Reduction of nitro aromatic compounds by zero-valent iron metal

Article abstract of DOI:10.1021/es950211h

The properties of iron metal that make it useful in remediation of chlorinated solvents may also lead to reduction of other groundwater contaminants such as nitro aromatic compounds (NACs). Nitrobenzene is reduced by iron under anaerobic conditions to aniline with nitrosobenzene as an intermediate product. Coupling products such as azobenzene and azoxybenzene were not detected. First-order reduction rates are similar for nitrobenzene and nitrosobenzene, but aniline appearance occurs more slowly (typical pseudo-first-order rate constants 3.5 x 10^-2, 3.4 x 10^-2, and 8.8 x 10^{- 3} min^-1, respectively, in the presence of 33 g/L acid-washed, 18-20 mesh Fluka iron turnings). The nitro reduction rate increased linearly with concentration of iron surface area, giving a specific reaction rate constant (3.9 ± 0.2 x 10^-2 min^-1 m^-2 L). The minimal effects of solution pH or ring substitution on nitro reduction rates, and the linear correlation between nitrobenzene reduction rate constants and the square-root of mixing rate (rpm), suggest that the observed reaction rates were controlled by mass transfer of the NAC to the metal surface. The decrease in reduction rate for nitrobenzene with increased concentration of dissolved carbonate and with extended exposure of the metal to a particular carbonate buffer indicate that the precipitation of siderite on the metal inhibits nitro reduction.

Full text of DOI:10.1021/es950211h

Environ. Sci. Technol. 1996, 30, 153-160
The first detailed studies of halocarbon degradation by
Reduction of Nitro Aromatic
Compounds by Zero-Valent Iron
Metal
iron metal in laboratory batch systems (4, 5) made several
fundamental considerations apparent: (i) halocarbon
degradation occurs by dechlorination through a surface
reaction with the metal as the ultimate electron donor; (ii)
the major determinants of degradation kinetics are mass
transfer to, area of, and condition of the metal surface; and
(iii) mediation by H2, metal impurities, or microorganisms
do not appear to be necessary for rapid degradation.
Concurrent studies performed in laboratory columns (5, 6)
showed that halocarbon degradation occurs similarly in
porous media, but that there is an important additional
consideration: geochemicalevolution ofthe matrixmaterial
by iron dissolution and precipitation over the course of
extended operation. Many additional studies of contami-
nant remediation with zero-valent metals have now been
reported (2, 7-9). These studies support the view that
contaminant degradation results from reduction coupled
to metal corrosion, and most address at least one additional
factor, such as the range of contaminants that may be
remediated with reducing metals, the effect of groundwater
geochemistry on remediation performance, or the pos-
sibilities for enhanced performance with derivative tech-
nologies.
A B I N A S H A G R A W A L † A N D
P A U L G . T R A T N Y E K *
Department of Environmental Science and Engineering,
Oregon Graduate Institute of Science & Technology,
P.O. Box 91000, Portland, Oregon 97291-1000
The properties of iron metal that make it useful in
remediation of chlorinated solvents may also lead to
reduction of other groundwater contaminants such
as nitro aromatic compounds (NACs). Nitrobenzene is
reduced by iron under anaerobic conditions to aniline
with nitrosobenzene as an intermediate product.
Coupling products such as azobenzene and azoxyben-
zene were not detected. First-order reduction rates
are similar for nitrobenzene and nitrosobenzene, but
aniline appearance occurs more slowly (typical
pseudo-first-order rate constants 3.5 × 10-2, 3.4 ×
10-2, and 8.8 × 10-3 min-1, respectively, in the presence
of 33 g/L acid-washed, 18-20 mesh Fluka iron turn-
ings). The nitro reduction rate increased linearly
with concentration of iron surface area, giving a specific
reaction rate constant (3.9 ( 0.2 × 10-2 min-1 m-2
L). The minimal effects of solution pH or ring substitution
on nitro reduction rates, and the linear correlation
between nitrobenzene reduction rate constants and the
square-root of mixing rate (rpm), suggest that the
observedreactionrates were controlledbymass transfer
of the NAC to the metal surface. The decrease in
reduction rate for nitrobenzene with increased con-
centration of dissolved carbonate and with extended
exposure of the metal to a particular carbonate buffer
indicate that the precipitation of siderite on the metal
inhibits nitro reduction.
