Reactivity of Fe(II)-Bearing Minerals toward Reductive Transformation of Organic Contaminants

Article abstract of DOI:10.1021/es0345569

Fe(II) present at surfaces of iron-containing minerals can play a significant role in the overall attenuation of reducible contaminants in the subsurface. As the chemical environment, i.e., the type and arrangement of ligands, strongly affects the redox potential of Fe(II), the presence of various mineral sorbents is expected to modulate the reactivity of surficial Fe(II)-species in aqueous systems. In a comparative study we evaluated the reactivity of ferrous iron in aqueous suspensions of siderite (FeCO ₃), nontronite (ferruginous smectite SWa-1), hematite (α-Fe₂O₃), lepidocrocite (γ-FeOOH), goethite (α-FeOOH), magnetite (Fe₃O₄), sulfate green rust (Fe^II₄Fe^III₂(OH)₁₂SO ₄·4H₂O), pyrite (FeS₂), and mackinawite (FeS) under similar conditions (pH 7.2, 25 m² mineral/L, 1 mM Fe(II)_aq, O₂ (aq) < 0.1 g/L). Surface-area-normalized pseudo first-order rate constants are reported for the reduction of hexachloroethane and 4-chloronitrobenzene representing two classes of environmentally relevant transformation reactions of pollutants, i.e., dehalogenation and nitroaryl reduction. The reactivities of the different Fe(II) mineral systems varied greatly and systematically both within and between the two data sets obtained with the two probe compounds. As a general trend, surface-area-normalized reaction rates increased in the order Fe(II) + siderite < Fe(II) + iron oxides < Fe(II) + iron sulfides. 4-Chloronitrobenzene was transformed by mineral-bound Fe(II) much more rapidly than hexachloroethane, except for suspensions of hematite, pyrite, and nontronite. The results demonstrate that abiotic reactions with surface-bound Fe(II) may affect or even dominate the long-term behavior of reducible pollutants in the subsurface, particularly in the presence of Fe(III) bearing minerals. As such reactions can be dominated by specific interactions of the oxidant with the surface, care must be taken in extrapolating reactivity data of surface-bound Fe(II) between different compound classes.

Full text of DOI:10.1021/es0345569

Environ. Sci. Technol. 2004, 38, 799-807
bound Fe(II) was found to enhance dramatically the trans-
Reactivity of Fe(II)-Bearing Minerals
toward Reductive Transformation of
Organic Contaminants
formation rates of many reducible pollutants, such as
nitroaromatic compounds (NACs) (1-3), polyhalogenated
alkanes (4-6), chromium(VI) (7), technetium(VI) (8), ura-
nium(VI) (9), oxamyl and related carbamate pesticides (10),
or disinfectants such as monochloramine (11). Considering
that surface-bound Fe(II) may be continuously regenerated
in the subsurface, for example, by microbial iron-reduction
(12, 13), it is one of the most potent natural mediators for
reductive in-situ transformation of contaminants. Field
studies have demonstrated that even in geochemically
complex polluted aquifers surface-bound Fe(II) was the
predominant reductant of nitroaromatic compounds (2). A
very recent study on reduction of polyhalogenated methanes
in iron- and sulfate-reducing sediments came up with similar
conclusions (14). The enhanced reactivity of surface-bound
Fe(II) can be rationalized within the framework of surface
complexation theory. Hydroxo ligands favor the oxidation of
Fe(II) (15) because of both increased electron density at the
Fe(II) center and stabilization of the product Fe(III) (16).
Previous work suggested the formation of a surface complex
upon adsorption of Fe(II) to iron oxides of the type
dFe(III)sOsFe(II)sOH (17), irrespective of mineral stoi-
chiometry and crystal structure of the iron oxide bulk phase.
From an environmental engineering point of view, the
existence of such defined Fe(II) surface species of uniform
reactivity would largely simplify predictions of the fate of
reducible pollutants at contaminated sites where iron-
reducing conditions prevail.
M A R T I N E L S N E R ,
R E N EÄ P . S C H W A R Z E N B A C H , A N D
S T E F A N B . H A D E R L E I N *
Swiss Federal Institute for Environmental Science and
Technology (EAWAG) and Swiss Federal Institute of
Technology Zurich (ETHZ), Ueberlandstrasse 133,
CH-8600 Duebendorf, Switzerland
Fe(II) present at surfaces of iron-containing minerals can
play a significant role in the overall attenuation of
reducible contaminants in the subsurface. As the chemical
environment, i.e., the type and arrangement of ligands,
strongly affects the redox potential of Fe(II), the presence
of various mineral sorbents is expected to modulate the
reactivity of surficial Fe(II)-species in aqueous systems. In
a comparative study we evaluated the reactivity of
ferrous iron in aqueous suspensions of siderite (FeCO ),
3
nontronite (ferruginous smectite SWa-1), hematite (R-Fe2O ),
3
lepidocrocite (γ-FeOOH), goethite (R-FeOOH), magnetite
(Fe3O ), sulfate green rust (FeII FeIII2(OH)12SO ‚4H O), pyrite
4
4
4
2
There are, however, several reasons why this assumption
may be an oversimplification for a wider range of minerals:
(i) Reactive ferrous iron does not only form on oxides, but
also on surfaces of other major iron minerals such as iron
sulfide (FeS), pyrite (FeS2), and siderite (FeCO3). (ii) The
reactivity of Fe(II) at mixed valent Fe(II)/ Fe(III) phases such
as magnetite (Fe3O4), green rust, or reduced ferruginous clay
minerals may differ in reactivity from Fe(II) at pure iron(III)
oxides, owing to the presence of structural Fe(II) within the
mineral lattice. (iii) Even various ferric (hydr)oxides such as
goethite (R-FeOOH), lepidocrocite (γ-FeOOH), or hematite
(R-Fe2O3) have distinctive crystallography and potentially
display different surface structures and reactivities. In
addition, evidence for reactive internal surfaces has been
found in layered minerals such as green rust and clay minerals
(18-20). Thus, the coexistence of an array of Fe(II) species
with differences in reactivity is very likely.
(FeS2), and mackinawite (FeS) under similar conditions
(pH 7.2, 25 m2 mineral/L, 1 mM Fe(II)aq, O (aq) < 0.1 g/L).
2
Surface-area-normalized pseudo first-order rate constants
are reported for the reduction of hexachloroethane
and 4-chloronitrobenzene representing two classes of
environmentally relevant transformation reactions of
pollutants, i.e., dehalogenation and nitroaryl reduction.
The reactivities of the different Fe(II) mineral systems varied
greatly and systematically both within and between the
two data sets obtained with the two probe compounds. As
a general trend, surface-area-normalized reaction rates
increased in the order Fe(II) + siderite < Fe(II) + iron oxides
< Fe(II) + iron sulfides. 4-Chloronitrobenzene was
transformed by mineral-bound Fe(II) much more rapidly
than hexachloroethane, except for suspensions of hematite,
pyrite, and nontronite. The results demonstrate that
abiotic reactions with surface-bound Fe(II) may affect or
even dominate the long-term behavior of reducible pollutants
in the subsurface, particularly in the presence of Fe(III)
bearing minerals. As such reactions can be dominated by
specific interactions of the oxidant with the surface,
care must be taken in extrapolating reactivity data of surface-
bound Fe(II) between different compound classes.
