Upload
others
View
10
Download
0
Embed Size (px)
Citation preview
Experimental study and steady-state simulation of
biogeochemical processes in laboratory columns
with aquifer material
Aria Amirbahmana,*, Rene Schonenbergerb, Gerhard Furrerc,Jurg Zobristb
aDepartment of Civil and Environmental Engineering, University of Maine, 5711 Boardman Hall, Orono, Maine,
ME 04469 5711, USAbSwiss Federal Institute of Environmental Science and Technology (EAWAG), Switzerland
c Institute of Terrestrial Ecology (ITO), Swiss Federal Institute of Technology (ETH), Zurich, Switzerland
Received 18 June 2001; received in revised form 5 August 2002; accepted 23 August 2002
Abstract
Packed bed laboratory column experiments were performed to simulate the biogeochemical
processes resulting from microbially catalyzed oxidation of organic matter. These included aerobic
respiration, denitrification, and Mn(IV), Fe(III) and SO4 reduction processes. The effects of these
reactions on the aqueous- and solid-phase geochemistry of the aquifer material were closely
examined. The data were used to model the development of alkalinity and pH along the column. To
study the independent development of Fe(III)- and SO4-reducing environments, two columns were
used. One of the columns (column 1) contained small enough concentrations of SO4 in the influent to
render the reduction of this species unimportant to the geochemical processes in the column.
The rate of microbially catalyzed reduction of Mn(IV) changed with time as evidenced by the
variations in the initial rate of Mn(II) production at the head of the column. The concentration of Mn
in both columns was controlled by the solubility of rhodochrosite (MnCO3(S)).
In the column where significant SO4 reduction took place (column 2), the concentration of
dissolved Fe(II) was controlled by the solubility of FeS. In column 1, where SO4 reduction was not
important, maximum dissolved Fe(II) concentrations were controlled by the solubility of siderite
(FeCO3(S)). Comparison of solid-phase and aqueous-phase data suggests that nearly 20% of the
produced Fe(II) precipitates as siderite in column 1. The solid-phase analysis also indicates that
during the course of experiment, approximately 20% of the total Fe(III) hydroxides and more than
70% of the amorphous Fe(III) hydroxides were reduced by dissimilatory iron reduction.
The most important sink for dissolved S(-II) produced by the enzymatic reduction of SO4 was its
direct reaction with solid-phase Fe(III) hydroxides leading initially to the formation of FeS.
0169-7722/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0169-7722(02)00151-1
* Corresponding author. Tel.: +1-207-581-1277; fax: +1-207-581-3888.
E-mail address: [email protected] (A. Amirbahman).
www.elsevier.com/locate/jconhyd
Journal of Contaminant Hydrology 64 (2003) 169–190
Compared to this pathway, precipitation as FeS did not constitute an important sink for S(-II) in
column 2. In this column, the total reacted S(-II) estimated from the concentration of dissolved sulfur
species was in good agreement with the produced Cr(II)-reducible sulfur in the solid phase. Solid-
phase analysis of the sulfur species indicated that up to half of the originally produced FeS may have
possibly transformed to FeS2.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Aquifer; Groundwater; Iron; Redox processes; Sulfur
1. Introduction
The presence of high concentrations of oxidizable organic matter in groundwater can
lead to the development of reducing conditions due to microbial activities. Geochemical
transformations brought about by the oxidation of organic matter can have important
implications for the quality of groundwater (Jacobs et al., 1988). These transformations
include reduction of electron acceptors such as dissolved O2, NO3, and SO4, reductive
dissolution of Fe- and Mn-(hydr)oxides, and precipitation or dissolution of various
carbonate and sulfide minerals. Alkalinity and pH of the groundwater also change due
to these transformations (von Gunten et al., 1994). The presence of a reduced zone can
have significant consequences for the mobility of toxic organic and inorganic chemicals.
We have simulated the development of Fe(III)- and SO4-reducing environments
downgradient from a municipal solid waste landfill (Riet, Switzerland) under well-
controlled laboratory conditions. This landfill has been the subject of previous studies
(Amirbahman et al., 1998; Abbaspour et al., 1999; Hoehn et al., 2000). Solid material
collected from the oxic part of the aquifer and inoculated with bacteria from the anaerobic
leachate was packed in two flow-through column reactors. Anaerobic conditions were
established by running solutions of known concentrations of organic carbon and electron
acceptors through the columns. Geochemical changes caused by anaerobic conditions in
aqueous and solid phases along the columns were studied. Of special interest was to study
the contributions of the enzymatic Fe(III)-reducing conditions and the dissolved S(-II)
species generated by the SO4-reducing bacteria to the oxidation capacity (OXC) of the
aquifer as defined by Heron and Christensen (1994, 1995) and Heron et al. (1994a). This
approach allows us to quantify the rates of geochemical transformations brought about by
the redox processes, and to evaluate the capacity of the aquifer solid phase to buffer
against the development of highly reducing conditions.
2. Experimental
2.1. Column design and operation
Columns were made of Plexiglas with an inner diameter of 5 cm and contained side
ports with Teflon-lined septa at 0.5, 1, 1.5, 2, 3, 4, 5, 6, 10, 14, 18, 22 and 26 cm. They
were filled with aquifer solid collected from the oxic zone of well 5 in the vicinity of
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190170
landfill Riet, Winterthur, Switzerland (Amirbahman et al., 1998). The aquifer solid was
first dried at 60 jC and then fractionated by nylon sieves. The fine sand fraction between
150 to 465 Am was filled into the columns up to a depth of 29 cm according to the
procedure of Stauffer and Dracos (1986) to insure homogeneous packing. The columns
were then filled with degassed water at ionic strength of 10 mM NaCl.
Two columns were operated separately in this study. Analysis of the effluent break-
through curves with tritium indicated that column 1 had a pore volume of 175 ml and a
porosity of 0.31. The pore volume and porosity for column 2 were 192 ml and 0.34,
respectively. Columns were operated at a flow rate of 0.45 ml min� 1 in the dark by
covering them with aluminum foil at room temperature.