The remediation of nitro aromatic compounds (NACs)
is of interest because the nitro aromatic moiety is among
the most characteristic of anthropogenic contaminants,
being second in this regard only to organochlorine func-
tional groups. NACs are common environmental con-
taminants because of their use as munitions, insecticides,
herbicides, pharmaceuticals, and industrial feed stock
chemicals for dyes, plastics, etc. (10). They also may be
formed in the environment from aromatic contaminants,
as is the case with the nitro-PAHs and nitrophenols found
in atmospheric waters (11, 12). Among the processes
contributing to the environmental fate of NACs (13),
reduction of the nitro group is certainly the most charac-
teristic. The transformation reaction generally produces
the corresponding aromatic amines, with minor amounts
of intermediates (hydroxylamines and nitroso compounds)
and coupling products (azo and azoxy compounds).
However, since aromatic amines are still of concern as
environmentalcontaminants, remediation ofNACs requires
transformation beyond nitro reduction. Two possible
means for obtaining complete removal of NAC transfor-
mation products are (i) biodegradation, which sometimes
occurs more rapidly for aromatic amines than for the parent
nitro compounds (14) and (ii) incorporation into natural
organic matter byenzyme-catalyzed coupling reactions (15).
In principle, either technique could be coupled to nitro
reduction with Fe0 and the combination applied to treat
environmental contamination by NACs.
Introduction
Over the last several years, a great deal of interest has
developed in the groundwater remediation communityover
the prospects of new treatment strategies based on
dechlorination by granular iron metal. The apparent
success of the first field demonstration at Base Borden,
Ontario (1), has led to the initiation of numerous feasibility
studies, pilot tests, and small to medium scale demonstra-
tion projects (2, 3). These developments have created a
need for more process-level insight into the chemistry of
these systems in order to explain, predict, and/ or enhance
their performance.
In this study, we investigated the reduction of nitroben-
zene by iron metal in bicarbonate-buffered batch systems
to (i) assess the potential utility of nitro reduction by Fe0
in groundwater remediation, (ii) further our understanding
of the reactivity of Fe0 with organic solutes in aqueous
systems, and (iii) gain insight into the role that precipitation
of solids such as FeCO3 may play in remediation perfor-
mance under geochemical conditions. The results indicate
that nitro reduction by Fe0 is significantly faster than
* Corresponding author e-mail address: tratnyek@ese.ogi.edu.
† Present address: Department ofGeologicalSciences, Wright State
University, Dayton, OH 45435.
9
0013-936X/96/0930-0153$12.00/0
1995 Am erican Chem ical Society
VOL. 30, NO. 1, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 5 3

dechlorination (which has already been shown to be
effective in field applications) but that the primary products
are aromatic amines. Thus, nitro reduction by iron may
be of direct use in treatment of waters contaminated with
NACs, ifthe resulting amines can be removed bysubsequent
treatment. Beyond the direct relevance of nitro reduction
to remediation objectives, the reaction was used as a probe
technique with which to study the behavior of Fe0 in
aqueous systems. For this purpose, the advantages of NAC
reduction over dechlorination include the following: (i)
volatilization of the organic reactant and its reduction
products is much less significant, (ii) the specific rate of
reaction is more rapid, and (iii) the reaction products do
not include chloride, which itself is a strong promoter of
corrosion.
FIGURE 1. Schematic diagram showing the various possible
transformation pathways for nitrobenzene in reducing environments.
Downward arrows (reactions I-III and VI-VIII) indicate reduction
reactions, and right-facing arrows (reactions IV-V) indicate
condensations. The dominant reduction pathway of the nitro group
to its amino equivalent may occur sequentially via nitroso and
hydroxylamine (reactions I-III).
Background
Oxidation of Fe0. Zero-valent iron metal, Fe0, is readily
oxidized to ferrous iron, Fe2+, by many substances. In
aqueous systems, this phenomenon leads to dissolution of
the solid, which is the primary cause of metal corrosion.