In contrast to solution chemistry, where the reactivity of
various Fe(II) complexes can be predicted using linear free
energy relationships (LFERs) and the reduction potential of
the complexes (21-24), predictions of the reactivity of Fe(II)
bound to surfaces is not feasible as the reduction potential
of such Fe(II) species is not known. In an alternative approach,
however, a probe compound might be used to measure the
reactivity of Fe(II) in the presence of different minerals so
that surface-bound Fe(II) species could be classified ac-
cording to the probe compound’s reaction rates. Two extreme
cases can be imagined if these rates are normalized to the
density of reactive Fe(II) sites: (a) Fe(II) on different minerals
reacts with a given probe compound at a uniform rate. This
would support the picture of one “representative” surface
complex such as dFe(III)sOsFe(II)sOH (Figure 1a). (b) The
reactivity of Fe(II) at different minerals differs greatly. This
would be indicative of various Fe(II) surface sites with
different intrinsic reactivities (Figure 1b). Such a reactivity
pattern can then be compared to one obtained with a second
reducible probe from a different compound class. Again, two
Introduction
Previous studies have pointed out the high reactivity of ferrous
iron when bound to ferric oxide minerals (“surface-bound
Fe(II)”) in comparison to that of Fe(II)aq in solution. Surface-
* Corresponding author phone: ++49 7071 2973148; fax: ++49
7071 295139; e-mail: haderlein@uni-tuebingen.de. Present address:
Center for Applied Geosciences (ZAG), Eberhard-Karls University
Tuebingen, Wilhelmstr. 56, D-72074 Tuebingen, Germany.
9
10.1021/es0345569 CCC: $27.50
Published on Web 12/23/2003
2004 Am erican Chem ical Society
VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 7 9 9

mined in such studies, owing to experimental difficulties in
performing lyophilization and BET-measurement under
anoxic conditions.
In this study we investigated the reactivity of surface-
bound Fe(II) for a series of environmentally relevant minerals
under similar and defined conditions, i.e., at pH 7.2, at 25
m2 mineral surface per liter and at 1 mM of dissolved Fe(II).
The background of dissolved Fe(II) was chosen to ensure
constant conditions during the experiments. The probe
compounds hexachloroethane and 4-chloronitrobenzene
represent polyhalogenated alkanes and nitroaromatic com-
pounds, respectively, two important classes of reducible
organic contaminants (32). The major goals of this study
were (1) to assess the reactivity of surface-bound Fe(II)
present on different minerals, and (2) to explore the predict-
ability of reaction rates of such Fe(II) species between
different classes of contaminants such as sketched in Figure
1.
Materials and Methods
FIGURE 1. Classification of the reactivity of Fe(II) at different minerals
based on reaction rates with probe compounds that are normalized
to reactive site densities. Possible scenarios: (a) “uniform
reactivity”; (b) different intrinsic reactivities; (c) similar and (d)
different patterns with different contaminants.
Maintenance of Anoxic Conditions. If not mentioned
otherwise, manipulations were carried out inside an oxygen
free glovebox (23 °C) with external regenerator, in which the
O2-level was continually controlled by an oxygen sensor (TOS
1.0, Ionic Systems) and kept below 0.1 ppm. As an additional
visual control, siderite (FeCO3) was periodically exposed to
the glovebox atmosphere. Siderite maintains its white color
under strict exclusion of oxygen, whereas it turns brown
within seconds and then red when exposed to small amounts
of oxygen (low ppm level). The fact that FeCO3 remained
white during all manipulations performed inside and (in
closed containers) outside the box, such as lyophilization,
BET-surface measurements, flame sealing of ampules or
agitating of vials, was taken as visual evidence for overall
strictly anoxic conditions.
Materials/Synthesis and Characterization of Minerals.
All air-sensitive minerals were synthesized in the glovebox,
washed repeatedly with deionized oxygen-free water (pre-
pared by heating to 90 °C and subsequently purging with
argon for 3 h), and freeze-dried. The BET-surface areas of
the resulting dry powders are listed in Table 1. Crystal
structures were confirmed by XRD-analysis (see Figure S1 in
the Supporting Information). Goethite, lepidocrocite, and
hematite were purchased as fine powders from Bayer
(Bayferrox 910, 943, and 105 M, respectively). Their crystal
structure has been confirmed in previous studies (6). Pyrite
was purchased in chunks (Alfa Aesar) and ground in a mortar
in the glovebox. Fe(II)-solutions were prepared by adding 28
g (0.5 mol) of iron powder (Merck) to 1 Lof1 M deoxygenated
HCl. The mixture was brought to reaction by heating to 70
°C under gentle stirring (60 rpm) during 2-2.5 h, until
evolution of H2 ceased (gas outlet for pressure release!). The
solution was then filtered within the glovebox through a 0.2-
µL PTFE filter to remove excess iron powder. The exact
concentration of dissolved Fe(II) was determined photo-
metrically according to the phenanthroline method (33).
Siderite was synthesized by mixing 500 mL of 1 M NaHCO3
and 500 mL of 0.5 M FeCl2 under stirring for 14 h in a closed
container. From time to time the lid was opened to release
CO2. After sedimentation, the supernatant was decanted, and
the mineral was repeatedly resuspended until the supernatant
had a pH of 7, followed by freeze-drying. Mackinawite was
synthesized by mixing 300 mL of 0.5 M Na2S (32 g ofNa2S‚
7-9 H2O (Fluka)) and 300 mL of FeCl2 under stirring for 48
h. The mixture was repeatedly centrifuged (6000 rpm) and
resuspended in 10 mM NaCl solution (2x) and water (1x).
The wet paste that remained after centrifugation was stored
for 2 years before lyophilization. X-ray diffractograms of the
lyophilized mineral were consistent with poorly crystalline
mackinawite. A different pattern indicative of a hydrated
cases can be imagined. (c) Reaction rates of Fe(II) with the
second probe follow very closely the trend of the first. Rates
would then be controlled by (1) the intrinsic reactivity of
Fe(II) on a given surface as “calibrated” with the first probe
compound and (2) the reduction potential of the probe
compounds. Different reaction mechanisms would not play
a major role, simplifying greatly the predictions of reductive
pollutant transformation (Figure 1c). Finally, (d) Reaction
rates of one probe compound do not follow the trend
observed with the other. This would indicate that specific
reaction mechanisms and/ or the accessibility to different
surface sites determined their reactivity, complicating ex-
trapolations of Fe(II) reactivities between different compound
classes (Figure 1d).
A quantitative picture of reactivity patterns of Fe(II) bound
to different minerals is, however, very difficult to derive from
the presently available literature data. Major obstacles are
(1) the lack of consistent data sets generated under com-
parable experimental conditions, and (2) the difficulty in
quantifying reactive Fe(II) surface-sites which is necessary
to scale the observed reaction rates. In addition to studies
of Fe(II) sorbed on Fe(III) oxides, the wide range of Fe(II)-
containing minerals investigated in reductive transformations
of organic pollutants includes magnetite (25-27), various
forms of FeS (troilite, mackinawite; (25, 28)), green rusts (27,
29), lattice-bound Fe(II) in layered silicates (e.g., biotite (30)
and smectites (20, 31)). However, these studies were con-
ducted with different reducible pollutants, at different pH
values, different mineral concentrations, and different de-
grees of surface saturation by Fe(II) so that quantitative
comparisons are not feasible. In studies dealing with the
reactivity of Fe(II) bound to the surface of pure iron(III)
minerals, observed rate constants were sometimes normal-
ized to the concentration of surface-bound Fe(II) determined
in adsorption experiments. However, because mixed valent
Fe(II)/ Fe(III)-minerals contain potentially reactive Fe(II) in
their mineral structure, this approach is not generally
applicable. In addition, the measured total concentration of
surface-bound Fe(II) may overestimate the concentration of
reactive sites, as in the case of nontronite, where experimental
evidence suggested that edge-bound Fe(II) was considerably
more reactive than Fe(II) sorbed at siloxane surfaces (20).
