 2 and Column Set 3 after 38 and 27 days respectively for all
conditions tested. Headspace CO with no supplement dropped from 15.8% initially to
2
2.5% for Column Set 2, and from 33.1% initially to 2.1% for Column Set 3 at 60 h.
Again, cellobiose and glucose always stimulated the greatest CO production, suggesting
2
that a rate-limiting step existed between cellulose and cellobiose in the degradation of
cellulose organic material.
Despite the lower rate of sulfate reduction observed in Column Set 2, sulfate-reducing
activity was at a comparable level to that of Column Set 3 when stimulated with lactate.
Thus, given an optimal carbon source, the sulfate reducers in Column Set 2 were as
active as the sulfate reducers in Column Set 3. This suggests that either the pathway
|
Colorado School of Mines
|
90
Table 5.5. Fermentation and respiration pathways for substrate supplements. (Adapted
from Gottschalk, 1986, and Rittman and McCarty, 2001)
Degradation Pathway: Reaction C02
Glucose and Glucose Yield per
Polymers Glucose
Hydrolysis Cellulose Cellobiose 0
Hydrolysis Cellobiose A 2 Glucose 0
Glucose Fermentation Glucose -> Lactate + C02 1
Ethanol Fermentation Glucose -> 2 Ethanol + 2 C02 2
Homofermentative Glucose -> 2 Lactate 0
Pathway
Heterofermentative Glucose A Lactate + Ethanol + C02 1
Pathway
Bifidum Pathway 2 Glucose -> 3 Acetate + 2 Lactate 0
Glucose Fermentation Glucose -> 3 Acetate + 3 H+ 0
Embden-Meyerhof- Glucose -> 2 C02 + 2H2 + Butyrate 2
Parnas Pathway
Acetone-Butanol 2 Glucose -> 4H2 + 5C02 + Acetone + Butanol 5
Fermentation
Mixed Acid Fermentation Glucose -> Succinate + Lactate + Formate + C02 1
+ H2 + Acetate + Ethanol
Butanediol Fermentation Glucose -> Lactate + Ethanol + Formate + 3 C02 3
+ H2 + Butanediol
Propionate Fermentation 3/2 Glucose -> 2 Propionate + Acetate + C02 2/3
Aerobic Respiration Glucose + 602 6C02 + 6H20 6
Denitrification Glucose + 4.8N03" + 4.8H+ ^ 6C02 + 8.4 H20 + 6
2.4N2
Degradation Pathway: Reaction C02 and H2S
Lactate Yield per
Lactate
Denitrification 3.4 H+ + 2.4 N03' + Lactate A 1.2 N2 + 4.2H20 + 3 C02
3 C02
Sulfate Reduction 2 Lactate + H2SG^ 2 Acetate + H2S + 2C02 + 1 C02 and 1/2
2H20 H2S
Acrylate Fermentation 3 Lactate 2 Propionate + Acetate + C02 + H20 1/3 C02
Pathway
Degradation Pathway: Reaction C02 and CH4
Acetate Yield per
Acetate
Ethanol-Acetate 1.2 Ethanol + 0.8 Acetate -> Butyrate + 0.4 H2 0
Fermentation
Aceticlastic Acetate -> CH4 + C02 1 C02 and
Methanogenesis 1CH4
Sulfate Reduction Acetate + H2SG^ A H2S + 2C02 + 2H2O 2 C02 and
1H2S
Methanogenesis C02 + 4H2 A CH4 + 2H20 1 CH4 per C02
|
Colorado School of Mines
|
98
0.22
- 0.2 Cellulose 1
g "g 0.18
—Cellulose 2
5 | 0.16
2 g 0.14
Cellobiose 1
%
1 °-12
X ^ 0.1 Cellobiose 2
g w 0.08
| § 0.06 41—Glucose 1
« 3 0.04
-A—Glucose 2
g ° 0.02
0
0 10 20 30 40 50 60 70
Time (h)
Figure 5.39. Normalized production of H S in the batch study of Column Set 3 after
2
99 days. Note: Each time point represents total moles summed from 0 h.
5.4.3 Discussion of Batch Study Method.
The major objectives of these experiments have been (1) to develop a method for
assessing microbial activities in anaerobic passive treatment systems, and ( ) to apply the
2
method to a column system for the purpose of discerning the rate-limiting step(s) in the
degradation of cellulose-based organic material as they influence sulfate reduction. The
method appears to be useful in quantifying the activities of sulfate reducers and
methanogens and providing information about rate-limiting steps in the degradation of
organic material within the column system. However, many considerations can be
addressed in refining the method.
|
Colorado School of Mines
|
100
reduction greater than 100 mg/L (10%) may have been measured using the IC, but the
other samples that resulted in lower levels of sulfate reduction may not have been
measured reliably. Thus, it would be difficult to compare the samples that received
substrate supplements to the samples that received no supplement to determine the
amount of sulfate reduction that was due to each supplement.
Gas Analyses. Gas analyses have proven to be a very sensitive tool with which to
measure overall activity in the system and quantify the activities of sulfate reducers and
methanogens as headspace gas composition in the serum bottles changes over time.
Quantifying important activities associated with degradation of cellulose-based organic
material provided information about the flow of carbon in the system, and the efficiency
with which it was proceeding in a direction that supported the desired activity, sulfate
reduction. Results of the gas analyses in batch experiments were correlated to
performance of the column system in terms of sulfate-reducing activity.
5.4.4 Conclusions and Implications for Passive Mine Drainage Treatment Systems.
The following is a brief summary of conclusions drawn from this study, followed by
suggestions for future research.
Gas Analyses. Gas analyses were very useful in quantifying microbial activities and
identifying the rate-limiting step(s) in the degradation of cellulose-based organic material
in the experimental system as they related to sulfate reduction.
Rate-Limiting Step. This research hypothesized that system performance was limited by
one or more microbial activities that function as rate-limiting steps in generating
substrates for sulfate-reducing bacteria. Data in this research supported limitation of
sulfate reduction by degradation of cellulose in cellulose-based organic carbon substrates.
|
Colorado School of Mines
|
102
carbon sources are required for passive treatment systems, as cellulosic wastes alone,
which are commonly used, would not be capable of sustaining sulfate-reducing bacteria
(Cocos et al. ).
2002
Alternatively, or in addition to the above approach, the data presented may be used to
inform efforts to design passive treatment systems of the appropriate size to sustain a
desired level of sulfate reduction. Larger passive treatment systems may not be ideal, but
“additional capacity in treatment systems can compensate for reduced biological activity”
(Hammack and Edenbom, 1992). Since sulfate reducers inhabit the surfaces of particles,
larger surface area would be conducive to greater populations of sulfate-reducing
bacteria, and systems could be designed with this in mind by using the smallest particle
sizes practical, particularly for the gravel added to the system (Lyew and Sheppard,
1997).
Most studies of sulfate-reducing passive treatment systems, including this one, have
been conducted at room temperature, which is within the optimal temperature range for
sulfate-reducing microbial activity. However, subsurface temperatures at some field sites
can be in the range of 5-10°C. Cold temperatures decrease biological and chemical
reaction rates (Benner et al. 1997). Sulfate reduction rates determined at room
temperature are not directly transferable to lower-temperature field conditions (Waybrant
et al. 1998). Therefore, the effects of cold temperatures on passive treatment systems are
in need of further characterization.
|
Colorado School of Mines
|
I ll
0.005
Cellulose 1
i! Cellulose 2
0.004
Cellobiose 1
Cellobiose 2
0.003
-m—Glucose 1
-A—Glucose 2
0.002
a> S
-m— Lactate 1
Lactate 2
-A—
I 1 0.001
-*—Acetate 1
-A— Acetate 2
0.000
10 20 30 40 50 60 70
Time (h)
Figure A-3. Normalized production of CH in the initial batch study of the reactive
4
mixture for Column Set 3. Note: Each time point represents total moles summed
from h.
0
Subsequent Batch Studies. After 27 days, the reactive mixture contained much less
labile organic carbon than was available initially. A stimulatory effect on the rate of
CO production was again observed with lactate, while glucose also increased the rate
2
of overall activity (Figure A-4). More lactate and glucose were consumed than
cellobiose, acetate, or cellulose, with cellulose degraded the least. Note the scale
difference between Figure A-l and Figures A-4, A-7, and A-10. A much smaller
percent of the carbon added was consumed compared to the initial batch study.
Lactate increased the rate of FES production after 27 days, followed by glucose and
cellobiose (Figure A-5). The cellulose and acetate added did not accelerate FES
production. The total moles of CFE produced were not significantly different with the
substrate supplements than with no supplement after 27 days (Figure A- ).