Metal corrosion is an electrochemical process in which
oxidation of Fe0 to Fe2+ is the anodic half-reaction. The
associated cathodic reaction may vary with the reactivity
of available electron acceptors. In anoxic pure aqueous
media, the acceptors include H+ and H2O, the reduction
of which yields OH- and H2. Thus, the overall process of
corrosion in anaerobic Fe0-H2O systems is classically
described by the following reactions:
this process consists of a series of two-electron additions,
proceeding through nitroso and hydroxylamino intermedi-
ates. However, the reduction potentials for reactions I and
II are very similar, so polarography performed on acid to
neutral aqueous solutions gives only two waves: the first
corresponding to a four-electron reduction for formation
of the hydroxylamine, and the second corresponding to a
two-electron reduction of the hydroxylamine to the amine.
At alkaline pH, only one six-electron wave is observed.
Accumulation of the nitroso compound is rarely found in
practice, and special techniques are often necessary to
obtain direct evidence for its formation (24, 25). Indirect
evidence for its formation is abundant, however, because
condensation reactions involving the nitroso intermediate
(reactions IV-V) give azoxy and azo compounds, which
are frequently observed in significant yields from nitro
reduction (26-28). Product distributions may be further
complicated by reduction of the condensation products
(reactions VI-VIII) (29) or rearrangements such as the
formation of benzidine (not shown) (30).
Reduction of NACs is well documented for aqueous
media containing zero-valent metals such as Fe, Zn, or Sn
because these reactions have been used widely in the
synthesis of amines (31). According to a method that
originated with an early study by Lyons and Smith (32),
high yields can be obtained using Fe0 and trace amounts
of salts containing Fe2+ or Cl-. In addition to the formation
of amines, dissolving metal reductions of NACs may
produce good yields of the other products represented in
Figure 1. For example, reduction of nitrobenzene with Zn
is reported to be useful for preparation of phenylhydroxyl-
amine (33). In general, however, yields of products other
than amines are usually small and variable, so the reaction
is of limited synthetic utility and has received little atten-
tion in the recent literature.
Fe0 + 2H+ h Fe2+ + H2
(1)
(2)
Fe0 + 2H2O h Fe2+ + H2 + 2OH-
The preferred cathodic half-reaction under aerobic condi-
tions involves O2 as the electron acceptor. In this case, the
primary reaction yields only OH- and not H2:
2Fe0 + O2 + 2H2O h 2Fe2+ + 4OH-
(3)
Other strong electron acceptors (oxidants), both inorganic
and organic, may offer additional cathodic reactions that
-
contribute to iron corrosion. Systems containing HCO3
and H2CO3 species may act as electron acceptors to oxidize
Fe0 and promote metalcorrosion (16-18). Organic oxidants
also react with Fe0, as illustrated by the dehalogenation of
chlorinated hydrocarbons, RCl:
Fe0 + RCl + H+ h Fe2+ + RH + Cl-
(4)
Reaction 4 can contribute significantly to the net dissolution
of iron, even in predominantly aqueous systems (19). NACs
also exhibit facile reduction by Fe0, but this reaction is not
a significant contributor to material damage by corrosion
and therefore has not been studied in detail from the
corrosion perspective.
Chem ical Transform ations of NACs. Under reducing
conditions, NACs may react by a variety of pathways, which
are summarized in Figure 1. Most of these reactions have
been studied in great detail either as preparative methods
in synthetic chemistry or as model reactions for electro-
chemical investigations, and numerous reviews on these
topics are available (20-23). In almost all cases, the major
process is reduction of the nitro functional group to the
corresponding amine (Figure 1, reactions I-III). Formally,
In environmental media, reduction of NACs occurs by
both biotic and abiotic means (34). The dominant reaction
pathway under anaerobic environmental conditions ap-
pears to be nitro reduction to the amine. However, there
is evidence for most of the other pathways in Figure 1,
including coupling of nitro reduction intermediates to form
azo and azoxy compounds, and reductive cleavage of azo
compounds to form amines (35). The former reaction is
ofparticular importance because azo and azoxycompounds
9
1 5 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 1, 1996

are of greater concern as environmental contaminants than
their precursors. The aromatic amines formed by nitro
reduction are subject to additional reactions in environ-
mental systems including covalent binding to natural
organic matter (36) or adsorption to mineral surfaces (37).
In addition to the reactions included in Figure 1, environ-
mental transformation of NACs can include further deg-
radation by microorganisms. This is likely to include
hydroxylation and cleavage of the aromatic ring with
eventualmineralization, but such reactions are not expected
by the direct action of reducing metals alone.
bottle was allowed 2 h to equilibrate on an end-over-end
rotary shaker at 10 rpm in a dark, 15 °C room.