Alternatively, reaction rates could be normalized to the total
BET-surface area of the minerals. The surface areas of oxygen-
sensitive minerals, however, have not always been deter-
9
8 0 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004

TABLE 1. Mineral Properties, Pseudo-First-Order Rate Constants, and Mass Transfer Estimates for Reactions of Hexachloroethane, 4-Chloronitrobenzene, and the Intermediate 4-Chlorophenyl
Hydroxylamine with Fe(II) Bound on Different Minerals
mass transfer constants
hexachloroethane
4-chlorophenyl hydroxylaminea
4-chloronitrobenzene
(a kL/25) b
BET
lower
estimate
upper
d
d
d
surface
Fe(II)ads
kobs
′
SD
kobs
′
SD
kobs
′
SD
estimate
b
b
mineral
area [m2/g]
[µmol/m2]c
log(kobs′)
-1.79
-1.80
-2.12
-2.94
-3.04
-3.12
-3.70
-3.74
-3.82
-4.76
-5.09
[h-1m-2L]
1.6 × 10-2
1.6 × 10-2
7.5 × 10-3
1.1 × 10-3
9.1 × 10-4
7.6 × 10-4
2.0 × 10-4
1.8 × 10-4
1.5 × 10-4
1.7 × 10-5
8.2 × 10-6
[h-1m-2L]
6.4 × 10-4
6.3 × 10-4
3.0 × 10-4
4.6 × 10-5
3.6 × 10-5
3.0 × 10-5
8.0 × 10-6
7.3 × 10-6
6.1 × 10-6
6.9 × 10-7
3.3 × 10-7
ne
2
2
2
3
6
1
2
4
6
1
1
log(kobs′)
1.04
[h-1m-2L]
1.1 × 101
-
[h-1m-2L]
ne
2
-
log(kobs′)
1.23
[h-1m-2L]
1.7 × 101
8.9 × 10-2
[h-1m-2L]
6.8 × 10-1
3.6 × 10-3
ne
2
[h-1
]
[h-1
]
Mackinawite
(FeS)
Pyrite
(FeS2)
12.6
0.88
45.7
45.7
16.2
19.2
103.4
17.6
13.7
38.3
10.4
-1.4
-15.2
-48
2.0 × 10-1
-
7.0 × 100
2.2 × 101
2.6 × 104
3.0 × 102
-
-1.05
4
Green rust
-0.52
-0.47
-0.34
-0.73
-1.88
-1.18
3.0 × 10-1
3.4 × 10-1
4.6 × 10-1
1.9 × 10-1
1.3 × 10-1
6.6 × 10-1
1.7 × 10-1
1.9 × 10-1
1.9 × 10-2
4.4 × 10-2
6.6 × 10-4
1.7 × 10-2
3
2
7
4
1
6
(GRSO
)
GRSO + 410 m M
4
MOPS
Goethite
11.2
0.28
0.08
1.9 × 100
1.2 × 100
3.8 × 10-1
1.3 × 100
1.9 × 10-4
2.0 × 10-3
8.8 × 10-5
7.6 × 10-2
4.9 × 10-2
1.5 × 10-2
5.2 × 10-2
7.5 × 10-6
8.0 × 10-5
3.5 × 10-6
5
1
3
6
1
1
1
4.5 × 100
5.2 × 100
1.2 × 103
1.2 × 102
(R-FeOOH)
Magnetite
(Fe3O4)
Nontronite
(clay m ineral)
Lepidocrocite
(γ-FeOOH)
Hem atite
(Fe2O3)
-53.6
3.64
-0.42
0.12
5.6
1.2
-3.72
-2.70
-4.05
Siderite
-10.8
(FeCO3)
Controlf
(only Fe(II),
no m ineral)
a
b
Rate constants for 4-chlorophenyl hydroxylam ine are only reported in cases where accum ulation of this interm ediate was found. Experim ents were conducted with 25 m 2/L m ineral surface; reported rate
constants and calculated lower and upper estim ates for m ass transfer coefficients, however, are norm alized to 1 m 2/L to im prove com parability; calculated lower and upper estim ates for m ass transfer constants
(see supplem entary inform ation) are reported as values divided by 25. For calculation see also Table S1 in the Supporting Inform ation; negative values indicate net release of Fe(II) from the m ineral into solution.
c
d
SD ) Standard deviation, either determ ined from the average of replicate experim ents (n) or from the standard deviation of pseudo-first-order plots in long-term experim ents using several independent batches.
n ) num ber of experim ents. f Rate constant of control is given as pseudo-first order constant h-1 divided by a factor of 25.
e

sodium iron sulfide similar to coyoteite (NaFe3S5‚2H2O) was
obtained when the mineral was resuspended in water for
one month (see Figure S1 in the Supporting Information).
In transformation experiments, however, mackinawite was
suspended no longer than 2 d prior to contaminant addition.
Sulfate green rust, GRSO4 was prepared by mixing 250 mL of
150 mM Fe(II)SO4 and 100 mL of 25 mM iron(III) sulfate
(prepared from desiccated Fe2(SO4)3), adding 200 mL of 0.3
M NaOH, and titrating with 25 mM iron(III) sulfate solution
to pH 7. The suspension was stirred for 7 d, and the pH was
repeatedly adjusted to pH 7. Finally, the suspension was
washed in a glass filter funnel, and thereafter it was quickly
freeze-dried. Magnetite was prepared following a procedure
by Regazzoni et al. (34). Briefly, 0.5 L of 0.25 M FeCl3 solution
and 0.5 L of 0.5 M FeCl2 solution were mixed, heated to 80
°C, and titrated with 2 M NaOH to pH 7. The mixture was
stirred for one month, and the pH was repeatedly adjusted
to 7 by addition of HCl. The mineral was washed several
times by sedimentation and decantation, before lyophiliza-
tion. Chemically reduced nontronite (ferruginous smectite
SWa-1) purchased from Source Clay Mineral Repository
(University of Missouri Columbia, MO) was prepared in the
glovebox according to a method adapted from Stucki et al.
(35), as described in detail by Hofstetter et al. (20). Briefly,
assays containing 0.25 g of finely ground clay mineral (size
fraction <2 µm) were suspended in 12.5 mL of citrate-
bicarbonate buffer and diluted with water to 50 mL. After
stirring for 15 min, 0.75 g of Na2S2O4 salt was added. After
8 d the clay was washed and homoionized by pressurized
filtration (80-mL ultrafiltration cell, Millipore AG, Volketswil,
Switzerland), 3× with 1 M NaCl and 1× with pure water (0.2
µm Fluoropore (PTFE) membrane filters, 47-mm diam,
Millipore).
Hexachloroethane (Fluka), tetrachloroethene (Aldrich),
bromoform (Fluka), 4-chloronitrobenzene (Fluka), and 4-
chloronitroaniline (Fluka) were of the highest purity available.
They were stored at 4 °C in the dark and used without further
purification. Spike solutions of 50-100 µM were prepared in
degassed, anoxic methanol (Scharlau).
Experim ental Conditions. Experiments were conducted
in 10-14 mL sample containers at 23 °C, with mineral
suspensions of 25 m2/ L surface concentration, contaminant
concentrations of 0.5-3 µM, 1 mM dissolved Fe(II), a pH of
7.2, and ionic strengths of 3.4-11.0 mM (see Table S1 in the
Supporting Information). Controls with no minerals present
(only Fe(II)) were included. A pH of 7.2 was chosen because
of its environmental relevance, notwithstanding the fact that
sorption of Fe2+ on mineral surfaces is strongly pH-dependent
and leads to different degrees of surface saturation for
different minerals (38). We chose small initial contaminant
concentrations of 0.5-3 µM to minimize limitations of
reaction kinetics by surface sites. Further, we added pre-
determined amounts of FeCl2 to the mineral systems to
obtain, after 48 h sorption time, concentrations of Fe(II) of
1 mM. This was (a) high enough to ensure maximum sorption
at this pH, (b) low enough to prevent precipitation of
Fe(OH)2 (K ) 4.8 × 10-17, (39)), and (c) reflects conditions
that may be encountered in iron-reducing natural environ-
ments (12). A relatively low concentration of mineral surface
(25 m2/ L) was chosen to prevent coagulation of particles
that might limit the accessibility of the surface for sorption
of Fe(II) and reaction (5). In experiments conducted with
hexachloroethane and goethite/ Fe(II), the ionic strength was
systematically varied from 5 to 25 mM, which corresponds
to a 20-fold to 100-fold excess in comparison to the volumetric
concentration of surface-bound Fe(II). No significant dif-
ferences in reaction rates could be observed (data not shown).