6
|
Colorado School of Mines
|
113
0.003
-a—Cellulose 1
ii 0.002 -A—Cellulose 2
Cellobiose 1
0.001 Cellobiose 2
-A—
O -*—Glucose 1
TT
o Ü 0 -A—Glucose 2
al i 20 30 40 -m— Lactate 1
0.001
I - -A— Lactate 2
I
-m— Acetate 1
0.002
- -A— Acetate 2
-0.003
Time (h)
Figure A- . Normalized production of CH in the batch study of Column Set 3 after
6 4
27 days. Note: Each time point represents total moles summed from 0 h.
After 41 days, CO production appeared very similar to the previous batch study. A
2
stimulatory effect on CO production was again observed with lactate and glucose, while
2
cellulose had no significant effect (Figure A-7). More lactate and glucose were
consumed than cellobiose, acetate, or cellulose, with cellulose degraded the least. Again,
only a fraction of the carbon added in the batch study was consumed. Lactate again
showed a stimulatory effect on H S production, followed by glucose and cellobiose,
2
while cellulose and acetate did not accelerate H S production (Figure A- ).
2 8
Methanogenesis after 41 days was accelerated by acetate and possibly by lactate (Figure
A-9).
|
Colorado School of Mines
|
115
0.003
-m—Cellulose 1
0.002 Cellulose 2
s -A—
3 ü Cellobiose 1
T 2J "3 0.001 Cellobiose 2
<
CL ü -e—Glucose 1
0 -A—Glucose 2
O ®
30 40
-*— Lactate 1
î i 0.001
!!- Lactate 2
-A—
-e— Acetate 1
0.002
- -A— Acetate 2
-0.003
Time (h)
Figure A-9. Normalized production of CH in the batch study of Column Set 3 after
4
41 days. Note: Each time point represents total moles summed from 0 h.
In the final batch study after 99 days, a stimulatory effect on CO production was
2
again observed with lactate and glucose, followed by cellobiose and acetate, while
cellulose showed no significant effect (Figure A-10). Lactate again showed a stimulatory
effect on H S production, followed by glucose and cellobiose, while cellulose and acetate
2
had no effect (Figure A-l 1). Methanogen activity after 99 days was again stimulated by
acetate, and to a lesser degree by lactate, with no effect observed from the other
supplements (Figure A-12). Again, only a fraction of the carbon added in the batch study
was consumed in the production of CO , FLS, and CH .
2 4
|
Colorado School of Mines
|
117
0.003
-m—Cellulose 1
0.002 -A—Cellulose 2
1
73 Cellobiose 1
3 0)
7 23 73 0.001 -A—Cellobiose 2
CL < -#—Glucose 1
O
X V) 0 -A—Glucose 2
O 0)
0 -m— Lactate 1
S
o — -0.001
-A— Lactate 2
s
0 -*— Acetate 1
î 1-
-0.002
-A— Acetate 2
-0.003
Time (h)
Figure A-12. Normalized production of CH in the batch study of Column Set 3 after
4
99 days. Note: Each time point represents total moles summed from 0 h.
The normalized data for all subsequent batch studies following the initial batch study
were very similar. This could be explained by the difference in microbial community
composition between the reactive mixture initially and the reactive mixture after 27 to 99
days in anaerobic columns. The initial community contained very active non-sulfate-
reducing populations that probably included aerobic respirers, since the mixture was not
anaerobic initially. The communities adapted over time to the anaerobic conditions in the
columns and the sulfate added in the influent. The amount of CO produced during the
2
initial batch study was consistent with consumption of 80% of the lactate added (Figure
A-l). Lactate and acetate were shown to be highly available for metabolism by the
microbial community in the initial batch study. However, in the subsequent batch
studies, all of the substrate supplements added were well in excess of the amount that was
degraded. At the most, only about 17% of the lactate added was actually consumed
during the subsequent batch studies (Figure A-4), and even smaller percentages of
|
Colorado School of Mines
|
ABSTRACT
Rare earth elements are in high demand in the United States. Independence from the
importation of rare earths is essential to alleviate dependence on China for these rare earth
elements. Bastnaesite, a rare earth fluorocarbonate, is one of the most abundant sources of rare
earths in the United States. It is a fluorocarbonate mineral containing primarily cerium and
lanthanum. The largest rare earth mine in the United States is Mountain Pass.
This research was done to find a way to combine flotation with novel collectors and
gravity separation techniques to reach an enhanced grade and recovery of rare earth elements
while rejecting the gangue minerals, calcite, barite and silicate minerals. The main economic
driving force is the price of hydrochloric acid in downstream processes, as calcite is an acid
consumer. Surface chemistry analysis was completed using adsorption density, zeta potential,
and microflotation on both gravity concentrates and run of mine ore samples. Four collectors
were examined. These were N,2.dihydroxybenzamide, N-hydroxycyclohexanecarboxamide, N,3.
dihydroxy-2.naphthamide, and N-hydroxyoleamide. Through this analysis it was determined
that, to obtain the desired results, that flotation would be the rougher stage and the gravity
separation would be utilized as the cleaner stage. Bench scale flotation tests were conducted on
the run of mine ore using conditions that were determined using a previously calculated Stat
Ease model. The bench tests that produced the most desirable results were then scaled up to a 10
kilogram float test. A concentrate from this test showed a rare earth oxide grade of 44%, while
rejecting 91% of the calcite. This concentrate was used for gravity separation. Through gravity
separation it was found that another 40% of the calcite could be rejected with a final rare earth
oxide grade of 47% in the concentrate that was produced.
iii
|
Colorado School of Mines
|
ACKNOWLEDGEMENTS
I would like to thank my advisor for this project, Dr. Corby Anderson for his support and
advice on this project. I would also like to thank my committee members, Dr. Patrick Taylor,
Professor Erik Spiller, and the members of the Kroll Institute for Extractive Metallurgy for their
assistance. Special thanks to Professor Brock O’Kelly for his expertise on this subject and to
Santa Jansone-Popova for synthesis of large quantities of the collectors. I also need to thank the
employees at Resource Development Inc. for allowing me to use their equipment. I am grateful
for the assistance of Grant Colligan during test work and analysis.
Support for this research came from the Critical Materials Institute, an Energy Innovation
Hub, which is funded by the U.S. Department of Energy, Office of Energy Efficiency and
Renewable Energy, and the Advanced Manufacturing Office.
I would also like to thank my friends and family. Finally, I would like to thank my fiancé,
Molly Reicher for her continued support throughout the pursuit of my degree.
xviii
|
Colorado School of Mines
|
some of what makes them desirable in technology today. A list of some of the uses for each of
the rare earth elements can be found in Table 2.1.
Table 2.1: List of rare earth elements and their uses. [3]
Z ELEMENT SYMBOL USE
21 Scandium Sc Aerospace framework, high-intensity street lamps, high
performance equipment
39 Yttrium Y TV sets, cancer treatment drugs, enhances strength of alloys
57 Lanthanum La Camera lenses, battery-electrodes, hydrogen storage
58 Cerium Ce Catalytic converters, colored glass, steel production
59 Praseodymium Pr Super-strong magnets, welding goggles, lasers
60 Neodymium Nd Extremely strong permanent magnets, microphones, electric
motors of hybrid automobiles, laser
61 Promethium Pm Not usually found in Nature
62 Samarium Sm Cancer treatment, nuclear reactor control rods, X-ray lasers
63 Europium Eu Color TV screens, fluorescent glass, genetic screening tests
64 Gadolinium Gd Shielding in nuclear reactors, nuclear marine propulsion,
increases durability of alloys
65 Terbium Tb TV sets, fuel cells, sonar systems
66 Dysprosium Dy Commercial lighting, hard disk devices, transducers
67 Holmium Ho Lasers, glass coloring, High-strength magnets
68 Erbium Er Glass colorant, signal amplification for fiber optic cables,
metallurgical uses
69 Thulium Tm High efficiency lasers, portable x-ray machines, high
temperature superconductor
70 Ytterbium Yb Improves stainless steel, lasers, ground monitoring devices
71 Lutetium Lu Refining petroleum, LED light bulbs, integrated circuit
manufacturing
3
|
Colorado School of Mines
|
2.2 Rare Earth Deposits
Rare earth elements do no occur in their elemental state naturally. They are found in other
mineral deposits. Rare earth deposits are found worldwide, but only a few are used for the
production of rare earths. China produces the majority of rare earths today. A summary of the
global rare earth production by country can be found in Table 2.2.