Early experiments in unbuffered Fe-H2O systems
resulted in the expected pH rise (>2 units in ∼4 h) due to
aqueous corrosion of the metal (eqs 2 and 3). Although
previous batch studies of dechlorination by Fe0 have not
shown evidence that the pH rise due to aqueous corrosion
significantly effects contaminant degradation rates (4), we
favored buffered systems for this study to gain greater
controlover experimentalconditions. Previous studies have
used various biological buffers in batch systems containing
Fe0 and not found any specific buffer effects (4, 7, 8). A
bicarbonate buffer with total dissolved carbonate ) 1.5 ×
10-2 M was adopted for routine experiments because it
provided a reasonable buffer capacity without visible
precipitation of iron carbonates during the course of a
typical degradation experiment (4-5 h).
A typical reduction experiment began by the addition
of 1 mL of 10-3 M deoxygenated aqueous stock solution of
nitrobenzene (or nitrosobenzene, or one of the substituted
nitrobenzenes) by needle injection through the septum.
Standard incubation conditions were maintained through-
out each experiment (i.e., dark, 15 °C, 10 rpm). Loss of
parent organic compound and formation of reduction
products were determined by periodically removing 200-
µL samples for immediate analysis by HPLC. The BET
surface area of the iron sample obtained from serum bottles
after the reduction experiment was 0.021 m2/ g.
Analyses. The concentrations of unreacted substrate
and its various reaction products were determined by
isocratic HPLC with UV absorbance detection. The ana-
lytical column (4 mm × 10 cm) consisted of Microsorb
packing (particle size, 3 µm) with a C-18 stationary phase.
A precolumn (4.6 mm × 1.5 cm) of the same material was
also used. The eluent consisted of 60/ 40 acetonitrile and
water (in most cases, buffered with Trizma at pH ) 7) with
a flow rate of 0.9 mL/ min. The absorbance was monitored
at 270 nm, which is close to the λmax for nitrobenzene (272
nm) and nitrosobenzene (280, 300 nm). Peaks were
identified by comparison with the retention times of
standard compounds, and concentrations were determined
from peak areas by comparison to standard curves. Surface
area was estimated by BET Kr-gas adsorption (Micromer-
itics) on samples that were prepared by rapidly drying the
metal with methanol and storing under N2 to avoid surface
oxidation.
Experimental Section
Chem icals. Nitro aromatic compounds and their likely
reaction products were obtained in high purity from Aldrich
and Chem Service and used as received. These included
nitrobenzene, 1-chloro-4-nitrobenzene, 1,3-dinitroben-
zene, 4-nitrotoluene, 4-nitroanisole, 2,4,6-trinitrotoluene,
parathion, nitrosobenzene, aniline, azobenzene, and azoxy-
benzene. Roughly 1 mM stock solutions were prepared in
deionized water (18 MΩ‚cm Nanopure) with up to 5% (v/ v)
HPLC-grade methanol. The bicarbonate buffer media were
prepared by flowing deoxygenated 1% CO2 in N2 or 100%
CO2 (i.e., pCO2 ) 0.01 and 1.0 atm, respectively) for roughly
2 h through a 15 mM NaOH solution in deionized water.
Analytical grade CO2 and other gases were deoxygenated
prior to use by passing through a heated column of reduced
copper. The pH range available from the above method
was between 5.5 and 8.0, as determined by pCO2 and the
aqueous alkalinity. To prepare more acidic buffers, the
initial NaOH solution was modified by the addition of
HCl prior to sparging with CO2.
The iron metal used in this study (Fluka electrolytic iron,
catalog no. 44905) was selected to give manageable nitro
reduction rates in the experimentalsystem described below.
Compared with the Fe0 used in our previous studies (4, 6),
this material has a coarser grain size (mostly >40 mesh)
and clean, smooth surfaces with a high metallic luster. The
nominal purity is 99.9% with the remainder consisting
primarily ofiron oxide. Trace impurities are less than 0.02%
C, 0.008% S, 0.003% Si, 0.002% P, 0.002% Mn, and 5 ppm
of Mg. The surface area of a 18-20 mesh sieved iron sample
prior to acid treatment was determined to be 0.005 m2/ g
by gas adsorption using BET analysis.