Therefore, we did not use an ionic strength buffer, except for
mackinawite, where sedimentation and filtration of particles
were hampered by the low ionic strength.
Preparation of Mineral Suspensions. An aliquot of
mineral corresponding to 25 m2/ L surface area was suspended
in anoxic, deionized water. The pH was adjusted to 7.2 (1 N
HCl/ NaOH), aliquots of FeCl2 were added, and the pH was
adjusted again until the concentration of Fe(II) in solution
was 1 mM as determined according to the phenanthroline
method (33). For magnetite and green rust, Fe(II)aq con-
centrations were slightly higher, because Fe(II) was released
from the surface of the freshly prepared minerals. Suspen-
sions were buffered at pH 7.2, either through the intrinsic
buffer capacity of the mineral surface or by small amounts
(0.4-1.2 mM) of added 4-morpholinopropanesulfonic acid
(MOPS). A pH adjustment to 7.2 was not feasible for sulfate
green rust, because addition of HCl or MOPS resulted in
rapid formation of a secondary black phase. For this mineral,
two different experimental conditions were used: (a) without
external buffer, at pH 8; under these conditions, green rust
maintained its green color for hours and was thus considered
sufficiently stable for our purposes; (b) with addition of MOPS,
at pH 7.5; under these conditions, a rapid color change to
black could be observed. The exact ionic composition of all
experiments is reported in Table S1 in the Supporting
Information.
Lyophilization. Lyophilization of all minerals was carried
out according to the Schlenk method (36). Mineral suspen-
sions were filled into a 200-mL glass finger connected to an
empty 1-L round-bottom flask via a U-shaped lyophilization
bridge, which was equipped with an outlet with stopcock for
evacuation. The whole apparatus was assembled inside the
glovebox, closed, and taken out. The suspension was frozen
by submerging the glass finger (not the fitting!) for 2 h in
liquid nitrogen. The outlet of the stopcock was then con-
nected via a metal hose to a vacuum pump (vacuum 2 × 10-3
mbar), the hose was flushed several times by quickly opening
and closing the stopcock, and the apparatus was finally
evacuated for 5 min. After disconnecting the pump, the glass
finger was taken out of the coolant, and instead the round-
bottom flask was immersed in liquid nitrogen for 2-3 d.
Thus, because of the low pressure, the ice sublimated
completely from the frozen sample suspension to the cooled
round-bottom flask leaving the mineral as a fine, dry powder
in the glass finger. The apparatus was subsequently trans-
ferred back into the glovebox and demounted, and the
powder was stored in a closed container in the glovebox.
Characterization. Surface areas of the minerals were
determined by adsorption of nitrogen according to the BET-
method (37) using a Sorptomatic 1990 Fisons device. Burets
were equipped with a stopcock, and could be filled and closed
inside the glovebox. They were then evacuated on an external
vacuum line, closed, and connected to the BET instrument.
Prior to opening the stopcock and starting the analysis, all
connections inside the instrument were evacuated and
flushed several times with nitrogen.
Kinetic Experim ents. Kinetic experiments were initiated
by addition of 0.1-0.5 mL of methanolic contaminant stock
solutions, and they were stopped by filtration through 0.2-
µm filters in the case of 4-chloronitrobenzene and by liquid/
liquid extraction with hexane in the case of hexachloroethane.
The sampling protocol was adapted to the time scale of the
experiments, which included half-lives of less than 10 s to
several weeks. Experiments with 4-chloronitrobenzene were
mostly performed inside the glovebox, and were typically
conducted in 14-mL glass flasks equipped with a stopcock
X-ray diffraction measurements were performed on an
XRD Siemens D 5000 and XDS 2000, Scintag, Cu KR source,
with a scanning rate of 1°/ min so that measurements took
about 70 min. The thin-layered samples were prepared in
the glovebox and kept in airtight containers until immediately
before the measurement.
9
8 0 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004

and a magnetic stirrer. Every 1-2 min, the flasks were quickly
opened, a 1-mL aliquot was withdrawn with a polypropylene
syringe (BD Plastipak), and the sample was subsequently
filtered into autosampler vials. Losses of the nitroaromatic
compound through volatilization and sorption were found
to be negligible. For some extremely fast reactions, a 10-mL
aliquot of mineral suspensions was drawn into a polypro-
pylene syringe (BD Plastipak). The reaction was initiated by
adding a contaminant stock solution. A PTFE filter (25 mm
× 0.2 µM, BGB Analytik) was then quickly emplaced and the
syringe was shaken for 5 s before the solution was filtered
into 1-mL autosampler vials with sampling intervals of about
5 s. In experiments lasting longer than 1 h batches were set
up in crimp serum-cap vials with Viton stoppers, removed
from the glovebox, and agitated on a reciprocating shaker at
23 °C in the dark until analysis. Nearly all experiments in this
study were therefore conducted in 9-16 identical batches of
10-mL suspensions. For each sampling point, one batch was
sacrificed and used for 2-3 replicates that were analyzed by
GC-ECD and HPLC, respectively (see below). Longer-lasting
experiments (duration one to several weeks) were conducted
in glass ampules (10-mL gold band, Wheaton). Inside the
glovebox, ampules were filled with 10 mL of mineral
suspension, spiked with contaminant stock solutions, and
closed with Viton stoppers (prepared from Viton strings).
The suspensions were frozen outside the box and then flame-
sealed. To avoid bursting of the glass during flame sealing,
the Viton stopper was pierced for pressure release by a
cannula (1.2-mm diam, Microlance 3) which was loosely kept
inside an argon-flushed hose.
FIGURE 2. Top: reaction of (a) green rust/Fe(II) and (b) siderite/
Fe(II) with ([) hexachloroethane to (9) tetrachloroethene; (O)
pentachloroethane; (4) trichloroethene). Bottom: reaction of (c)
lepidocrocite/Fe(II) and (d) siderite/Fe(II) with (0) 4-chloronitroben-
zene (4-Cl-NB) to (9) 4-chloroaniline; (O) deficit in molar balance;
(4) (UV-absorption (intermediate)245 nm/UVabsorption (4-Cl-NBt)0)-
276 nm) for HPLC peak at tret ) 7.7 min. (see Materials and Methods).
Curve lines are modeled with pseudo-first-order rate constants
given in Table 1.
Analytical Procedures. For analysis of chlorinated hydro-
carbons, 0.4 mL of mineral suspension was extracted with
0.35 mL of hexane (Merck), which contained 5 µM CHBr3 as
internal standard. Extraction took place for 45 s on a vortex
shaker in 0.8-mL autosampler vials with PTFE-coated butyl
rubber septa. Six calibration standards were prepared in water
and extracted in an identical way, except for pentachloro-
ethane, which undergoes â-elimination of HCl in aqueous
solution so that standards were prepared directly in hexane/
CHBr3. Samples and standards were either analyzed within
2 d or stored at -30 °C until analysis. Analysis took place on
a gas chromatograph with electron capture detector (GC/
ECD, Carlo Erba HRGC 5160 with autosampler AS200 and a
Carlo Erba ECD 400 with a Ni-63 source) equipped with a
DB-624 column (i.d. 0.32 mm, 1.8-µm film thickness, J&W).
The injector and detector temperatures were 250 °C and 320
°C respectively, carrier gas flow was 1.5 mL/ min (hydrogen),
and makeup gas flow was 12 mL/ min (90% argon, 10%
methane). The parameters for measurement were as fol-
lows: injection of 1.5 µL splitless, 60 °C (1 min), ramp 10
°C/ min to 175 °C (0 min), ramp 30 °C/ min to 250 °C (6 min).