Table 2.2: Rare earth production by country in 2015 and 2016. [4]
Mine production
Country 2015 2016 Reserves
United States 5,900 - 1,400,000
Australia 12,000 14,000 3,400,000
Brazil 880 1,100 22,000,000
Canada - - 830,000
China 105,000 105,000 44,000,000
Greenland - - 1,500,000
India 1,700 1,700 6,900,000
Malaysia 500 300 30,000
Malawi - - 136,000
Russia 2,800 3,000 18,000,000
South Africa - - 860,000
Thailand 760 800 NA
Vietnam 250 300 22,000,000
World total (rounded) 130,000 126,000 120,000,000
One such mineral deposit is bastnaesite, a rare earth fluorocarbonate. Bastnaesite is found
in vein deposits and contains as much as 75% rare earth oxide. It is primary composed of cerium
and lanthanum. [1] There are two major bastnaesite deposits in the world, Bayan Obo in China
and Mountain Pass in the United States. Bayan Obo is a deposit of approximately 800 million
metric tons, while Mountain Pass is a deposit of approximately 3.3 million metric tons. The
Mountain Pass Mine is located in Southern California. It was found in 1949 as the largest rare
4
|
Colorado School of Mines
|
Rare earth prices depend on the purity and state of the element. A pure metal is worth
more than the metal oxide. Because of the amount of rare earths that China produces, the price is
controlled by their export quota. In 2010 the price jumped, possibly due to a reduction in China’s
rare earth export quota. [10] The current rare earth metal prices are shown in Table 2.3.
Table 2.3: Rare earth metal prices. [11]
Rare Earth Metal Price (USD)/kg
Lanthanum (>99%) $7.00
Cerium (>99%) $7.00
Praseodymium (>99%) $85.00
Neodymium (>99.5%) $60.00
Samarium (>99.9%) $7.00
Gadolinium (>99.9%) $55.00
Terbium (>99.5%) $550.00
Dysprosium (>99%) $350.00
Erbium (>99.9%) $95.00
Yttrium (>99.9%) $35.00
Scandium (99.9%) $15,000.00
2.4 Rare Earth Mineral Processing
Rare earths are processed based on their ore body and specific minerals in which they are
contained. Mines typically upgrade the ore before a leach is done to obtain the rare earths. Two
common upgrading methods are flotation and physical separation.
The primary rare earth mineral at the Mountain Pass mine in bastnaesite. The mine used a
rougher and cleaner flotation process to produce a 60% rare earth oxide concentrate, which was
further processed to produce pure rare earth oxides. [12] The process begins with a crushing and
grinding circuit. The ore is crushed with an impact, jaw and cone crusher, followed by grinding
in a ball and rod mill. Cyclones are used to separate out material of the desired size for flotation.
6
|
Colorado School of Mines
|
The oversized material gets reground in a ball mill. [13] A simplified flow sheet is shown in
Figure 2.3, while the full old version of the flow sheet is shown in Figure 2.4. If the United
States is to regain footing in the rare earths market, then Mountain Pass needs to be reopened.
Currently the Mountain Pass mine is in the process of being purchased by an investor group with
ties to China. [7]
Bear Lodge, an upstart rare earth carbonite mine in Wyoming, has made attempts to
produce a rare earth concentrate using physical separation methods. Their plan is to use gravity
and magnetic separation as upgrading methods, then use an acid leach to extract the rare earths.
It was reported that while using the physical separation methods the mine would be able to
obtain an 88% recovery of rare earth oxides with a 55% mass pull. [14]
2.5 Flotation Surface Chemistry
True flotation is to selectively render desired minerals hydrophobic, allowing them to
attach to air bubbles. It is a three phase process with many variables within each phase. [15] An
illustration of the principle of froth flotation is shown in Figure 2.5. From the figure one can see
the three phases of flotation, the pulp, air and the mineralized froth. A surface chemistry analysis
is needed to understand the ability of reagents, collectors and depressants to selectively attach to
their desired surfaces. Common analyses include, adsorption density, zeta potential and
microflotation studies.
7
|
Colorado School of Mines
|
Figure 2.5: An illustration of froth flotation. [15]
2.5.1 Mineralogical Analysis
The composition of the ore body needs to be understood before any surface chemistry
analysis can be completed. A mineral liberation analysis (MLA) of a representative sample can
be used to determine the size that the particles need to be for the desired mineral to be liberated
enough for flotation to be effective. Fine particles can negatively impact the effectiveness of
flotation, while larger particles will not have enough liberation to be selective.
X-Ray fluorescence (XRF) is an analytical technique used to determine the composition
of representative samples on an elemental level. It is a non-destructive technique that measures
the secondary X-ray when it is excited from a primary X-ray. When an electron from the atom’s
inner shell is removed by an X-ray another electron fills the spot and drops to a lower energy
state. This drop releases a secondary X-ray which is captured by the XRF machine. The spectra
is analyzed by the characteristic XRF peaks associated with each element. [16]
10
|
Colorado School of Mines
|
Another necessary mineralogical analysis for understanding flotation is the Brunauer-
Emmett-Teller (BET) surface area analysis. It measures the specific surface area and pore size in
particles by adsorption of nitrogen onto the mineral surface. Nitrogen gas is adsorbed onto the
surface of the particle and used to determine the surface area using an adaptation of the
Langmuir theory describing monolayer and multilayer adsorption. [17]
2.5.2 Reagents
Collectors are used to adsorb onto the surface of desired minerals rendering them
hydrophobic. They typically have a polar and a non-polar group. The non-polar group is often a
long chain of hydrocarbons that render the particle hydrophobic because of the change in surface
charge. As the length of the hydrocarbon tail increases the hydrophobicity also increases, but the
solubility of the collector decreases, limiting the chain length. The polar group determines the
selectivity of the collector due to their chemical, electrical or physical attraction to the particle.
[15] This relationship can be seen in Figure 2.6. They are used in particular amounts so as to not
oversaturate the solution and float undesirable minerals. Because of the development of
multilayer adsorption, increased concentration of a collector can adversely affect the recovery of
the desired mineral.
Figure 2.6: Adsorption of a collector onto a mineral surface. [15]
11
|
Colorado School of Mines
|
Collectors are broken up into two groups, anionic collectors and cationic collectors.
Anionic collectors are the most commonly used for bastnaesite flotation. Collectors used or
researched for the Mountain Pass mine material include fatty acids and hydroxamates. [1] Fatty
acids are inexpensive, but require the use of depressants and heat because their selectivity is low.
Hydroxamic acids have been researched for the flotation of bastnaesite because they are more
selective than fatty acids and do not require the addition of heat to effectively recover
bastnaesite. [18] Results from testing have shown that hydroxamates are effective for the
flotation of cerium and lanthanum, while they are less effective for niobium and yttrium. [19] It
is theorized that the reason for hydroxamates being so effective is chelation. Chelation is a type
of bonding that allows two or more bonds to form between a ligand and a separate metal ion.
[20] Another reason that hydroxamates are more selective is that the gangue minerals associated
with bastnaesite do not form as stable complexes as the rare earth bearing minerals with the
hydroxamates. [21]
Common depressants for this system include soda ash and ammonium lignin sulfonate.
Soda ash is used as a pH modifier, but also acts to control the carbonate anions. [1] In the
presence of barite the carbonate anions coat the surface of barite, changing it into barium
carbonate which acts as a depressant for barite in the process. [21] Ammonium lignin sulfonate
acts as a depressant for barite as well at high pH. At a higher pH the potential on the surface
barite is positive, while calcite and bastnaesite are negative, making the ammonium lignin
sulfonate attach more easily to its surface. [22]
2.5.3 Adsorption
Adsorption is studied to determine how a collector adsorbs onto the surface of a particular
mineral. Adsorption can be chemical or physical. During physical adsorption no activation
12
|
Colorado School of Mines
|
energy is required, so equilibrium is reached quickly and it is easily reversible. An activation
energy is required for chemical adsorption and it is limited to a monomolecular adsorption later.
[23] Multilayer adsorption occurs, until a critical micelle concentration is reached, only by
physical adsorption as seen in Figure 2.7.
Figure 2.7: An illustration of multilayer adsorption. [23]
Because physical adsorption requires less energy, many of the molecules will be attracted
to the surface of the mineral. Chemical adsorption occurs when the molecules get close enough
to the mineral surface that chemical adsorption becomes a favorable reaction. The molecules
become attracted to the surface because of van der Waals interactions, which develop an energy
minimum allowing chemical adsorption to more easily occur. [23] Physical adsorption is
necessary for chemical adsorption to happen because of this energy minimization developed
because of the physical adsorption. Adsorption is generally an endothermic process, and this is
no different for the adsorption of hydroxamates onto a bastnaesite surface. [21]
Increased concentration of the collector in the solution increases the probability that the
collector will form multilayer adsorption onto the mineral surface. Multilayer adsorption is
13
|
Colorado School of Mines
|
advantageous when the critical micelle concentration it reached, because it will drastically
increase the recovery of the desired mineral. During monolayer formation the surface charge is
not reversed so the surface will remain hydrophobic, when multilayer adsorption occurs the
polarity can be reversed. [23] This phenomena can be seen in Figure 2.8.