Iron Pretreatm ent. Prior to use, the Fe0 grains were
hand-sieved to constrain grain size to 18-20 mesh and
sonicated in 10% HCl (v/ v) for 20 min to remove surface
oxides or other contaminants. The cleaned metal was
washed four times with the bicarbonate buffer to remove
residual acidity or chloride remaining after the acid
treatment. The entire acid-washing procedure was per-
formed under a stream of deoxygenated N2 with freshly
deoxygenated bicarbonate buffer to avoid further surface
oxidation to metal oxides. A series of control tests indicated
that the mass of Fe0 lost due to this pretreatment was 4.9
( 0.2%. The BET surface area of the iron sample following
the acid-treatment increased to 0.038 m2/ g.
Results and Discussion
Pathway of Nitro Reduction by Fe0. The reaction of
nitrobenzene by Fe0 under the conditions of this study
produced aniline with nitrosobenzene as an intermediate
product. Accounting for these three species gives good
mass balance at pH > 5.5 (>85%), but with a reproducible
dip at intermediate times (Figure 2), which suggests that
all of the nitrosobenzene formed is not in solution and/ or
that there is an additional reaction intermediate. Phenyl-
hydroxylamine is the most likely intermediate and may
correspond to an unidentified peak with retention time )
1.2 min (cf. ∼1.0 min for the solvent peak, 2.2 min for aniline,
3.7 min for nitrobenzene, and 4.2 min for nitrosobenzene).
Similar elution patterns have been reported for nitroben-
zene reduction in other systems (24, 38). The variation in
size of this peak with reaction time suggests an intermediate
in reduction from nitrosobenzene to aniline, and similar
behavior was observed when nitrosobenzene was used as
Model Reaction System s. Individual degradation ex-
periments were performed in anaerobic (sealed) batch
systems prepared in 60-mL serum bottles. In most cases,
bottles received 2 g of dry, sieved iron. Following the acid
wash, the bottles were filled with fresh bicarbonate buffer
medium and crimp-sealed using thick butyl rubber septa
(Pierce). Before initiating degradation experiments, each
9
VOL. 30, NO. 1, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 5 5

FIGURE 3. Pseudo-first-order disappearance of nitrobenzene (same
experiment as Figure 2) and nitrosobenzene (separate experiment
under identical conditions). Linear regression gives reduction rate
constants of k1 ) 0.035 ( 0.001 min-1 (t1/2 ) 19.7 min) for nitrobenzene
and k2 ) 0.034 ( 0.001 min-1 (t1/2 ) 20.4 min) for nitrosobenzene.
FIGURE 2. Kinetics of nitrobenzene reduction to nitrosobenzene
and then to aniline under standard conditions: 33.3 g/L 18-20 mesh
acid-washed Fluka iron turnings, pretreated in 1.5 × 10-2
M
bicarbonate buffer (pCO ) 0.01 atm) for 2 h, mixed throughout by
2
rotation at 10 rpm and 15 °C. Data plotted relative to the initial
concentration of nitrobenzene (1.5 × 10-5 M).
regression on these data was used for routine determination
of k1, the pseudo-first-order rate constant for eq 6. The
average standard deviation for nitro reduction rate con-
stants determined for severalseries ofreplicate experiments
was <2.5%. As was concluded previously for the dechlo-
rination of CCl4 (4), the lack of deviation from first-order
kinetics indicates that changes in reactivity of the Fe0 due
to corrosion and precipitation are not significant over
several hours. However, nitro reduction experiments that
were allowed to run to completion did exhibit a smalldecline
in k1 at exposure times over 1 h. A similar decline in rates
was observed by Gillham and O’Hannesin for the reduction
of chlorinated aliphatics in batch experiments (5).
First-order rate constants for the appearance of nitroso-
benzene were obtained by nonlinear regression, assuming
a kinetic model for sequential first-order reactions is
appropriate (4). For the data in Figure 2, the rate constant
is 0.069 ( 0.010 min-1, which is roughly twice the value k1
) 0.035 ( 0.001 min-1 obtained by fitting the data for
nitrobenzene disappearance. The discrepancy is modest,
but appears to be significant, and suggests an additional
reaction for nitrosobenzene formation. One such reaction
might be the dismutation of phenylhydroxylamine (39),
but no effort was made to test this hypothesis experimen-
tally.
the initial substrate. However, a standard for phenylhy-
droxylamine was not available, so this assignment was not
confirmed. Neither azoxybenzene or azobenzene were
observed as products of nitrobenzene reduction, indicating
that condensation reactions (pathways IV-V, Figure 1) are
insignificant at the substrate concentrations used in this
study. The absence of detectable azo and azoxy interme-
diates suggests that reduction of these compounds (i.e.,
pathways VI-VIII, Figure 1) was not a significant contributor
to the formation of aniline.