276 nm, and 4-chloroaniline (tret ) 8.7 min) was quantified
at 245 nm. Frequently an additional peak (probably 4-chloro-
phenyl hydroxylamine, see Results and Discussion, λmax
)
240 nm) appeared at tret ) 7.7 min. Its area increased and
decreased with the missing molar balance observed in the
experiments (see Figure 2c).
Results and Discussion
Reaction Kinetics and Product Form ation. The reactions of
hexachloroethane and 4-chloronitrobenzene with mineral-
bound Fe(II) generally followed pseudo-first-order behavior
(see examples in Figure 2). Reaction of hexachloroethane
produced quantitatively tetrachloroethene (C2Cl4) for all
minerals. No further dechlorination of C2Cl4 was observed
during the (relatively short) period of our measurements (2-3
half-lives of the reaction of hexachloroethane). In the case
of fast transformation (half-lives of a few hours) small
amounts of pentachloroethane (C2HCl5) and traces of
trichloroethene (C2HCl3), its dechlorination product, were
detected (see Figure 2a and Scheme 1).
Aqueous filtrates for analysis of 4-chloronitrobenzene and
4-chloroaniline were kept in 1-mL autosampler vials with
PTFE septa at -30 °C and then analyzed without further
treatment. Analysis took place by gradient high-performance
liquid chromatography (HPLC) on an RP-8 reversed-phase
column (125 × 4 mm) equipped with a precolumn (4 × 4
mm; both LiChroCart stainless steel cartridge, 5-µm spheres;
Merck) connected to a pumping system (Gynkotek M480
with gradient pump Henggeler DG4) supplemented with an
autosampler (Gina 50; Gynkotek) and a diode array UV
detector (340S, Gynkotek). The injected volume was 100 µL,
the flow was 1 mL/ min, the eluents were (A) methanol, (B)
5 mM phosphate (pH 7) in H2O, (C) H2O, and the gradient
program was as follows: 5 min. prior to injection 20% A, 20%
B, 60% C; ramp to 4 min prior to injection 20% A, 50% B, 30%
C; ramp to 0 min (injection) 20% A, 80% B; ramp to 5 min
60% A, 20%B, 20% C; ramp to 10 min 60% A, 40% C; ramp
to 13 min 90% A, 10% C; ramp to 15 min (end) 20% A, 80%
C; 4-chloronitrobenzene (tret ) 10.9 min) was quantified at
SCHEME 1
Roberts et al. (40, 41) have argued that the pentachloro-
ethyl carbanion decomposes quickly without significant
protonation in aqueous solution, indicating that formation
of C2HCl5 may be diagnostic for formation of pentachloroethyl
radicals. At our experimental conditions, i.e., in the absence
of efficient H-donors, we do not expect nor find significant
formation of pentachloroethane, even if the reaction mech-
anism involves pentachloroethy radicals.
In most experiments with 4-chloronitrobenzene, disap-
pearance of the parent compound was more rapid than the
formation of 4-chloroaniline (Figure 2c). We attributed a peak
9
VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 8 0 3

to that of 4-chloronitrobenzene. Moreover, they show that
the expression in the denominator in the Langmuir-
Hinshelwood expression can become significantly different
from unity already for low substrate concentrations (see also
Figure S2 in the Supporting Information). Still, the estimate
SCHEME 2
in the HPLC chromatogram which appeared at intermediate
reaction times (see Materials and Methods) to transient
accumulation of the intermediate 4-chlorophenyl hydroxyl-
amine based on its elution time in proximity to that of
4-chloroaniline and its UV spectrum. In analogy to earlier
studies by Dunnivant et al. (42) and Klausen et al. (1) (see
Scheme 2), no indication for 4-chloronitrosobenzene was
found, except for possible traces in reactions with magnetite
(data not shown).
of (kintrinsic‚cFe(II),ads‚KAr-NO ) ) 2.5 L h-1 m-2 is of the same
2
order of magnitude as the pseudo-first-order rate constant
of 1.3 L h-1 m-2 given in Table 1. This indicates that rate
constants reported in this study can be qualitatively inter-
preted as lumped parameters including the intrinsic rate
constant, kintrinsic, the concentration of reactive Fe(II) surface
sites, cFe(II), ads, and the affinity of the substrate to the mineral
surface, KAr-NO . Modeling of the two scenarios assuming (a)
2
a constant concentration of surface bound Fe(II) and (b) a
decreasing concentration of reactive Fe(II) surface species,
showed that depletion of Fe(II) sites did not play a major role
in the reaction of lepidocrocite/ Fe(II) with 4-chloroni-
trobenzene (see Supporting Information).
Pseudo-first-order rate constants, kobs, for 4-chloronitro-
benzene (Ar-NO2) were determined from ln([Ar-NO2]/
[Ar-NO2]t ) 0) vs time plots using data points at the beginning
of the experiment. Using data after complete disappearance
of 4-chloronitrobenzene at a time t′, pseudo-first-order rate
constants were also calculated for 4-chlorophenyl hydroxyl-
amine (Ar-NOH), from ln([Ar-NOH]/ [Ar-NOH]t ) t′) vs time
plots, starting at time t′ and using calculated concentrations
of the intermediate by assuming a complete molar balance.
As can be seen from Figure 2c, experimental data could be
fitted well using this assumption of consecutive reactions.
Exceptions are experiments with siderite, hematite, and
pyrite, and with Fe(II) in homogeneous solution, where no
accumulation of 4-chlorophenyl hydroxylamine was observed
(Figure 2d), and with green rust, where initial disappearance
of 4-chloronitrobenzene was too rapid to be quantified.
Experimental reaction rates constants kobs were checked
for artifacts such as (i) mass-transfer limitations or (ii) effect
of higher substrate concentrations. (i) Mass transfer may
become rate-limiting if surface reactions are so fast that
contaminant diffusion to the surface becomes the slow step
in the overall process (43). Values of kobs are then of similar
magnitude as estimates for a‚kL, where kobs is the observed
pseudo-first-order rate constant, a is the ratio of external
(geometric) particle surface area to volume of solution (m-1),
and kL is the calculated mass transfer coefficient (ms-1). Our
estimates of a‚kL (see Supporting Information and Table 1)
indicate that mass transfer was possibly rate-limiting only in
the reaction of 4-chloronitrobenzene with mackinawite (FeS)
(100%), goethite (40%), and magnetite (30%), but not in
reactions of hexachloroethane. (ii) At higher substrate
concentrations, deviations from pseudo-first-order kinetics
can occur for two reasons: (a) competition between sub-
strates for a restricted number of surface sites, which may
lead to a decrease of observed reaction rates, and (b) depletion
of reactive Fe(II) surface sites which can pose a limit to the
overall reaction. These effects were investigated exemplary
for the reaction of 4-chloronitrobenzene with lepidocrocite/
Fe(II) (see Supporting Information). Concentration/ time
graphs were evaluated according to Langmuir-Hinshelwood
kinetics assuming a rate-determining surface reaction (44)
Reactivity of Different Fe(II)-Containing Minerals with
Hexachloroethane. Surface-area-normalized reaction rate
constants kobs′ (kobs′ ) kobs/ {surface area [m2/ L]}) of hexa-
chloroethane differed greatly for Fe(II) bound at different
minerals (see Figure 3a), in a fashion similar to the pattern
sketched in Figure 1b. The reactivity increased in the order
Fe(II)aq + siderite < Fe(II) + iron oxides < Fe(II) + iron
sulfides, with the ferruginous clay mineral nontronite ranking
between the iron oxides. This observed reactivity pattern
might indicate different intrinsic reactivities of surface-bound
Fe(II), but it is also consistent with (a) different surface
densities of reactive Fe(II) or (b) different specific interactions
of the probe compounds with various surfaces (e.g., precursor
complex formation). In addition, owing to the presence of
interlayers in minerals such as nontronite or green rust, the
determined BET surface may not necessarily reflect their
total effective surface. The normalization with respect to the
BET surface can therefore be regarded as an operational
approach, with the aim to categorize all minerals according
to the same procedure with parameters that can be easily
obtained.