Figure 2.8: The effect of multilayer adsorption on a mineral surface. [23]
To find the ideal concentration of a collector for flotation adsorption isotherms are
developed. An example of an adsorption isotherm can be seen in Figure 2.9. As seen in the
figure, not only does concentration have an effect on multilayer formation, but temperature does
as well. This is because of the low energy requirement of physical adsorption onto the ends of
the collectors.
Another important factor when considering adsorption is pH. The relationship between
recovery and pH can be seen in Figure 2.10, along with the relationship of temperature and
recovery in Figure 2.11. Both pH and temperature have a large effect on the recovery and grade
of bastnaesite in flotation. This effect can be predicted through adsorption studies, as shown in
Figure 2.9. As seen in Figure 2.11, the temperature at which flotation occurs can drastically
increase the recovery and selectivity of a collector to a specific mineral surface. During
adsorption studies the optimal pH range can be determine before beginning flotation test work.
14
|
Colorado School of Mines
|
2.5.4 Zeta Potential
The electrical change on a mineral’s surface needs to be understood in surface chemistry
studies. Zeta potential is the charge that forms at the solid/liquid interface. It causes a region of
counterions to form around the particle. This forms an electric double layer, the Stern layer and
the diffuse layer, as illustrated in Figure 2.12. In the Stern layer the counterions attach to the
particle to neutralize the surface of the particle. In the diffuse layer there is a mix of positive and
negative ions. In Figure 2.12 a negatively charged particle is surrounded by positive ions in the
Stern plane. While more positively charged ions are still attracted to the particle, they are
repelled by the positive ions in the Stern plane. Shear that takes place on the Stern plane when
the particle moves creates a potential at that plane which is known as the zeta potential. [15]
Even though the potential at the Stern plane is less than the surface potential it is considered
significant because it can be measured and it reflects interactions with particles within the
solution surrounding it. [24]
The point of zero charge (PZC) is the pH at which the zeta potential of a specific material
is zero. The Iso-Electric Point (IEP) is the point at which the zeta potential is zero when the
mineral is in the presence of an electrolyte with potential determining ions. If an ion shifts the
PZC to an IEP then that ion is called a potential determining ion (PDI). PDIs create the electric
double layer, thus allowing the pH at which the IEP occurs to shift. Table 2.4 shows various
reported IEPs for bastnaesite.
17
|
Colorado School of Mines
|
Figure 2.12: An illustration of the electrical double layer. [26]
Figure 2.13 is an illustration of the effect of PDIs on the surface chemistry of a particle in
the presence of an indifferent electrolyte. Point A is the PZC for the hydrophilic surface in the
presence of the indifferent electrolyte. The dashed line is an example of how the zeta potential
changes in the presence of a physical adsorbing anionic surfactant. Because there is no
adsorption at an increased pH, the IEP is the same as the PZC. As the pH is decreased to point C
there is adsorption of the surfactant on the surface which yields an IEP. Points B’ and B” are
both IEPs for physically adsorbed and chemically adsorbed surfactants, respectively. Point B’ is
at a pH only slightly lower than the PZC shown at point A, while point B” is at a significantly
lower pH. E’ is the point at which the mineral surface is so negative that the adsorption potential
is overcome. Point E” shows where the chemical contribution to the free energy of adsorption is
overcome by the electrostatic repulsion. This point is generally the upper limit of flotation with a
chemisorbing collector. [28]
18
|
Colorado School of Mines
|
2.6 Microflotation
Microflotation is used to test flotation conditions on a small scale so that excess materials
are not consumed needlessly. In microflotation studies the variables affecting flotation can be
studied quickly and effectively. Some of the variables that can be tested are collector type, other
reagents, collector concentrations, pH, temperature, and conditioning time. On the microflotation
scale it is easier to control these variables. It is also beneficial to see how changing potential
determining ions effect the selectivity of collectors. Since microflotation does not require a large
amount of material, it can be completed and analyzed quickly compared to bench or pilot scale
flotation tests. Microflotation experiments usually consume 0.5-3.0 grams of material within a
known size fraction, usually between 100 and 325 US mesh. [24] Since bastnaesite is hydrophilic
the flotation process will render the particle hydrophobic and allow the desired mineral to attach
to air bubbles and form a froth. The froth can then be collected and analyzed soon after the
completion of the test.
2.7 Bench and Large Scale Flotation
After the completion of microflotation, bench scale flotation can be done to test the
effectiveness of the flotation process on a larger scale. The pulp density is increased significantly
(1 wt% to 25 wt%), which alters the concentrations of the reagents in the solution. The solution
is increased from 50 mL in microflotation to 1 liter on the bench scale. Bench scale test work
also allows more variables to be easily tested, such as depressant addition. Temperature can more
easily be controlled on this scale as well. Even though more material is used for bench flotation it
is still a quicker test than a pilot scale test, and the material can be analyzed quickly.
After bench flotation is completed the process can be up-scaled to a 30 liter flotation test.
The advantage of this test is that it more closely resembles what will happen on the plant scale.
20
|
Colorado School of Mines
|
The same variables that are tested on bench scale are tested here. The disadvantage is that it
consumes reagents and other materials quickly. Because of this the flotation conditions that are
used on this scale should already be proven on the bench scale.
2.8 Gravity Separation
Gravity separation is a technique involving the manipulation of particle densities to
separate the less dense particles from the more dense ones. To see if gravity separation is
possible, the concentration criterion is calculated. The equation for concentration criterion (CC)
is as follows:
(2.2)
(cid:23)(cid:24)(cid:25)(cid:23)(cid:26)
(cid:22)(cid:22) = (cid:23)(cid:27)(cid:25)(cid:23)(cid:26)
Where D is the density of the heavy particles, D is the density of the fluid and D is the
h f l
density of the light particles. If the concentration criterion is greater than 2.5 the gravity
separation is viable and below 1.25 it is impossible. If the concentration criterion is less than 2.5
and greater than 1.25 then the separation is possible, but difficult. [29] Table 2.5 shows the
concentration criterion for major minerals present in the ore from Mountain Pass. From the
concentration criterion it appears that calcite, dolomite and quartz should be separable from the
rare earth bearing minerals. Because barite and the rare earth bearing minerals have similar
densities they are inseparable using gravity techniques.
Table 2.5: Specific gravities and concentration criterion for components of the Mountain Pass
ore. [30]
Mineral Density (g/cm3) Concentration Criterion
Bastnaesite (REE)CO F 4.97 0.97
3
Parasite
(Ca(Ce,La) (CO ) F 4.36 1.15
2 3 3 2
21
|
Colorado School of Mines
|
Table 2.5: Continued
Monazite ((Ce,La)PO 5.15 0.93
4
Synchysite (CaCe(CO ) F 4.02 1.28
3 2
REE Bearing Minerals 4.87 1.00
Calcite (CaCO ) 2.71 2.26
3
Dolomite (CaMg(CO ) ) 2.84 2.10
3 2
Barite (BaSO ) 4.48 1.11
4
Celestine (SrSO ) 3.95 1.31
4
Quartz (SiO ) 2.62 2.39
2
There are a wide range of gravity separation technologies including, shaking table,
Knelson and Falcon concentrators. Based on work done by Alex Norgren, the ultrafine falcon
concentrator worked well on the Mountain Pass ore. [31] The ultrafine falcon allows for
separations to occur at low particles sizes (<38 microns). [32] The advantage to the ultrafine
falcon over the falcon concentrator is that no additional process water is required and it is able to
separate particles of a lesser size. A cross section diagram of the ultrafine falcon can be seen in
Figure 2.14. With the Falcon concentrator, the feed enters from the top of the spinning bowl. The
slurry then flows over the bowl. As this happens the more dense particles are left behind in the
bowl while the less dense particles flow out of the top of the bowl and are collected as tailings.
22
|
Colorado School of Mines
|
CHAPTER 3: EXPERIMENTAL METHODS
3.1 Sample Preparation
A run of mine (ROM) ore sample was obtained from the Mountain Pass mine. The
sample was split using shovels and the cone and quarter method to preserve the homogeneity of
the sample. The samples were split into approximately 35 kg fractions and placed into buckets
for further use. The ore in the buckets was then crushed using a roll crusher. The roll crusher was
set to specific gap sizes of 4.3 mm for the first pass and 2.3 mm for the second pass to obtain the
desired sample size for grinding. A size analysis was done after the samples were crushed and is
shown in Table 3.1. In preparation for grinding the samples were split, using a Jones splitter, into
1 kg and 10 kg charges.
Table 3.1: The particle size analysis of the roll crushed ore.