The background influent solution was prepared twice a week by bubbling 100% CO2(g)
for 10–30 min and then 1% CO2(g) for 60 min through a suspension of calcite. This
procedure produced an alkalinity of 3.72F 0.65 meq l� 1. The average influent concen-
trations of sodium propionate (NaC3O2H5) and NaNO3 in both columns were 0.61 and
0.13 mM, respectively. The concentration of propionate in column 2 was increased to 1.0
mM after day 78 of operation. To suppress the dissimilatory SO4 reduction in column 1, no
SO4 was added to this column. However, the influent of column 1 contained a small
amount of SO4 due to impurities in CaCO3 and other salts used. This SO4 did not produce
any appreciable concentration of S(-II) that influenced the dissolved Fe concentration in
column 1. The influent solution in column 2 consisted of an average of 0.52 mM Na2SO4.
The influent solutions in both columns also contained 10–100 nM concentrations of
Co(II), Zn(II), Cu(II), Ni(II) and Mo(VI), 0.6 AM PO4 as nutrient and 10 mM NaCl as
ionic strength buffer.
Both columns were inoculated at the inlet with anaerobic bacteria which were present in
groundwater samples collected under anoxic conditions from well 1 of the landfill Riet
(Amirbahman et al., 1998). Anoxic condition was established in both columns within the
first 7 days of operation. All chemicals used were of reagent grade or better and the water
employed was double distilled.
2.2. Analysis of the aqueous phase
Dissolved oxygen was measured using a combined LF-196 instrument (WTW, Well-
heim, Germany) in a closed flow-through cell. A combination pH electrode (Metrohm)
and a pH meter (Metrohm 654) were used to measure the pH. Alkalinity was measured by
standardized 10 mM HCl to pH 4.5. Alkalinity and pH were measured immediately after
the samples were collected. The concentration of the dissolved inorganic carbon (DIC)
was calculated using pH and alkalinity measurements and by considering the contributions
of propionate, acetate and dissolved S(-II) to alkalinity. Propionate, acetate, SO4 and NO3
were measured simultaneously by ion chromatography (Dionex) with an AS11 anion
exchange column using the gradient technique (Ammann and Ruttimann, 1995). The
dissolved organic carbon (DOC) analysis of the effluent for a selected few samples
indicated that more than 95% of the effluent DOC was due to propionate and acetate.
Effluent samples for metal and S(-II) analyses were collected in 1 M HNO3 and 0.1 M zinc
acetate solutions, respectively, such that oxidation or volatilization would be minimized.
Dissolved Fe, Mn, K, Na, Mg, and Ca were measured using ICP-OES (Spectro Analytical
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 171
Instruments). Dissolved S(-II) concentrations were measured using the photometric
method with methylene blue (Zhabina and Volkov, 1978).
2.3. Analysis of the solid phase
Chemical extraction techniques were used to analyze the solid samples for Fe and
reduced sulfur. The solid-phase Fe(III) was extracted by the Ti(III)–EDTA method, which
represents the total reducible Fe(III) phases (Heron et al., 1994a). This technique is
capable of completely reducing amorphous synthetic hydrous ferric oxide (HFO) and
lepidocrocite, and reducing more than 90% of synthetic hematite and goethite, but extracts
only 5% of magnetite and less than 5% of siderite (Heron et al., 1994b).
The concentration of the amorphous Fe(III) hydroxide was determined by chemical
extraction using ascorbate at pH 8 (Kostka and Luther, 1994). Extraction of synthetic
Fe(III) hydroxides showed that nearly all of amorphous HFO and less than 2% of
lepidocrocite and goethite are extracted by this technique (Amirbahman et al., 1998).
The ion exchangeable Fe(II) was extracted by mixing a suspension of sand in a 1 M CaCl2solution at pH 7 for 24 h (Heron et al., 1994b).
Acid volatile sulfide (AVS), which primarily reflects the amorphous FeS, was measured
using the technique developed by Zhabina and Volkov (1978). Cr(II)-reducible sulfide
(CRS) technique was used to measure the total reducible S (Canfield et al., 1986). In
general, CRS is a measure of FeS, FeS2, and elemental and organic S. Our previous CRS
measurements of similar soil under a landfill indicate that the elemental and organic S are
very small fractions of total reducible S (Amirbahman et al., 1998). A detailed description
of the extraction methods used is given previously (Amirbahman et al., 1998).
Quantitative analysis of the solid before packing was performed with a scanning
electron microscope (JEOL JSM-840) using an energy dispersive X-ray spectrometer
TRACOR 5402. The results were in agreement with X-ray diffraction patterns and
indicated that the main constituents are Si (25.3F 2.0% wt.) from quartz and silicates
(primarily feldspar and clay minerals) and Ca (23.5F 2.9% wt.) from calcite. Metals such
as Fe (3.0F 0.7% wt.) and Al (4.5F 0.4% wt.) were detected in smaller amounts. Due to
their relatively small amounts, which is typical of most natural samples, crystalline and
amorphous Fe(III) hydroxides were not distinguished.
3. Steady-state geochemical modeling
We have used the computer program STEADYQL (Furrer et al., 1989) to perform
geochemical calculations and sensitivity analysis of the results of our column experiments.
This program considers chemical speciation and reaction kinetics in a continuous-flow
stirred tank reactor. The flow domain is subdivided into several box reactors in series, where
the outflow from the first reactor constitutes the inflow for the second reactor. Therefore, the
column system is modeled as several continuously stirred tank reactors in series. Reaction
rates for geochemical processes are obtained by combining the concentration gradient of a
given component with the flow velocity. Reaction rates are assumed to be at steady-state
conditions during the course of sample collection. STEADYQL has been previously used in
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190172
other studies involving biogeochemical changes in flow-through systems (Furrer et al.,
1996; Amirbahman et al., 1998). The compartment between two consecutive sampling
ports were considered one box reactor, where the concentrations of components in the
inflow and the outflow are known, and are used to determine the reaction rate coefficients.
The processes in the columns were divided into ‘‘fast’’ and ‘‘slow’’ categories. The
‘‘fast’’ processes consist of equilibrium acid–base and aqueous complexation reactions,
and are reported elsewhere (Amirbahman et al., 1998). The ‘‘slow’’ processes are kineti-
cally controlled and consist of all of the microbially catalyzed redox reactions (Table 1), and
reaction of dissolved S(-II) and solid-phase Fe(III) hydroxides. The reaction rate coef-
ficients were determined based on concentration of the independent components, such as
dissolved O2, NO3, and SO4. The dependent species are those species whose turnover rates
and concentrations are affected by all of the biogeochemical processes involving the
independent species. The dependent species here are propionate, acetate, DIC and pH. The
modeled values of the dependent species can be compared to the actual experimental
measurements to determine the validity of the modeling approach and assumptions.