Based on these results, sequential nitro reduction to
aniline via intermediate nitroso and hydroxylamine com-
pounds appears to be the only important transformation
process occurring in the Fe0-H2O-CO2 systems studied.
Formally, the overall reaction
ArNO2 + 3Fe0 + 6H+ f ArNH2 + 3Fe2+ + 2H2O (5)
occurs in three steps, each involving a two-electron
reduction, with Fe0 as the ultimate electron donor. The
contributing two-electron reactions can be written as
k1
ArNO2 + Fe0 + 2H+ 98 ArNO + Fe2+ + H2O (6)
The kinetics of nitrosobenzene reduction (eq 7) were
also investigated in two ways. Using nitrosobenzene as
the initial substrate, pseudo-first-order disappearance was
observed, and therefore k2 could be obtained simply by
linear regression of the natural log concentration versus
time data. Alternatively, the concentration of nitrosoben-
zene as an intermediate in nitrobenzene reduction could
be fit by nonlinear regression to the kinetic model for
sequential first-order reactions. For similar conditions, the
value ofk2 obtained from sequentialnitrobenzene reduction
(0.006 ( 0.001 min-1, Figure 2) was significantly less than
that obtained from nitrosobenzene as the initial substrate
(0.034 ( 0.001 min-1, Figure 3). This difference may be
due to competition for reactive sites on the metal by
nitrobenzene and its sequential reduction products. The
reduction of phenylhydroxylamine (eq 8) was accessible
only in terms of aniline appearance. The kinetics of aniline
appearance in Figure 2 fit a first-order model with a rate
constant, k3, of 0.009 ( 0.001 min-1. The relatively slow
rate of aniline appearance is consistent with previous
k2
ArNO + Fe0 + 2H+ 98 ArNHOH + Fe2+
(7)
k3
ArNHOH + Fe0 + 2H+ 98 ArNH2 + Fe2+ + H2O (8)
where each k represents the observed first-order rate
constant for the associated reduction step. The distribution
of products depends on the relative values of these rate
constants, which, in turn, will vary with reaction conditions.
Kinetics of Transform ation. Experiments were gener-
ally performed with nitrobenzene at an initialconcentration
of about 1.5 × 10-5 M, and the reaction was monitored
until >98% complete (∼3 h). Loss of nitrobenzene was
found to be <5% over 48 h in control experiments without
Fe0, so all of the disappearance during the first few hours
was attributed to nitro reduction. Naturallog concentration
versus time plots for nitrobenzene reduction were unam-
biguously linear over at least 3 half-lives (Figure 3), so linear
9
1 5 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 1, 1996

TABLE 1
Effect of Substitution on Pseudo-First Order Rate
Constant of Substrate Reductiona
k1
t1/2
(min)
substrateb
nitrobenzene
nitrosobenzene
1,3-dinitrobenzene
D (cm2 s-1 c
)
E
1/2 (V)d
(min-1
)
6.8 × 10-6 -0.55 (49)
7.1 × 10-6 -0.63 (50) 0.0339 20.4
6.2 × 10-6 -0.25 (51) 0.0339 20.4
4-chloronitrobenzene 6.2 × 10-6 -0.40 (52) 0.0336 20.6
4-nitroanisole
4-nitrotoluene
2,4,6-trinitrotoluene
parathion
5.9 × 10-6 -0.55 (51) 0.0327 21.2
6.2 × 10-6 -0.46 (52) 0.0335 20.7
5.6 × 10-6
0.0330 21.0
4.3 × 10-6 -0.21 (53) 0.0250 27.7
a
Conditions: 33.3 g/L acid-washed 18-20 m esh Fluka granular iron,
b
1.5 × 10-2 M carbonate buffer at pH 5.9, 10 rpm , and 15 °C. Initial NAC
concentration ) 1.5 × 10-5 M. Molecular diffusivity estim ated for
c
d
aqueous solution at 15 °C using the m ethod in ref 54. Polarographic
half-wave potentials for aqueous solution at pH 7; references given in
parentheses.