Microns Weight (g) Percent Passing
+2380 51.2 89.2%
-2380 +1410 141.0 59.5%
-1410 +595 127.9 32.5%
-595 +210 70.1 17.7%
-210 +150 17.9 13.9%
-150 66.0 0.0%
All the material that was ground for large scale flotation test work was completed using a
rod mill. The feed for the rod mill consisted of 50 weight percent ore and 50 weight percent
water. All grinding was completed at Resource Development Inc. In preparation for grinding a
10 kg charge, a series of 1 kg test were run in an attempt to scale the tests up easier. The 80%
passing size required for flotation was 50 microns. Through these 1 kg tests it was determined
24
|
Colorado School of Mines
|
that the time needed to grind 10 kg of crushed ore to the desired size was 32 minutes. The rod
mill used is shown in Figure 3.1. A particle size analysis of the 10 kg rod mill product was done
and is shown in Table 3.2. For the particle size analysis a sample was wet sieved through a 400-
mesh screen. The respective size fractions were dried and the +400-mesh material was run
through a rotap for 20 minutes using 100, 115, 200, 270, 325 and 400 mesh screens. The P for
80
the sample was determined to be 52 microns.
Figure 3.1: The rod mill used for grinding 10 kg samples.
Table 3.2: The particle size analysis of the milled sample.
Microns Weight (g) Percent Passing
+150 0.9 99.9%
-150 +125 1.0 99.8%
-125 +100 2.2 99.5%
-100 +75 17.7 97.6%
-75 +53 76.4 89.1%
-53 +44 82.2 80.0%
-44 +37 32.2 76.5%
-37 410.4 31.0%
The mineralogy of the sample was determined using XRF, XRD and MLA. A sample of
the ROM ore was sent to Montana Tech for the XRD and MLA to be completed. Figure 3.2
25
|
Colorado School of Mines
|
collectors. The points were fit to a linear line and used in determining the concentration of the
collector in solution corresponding to each absorbance. The adsorption density was derived from
Equation 3.1: [17]
(3.1)
∆(cid:20)∗(cid:30)
Γ = (cid:31)∗(cid:32)
Γ is the adsorption density in mol/m2, ΔC is the change in concentration of the solution in
moles, V is the solution volume in L, m is the mass of the sample placed into solution in grams,
and A is the specific surface area of the mineral in m2/g. [24]
3.5 Zeta Potential
A Microtrac Stabino was used for zeta potential measurements, as shown in Figure 3.5.
From these measurements the iso-electric points (IEP) for each condition were found. The same
samples that were used for the adsorption study were used for the zeta potential measurements.
The samples were placed into deionized water with a concentration of 0.5 g/L. Experiments were
run in water only to determine the point of zero charge (PZC). Other experiments were analyzed
with the collectors added in to determine how the collector affects the IEP. The collector
concentrations were 1 mM for collectors 2 and 5, and 0.1 mM for collectors 8 and 14.
Figure 3.5: A Microtrac Stabino instrument used for Zeta Potential measurements. [35]
31
|
Colorado School of Mines
|
3.6 Microflotation
Microflotation studies were done using the same materials used for the zeta potential
tests. A Partridge Smith Cell was used for all experiments, as illustrated in Figure 3.6. The
solution was made up of 0.52 grams solids and 52 mL water. For each experiment the sample
was added to the solution with a specific collector concentration. The pH was changed after the
addition of the collector using potassium hydroxide. The slurry was conditioned for 15 minutes
in a 100 mL beaker. After 13 minutes, a drop of methyl isobutyl carbinol (MIBC) frother was
added to the slurry. After conditioning was completed, the slurry was placed into the Partridge
Smith cell for flotation. Compressed air was added into the system at a flow rate of 26.6
cm3/min. The concentration of the collector was not varied for any tests involving that collector.
The only variable was pH. The pH range was 9.5-10.5. The concentrates and tailings were
analyzed using the XRF machine.
Figure 3.6: A schematic of a Partridge Smith cell used for microflotation studies. [36]
32
|
Colorado School of Mines
|
3.7 Bench Flotation
Bench flotation tests were conducted using a Metso Denver D-12 Legacy cell, shown in
Figure 3.7. These tests were conducted using run of mine material with an 80% passing size of
50 microns. 333 grams of material were added into a 1 liter slurry for flotation, forming a slurry
concentration of 25 weight percent solids. If heat was required, then the water was heated before
it was combined with the ore. For this study more reagents were used than in microflotation. If
depressant was needed for the test, the ore, water and depressant were combined and allowed to
condition for 5 minutes. If soda ash was being used as a pH modifier then once it was added to
the slurry it was allowed to condition for 3 minutes. Once the pH was adjusted, then the collector
was added and allowed to condition for 10 minutes. The collectors that needed to be dissolved in
ethanol were added to the slurry, then emulsified for 3 minutes by a Hamilton Beach
Commercial HMI200 Immersion Blender. If the pH needed to be modified further, it was done
so during the next conditioning stage. Conditioning was done at 900 rpm. Two minutes before
flotation was started one drop of MBIC frother was added. After conditioning, air was allowed
into the system and the material was allowed to float for two minutes. The concentrates and
tailings were collected, filtered and then dried for analysis.
3.8 10 Kilogram Flotation
The 10 kilogram flotation tests were conducted using a flotation cell at Resource
Development Inc. (RDI), pictured in Figure 3.8. The sample used for the tests was removed from
the rod mill and placed directly into the flotation cell. The slurry concentration was
approximately 25 weight percent solids. If the test required heat, then the slurry was heated using
a heating coil, placed directly into the slurry. Once temperature was reached, the depressant was
added, if needed. The slurry was then allowed to condition for 5 minutes. If soda ash was used as
33
|
Colorado School of Mines
|
CHAPTER 4: SURFACE CHEMISTRY
Surface chemistry experiments were conducted on the run of mine ore and the gravity
separated concentrate to determine the conditions for flotation for the gravity concentrate. It was
also used to determine the differences in the two samples. The adsorption and zeta potential
results are discussed below.
4.1 Adsorption Density
Adsorption studies are used to determine how the collector adsorbs onto the mineral
surface. Experiments were conducted using the run of mine (ROM) ore and the gravity separated
sample. Only a comparative study was done between the ROM ore and the gravity concentrate
sample because these collectors had already been studied by Dylan Everly. [17] The adsorption
studies were completed with collectors 2, 5 and 8. Collector 14 was not examined because the
same method could not be used as collectors 2, 5 and 8. Equilibrium time was determined before
all other studies were started. The conditions for the equilibrium study were 0.001 M and 9.5 pH
for each collector. All experiments were conducted at room temperature. A pH range was
studied, between 3.5 and 11.5, with a constant initial collector concentration of 0.001 M for each
test. A study of the effect of collector concentration was also conducted with initial collector
concentrations between 0.00025 M and 0.0025 M with a constant pH of 9.5. Reference lines are
provided on each of the graphs for the horizontal and vertical monolayer adsorption densities for
hydroxamic acid, since the collectors being studied do not necessarily have known horizontal or
vertical monolayer adsorption densities. The surface areas and adsorption densities in each
orientation of hydroxamic acid are outlined in Table 4.1. Hydroxamic acid is used as an
37
|
Colorado School of Mines
|
4.1.2 pH vs. Adsorption Density
Experiments were run using the equilibrium time determined in the previous section to
show how the adsorption density of each collector changes with respect to pH. All the collectors
had an initial concentration of 0.001 M. Figure 4.3 illustrates how the ROM ore sample and the
gravity concentrate reacted with changing pH with the addition of collector 2. From this plot it
can be seen that multilayer adsorption occurs for the gravity sample around a pH of 8-10. This
corresponds to the pH range of bastnaesite, meaning that the collector is adsorbing to the desired
mineral surface because of the increased driving force for adsorption. The ROM ore sample has a
peak at a pH of 9.5, but the adsorption densities indicate that it never has more than a single
horizontal adsorption layer.
25.00
20.00
15.00
10.00
5.00
0.00
0 2 4 6 8 10 12
pH
ROM Ore Gravity Concentrate Horiontal Vertical
Figure 4.3: The pH vs. adsorption density plot for collector 2.
Figure 4.4 shows the pH vs. adsorption density for collector 5. For the gravity
concentrate there is a multilayer adsorption around pH 10, as expected, but for the ROM ore
40
)601*2m/lom(
ytisneD
noitprosdA
|
Colorado School of Mines
|
4.1.3 Equilibrium Concentration vs. Adsorption Density
A study was done for compare how the gravity concentrate differed from the ROM ore
sample with increasing equilibrium concentrations. Figure 4.6 shows how the adsorption density
changes with respect to the equilibrium concentration of the collector on the mineral surface. The
ROM ore sample and the gravity concentrate both follow the same general trend, but the gravity
concentrate has a much higher adsorption density, as expected. At lower concentrations the
adsorption density for the gravity concentrate is slightly higher than that of a vertical monolayer,
while the ROM ore is slightly higher than that of a horizontal monolayer.
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
0.00 0.50 1.00 1.50 2.00 2.50
Equilibrium Concentration (M*103)
ROM Ore Collector 2 Gravity Concentrate Collector 2
Horizontal Vertical
Figure 4.6: Equilibrium concentration of the collector vs. adsorption density for collector 2.