Zero-order rate coefficients were used to describe the geochemical processes. For
microbially catalyzed processes, the rate coefficients were calculated by dividing the
change in the concentration of the corresponding independent components in a given
reactor by the mean residence time in that reactor. The computer program STEADYQL
performs sensitivity analysis with respect to the rate coefficients that control the
concentrations of the dependent species. This provides us with useful information on
changes in concentrations of the dependent species brought about by changes in
concentrations of the independent components. The reported sensitivity coefficients are
normalized with respect to the rate coefficients, which are directly calculated from
concentrations of the independent components.
4. Results and discussion
4.1. Reduction of dissolved oxygen and nitrate
The microbially catalyzed redox processes for both columns are listed in Table 1. At the
end of the first week, all of dissolved O2 (0.25 mM) was consumed within the first 0.5 cm
Table 1
Stoichiometry of microbially catalyzed kinetic processes
Processes Stoichiometry
Aerobic respiration 2C3H5O2� + 7O2 + 2H
+! 6CO2 + 6H2O
Denitrification 5C3H5O2� + 14NO3
� + 4H+! 15HCO3� + 7N2 + 7H2O
Mn(IV) reduction C2H3O2� + 4MnO2 + 7H
+! 4Mn2 + + 2HCO3� + 4H2O
Fe(III) reduction C2H3O2� + 8FeOOH+ 15H+! 8Fe2 + + 2HCO3
� + 12H2O
Sulfate reduction 4C3H5O2� + 3SO4
2�! 4C2H3O2� +H+ + 4HCO3
� + 3HS�
Fermentation of propionate to acetate C3H5O2� + 3H2O!C2H3O2
� +H+ +HCO3� + 3H2
C3H5O2�: propionate
C2H3O2�: acetate
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 173
of the columns. Since O2 is likely to have been depleted at a shorter distance than the first
sampling port at 0.5 cm, the estimated zero-order rate coefficient of 9 nM s� 1 constitutes a
minimum value.
Nearly all added NO3 (average inlet concentration of 0.13 mM) was reduced in the
first 0.5 cm of both columns, even though at different initial rates in each column. The
onset of denitrification in column 1 was between days 7 and 14 after inoculation. By
day 14, nearly 95% of NO3, and by day 35, more than 98% of NO3 was reduced within
the first 0.5 cm. In column 2, a total of approximately 35% of NO3 was reduced by day
7. This reduction was almost all limited to the first 0.5 cm. By day 21, more than 90%
of NO3 was reduced in the first 0.5 cm with the rest being reduced in the following 1.5
cm. By day 35, nearly 97% of NO3 was reduced in the first 0.5 cm of column 2. The
increase in the rate of denitrification with time is due to the establishment and
acclimation of the NO3-reducing bacteria. In a similar system, von Gunten and Zobrist
(1993) have observed the onset of denitrification between days 1 and 6 after the
inoculation. They observed a virtually complete NO3 reduction within the first 1 cm of
the column.
4.2. Manganese reduction
Fig. 1 shows the concentration distribution of dissolved Mn(II) on different days in
both columns. Solid-phase Mn(IV) undergoes early reduction following the establishment
of anoxic environment in both columns. The onset of Mn(IV) reduction in both columns
was before day 7 of operation. For the first 14 days, when still negligible concentrations of
SO4 were reduced in column 2, nearly equivalent amounts of dissolved Mn(II) were
observed in both columns. In column 1, the initial zero-order rate coefficients of dissolved
Mn(II) production on days 7 and 14 were 0.28 and 1.02 nM s� 1, respectively.
Corresponding values of 0.27 and 1.11 nM s� 1 were observed in column 2. By day 35,
the initial dissolved Mn(IV) reduction rate coefficients diminished to 0.32 and 0.48 nM
s� 1 in columns 1 and 2, respectively. These rates were observed close to the inlet of the
columns, where initially the bulk of the bacterial biomass was located, the solution was
still undersaturated with respect to rhodochrosite (MnCO3(S)), and the adsorption sites at
the soil surface were likely to be saturated with respect to Mn(II). Saturation of surface
sites with respect to reduced metals has been substantiated by independent adsorption
experiments for the same aquifer material (results not shown). Therefore, we believe that
these rates are those of the overall Mn(II) production. The initial dissolved Mn(II)
production rate coefficients diminished with time to below 0.05 nM s� 1 by day 219 in
both columns.
4.3. Iron reduction
Dissolved Fe(II) was observed in both columns before day 7. It is assumed that the
Fe(II) release is due to the dissimilatory reduction of solid-phase Fe(III) hydroxides in the
presence and absence of SO4 (Jakobsen and Postma, 1999). First, we present the results
from column 1, where precipitation of FeS did not take place due to the presence of small
concentrations of SO4 in the influent.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190174
Fig. 2 exhibits the distribution of dissolved Fe(II) along the length of column 1 at
different times. This distribution is characterized by relatively higher Fe(II) release rates
close to the inlet than at the outlet of the column. Dissolved Fe(II) concentrations along the
length also increased up to day 85, after which they decreased.
The observed release rate of dissolved Fe(II) reflects not only the reduction rate of
Fe(III) hydroxides, but also adsorption, desorption and precipitation of Fe(II). To study
variations in the reduction rate of Fe(III), we have considered the dissolved Fe(II)
Fig. 1. Distribution of dissolved Mn(II) in (a) column 1, and (b) column 2. The numbers refer to the days samples
were collected for measurement.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 175
distribution at the first 2 cm of column 1. Since at this location the adsorption sites are
likely to be saturated with respect to Fe(II) and the alkalinity and Fe(II) concentrations are
not high enough to induce precipitation of siderite, the measured initial production rates of
dissolved Fe(II) may be the best approximation for the dissimilatory Fe(III) reduction
rates. Close to the inlet of column 1, a zero-order rate of 0.1 nM s� 1 for Fe(III) reduction
is observed on day 7 (data not shown). However, the concentration of dissolved Fe(II)
decreases after the first 2 cm due to precipitation of siderite (Fig. 3a). The initial Fe(II)
release rate increases to 4.5 nM s� 1 by day 70, after which it decreases perhaps due to the
depletion of the more reactive (amorphous) solid Fe(III) hydroxide species. By day 85 of
operation, the initial rate of Fe(II) production in column 1 is 2.5 nM s� 1 and continues at
an initial rate of 2.1 nM s� 1 until the end of column operation on day 219. Even though
the rate of dissimilatory Fe(III) reduction cannot be directly determined for downgradient
Fig. 2. Distribution of dissolved Fe(II) in column 1. The numbers refer to the days samples were collected for
measurement.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190176
Fig. 3. Saturation index with respect to (a) siderite in column 1, (b) rhodochrocite in column 1, and (c)
rhodochrocite in column 2.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 177
locations in column 1, we expect that time variation of this rate would be similar to that
measured close to the inlet of this column.