To check the validity of this approach, in cases of minerals
where Fe(II) sorption had been quantified (see Table 1),
reaction rates were alternatively normalized to the sorption
density of Fe(II) (Figure 3b). In the case of iron oxides and
hexachloroethane this normalization procedure also led to
rate constants that varied only by half an order of magnitude
(see Figure 3a and b). This indicates that (a) the reactivity of
Fe(II) at various iron oxides may indeed be described by
similarly reactive dFe(III)sOsFe(II)sOH surface complexes
such as sketched in Figure 1a and that (b) normalizations on
surface area and/ or Fe(II)-sorption-site-density both give
practical approximations to the true concentration of reactive
sites that cannot be measured directly. The orders of
magnitude differences in reactivity observed for reactions of
hexachloroethane (Figure 3a) with Fe(II) present at iron
minerals of different chemical composition thus are likely
due to different intrinsic reactivities of Fe(II) at the surfaces
of these minerals such as sketched in Figure 1b rather than
by variations in the Fe(II) sorption densities. Reactivity
differences between the mineral classes may therefore be
rationalized as follows.
∆cAr-NO
∆t
kintrinsic‚cFe(II),ads‚(KAr-NO ‚cAr-NO )
2 2
2
)
1 + (KAr-NO ‚cAr-NO ) + (KAr-NOH‚cAr-NOH
)
2
2
where kintrinsic is the intrinsic rate constant of 4-chloronitro-
benzene in the surface reaction, cFe(II), ads is the concentration
of Fe(II) surface sites, and cAr-NO and cAr-NOH are the
concentrations of 4-chloronitrobenzene and 4-chlorophenyl
hydroxylamine. The constants KAr-NO and KAr-NOH represent
the affinity of the compounds toward sorption to reactive
surface sites and thus describe the competition between
different substrates at the surface. Values of KAr-NO2 (0.13 (
0.05 µM-1) and KAr-NOH (1.78 ( 0.79 µM-1) were obtained,
and indicate a 10× higher affinity of the intermediate
4-chlorophenyl hydroxylamine to reactive sites in comparison
Fe(II) on FeCO3. Earlier studies (45) found that carbonate
ligands in solution enhanced oxidation of Fe(II). In contrast,
our data suggest that Fe(II) within the siderite lattice was a
poor promotor of electron transfer to organic pollutants. The
low reactivity of Fe(II) at the siderite surface may be explained
by stabilization of Fe(II) compared to Fe(III) within FeCO3.
Conversely, the high reactivity of FeS or FeS2 may also be
attributable to reduced sulfide or disulfide ligands. These
sulfide entities may act either as promoters of electron transfer
by Fe(II)) in analogy to iron-sulfur clusters in non-heme
2
2
9
8 0 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004

FIGURE 3. Logarithmic pseudo-first-order rate constants for the reaction of different Fe(II) containing minerals with the reducible probe
compounds (0) hexachloroethane, (b) 4-chloronitrobenzene, and (O) 4-chlorophenyl hydroxylamine. (a) Surface area-normalized, (b)
normalized to sorbed Fe(II). The asterisk (*) indicates likely mass transfer limitation.
iron proteins (46), as reductants of Fe(III) centers formed
after electron transfer from Fe(II), or as direct conveyers of
electrons to the contaminant.
erature data, however, indicates that normalization to the
BET surface is an expedient approach to comparing such
data.
The intermediate reactivity of iron oxides may be at-
tributable to the extent by which the coordination at the
oxide surface promotes electron transfer from dFe(III)sOs
Fe(II)sOH complexes. The comparatively low reactivity of
nontronite, finally, suggests that structural Fe(II) in interlayers
played only a minor role in the transformation of hexa-
chloroethane. Recent data (19) suggest that Fe(II) coordinated
to edge surface hydroxyls dominates the reactivity of nontro-
nite. Conversely, green rust at pH 8 was very reactive, which
points to a simple pH effect, to a high reactivity of Fe(II)
associated with the mineral, or to a large number of surface
sites, possibly located in readily accessible mineral interlayers
(18). An X-ray diffractogram of resuspended sulfate green
rust after one month showed peaks at smaller interlayer
distances, which could indicate a partial anion exchange in
the interlayer, and, thus, a certain interlayer accessibility for
contaminants (see Figure S1 in the Supporting Information).
The overall high reactivity of green rust is in agreement with
results from Pecher et al. who observed increased reactivity
of goethite/ Fe(II) at higher pH values which they attributed
to the formation of highly reactive secondary mineral
precipitates such as green rust (6). Conversely, when
suspended green rust was buffered at pH 7.5 it turned black,
and the reaction rate dropped to the level of magnetite and
goethite. This indicates that a surface layer of a less reactive
secondary precipitate (probably magnetite) formed which
dominated the overall reactivity, irrespective of the presence
of green rust in the bulk phase underneath. Green rust could
thus be identified as an exceptionally reactive iron (hydr)-
oxide phase.
Reactivity of Different Fe(II)-Containing Minerals with
4-Chloronitrobenzene. For a given mineral suspension, rate
constants of 4-chloronitrobenzene were consistently higher
than those for hexachloroethane and showed a similar
reactivity pattern (Fe(II) + siderite < Fe(II) + iron oxides <
Fe(II) + iron sulfides; Figure 3a and b). Deviations from the
reactivity pattern found for hexachloroethane were observed
for pyrite/ Fe(II), hematite/ Fe(II), and, to a lesser degree, for
nontronite/ Fe(II), in a fashion similar to the sketch in Figure
1d. Specific interactions with reactive surface-bound Fe(II)
and/ or differences in the accessibility to different surface
sites must therefore play a role in determining the overall
reactivity of nitroaromatic compounds. Possibly, 4-chlo-
ronitrobenzene was transformed at reactive sites different
from those of hexachloroethane and such “4-chloroni-
trobenzene-specific” sites were less abundant in hematite
and pyrite suspensions. This explanation would be consistent
with previous observations made in iron-reducing aquifer
columns (47, 48) and a long-term study with granular iron(0)
(49), which suggested that chlorinated hydrocarbons and
nitroaromatic compounds reacted at least partially inde-
pendent from each other at different reaction sites. An
alternative explanation is specific sorption of 4-chloronitro-
benzene, either (a) at unreactive sites, which would retard
the reaction in some cases, or (b) at reactive sites, which
would lead to accelerated rates for the majority of minerals.
Recent evidence for sorption of nitroaromatic compounds
at nonreactive sites comes from experiments with ferruginous
clay minerals, where an observed retarded transformation
was attributed to sorption at nonreactive siloxane surfaces
(20). This may explain why in our study 4-chloroaniline was
formed much more slowly in suspensions of nontronite as
compared to batches with iron oxides (Figure 3). In suspen-
sions of all other minerals studied, however, we did not notice
significant sorption of 4-chloronitrobenzene. The cases of
atypically low reactivity of 4-chloronitrobenzene in our study
may thus be an interesting starting point for further
investigations of specific surface reactivities of nitroaromatic
compounds. They compromise, however, attempts to es-
tablish reactivity rankings (see Figure 1c) in order to estimate
Com parison of Reaction Rates with those of Earlier
Studies. Whereas the measured reaction rate of hexa-
chloroethane with goethite/ Fe(II) (0.91‚10-3 Lm-2h-1) com-
pares reasonably well with data by Pecher et al. (6) (2.0‚10-3
Lm-2h-1), the value for mackinawite (0.016 Lm-2h-1) is six
times smaller than that reported by Butler et al. (28) (0.1
Lm-2h-1 at pH 7.1). The difference may be caused by different
experimental conditions (mineral loading, total surface area,
and/ or concentration of Fe(II) in solution). The overall fairly
good agreement between our values and comparable lit-
9
VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 8 0 5

A back-of-the-envelope calculation for the case of iron-
oxide-rich sediments follows. If we consider an iron-reducing
model sediment with a porosity of 0.3, an average mineral
density of 2.5 g/ cm3, and a content of 2% iron oxide with a
specific surface area 10 m2/ g, we arrive at (700 cm3 mineral/
L‚2.5 g/ cm3‚2/ 100‚10 m2/ g ) ) 350 m2 iron-oxide-surface/ L
sediment. Kenneke and Weber (14) have used material of
similar bulk properties originating from an iron-reducing
pond. After 10-fold dilution their suspensions contained 16
g/ L solids of a specific surface area of 22 m2/ g, corresponding
to (10‚16 g/ L‚22 m2/ g ) ) 3520 m2 silt-and-clay-surface per
L of suspension. This material transformed halogenated
methanes about 10× more slowly than goethite/ Fe(II) (based
on surface-normalized reaction rates), indicating that its
reactivity was equivalent to 352 m2 iron oxide/ L sediment.