Figure 4.7 illustrates how the adsorption density changes as equilibrium concentration
increases for collector 5. Again, both samples follow a similar increasing trend, but with
collector 5 the ROM ore sample has a higher adsorption density than that of the gravity
concentrate. This could mean that collector 5 has a higher affinity to the gangue minerals than
for bastnaesite.
42
)601*2m/lom(
ytisneD
noitprosdA
|
Colorado School of Mines
|
4.1.4 Adsorption Thermodynamics
The Gibbs free energy of adsorption was calculated for the samples at room temperature.
The results are shown in Table 4.2. The calculated Gibbs free energies are for physical
adsorption because they were calculated using information from the critical micelle
concentration. At the critical micelle concentration the barrier to chemical adsorption has been
overcome and the collector attaches itself to the mineral surface more easily by physical
adsorption. The calculated Gibbs free energy is negative for all the collectors. A negative Gibbs
free energy indicates that the adsorption onto the mineral surface is spontaneous and will proceed
easily without an energy added into the system. The free energy indicates that for collectors 2
and 5 adsorption is slightly more favorable for the gravity concentrate. This could be because
there is a higher concentration of bastnaesite in the sample, while the adsorption of collector 8 is
more favorable with the ROM ore sample.
Table 4.2: The Gibbs free energy of adsorption for the collectors onto the ROM ore sample and
gravity concentrate.
ΔGo
ads
(298 K) [kCal/mol]
Sample Collector 2 Collector 5 Collector 8
ROM Ore -4.83 -5.95 -6.16
Gravity Concentrate -5.49 -6.01 -5.64
4.2 Zeta Potential
Zeta potential measurements were conducted on the ROM bastnaesite ore and the gravity
concentrate material. These measurements were used to determine how the each of the collectors
interacted with the mineral surface and the electrical nature in the solution, to help determine the
best pH range to conduct flotation studies in. Collectors 2, 5 8 and 14 were all evaluated. It was
44
|
Colorado School of Mines
|
samples after the addition of the collector. Also, the chemical contribution of the collector is not
negated until a pH of 6.5 for both samples.
0
2 3 4 5 6 7 8 9 10 11
-5
-10
-15
-20
-25
-30
-35
pH
Gravity Concentrate ROM Ore
Figure 4.10: pH vs. Zeta Potential for the gravity concentrate and ore with the addition of
collector 2 at a concentration of 1E-3M.
The zeta potential with the addition of collector 5 is shown in Figure 4.11. The IEP of the
gravity concentrate is lower than that of the ROM ore, similar to the behavior of the PZCs of
each sample. As the pH was increased the zeta potential was decreased and even at pH 10 the
zeta potential does not match that of the samples in water. This indicates that the chemical
contribution is strong for this collector on the mineral surface.
With the addition of collector 8 the IEP is only slightly decreased from the PZC of the
samples, as seen in Figure 4.12. The zeta potential never matches up with the zeta potential of
the samples in water. This means that the zeta potential with the addition of this collector never
overcomes the chemical contribution of collector 8.
46
)Vm(
laitnetoP
ateZ
|
Colorado School of Mines
|
ratio is a measure of selectivity. A higher ratio indicates that the calcite is being rejected while
the rare earth oxides are being recovered more easily.
5.1 Collector 2
Three tests were run in duplicate to get enough material to use for XRF analysis for each
ore sample. The best results from the run of mine (ROM) ore sample and the gravity sample are
shown in Figure 5.1 from the tests using collector 2. From the figure it can be seen that, as
expected, the recovery of the gravity concentrate is lower than that of the ore that has only been
floated. The gravity concentration stage results in better selectivity, shown in the REO/CaO ratio,
the floated gravity concentrate maintains that higher selectivity. However, the treated gravity
concentrate has a lower overall grade compared to the treated ROM ore, 19% rare earth oxide
(REO) compared to 24% REO.
100% 3
90%
2.5
80%
70%
2
60%
50% 1.5
40%
1
30%
20%
0.5
10%
0% 0
Untreated ROM Untreated Collector 2 ROM Collector 2
Ore Gravity Ore Gravity
Concentrate Concentrate
Grade (%) Recovery (%) REO/CaO
Figure 5.1: Results of microflotation from collector 2, showing cumulative recovery for the
floated gravity concentrate.
50
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
5.2 Collector 5
Three tests were run in duplicate varying the pH of each of the tests. The most selective
results are shown in Figure 5.2. Again, the results indicate that with gravity separation before
flotation, the selectivity is increased above that of flotation first. But with this collector the
selectivity is not enhanced through flotation and the REO grade is only slightly increased above
that of the feed grade. Both samples increase their respective grades 2% through flotation. The
REO/CaO ratio does not increase significantly after flotation for either of the samples. The
recovery and grade are greater for the gravity pre-concentrated material compared to the ROM
ore sample that was floated. Overall, this collector displays very little selectivity, and results in
grades and recoveries that are less than those of the other collectors.
100% 1.2
90%
1
80%
70%
0.8
60%
50% 0.6
40%
0.4
30%
20%
0.2
10%
0% 0
Untreated ROM Untreated Collector 5 ROM Collector 5
Ore Gravity Ore Gravity
Concentrate Concentrate
Grade (%) Recovery (%) REO/CaO
Figure 5.2: Microflotation results from using collector 5, showing selectivity, REO grade and
recovery.
51
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
5.3 Collector 8
Three microflotation tests were run in duplicate for this collector at a concentration of
5E-4 M. The best results of these tests can be seen in Figure 5.3. The recoveries from each
sample are similar, but the grade is increased in the test that had gravity concentration first. The
REO/CaO ratio increased with the use of this collector, but only slightly above that of the feed
samples. The selectivity is higher for the floated gravity concentrate than for the ROM ore
sample that was floated. Again, this collector shows only a slight increase in the grades of the
concentrate over their respective feed grades.
100% 1.2
90%
1
80%
70%
0.8
60%
50% 0.6
40%
0.4
30%
20%
0.2
10%
0% 0
Untreated ROM Untreated Collector 8 ROM Collector 8
Ore Gravity Ore Gravity
Concentrate Concentrate
Grade (%) Recovery (%) REO/CaO
Figure 5.3: Results of the microflotation tests using collector 8, showing the resulting grades,
recoveries and selectivity.
5.4 Collector 14
The pH was varied between three tests for each sample that were run in duplicate. The
best results of which can be seen in Figure 5.4. After the flotation of the gravity concentrate, the
REO/CaO ratio was decreased. This was true for the ROM ore sample as well. The REO grade
52
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
of the concentrates was no greater than their respective feed grades either. The grade and
recoveries for the floated gravity concentrate are higher than that of the ROM ore sample that
was floated, but overall this collector exhibited low selectivity on the microflotation level.
100% 1
90% 0.9
80% 0.8
70% 0.7
60% 0.6
50% 0.5
40% 0.4
30% 0.3
20% 0.2
10% 0.1
0% 0
Untreated ROM Untreated Collector 14 Collector 14
Ore Gravity ROM Ore Gravity
Concentrate Concentrate
Grade (%) Recovery (%) REO/CaO
Figure 5.4: Microflotation results of the tests run with collector 14, with resulting grades,
recoveries and REO/CaO ratios.
5.5 Conclusions
A summary graph of the results above can be seen in Figure 5.5. From the experimental
data obtained for both samples that were used for microflotation test work, the most selective
collector is collector 2. In all the tests the floated gravity concentrates exhibited the highest
selectivity, but also had decreased recovery and sometimes decreased grade compared to the
grade of the ROM ore flotation concentrate. Although gravity pre-concentration is more
selective, the general trend seems to be that there is no appreciable increase in grade. From these
results it was determined that further test work would focus on rougher flotation followed by a
cleaner gravity stage.
53
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
CHAPTER 6: BENCH SCALE FLOTATION
Bench flotation tests were conducted to determine the most effective conditions to run
large scale tests on. All the collectors that were used in microflotation were used for bench scale
flotation. These tests added more variables including, collector, collector concentration,
depressant concentration, temperature, pH, pH modifier and continuous pH modification. The
depressant used was ammonium lignin sulfonate. The pH modifiers were potassium hydroxide
and sodium carbonate. The experiments were completed at a slurry density of 25 weight percent
solids (333 grams ore and 1 liter distilled water). The conditions tested were determined from
previous research done by Dylan Everly. [17] From his design of experiments a minimum of four
conditions for each collector were chosen for further study. For collector 2 six conditions were
tested. Two of the test conditions were ones that Dylan Everly had obtained the best results of
each collector with and the remaining tests were determined by the Design Expert 10
optimization study software developed by Stat-Ease. The conditions obtained from Stat-Ease
were predicted to either have increased recovery or increased grade and recovery, while also
showing selectivity for bastnaesite.