4.3.1. Precipitation of siderite
Solubility calculations indicate that siderite should be considered as a sink for the
dissolved Fe(II) in column 1. Fig. 3a shows the saturation index of siderite as a function
of distance at different times in this column. Precipitation of siderite in not expected close
to the inlet of this column at any time. Therefore, at a distance no longer than 1.5–2 cm,
concentration of dissolved Fe(II) is controlled only by dissimilatory Fe(III) reduction.
Initially, the solution is undersaturated with respect to siderite throughout the column, due
to low concentrations of dissolved Fe(II) and DIC. By day 41, however, precipitation may
occur toward the end of the column (Fig. 3a). This increase in saturation index is due to
the increasing production of DIC and dissolved Fe(II) during this time. The pore water
remains close to saturation along most of the length of the column until the end of the
experiment. Toward the end of the operation of column 1, where high concentrations of
DIC are produced, precipitation of siderite is expected at distances as short as 1.5 cm
from the inlet. Unlike field conditions, complexation of Fe with organic ligands is
negligible in this study. Therefore, in chemical speciation modeling, all measured Fe
species in solution have been considered as either free Fe(II) or Fe(II) complexed with
carbonate species.
4.3.2. Solid-phase analysis of column 1
Solid-phase extractions of both columns were performed before and after the experi-
ment. Fig. 4a illustrates the distribution of the solid-phase Fe(III) hydroxides in column 1
obtained using Ti(III)–EDTA extraction. These are the residual Fe(III) hydroxides that
were measured after the completion of the experiment. The data in Fig. 4 have been
corrected for the Fe(II) adsorbed to aquifer matrix by subtracting the concentration of
Fe(II) extracted by CaCl2 (results not shown here).
Ti(III)–EDTA extraction results indicate that initially the solid phase contained a total
concentration of 21 Amol Fe(III) g� 1. The total residual concentration of Ti(III)–EDTA
extractable Fe(III) averaged over the entire column length is 17 Amol g� 1. Therefore,
according to these measurements, a total of 4 Amol Fe(III) g� 1, or approximately 19% of
the total Ti(III)–EDTA extractable Fe(III) hydroxides, has been reduced. The total
adsorbed Fe(II) concentration, as estimated by CaCl2 extraction of the solid phase and
averaged over the entire length of column 1, is only 0.06 Amol g� 1, which is a very small
part of the total produced Fe(II).
Fig. 4b shows the distribution of the residual Fe(III) hydroxides extracted using
ascorbate along the length of column 1. This fraction, which consists largely of the
amorphous Fe(III) phases, had an initial concentration of 1.1 Amol g� 1. The residual
concentration of the ascorbate extractable Fe(III) hydroxides averaged over the length of
the column is 0.3 Amol g� 1. This concentration may even be lower due to the possible
oxidation of small amounts of adsorbed Fe(II) during sample handling and measurements.
Previous studies of biotic and abiotic reductive dissolution of solid Fe(III) hydroxides
have shown that amorphous phases dissolve more readily than the crystalline phases
(Lovley et al., 1991; Amirbahman et al., 1997). The same behavior is also observed in this
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190178
study. Comparison of the initial and residual masses of ascorbate extractable Fe(III)
indicates that more than 70% of the amorphous Fe(III) phase has been reductively
dissolved in column 1. Even though only 5% of the total extractable Fe(III) hydroxides
are amorphous, they constitute nearly 20% of the total reduced Fe(III). During the course
of the experiment, higher concentrations of the ascorbate-extractable Fe(III) phases have
been dissolved closer to the inlet of the column.
Table 2 summarizes solid- and liquid-phase measurements of total Fe in column 1.
Total initial and residual Ti(III)–EDTA extractable amounts of Fe(III) hydroxides are 22.0
and 17.8 mmol, respectively. The total mass of reduced Fe(III) hydroxides estimated from
solid-phase data is, therefore, 4.2 mmol. The total mass of aqueous phase Fe(II) that has
been released from column 1 during the course of experiment is 3.3 mmol. Therefore, the
total mass of Fe(II) reacted in the column, which is the difference between the total
reduced Fe(III) hydroxides and the total aqueous Fe(II) output, is 0.9 mmol. Accounting
for the total adsorbed mass of Fe(II) as estimated by CaCl2 extractions, the total mass of
Fig. 4. Concentration of solid-phase Fe(III) hydroxides in column 1 extracted by (a) Ti(III)–EDTA, and (b)
ascorbate. A background concentration of 3.8F 0.7 Amol Fe/g solid before the experiment has been subtracted
from the Ti(III) –EDTA extractable Fe data.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 179
Fe(II) precipitated as siderite is approximately 0.8 mmol (Table 2). Adsorbed Fe(II)
constitutes only a small part of the total Fe(II) budget.
Given that the reduced Fe(III) in column 1 is only a small fraction of the total Ti(III)–
EDTA extractable Fe(III) hydroxides, any uncertainties in analytical measurements could
lead to significant errors in the estimation of precipitated siderite. According to the data
presented in Table 2, nearly 20% (F 10%) of the total produced Fe(II) precipitates as
siderite, while the rest leaves the column.
4.3.3. Average rates of Fe(III) reduction
Analysis of the rates of release of dissolved Fe(II) close to the inlet of column 1
discussed above suggests that these rates change with time due to the preferential depletion
of the amorphous compared to the more crystalline Fe(III) hydroxides. We have used the
initial and residual concentrations of Ti(III)–EDTA extractable Fe(III) hydroxides to
estimate the zero-order rate coefficients for the reduction of Fe(III) at different locations
along the column, by taking into account the flow rate and the mean residence time. These
reduction rate coefficients are averaged over the entire duration of operation of column 1
and are shown in Fig. 5. According to Fig. 5, the average zero-order Fe(III) reduction rate
decreases approximately three-fold from the inlet to the outlet of the column. Given the
uniformly available surface area of Fe(III) hydroxides along the column, the most likely
explanation for the decrease in the Fe(III) reduction rate in the direction of flow would be
the non-uniform distribution of the density of the Fe(III)-reducing bacteria.