We can now apply the rate constant determined with
goethite/ Fe(II) in our study to calculate the half-life of
hexachloroethane in the iron-reducing model sediment as
ln(2)/ (9.1‚10-4 Lm-2h-1‚350 m2/ L) ) 2.2 h at 23 °C. Results
of Pecher et al. (6) with goethite/ Fe(II) may further be used
to estimate also transformation rates of polyhalogenated
methanes. Tetrachloromethane, for example, was found to
react 3.2 times slower than hexachloroethane in their study,
so that the half-life in the model sediment can be calculated
to (3.2‚2.2 h ) ) 6.9 h at 23 °C. An educated guess for
trichloroethene transformation can be made using data of
Lee and Batchelor (27). As can be seen from Figure 4, their
observed reaction rate for trichloroethene with magnetite
was about 2.5 × 104 times smaller than the one for
hexachloroethane, resulting in an estimated half-life of about
(2.5 × 104‚2.2 h ) ) 5.6 × 104 hours or 6.4 years at 23 °C in
the iron-reducing model sediment.
Note that an unrealistically high surface of 20 000 m2/ L
would be required in the case of FeCO3 for equal reactivity
but a surface of only 20 m2/ L would be sufficient in the case
of iron sulfides. Kenneke and Weber (14) compared the
reactivity of sulfate- and iron-reducing sediment samples
and found little differences in halomethane reduction rates.
In light of our data, their conclusion that Fe(II) on iron oxides
was the dominating reductant in both sediments suggests
that the exposed surface area of sulfide minerals was very
low.
Such estimates illustrate how data from this study may
be used to calculate the persistence of structurally related
compounds such as halogenated alkanes, and, possibly, also
alkenes, in natural environments. However, because reactivity
patterns in our laboratory experiments showed different
trends for different classes of contaminants (4-chloronni-
trobenzene/ hexachloroethane), such calculations may not
be applied to estimate reaction rates of different contaminant
classes (e.g., nitroaromatic compounds). The influence of
specific reaction mechanisms and/ or specific contaminant/
surface interactions, which may contribute to the overall
reactivity, prevents such simple extrapolations. When making
such back-of-the-envelope calculations, finally, care has to
be taken that the surface reaction is really the rate-
determining step in the overall process. Heijman et al. (26)
investigated the transformation of highly reactive nitro-
aromatic compounds at elevated concentrations in iron-
reducing aquifer columns, and found that the overall
reactivity was limited by microbial regeneration of a limited
number of reactive sites, rather than by the actual abiotic
transformation itself. The transformation rates reported in
this study should, therefore, be regarded as maximum values
per surface area, for the case that dissolved ferrous iron is
provided in excess by in-situ microbial activity.
FIGURE 4. Logarithmic pseudo-first-order rate constants for
reactions of (0) hexachloroethane (this study), and trichloroethene
(4) Sivavec et al. (25)/2 Lee et al. (27, 50)) with different Fe(II)
bearing minerals.
reaction rates of one type of contaminant (e.g., a nitroaro-
matic compound) from data of other compound classes (e.g.,
chlorinated alkanes).
Extrapolation of Reaction Rates between Structurally
Related Com pounds. To address the question whether
reactivity patterns as discussed above may be similar for
structurally related compounds, such as sketched in Figure
1c, our data for the aliphatic chlorinated hydrocarbon
hexachloroethane may be compared to literature data of an
olefinic chlorinated hydrocarbon, trichloroethene (see Figure
4). Recently published (27, 50) rate constants for trichloro-
ethene (green rust 8.55 × 10-5 Lm-2 day-1, pyrite 2.53 × 10-5
Lm-2 day-1, magnetite 7.21 × 10-7 Lm-2 day-1) show a similar
reactivity trend as observed with hexachloroethane. Earlier
data by Sivavec et al. (25) for reactions of trichloroethene in
reaction with troilite (an iron sulfide, 5.52 × 10-2 Lm-2 day-1
)
and magnetite (4.56 ×10-4 Lm-2 day-1) show a similar relative
trend, despite large differences in absolute rate constants. (A
discussion of these differences is difficult, because the
experimental conditions of that study are not well specified.)
These trends suggest that it may be possible to extrapolate
reactivity patterns with different Fe(II)-bearing minerals
within the class of aliphatic/ olefinic hydrocarbons. In
contrast, a recent study addressing the reactivity of poly-
halogenated methanes in sulfate- and iron-reducing sedi-
ments (14) found reactivity patterns in both types of
sediments were similar to those of Fe(II)/ goethite suspen-
sions, but differed from those of FeS batches. However, as
the commercial FeS used contained impurities of Fe(0), the
pattern attributed to FeS may have been masked by the
reactivity of zerovalent iron. In conclusion, the reactivity
trends observed for hexachloroethane in the present study
may be valuable for predicting relative reactivities of struc-
turally related chlorinated hydrocarbons. Prospects and
limitations of such predictions, however, need to be validated
by future studies.
Environmental Significance
The strong dependence of reaction rates of surface-bound
Fe(II) on the chemical composition of iron-containing
minerals indicates that natural abiotic transformation reac-
tions involving such species in the field may vary strongly
with (bio)geochemical conditions. On the basis of our data,
one would expect reductive transformation of hexachloro-
ethane by Fe(II) to be fastest either at coexisting sulfate- and
iron-reducing conditions, where Fe(II) sulfides can form, or
under rapidly changing oxic/ anoxic redox conditions, where
green rusts may form. Significant reactivity would also be
expected for iron(III)oxide-rich soils, when dissolved Fe(II)
is present in the pore water. Low transformation rates are
anticipated in strictly anoxic calcite-rich grounds, where
siderite-formation takes place.
Acknowledgments
M. Huber contributed part of the experimental data on
reduction of hexachloroethane. We thank L. Roberts, S. Hug,
9
8 0 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 3, 2004

(21) Buerge, I. J.; Hug, S. J. Environ. Sci. Technol. 1998, 32, 2092-
2099.
(22) Wehrli, B. In Aquatic Chemical Kinetics; Reaction Rates of
Processes in Natural Waters; Stumm, W., Ed.; Wiley-Inter-
science: New York, 1990.
(23) Strathmann, T. J.; Stone, A. T. Environ. Sci. Technol. 2002, 36,
653-661.
(24) Strathmann, T. J.; Stone, A. T. Environ. Sci. Technol. 2002, 36,
5172-5183.
(25) Sivavec, T. M.; Horney, D. P. In ACS National Meeting; Division
of Environmental Chemistry ACS, 1997; American Chemical
Society: Washington, DC, 1997; pp 115-117.
(26) Heijman, C. G.; Grieder, E.; Holliger, C.; Schwarzenbach, R. P.
Environ. Sci. Technol. 1995, 29, 775-783.
(27) Lee, W.; Batchelor, B. Environ. Sci. Technol. 2002, 36, 5147-
5154.