6.1 Collector 2
Six tests were attempted using collector 2. The test conditions are shown in Table 6.1. The
predicted grades and recoveries for rare earth oxide (REO) and the gangue minerals are shown in
Table 6.2. Tests 3.6 were suggested conditions by the Design Expert 10 software. This collector
was proven to have the highest selectivity of bastnaesite, while maintaining a high grade and
recovery of REO. As seen in Table 6.2, it was expected that the REO grade could be higher than
what had previously been proven. All successful tests completed with the collector had a unique
55
|
Colorado School of Mines
|
froth that would form before air was added into the system. This froth seemed to contain the bulk
of the material that would be removed to form the concentrate.
Table 6.1: Test conditions for bench tests using collector 2.
Continued Depressant
Test Concentration pH pH Temperature Concentration
Number (M) pH modifier Modification (oC) (mM)
2.1 7.50E-03 8.5 Soda Ash Yes 82 0
2.2 7.50E-03 9.5 KOH No 50 0
2.3 7.50E-03 10.5 Soda Ash Yes 50 0
2.4 1.00E-02 9.5 KOH No 82 0.75
2.5 1.00E-02 9.5 KOH No 20 0
2.6 1.00E-02 9.5 KOH No 20 0.75
Table 6.2: The predicted grades and recoveries for REO and the predicted grades of the gangue
minerals for tests run with collector 2.
Test REO REO CaO BaO
Number Grade Recovery Grade Grade
2.1 41.14 77.64 10.77 4.73
2.2 26.98 93.16 15.55 6.39
2.3 35.17 104.17 9.14 5.58
2.4 45.01 96.16 15.64 6.90
2.5 33.09 99.14 13.80 10.52
2.6 37.68 83.69 14.95 8.64
Figure 6.1 shows the test results from the experiments run with collector 2. Test 2.3 is not
shown because no froth was formed, and therefore no concentrate was made. Although the
Design Expert 10 software predicted the highest REO grade from test 2.4 and an REO/CaO ratio
of 3, the results did not reflect this. All of the tests suggested by the software did not work as
expected. This could be due to the fact that the software is mathematical, and chemical models
are more unpredictable than mathematical ones. Even though the new tests did not shown
anything promising, tests 2.1 and 2.2 exhibited high grades and recoveries which matched the
56
|
Colorado School of Mines
|
expected grades and recoveries. Test 2.1 showed the highest selectivity with a REO/CaO ratio of
4. The conditions for this test allowed it to reject 91% of the calcite, 95% of the barite and 95%
of the silicates, while maintaining a 70% recovery of REO.
100% 4.5
90% 4
80%
3.5
70%
3
60%
2.5
50%
2
40%
1.5
30%
1
20%
10% 0.5
0% 0
Untreated Ore 2.1 2.2 2.4 2.5 2.6
Test Number
REO Grade REO Recovery REO/CaO
Figure 6.1: The bench flotation results of the tests using collector 2.
6.2 Collector 5
Four bench scale tests were completed using collector 5. The test conditions are shown in
Table 6.3. Tests 5.3 and 5.4 used conditions that were predicted to have high REO grades and
recoveries as suggested by the Design Expert 10 software. The predicted grades and recoveries
are shown in Table 6.4. The Design Expert 10 software predicted that the tests using elevated
temperatures for this collector would result in higher grade and recovery of REO, while also
rejecting the gangue minerals. Further testing the collector at elevated temperatures was also
recommended by Everly. [17]
57
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
Table 6.3: Bench flotation conditions for tests using collector 5.
Continued Depressant
Test pH pH Temperature Concentration
Number Concentration pH Modifier Modification (oC) (mM)
Soda
5.1 1.00E-2 11 Ash Yes 20 0.75
Soda
5.2 5.00E-3 8.5 Ash No 20 0.75
5.3 5.00E-3 9.5 KOH No 50 0.75
5.4 5.00E-3 9.5 KOH No 82 0.75
Table 6.4: Predicted REO grades and recoveries and gangue mineral grades for bench tests using
collector 5.
Test Number REO Grade REO Recovery CaO Grade BaO Grade
5.1 10.03 98.06 18.60 13.98
5.2 10.51 97.05 17.71 14.57
5.3 18.59 116.58 14.27 14.29
5.4 17.56 167.10 16.56 12.92
The results of the bench test work featuring collector 5 can be seen in Figure 6.2. The
selectivity of this collector is low. Although the predicted grades and recoveries matched the
predicted ones, with the exception of test 5.3, the collector did little to upgrade the REO grade
compared to the other collectors. Test 5.4 was the most selective with a REO grade of 21%,
calcite recovery of 36%, barite recovery of 29% and silicate mineral recovery of 12%.
6.3 Collector 8
Four bench scale flotation tests were completed using collector 8. The test conditions can be
seen in Table 6.5. Tests 8.1 and 8.2 were suggested conditions by the Design of Experiments
software Design Expert 10. The predicted grades and recoveries for each of these tests can be
found in Table 6.6. The predicted grades are similar to those of the tests that had already been
completed, but the REO recovery was expected to be increased under these conditions.
58
|
Colorado School of Mines
|
The results of these tests can be seen in Figure 6.3. These tests most closely resembled the
predicted values for grade and recovery for both REO and the gangue minerals, with the
exception of test 8.3. Test 8.3 had an increased REO grade than what had been predicted. The
grade of the concentrate was 31%, while the predicted grade was 12%. This test condition had
been run previously by Dylan Everly. It is unknown what caused this discrepancy between the
results that he had obtained from these conditions and the results from this test. The collector
may have been better dissolved into the solution in this test or it is possible that less soda ash had
been used as a pH modifier. This test resulted in a recovery of 15% for calcite, 10% for barite
and 12% for silicates, while maintaining a 68% recovery of REO.
100% 3
2.5 80%
2
60%
1.5
40%
1
20%
0.5
0% 0
Untreated Ore 8.1 8.2 8.3 8.4
Test Number
REO Grade REO Recovery REO/CaO
Figure 6.3: Bench scale flotation results for the tests that utilized collector 8.
6.4 Collector 14
Four flotation tests were conducted using collector 14. The experimental conditions that were
used can be found in Table 6.7. Table 6.8 shows the predicted grade and recoveries for selected
minerals. Tests 14.1 and 14.4 were done using conditions suggested by the Design Expert 10
software. With these conditions it was predicted that REO recovery would be increased
significantly.
60
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
Table 6.7: The test conditions for the bench flotation tests conducted with collector 14.
Continued Depressant
Test Concentration pH pH Temperature Concentration
Number (M) pH Modifier Modification (oC) (mM)
14.1 3.00E-04 8.5 KOH No 82 0
14.2 3.00E-04 8.5 Soda Ash No 82 0.75
14.3 1.00E-04 8.5 Soda Ash Yes 82 1.5
14.4 1.00E-04 9.5 KOH No 82 0.75
Table 6.8: The predicted REO grade and recovery along with the predicted grade of gangue
minerals for bench tests using collector 14.
Test Number REO Grade REO Recovery CaO Grade BaO Grade
14.1 20.10 98.23 15.16 7.97
14.2 19.15 88.88 16.55 9.04
14.3 30.75 60.86 13.18 8.60
14.4 34.94 112.04 17.07 10.05
Figure 6.4 shows the results from the bench flotation test work conducted using collector
14. The tests conducted with collector 14 did not meet expectations. The only test that appeared
to have any selectivity was test 14.3. It had a grade of 20% REO. Again, there was a discrepancy
between the work that Dyaln Everly had done and these tests. With this collector it resulted in a
lower grade and recovery than what had been expected.
6.5 Conclusions
From these bench scale tests, a few stood out as promising for large scale test work. A
summary of the best tests can be seen in Figure 6.5. No tests from collector 5 were chosen
because it showed no selectivity for bastnaesite over the gangue minerals. Test 2.6 was chosen
because it had the best results of any test completed at a lower temperature. The tests for
collector 14 were chosen because they exhibited selectivity in the work done by Dylan Everly.
61
|
Colorado School of Mines
|
CHAPTER 7: LARGE SCALE FLOTATION TESTS
Large scale flotation tests were conducted to prove the process on a larger scale. They were
conducted using 10 kilograms of ore. The crushed ore was placed into a rod mill for 32 minutes
then taken from the rod mill and placed directly into the flotation cell. The cell was filled with
water to reach an approximate slurry density of 25 weight percent solids (10 kilograms ore, 30
liters water). If the test required heat, then it was heated using a heating coil. Once the
temperature was reached, the conditioning stages were started. The reagent additions were scaled
up at a 1:1 ratio with the exception of frother. Through practice tests that were conducted, it was
determined that for the 10 kilogram flotation tests only one drop of frother was required. The
flotation occurred for 2.8 minutes.