The average rate of reduction of Fe(III) hydroxides at 1.5 cm estimated from the solid-
phase analysis is 1.9 nM s� 1 . Using the aqueous phase data, this rate is 2.4 nM s� 1 at the
first 2 cm of column 1. The latter rate was estimated by integrating and averaging the
initial zero-order rates with respect to time. The agreement between the data estimated
Table 2
Mass balance of iron species in column 1
Measurements and procedures Mass (mmol)
Total initial Ti(III) –EDTA extractable Fe(III)a 22.0
Total residual Ti(III) –EDTA extractable Fe(III)a,b 17.8
Total reduced Fe(III) hydroxidesc 4.2
Initial ascorbate extractable Fe(III)a 1.2
Residual ascorbate extractable Fe(III)a,b 0.3
Total reduced amorphous Fe(III) hydroxides 0.9
Residual CaCl2 extractable Fe(II)d 0.06
Total dissolved Fe(II) outpute 3.3
Total precipitated Fe(II)f 0.8
a Total reducible Fe(III) in the solid phase extracted directly and integrated over the entire length of column.b These values have been corrected for the CaCl2 extractable Fe(II).c The difference between total initial Fe(III) and total residual Fe(III) obtained by Ti(III) –EDTA extraction.d The ion exchangeable Fe(II) adsorbed to the solid surface.e Total dissolved Fe(II) released from the column over the course of experiment.f Total precipitated Fe(II) = Total reduced Fe(III) hydroxides�Total dissolved Fe(II) output�Total residual
CaCl2 extractable Fe(II).
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190180
from the solid and aqueous phases is reasonable considering that sensitivity analysis
indicates an approximately 40% variation in the values of the average reduction rates with
a 20% change in the concentration of residual Ti(III)–EDTA extractable Fe(III).
4.4. Comparison of Mn and Fe turnover
During the first 3 weeks of operation of the columns, the dissolved Mn(II) concen-
trations were higher than those of the dissolved Fe(II), indicating a higher rate of reduction
of Mn(IV) oxides. By day 35, however, higher concentrations of dissolved Fe(II) were
observed in column 1, where there was no FeS precipitation. The decrease in the initial
rate of Mn(II) production with time is likely due to the depletion of the more available
fractions of Mn(IV) oxides by the Mn-reducing bacteria, as suggested previously (von
Gunten and Zobrist, 1993). The removal of dissolved Mn(II) can be explained by two
possible mechanisms: precipitation as rhodochrosite and/or adsorption to FeS and other
solids, with adsorption to FeS possible only in column 2. As indicated by equilibrium
calculations, after day 14, precipitation of Mn(II) as rhodochrosite took place (Fig. 3b and
c). Equilibrium with respect to rhodochrosite was maintained throughout the duration of
this study in both columns. The overall decrease in the dissolved Mn(II) concentration
with time is, therefore, explained by the gradual increase in alkalinity, which was in turn
generated by the microbially catalyzed reactions.
Other likely mechanisms for the removal of Mn(II) are its adsorption to and
incorporation into FeS and other solids. Morse and Luther (1999) have argued that
Fig. 5. The average rate of dissimilatory Fe(III) hydroxide reduction in column 1 estimated from the solid-phase
measurements.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 181
despite the higher rate of water exchange for Mn2 + than for Fe2 +, Mn2 + does not
readily form a MnS phase. Instead, they have proposed incorporation of Mn(II) into FeS
and eventually into pyrite. However, the latter mechanism may be kinetically con-
strained, as observed for the dissolution of sulfide minerals (Morse and Luther, 1999).
Kornicker and Morse (1991) and Arakaki and Morse (1993) have shown that dissolved
Mn(II) can also adsorb onto FeS phases and be pyritized. From our data, we cannot
estimate the fraction of Mn(II) removal by each mechanism. However, the relatively
constant concentrations of this species at any given time throughout the length of both
columns suggest that precipitation as rhodochrosite would be the dominant removal
mechanism.
4.5. Sulfate reduction
This section discusses the geochemical changes brought about due to the reduction of
SO4 in column 2. Production of S(-II) from SO4 (average influent concentration of 0.52
mM in column 2) was completely catalyzed by the SO4-reducing bacteria, with propionate
as the sole electron source for the reduction of SO4 (Table 1).
The diagenetic sulfur cycle has been described in detail elsewhere (Davison, 1991;
Luther et al., 1992). Reaction of S(-II) with iron species generally follows two major
reaction pathways. The first pathway involves the direct attack on the solid-phase Fe(III)
by aqueous S(-II) to readily form FeS at the surface. This is the main sink for the dissolved
S(-II) in most natural systems, where a sufficient pool of available solid-phase Fe(III)
hydroxides exists. The FeS mineral formed via this pathway undergoes further reaction to
form pyrite (FeS2) (Berner, 1984). The stoichiometry of direct S(-II) attack is (von Gunten
and Zobrist, 1993)
2FeOOHþ 3HS� ! 2FeSþ ^ Sð0Þ]]þ 3OH� þ H2O ð1Þ
FeSþ ^ Sð0Þ]] ! FeS2 ð2Þ
The second pathway is the reaction between dissolved S(-II) and Fe(II) and precipitation as
FeS, provided that the solubility limit of the mineral is reached:
Fe2þ þ HS� ! FeSþ Hþ ð3Þ
Typical profiles for the time-dependent distribution of aqueous S(-II) along the length
of column 2 are shown in Fig. 6. On day 7, the measured concentrations of S(-II) were
below the detection limit (0.15 AM) along the column. Dissolved S(-II) was detected at 0.5
cm from the inlet on day 14. The total concentration of dissolved S(-II) along the column
increased in the downgradient direction with time. After day 85, S(-II) reached the end of
the column and was detected in the effluent. By the end of operation of column 2,
dissolved S(-II) traveled nearly conservatively (Fig. 6).
Fig. 6 also shows the change in the concentration of SO4 for day 85. A similar pattern
was observed throughout the duration of the experiment, where SO4 reduction initiated
close to the column inlet because of the high concentration of the organic substrate and
density of the SO4-reducing bacteria. For example, on day 85, more than 98% of SO4 was
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190182
reduced within the first 4 cm of column 2. Kinetics of SO4 reduction process in this zone
may be characterized by a pseudo-first-order reaction with a rate coefficient of 7.4� 10� 4
s� 1 . The limiting factor in the dissimilatory reduction of SO4 was the availability of
propionate. During the course of this experiment, the bacteria did not use acetate as an
electron acceptor to reduce SO4, despite its presence in column 2.