(28) Butler, E. C.; Hayes, K. F. Environ. Sci. Technol. 1998, 32, 1276-
1284.
(29) Erbs, M.; Hansen, H. C. B.; Olsen, C. E. Environ. Sci. Technol.
1999, 33, 307-311.
B. Nowack, T. Hofstetter, M. Huber, M. Erbs, and N. Tobler
and three anonymous reviwers for helpful comments on the
manuscript. This work was funded by the Swiss National
Science Foundation (Grant 21-57171.99), by the DOW
Foundation Supporting Public Health and Environmental
Research Effects (SPHERE) program, and by the Swiss Federal
Office for Education and Science (grant BBW 99.0396) as
part of the European research project “Protection of Ground-
water Resources at Industrially Contaminated Sites” (PURE,
EVK1-CT-1999-00030).
Supporting Information Available
X-ray diffractograms of synthesized minerals (Figure S1); ionic
composition and ionic strength in the experiments (Table
S1); calculations of mass transfer parameters; effect of
compound concentrations on kobs (Figure S2); correlations
between logarithmic rate constants of hexachloroethane and
4-chloronitrobenzene/ 4-chlorophenyl hydroxylamine (Fig-
ure S3). This material is available free of charge via the Internet
at http:/ / pubs.acs.org.
(30) Kriegmann-King, M. R.; Reinhard, M. Environ. Sci. Technol. 1992,
26, 2198-2206.
(31) Cervini-Silva, J.; Larson, R. A.; Wu, J.; Stucki, J. W. Environ. Sci.
Technol. 2001, 35, 805-809.
(32) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M.
Environmental Organic Chemistry, 2nd ed.; John Wiley & Sons:
New York, 2002.
(33) Fadrus, H.; Maly, J. The Analyst 1975, 100, 549-554.
(34) Regazzoni, A. E.; Urrutia, G. A.; Blesa, M. A.; Maroto, A. J. G. J.
Inorg. Nucl. Chem. 1981, 43, 1489-1493.
(35) Stucki, J. W.; Golden, D.; Roth, D. C. Clays Clay Miner. 1984, 32,
191-197.
(36) Shriver, D. F.; Drezdzon, M. A. The Manipulation of Air-Sensitive
Compounds; Wiley: New York, 1986.
Literature Cited
(1) Klausen, J.; Tro¨ ber, S. P.; Haderlein, S. B.; Schwarzenbach, R.
P. Environ. Sci. Technol. 1995, 29, 2396-2404.
(2) Ru¨ gge, K.; Hofstetter, T.; Haderlein, S. B.; Bjerg, P. L.; Knudsen,
S.; Zraunig, C.; Mosbaek, H.; Christensen, T. Environ. Sci.
Technol. 1998, 32, 23-31.
(3) Hofstetter, T.; Heijman, C. G.; Haderlein, S. B.; Holliger, C.;
Schwarzenbach, R. P. Environ. Sci. Technol. 1999, 33, 1479-
1487.
(4) Haderlein, S. B.; Pecher, K. In Kinetics and Mechanisms of
Reactions at the Mineral/Water Interface; Sparks, D. L., Grundl,
T., Eds.; ACS Symposium Series No. 715; American Chemical
Society: Washington, DC, 1998; Chapter 17, pp 342-357.
(5) Amonette, J.; Workman, D. J.; Kennedy, D. W.; Fruchter, J. S.;
Gorbi, Y. A. Environ. Sci. Technol. 2000, 34, 4606-4613.
(6) Pecher, K.; Haderlein, S. B.; Schwarzenbach, R. P. Environ. Sci.
Technol. 2002, 36, 1734-1741.
(37) Brunauer, S.; Emmet, P. H.; Teller, E. J. Am. Chem. Soc. 1938,
60 (2), 309-319.
(38) Coughlin, B. R.; Stone, A. T. Environ. Sci. Technol. 1995, 29,
2445-2455.
(39) Lide, D. R. CRC Handbook of Chemistry and Physics, 76 ed.; CRC
Press: Boca Raton, FL, 1994.
(40) Roberts, A. L.; Gschwend, P. M. Environ. Sci. Technol. 1991, 25,
76-86.
(41) Roberts, A. L.; Gschwend, P. M. J. Contam. Hydrol. 1992, 16,
157-174.
(42) Dunnivant, F. M.; Schwarzenbach, R. P.; Macalady, D. L. Environ.
Sci. Technol. 1992, 26, 2133-2141.
(43) Arnold, W. A.; Ball, W. P.; Roberts, A. L. J. Contam. Hydrol. 1999,
40, 183-200.
(44) Fogler, H. S. Elements of Chemical Reaction Engineering; Prentice
Hall: Engelwood Cliffs, NJ, 1992.
(45) King, D. W. Environ. Sci. Technol. 1998, 32, 2997-3003.
(46) Spiro, T. G. Iron-Sulfur Proteins; Wiley-Interscience: New York,
1982.
(47) Hofstetter, T. B. Ph.D. Thesis, Swiss Federal Institute of
Technology; ETH Zu¨ rich, 1999.
(48) Schwarzenbach, R. P.; Hofstetter, T. B.; Elsner, M.; Haderlein,
S. B. In 220th National Meeting of the American Chemical Society;
American Chemical Society: Washington, DC, 2000; Vol. 40 (2),
pp 311-313.
(49) Klausen, J.; Vikesland, P. J.; Kohn, T.; Burris, D. R.; Ball, W. P.;
Roberts, A. L. Environ. Sci. Technol. 2003, 37, 1208-1218.
(50) Lee, W.; Batchelor, B. Environ. Sci. Technol. 2002, 36, 5348-
5354.
(7) Buerge, I. J.; Hug, S. J. Environ. Sci. Technol. 1999, 33, 4285-
4291.
(8) Cui, D.; Eriksen, T. E. Environ. Sci. Technol. 1996, 30, 2259-
2262.
(9) Liger, E.; Charlet, L.; Van Capellen, P. Geochim. Cosmochim.
Acta 1999, 63, 2939-2955.
(10) Strathmann, T. J.; Stone, A. T. In 219th ACS National Meeting;
Division of Environmental Chemistry, American Chemical
Society: Washington, DC, 2000; Vol. 40 (1), pp 141-144.
(11) Vikesland, P. J.; Valentine, R. L. Environ. Sci. Technol. 2000, 34,
83-90.
(12) Bjerg, P. L.; Ru¨ gge, K.; Pedersen, J. K.; Christensen, T. H. Environ.
Sci. Technol. 1995, 29, 1387-1394.
(13) Christensen, T. H.; Kjeldsen, P.; Albrechtsen, H.-J.; Heron, G.;
Nielsen, P. H.; Bjerg, P. L.; Holm, P. E. Crit. Rev. Environ. Sci.
Technol. 1994, 24, 119-202.
(14) Kenneke, J. F.; Weber, E. J. Environ. Sci. Technol. 2003, 37, 713-
720.
(15) Millero, F. Geochim. Cosmochim. Acta 1985, 49, 547-554.
(16) Stumm, W.; Morgan, J. J. Aquatic Chemistry, 3 ed.; John Wiley
& Sons: New York, 1996.
(17) Charlet, L.; Silvester, E.; Liger, E. Chem. Geol. 1998, 151, 85-93.
(18) Hansen, H. C. B.; Koch, C. B. Clay Miner. 1998, 33, 87-101.
(19) Hansen, H. C. B.; Guldberg, S.; Erbs, M.; Koch, C. B. Appl. Clay
Sci. 2001, 18, 81-91.
Received for review June 4, 2003. Revised manuscript re-
ceived October 17, 2003. Accepted November 1, 2003.
(20) Hofstetter, T.; Haderlein, S. B.; Schwarzenbach, R. P. Environ.
Sci. Technol. 2003, 37, 519-528.
ES0345569
9
VOL. 38, NO. 3, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 8 0 7