The conditions for the tests were chosen from the best results from the bench scale flotation
test work. None of the bench tests using collector 5 yielded results that indicated selectivity, so it
was not used for any large scale test work. The test conditions can be found in Table 7.1. Two of
the tests required reduced heat, while the other four required the slurry to be heated to 82oC. The
tests were conducted at a wide pH range (8.5-11) with the three of the tests being conducted at
8.5 pH.
7.1 Test 2.1
This test was the only one conducted at room temperature. The flotation time was two
minutes. The results can be seen in Figure 7.1. The results from this test indicated that there is a
difference between the bench scale tests and the large scale tests. The concentrate showed no
improvement in grade from the ore and there was no selectivity of calcite over bastnaesite. The
63
|
Colorado School of Mines
|
REO recovery was extremely low at 19%. Because there was a limited quantity of the collector
this test could not be conducted again for a longer time period, but the results indicate that this
would have no effect on the overall grade or recovery of REO. The reason for this having such a
low grade and recovery could be due to the fact that the particle size may have been larger than
anticipated or there may have been some precipitants in the water that was used, since it was not
distilled water.
Table 7.1: The test conditions for the large scale flotation test work.
Concent pH Continued Depressant
Temperature
Colle ration Modi pH Concentration
(oC)
Test ctor (M) pH fier Modification (mM)
1.00E-
2.1 2 02 9.5 KOH No 20 0.75
7.50E- Soda
2.2 2 03 8.5 Ash Yes 82 0
7.50E-
2.3 2 03 9.5 KOH No 50 0
2.50E- Soda
8.1 8 03 11 Ash No 82 0.75
3.00E- Soda
14.1 14 04 8.5 Ash No 82 0.75
1.00E- Soda
14.2 14 04 8.5 Ash Yes 82 1.5
7.2 Test 2.2
Test 2.2 was conducted using collector 2 at elevated temperature. This test was conducted for
two minutes. After two minutes, the froth was visibly less concentrated with mineralization. The
results can be seen in Figure 7.2. This test exhibited extremely high selectivity of bastnaesite
over the gangue minerals, the REO/CaO ratio is 7.
64
|
Colorado School of Mines
|
Where t is time in minutes and the grade and recovery are in percent. The REO grade
after one minutes was 15%, which was close to what was expected, but the REO recovery was
only 12%. This test showed some selectivity to bastnaesite after the first minute, but after eight
minutes the REO recovery was 23.3% while the recovery of the gangue minerals were 19.9%
barite, 19.8% calcite and 13.9% of the silicates. Overall, the test results were poorer than
expected. It is possible that the collector was unable to disperse properly in the slurry to assist in
flotation or the grind size was too large.
100% 1.2
90%
1
80%
70%
0.8
60%
50% 0.6
40%
0.4
30%
20%
0.2
10%
0% 0
Grade Recovery REO/CaO
Figure 7.5: The results of the large scale test 14.1. The concentrates are cumulative by the
minute.
7.6 Test 14.2
This was the final test completed with collector 14. The bench scale tests indicated that
this test should have a REO grade around 20%. The large scale test was run for eight minutes to
obtain enough material for analysis. The results of this test can be seen in Figure 7.6. This test
acted nothing like what the bench scale test had indicated. The REO grade after one minute was
69
yrevoceR
&
edarG
OaC/OER
|
Colorado School of Mines
|
needed to be dissolved in ethanol, they may not have been able to disperse properly on the large
scale. On the bench scale they were emulsified once they were place into solution, but on the
large scale this was not possible. It was theorized that since the agitation was so much greater,
that this would not be a problem for these collectors. The grind size may not have had a P of 50
80
microns either. Grinding is overall unpredictable, a general rule can be established, but with
different feeds the grind size could change drastically. This could cause problems in flotation on
any scale. One test from collector 2 was by far the best result obtained from this test work. It had
a REO grade of 44.4% after two minutes of flotation along with a REO recovery of 80.8%. It
simultaneously proved its selectivity by recovering only 5.0% of the barite, 9.2% of the calcite
and 5.6% of the silicates. The material from this test was used for further study on gravity
separations.
71
|
Colorado School of Mines
|
CHAPTER 8:GRAVITY SEPARATION
The gravity separation test work was completed using an ultrafine Falcon concentrator. This
centrifugal concentrator is used for materials with a small particle sizes. The flotation
concentrate from the large scale test 2.2 was used for this test. To be able to scale the Falcon
results up to plant scale a concentrate weight is required. Since the bowl is only able to hold a
limited amount of material, multiple passes needed to be made. For this test 1462.1 grams of the
flotation concentrate were used. Each pass was run in succession, with the tailings from the
previous pass being used as the feed for the next one. The flow rate was kept constant between
all the passes at 5 L/min., although it was difficult to keep it consistent. The slurry density was
15 weight percent solids for each pass and the rpms of the Falcon were kept at 1313 for all the
passes.
The rare earth oxide (REO) head grade of the feed was 39.0% and for calcite it was
11.3%. The difference in feed grade compared to the flotation product could be due to
preferential splitting. The sample was split out using a Jones splitter. The results of each pass can
be seen in Figure 8.1. As expected the REO recovery increases as the REO grade decreases. The
CaO rejection is good, but the grade of the CaO is greater after pass three than what it was for
the feed. Since CaO separation is the driving force for this project and the main gangue mineral
that should be removed in this process, it was the only one that was closely examined.
72
|
Colorado School of Mines
|
CHAPTER 9:ECONOMIC ANALYSIS
The goal of this project was to test a process that decreases costs associated with reagent
use. One of the largest costs comes from hydrochloric acid leaching of bastnaesite concentrates.
Because the calcite is a high acid consumer, providing a process that selectively separates it is
advantageous. This economic analysis was completed to compare the two processes. It is not a
full economic analysis. The comparative processes chosen were rougher, cleaner and scavenger
flotation, and just rougher flotation followed by a cleaner gravity stage.
9.1 Assumptions
A number of assumptions were made for this model. It was assumed that the
infrastructure would not be needed, but new equipment would be for both processes. The capital
costs were estimated based on a throughput of 100 tonnes of ore per hour, 2400 tonnes per day.
Both circuits used collector 2 as the primary reagent for flotation. After the cleaner stages only
the hydrochloric acid costs were considered. Hydrochloric acid’s primary purpose is to leach the
bastnaesite, but calcite is also an acid consumer. For calculations involving acid consumption the
bastnaesite leaching was not considered because it is necessary and consistent with how much
rare earth oxide is recovered. For this analysis it was assumed that it was consumed
stoichiometrically based on equation 9.1 and that no additional costs were incurred by the acid
consumption:
(9.1)
2(cid:60)(cid:22)(cid:61) +(cid:22)(cid:54)(cid:22)(cid:34)(cid:62) → (cid:22)(cid:54)(cid:22)(cid:61)(cid:18) +(cid:60)(cid:18)(cid:34)+(cid:22)(cid:34)(cid:18)
It was also assumed that no acid was recycled. The feed grade for both scenarios was
assumed to be 8% REO and 17% calcite. For the flotation circuit it was assumed that the grade
of the concentrate was 45% REO with a recovery of 80%, while for the float and gravity circuit
76
|
Colorado School of Mines
|
it was assumed that the concentrate grade was 50% REO with a recovery of 90% from the
flotation concentrate. It was also assumed that for the flotation circuit concentrate the calcite
grade was 9% with a recovery of 9% and for the flotation and gravity circuit the calcite grade
was 7% with an additional rejection of 40%. The grades and their recoveries were calculated
from the results shown in chapters 7 and 8. It was assumed that 90% of REO in the final
concentrate was recovered from leaching and solvent extraction. The final products were high
purity cerium oxide (>99.5%), lanthanum oxide (>99.5%), praseodymium (>99%) and
neodymium oxide (>99.5%). It was assumed that the final REO product consisted of 50%
cerium, 33% lanthanum, 4% praseodymium, and 12% neodymium. Taxes and other downstream
capital and operating costs were not calculated for the final analysis.
9.2 Capital Costs
The capital equipment costs were estimated using CostMine 2017. The equipment and
their associated costs are outlined in Table 9.1 for the flotation circuit, and Table 9.2 for the
flotation and gravity circuit.
Table 9.1: Capital equipment costs for the flotation circuit.
Equipment Quantity Cost Per Unit
Feeder 2 $16,005.59
Jaw Crusher Double Toggle 1 $305,600.00
SAG Mill (6.7x2.1 m) 1 $2,975,176.09
SAG Motor 1 $87,701.86
Conveyor Belt 1 $11,192.95
Wet Ball Mill (4.3x7.9 m) 1 $2,412,897.52
Ball Mill Motor 1 $186,278.76
Cyclone (91.4 cm Diameter) 1 $18,691.46
Regrind Ball Mill (3x3.7 m) 1 $955,402.17
Rougher Tank Mixers (88.9 cm impeller diameter) 4 $25,214.29
77
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.