Lack of stoichiometric correspondence between the distribution of SO4 and dissolved
S(-II) is evident (Fig. 6) and is attributed to the reactions of S(-II) with Fe(II) and Fe(III) as
depicted in Eqs. (1)–(3). The kinetics of production of dissolved S(-II) may be followed
by changes in SO4 concentration.
4.5.1. Solid-phase analysis of column 2
Solid-phase extraction of sulfur species in column 2 was performed before and after the
experiment. These extractions included AVS and CRS as described in the Experimental
section. Fig. 7 shows the results from AVS and CRS extractions of solids after the
experiment in column 2. Solid-phase sulfur forms initially as amorphous FeS. Further
transformation to pyrite takes place only later in the solid phase (Eq. (2)).
Fig. 6. Distribution of dissolved sulfur species (S(-II) and SO4) in column 2. The numbers refer to the days
samples were collected for measurement. The empty symbols represent S(-II) and the filled circle represents SO4
on day 85.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 183
Integration of concentrations of the solid-phase sulfur species along the length of the
column indicates that a total of 30.4 mmol of CRS and a total of 15.2 mmol of AVS were
generated in column 2 during the course of the experiment. This means that half of the
originally formed FeS has been transformed into pyrite. Given our analytical procedure for
determination of AVS, however, it may be difficult to resolve the exact extent of
transformation of FeS to pyrite. This is due to the possible enhanced oxidation of H2S
by dissolved Fe(III) under acidic condition and at an elevated temperature during the AVS
extractions (von Gunten and Zobrist, 1993). Such artifact, if indeed effective, would
underestimate the concentration of AVS. Therefore, the measured AVS reported here
should be considered as the lower limit. Likewise, the extent of transformation into pyrite
would perhaps be less than half of the originally formed FeS.
Table 3 presents the mass balance of sulfur species in column 2. During the 219 days of
experiment, the total SO4 input into column 2 was 72 mmol, and the measured total SO4
output from the end of this column was 16 mmol. The latter value was estimated by
integrating and averaging the total SO4 output with respect to time. The difference between
total SO4 input and output (56 mmol) is equivalent to the total S(-II) produced in column 2.
A total dissolved S(-II) output mass of 23 mmol was estimated from our measurements. The
total reacted S(-II) of 33 mmol during the course of the experiment is simply the difference
between the total S(-II) produced and the S(-II) released from the column. This is in good
agreement with the total mass of solid-phase sulfur species produced from the dissimilatory
Fig. 7. Concentration of solid-phase reduced sulfur in column 2 extracted by AVS and CRS techniques. The
background concentrations of AVS and CRS in oxic solid phase are 0.3F 0.1 and 16.0F 7.1 Amol S/g solid,
respectively.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190184
SO4 reduction as determined by the CRS method (30 mmol). The two values are
highlighted in bold in Table 3.
Heron et al. (1994a) defined the OXC of an aquifer as its ability to restrict the
migration of a reduced plume. Operationally, they quantified the OXC as the equivalent
concentration of Ti(III)–EDTA reducible Fe(III). Assuming that all of the AVS (15.18
mmol) consists of sulfur in FeS and all of the CRS (30 mmol) consists of sulfur in FeS
and FeS2, a total mass of 22.5 mmol of available Fe(III) is estimated in column 2. This is
in very good agreement with the initial mass of Ti(III)–EDTA reducible Fe(III) of 22.0
mmol (Table 2). This observation suggests that the Ti(III)–EDTA reducible Fe(III) is
indeed a good indicator for the OXC of an aquifer where the principal reductant is
dissolved S(-II).
4.6. Geochemical modeling of column 2
The geochemical processes in column 2, where significant SO4 reduction takes place,
have been quantified using the computer program STEADYQL. One of the aims of this
modeling effort is to evaluate the sensitivity of the modeled dependent species (DIC, pH,
propionate and acetate) with respect to the kinetics of the biogeochemical processes. Of
special interest here is to evaluate the effect of sulfur cycling processes such as reduction
of SO4 and direct reaction of S(-II) with surface Fe(III) hydroxides (Eq. (1)) on the
development of pH and alkalinity. Comparison between measured and simulated results
also allows us to validate our assumptions.
We present here the modeling results of the aqueous phase data of column 2 collected
on day 85. The concentrations of O2, NO3, SO4 and propionate in the influent were 0.25,
0.10, 0.55 and 1.0 mM, respectively. The influent DIC concentration was 4.7 mM and pH
was 7.7. Equilibrium reactions reported elsewhere have been used for speciation
calculations (Amirbahman et al., 1998). The microbially catalyzed kinetic processes
relevant to this data are listed in Table 1.
Zero-order rate coefficients for the microbially catalyzed redox processes were obtained
from the changes in concentrations of the relevant independent species and by taking into
account the mean residence time in each box reactor as described previously. Rates of
Table 3
Mass balance of sulfur species in column 2
Measurements and procedures Mass (mmol)
Sulfate inputa 72
Sulfate outputa 16
S(-II) producedb 56
S(-II) outputa 23
S(-II) reactedc 33
Acid volatile sulfide producedd 15
Cr(II)-reducible sulfur producedd 30
a Total dissolved mass over the course of experiment.b The difference between sulfate input and output.c The difference between S(-II) produced and S(-II) released from the column, initially forming FeS.d Total sulfur in the solid phase measured directly and integrated over the entire length of column.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 185
change of concentrations of O2, NO3 and SO4 were used to predict the concentrations of
propionate and acetate. Aerobic respiration, denitrification and SO4 reduction all use
propionate as the carbon source (Table 1). Aerobic respiration was modeled only in the
first 0.5 cm of column 2, where complete reduction of 0.25 mM of O2 was observed.
Denitrification process was observed up to 2 cm from the inlet, where the entire added 0.1
mM NO3 was reduced. Sulfate reduction was considered for the entire length of the
column, even though SO4 was reduced mainly close to the inlet. This process produces 1
mol of acetate for every mole of oxidized propionate (Table 1).
Dissimilatory Fe and Mn reduction processes use acetate as the organic carbon source
in this study. However, no decrease in acetate concentration was observed due to Fe and
Mn reduction processes, since the concentrations of produced metals are nearly two orders
of magnitude less than the concentration of available acetate. Rates of Fe and Mn
reduction processes measured in column 1 in the absence of significant SO4 reduction
were used for column 2.
The measured and simulated concentrations of the two organic acids are shown in Fig.
8. Sulfate reduction was originally considered as the only dissimilatory process here that
produces acetate by oxidation of propionate. However, changes in SO4 concentration
could not account for the entire production of acetate. To model the distribution of acetate,
fermentation of propionate to acetate and hydrogen gas was considered (Table 1; Thauer et
al., 1977). This process was considered only between 1 and 1.5 cm from the inlet and
accounts for the production of nearly 15% of acetate on day 85. Fermentation of
propionate to acetate in column 2 started sometime after the 10th week of operation
and increased in rate with time. By the end of operation on day 219, this process was
Fig. 8. Concentrations of propionate and acetate in column 2 on day 85. Circles are experimental measurements
and lines are STEADYQL simulations.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190186
complete at the end of column 2 with all of propionate converted to acetate (data not
shown).
The modeled aerobic respiration, denitrification, SO4 reduction and propionate
fermentation processes together account for 92% of consumption of propionate, whereas
the measured data indicate a 98% consumption. This difference may be due to the
consumption of propionate by the bacteria to generate biomass. Such a process is expected
to take place close to the inlet of the column, due to higher density of bacteria.
Even though the saturation index of solution with respect to calcite varies from 0.5
close to the inlet to 0.9 at the end of the column, precipitation of CaCO3 is not observed.
Using typical rate coefficients for precipitation of CaCO3(S) estimated under field
conditions in a similar porous media (Amirbahman et al., 1998), the expected decrease
in the concentration of dissolved Ca from the inlet to the outlet of column 2 would be in
the order of 1 AM. This is more than three orders of magnitude smaller than the measured
Ca concentrations at any location along the columns.
The simultaneous reduction of SO4 and Fe in the same zone has been previously
reported and may be attributed to the low availability of solid-phase Fe(III) hydroxides
(Canfield, 1989). In column 2, the low dissolved concentrations of Fe(II) indicate that the
distribution of this species is controlled by the solubility of amorphous FeS. This low
concentration of Fe(II) keeps the solution well below the solubility of siderite.
Reaction of dissolved S(-II) with the surface Fe(III) hydroxides was considered as a
kinetically controlled process. Due to the small concentrations of dissolved Fe(II) (0.1–1
AM), precipitation of FeS was considered as an equilibrium reaction (Eq. (3)). von Gunten
and Furrer (2000) have observed kinetically controlled precipitation of FeS under similar
conditions. As discussed below, however, inclusion of such a process has a very small
effect on the overall S(-II) balance in column 2 and development of pH and alkalinity.
The rate of reaction of dissolved S(-II) with the surface Fe(III) hydroxides was
calculated for each box by taking into account the total reduced SO4 concentration, the
change in the dissolved S(-II) concentration and the concentration of S(-II) that is
precipitated as FeS (Furrer et al., 1996). To calculate the latter concentration in column
2, a rate coefficient of 2.5 nM s� 1 for the dissimilatory Fe reduction was considered. This
rate was observed close to the inlet of column 1 on day 85, where SO4 reduction was
negligible, and was applied uniformly along the length column 2. Given the data presented
in the previous section, this rate represents a high end estimate for the rate of production of
dissolved Fe(II) in both columns. The difference in the estimated total and measured Fe(II)
concentrations is the precipitated concentration of Fe(II), which is also equivalent to the
concentration of precipitated S(-II).
Direct reaction with surface Fe(III) hydroxides constitutes a more important sink for the
dissolved S(-II) than precipitation of FeS in column 2. Experimental measurements
indicate that a total of 0.53 mM of SO4 was reduced to S(-II) on day 85. Approximately
99% of the produced S(-II) reacts in the column on day 85. Considering a rate coefficient
of 2.5 nM s� 1 for the dissimilatory Fe(III) reduction in column 2, a total maximum
concentration of 63 AM of Fe(II) would be generated in this column by this mechanism.
Given that 21 AM of Fe(II) leaves the column, a total of 42 AM of Fe(II) has been
precipitated as FeS. The precipitated FeS is therefore, more than an order of magnitude
smaller than the total reacted S(-II).
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 187
Results of STEADYQL simulations of column 2 are shown in Fig. 9. This figure
exhibits the measured and modeled values of pH and DIC for day 85. The modeled values
are in reasonable agreement with the measured values, indicating that our basic assump-
tions are correct. Sensitivity analysis shows that close to the inlet of the column, aerobic
respiration, denitrification and SO4 reduction processes largely control the evolution of pH
and DIC. Propionate fermentation to acetate, where considered, has a large effect on
lowering the pH. Toward the end of column2, where the direct reaction of dissolved S(-II)
with surface Fe(III) hydroxides is important, this process controls almost all of the increase
in pH. In column 1, the Mn and Fe reduction processes do not contribute to the evolution
of pH and DIC significantly. In this column, pH and DIC evolution are controlled by
aerobic respiration and denitrification processes.
Fig. 9. (a) pH, and (b) dissolved inorganic carbon concentrations in column 2 on day 85. Circles are experimental
measurements and lines are STEADYQL simulations.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190188
5. Conclusions
We have studied the evolution of reducing environments in two separate flow-through
packed bed reactors. All geochemical transformations were directly or indirectly related to
the microbial oxidation of organic matter by various electron acceptors. The rates of
enzymatic reduction of Mn(IV) and Fe(III) exhibited an initial increase and a gradual
decrease in time, due to the preferential dissolution of the more reactive mineral phases.
Regardless of the presence of dissolved S(-II), dissolved Mn(II) concentration was
controlled by the solubility of rhodochrosite. Dissolved Fe(II) concentration was con-
trolled by the solubility of siderite or amorphous FeS, depending on the concentration of
dissolved S(-II). However, in the column where Fe(II) concentration was controlled by
dissolved S(-II), the most important mechanism for the reduction of Fe(III) hydroxide was
the direct attack by the dissolved S(-II). For the duration of our experiment, enzymatic
Fe(III) reduction contributed only to a small extent to the total mass of reduced Fe (19%)
due to the slower kinetics of this process than the enzymatic sulfate reduction process and
the reactions depicted in Eqs. (1)–(3).
Acknowledgements
The authors wish to thank Dr. A. Ammann and T. Ruttimann of EAWAG for their help
in analyzing the samples. Professor R. Giovanoli of the University of Bern performed the
SEM and XRD analysis of the soil samples. Critical comments of three reviewers are
greatly acknowledged.
References
Abbaspour, K., Matta, V., Huggenberger, P., Johnson, C.A., 1999. A contaminated site investigation: comparison
of information gained from geophysical measurements and hydrogeological modeling. J. Contam. Hydrol. 40,
365–380.
Amirbahman, A., Sigg, L., von Gunten, U., 1997. Reductive dissolution of Fe(III) (hydr)oxides by cysteine:
mechanism and kinetics. J. Colloid Interface Sci. 194, 194–206.
Amirbahman, A., Schonenberger, R., Johnson, C.A., Sigg, L., 1998. Aqueous- and solid-phase biogeochemistry
of a calcareous aquifer system downgradient from a municipal solid waste landfill (Winterthur, Switzerland).
Environ. Sci. Technol. 32, 1933–1940.
Ammann, A.A., Ruttimann, T.B., 1995. Simultaneous determination of small organic and inorganic anions in
environmental water samples by ion-exchange chromatography. J. Chromatogr., A 706, 259–269.
Arakaki, T., Morse, J.W., 1993. Coprecipitation and adsorption of Mn2 + with mackinawite (FeS) under con-
ditions similar to those found in anoxic sediments. Geochim. Cosmochim. Acta 57, 9–15.
Berner, R.A., 1984. Sedimentary pyrite formation: an update. Geochim. Cosmochim. Acta 48, 605–615.
Canfield, D.E., 1989. Reactive iron in marine sediments. Geochim. Cosmochim. Acta 53, 619–632.
Canfield, D.E., Raiswell, R., Westrich, J.T., Reaves, C.M., Berner, R.A., 1986. The use of chromium reduction in
the analysis of reduced inorganic sulfur in sediments and shales. Chem. Geol. 54, 149–155.
Davison, W., 1991. The solubility of iron sulfides in synthetic and natural waters at ambient temperature. Aquat.
Sci. 53, 309–329.
Furrer, G., Westall, J., Sollins, P., 1989. The study of soil chemistry through quasi-steady-state models: I.
Mathematical definition of model. Geochim. Cosmochim. Acta 53, 595–601.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190 189
Furrer, G., von Gunten, U., Zobrist, J., 1996. Steady-state modelling of biogeochemical processes in columns
with aquifer material: 1. Speciation and mass balances. Chem. Geol. 133, 15–28.
Heron, G., Christensen, T.H., 1994. The role of aquifer sediment in controlling redox conditions in polluted
groundwater. In: Dracos, J., Stauffer, F. (Eds.), Transport and Reactive Processes in Aquifers. Balkema,
Rotterdam, The Netherlands, pp. 73–78.
Heron, G., Christensen, T.H., 1995. Impact of sediment-bound iron on redox buffering capacity in a landfill
leachate polluted aquifer (Vejen, Denmark). Environ. Sci. Technol. 29, 187–192.
Heron, G., Christensen, T.H., Tjell, J.C., 1994a. Oxidation capacity of aquifer sediments. Environ. Sci. Technol.
28, 153–158.
Heron, G., Crouzet, C., Bourg, A.C.M., Christensen, T.H., 1994b. Speciation of Fe(II) and Fe(III) in contami-
nated aquifer sediments using chemical extraction techniques. Environ. Sci. Technol. 28, 1698–1705.
Hoehn, E., Johnson, C.A., Huggenberger, P., Amirbahman, A., Peter, A., Zweifel, H.R., 2000. Investigative
strategies and risk assessment of old unlined municipal solid waste landfills. Waste Manag. Res. 18, 577–589.
Jacobs, L.A., von Gunten, H.R., Keil, R., Kuslys, M., 1988. Geochemical changes along a river-groundwater
infiltration flow path: Glattfelden, Switzerland. Geochim. Cosmochim. Acta 52, 2693–2706.
Jakobsen, R., Postma, D., 1999. Redox zoning, rates of sulfate reduction and interactions with Fe reduction and
methanogenesis in a shallow sandy aquifer, Romo, Denmark. Geochim. Cosmochim. Acta 63, 137–151.
Kornicker, W.A., Morse, J.W., 1991. The interactions of divalent cations with the surface of pyrite. Geochim.
Cosmochim. Acta 55, 2159–2172.
Kostka, J.E., Luther, G.W., 1994. Partitioning and speciation of solid phase iron in saltmarsh sediments. Geo-
chim. Cosmochim. Acta 58, 1701–1710.
Lovley, D.R., Phillips, E.J.P., Lonergan, D.J., 1991. Enzymatic versus nonenzymatic mechanisms for Fe(III)
reduction in aquatic sediments. Environ. Sci. Technol. 25, 1062–1067.
Luther, G.W., Church, T.M., Kostka, J.E., Sulzberger, B., Stumm, W., 1992. Seasonal iron cycling in the marine
environment: the importance of ligand complexes with Fe(II) and Fe(III) in the dissolution of Fe(III) minerals
and pyrite, respectively. Mar. Chem. 40, 81–103.
Morse, J.W., Luther, G.W., 1999. Chemical influences of trace metal sulfide interactions in anoxic sediments.
Geochim. Cosmochim. Acta 63, 3373–3378.
Stauffer, F., Dracos, T., 1986. Experimental and numerical study of water and solute infiltration in layered porous
media. J. Hydrol. 84, 9–34.
Thauer, R.K., Jungermann, K., Decker, K., 1977. Energy conservation in chemotrophic anaerobic bacteria.
Bacteriol. Rev. 41, 100–180.
von Gunten, U., Furrer, G., 2000. Steady-state modelling of biogeochemical processes in columns with aquifer
material: dynamics of iron–sulfur interactions. Chem. Geol. 167, 271–284.
von Gunten, U., Zobrist, J., 1993. Biogeochemical changes in groundwater-infiltration systems: column studies.
Geochim. Cosmochim. Acta 57, 3895–3906.
von Gunten, H.R., Karametaxas, G., Keil, R., 1994. Chemical processes in infiltrated riverbed sediment. Environ.
Sci. Technol. 28, 2087–2093.
Zhabina, N.N., Volkov, I.I., 1978. A method of determination of various sulfur compounds in sea sediments and
rocks. In: Krumbein, W.E. (Ed.), Environmental Biogeochemistry; Methods, Metals and Assessments, vol. 3.
Ann Arbor Science Publishers, Ann Arbor, MI, pp. 735–745.
A. Amirbahman et al. / Journal of Contaminant Hydrology 64 (2003) 169–190190