FIGURE 4. Effect of Fe0 surface area concentration on the pseudo-
first-order rate constant for nitrobenzene reduction. Experiments
were performed by varying mass of 18-20 mesh, acid-washed Fluka
iron turnings. All were performed in 1.5 × 10-2 M bicarbonate buffer
at 10 rpm and 15 °C, after 2 h for buffer equilibration. The regression
line corresponds to eq 9.
reports that aniline inhibits corrosion of steel by adsorbing
strongly and covering surface sites (40).
Effect of Substrate Properties. The similarity in disap-
pearance rates for nitrobenzene and nitrosobenzene (Figure
3) suggests a fundamental similarity in the rate controlling
processes. However, the diffusion coefficients and reduc-
tion potentials for these two compounds are very similar
(Table 1), so this result alone cannot be used to determine
whether the kinetics are limited by diffusion to/ from
reactive surface sites or by chemical reaction steps (activa-
tion control). Therefore, additional experiments were
performed with a series of para-substituted nitrobenzenes
selected to exhibit a range of diffusive and electronic
properties. The resulting values of k1 (Table 1) show little
if any change in nitro reduction rates over a wide range of
redox potentials as long as substrate diffusivities are similar.
Parathion is the only compound that exhibits a significantly
different value of k1, consistent with the lower diffusion
coefficient that results from the bulky p-(O,O-diethyl
phosphorothioate) substituent on the nitrophenyl ring.
Effects of the Iron Surface. Since the dechlorination of
halocarbons by Fe0 involves reaction at the metal surface
(4, 5), it was anticipated that the condition and quantity of
metal surface area would also influence the kinetics of nitro
reduction. Previous work has shown that treatment of the
metal with dilute HCl prior to each experiment accelerates
dechlorination rates and improves the reproducibility of
reaction rate determinations (4). In this study, preliminary
experiments showed similar effects for nitrobenzene re-
duction. The greatest increase in k1 resulted from the
combined action of HCl and sonication, so a protocol based
on this treatment was adopted for all subsequent experi-
ments. Aside from the practical advantages of increasing
the magnitude and reproducibility of the desired reaction
rates, our investigations ofpretreatment effects also provide
insight into the role of the metal surface in contaminant
transformation. The observed effect of acid pretreatment
may be due to one or more of the following changes: (i)
cleaning of the surface by dissolution of metal and
breakdown of the passivating oxide layer; (ii) increasing
the metal surface area by etching and pitting through
corrosion or by mechanical abrasion during sonication;
(iii) increase in the density of highly reactive “sites”
consisting of steps, edges, and kinks, following corrosion
by acid; and (iv) increased concentrations of adsorbed H+
and Cl- that persist after pretreatment with HCl.
Under circumstances where the metal surface condition
is effectively constant, the quantity of available surface area
is among the most significant experimental variables
affecting contaminant reduction rates. Figure 4 shows that
rate constants for nitrobenzene reduction, k1, increase
linearly with the mass of Fe0 per unit reaction volume (g/ L)
over the range of conditions studied. Assuming a constant
specific surface area of 0.021 m2/ g, the more general
independent variable ofsurface area concentration, [surface
area], can be derived. Linear regression on the data for k1
(min-1) and [surface area] (m2/ L) gives
k1 ) (0.039 ( 0.002) × [surface area] +
(0.003 ( 0.001) (9)
with r2 ) 0.997 for n ) 5, and the uncertainties represent
one standard deviation. The resulting intercept is statisti-
cally significant but negligible under realistic experimental
conditions, indicating that other loss processes are minor
and essentially all disappearance is attributable to reaction
with Fe0. The slope of eq 9 is the specific reaction rate
constant; i.e., where k1 has been normalized to 1 m2/ L iron.
Reduction rates characterized in these terms should be
independent of the mass and specific surface area of the
metal used and the volume of the reaction system. At
present, however, there is no practical way to factor out
variation in the density ofreactive sites on the metal surface.
The latter correction will be particularly important if
meaningful comparisons are to be made between iron
samples from different sources or with different histories
that affect the type and density of surface precipitates. For
example, the 16-fold difference between the values 2.5 (
0.2 × 10-3 min-1 m-2 L reported previously for CCl4
reduction by Fisher Fe0 (4) and 3.9 ( 0.2 × 10-2 min-1 m-2
L determined for nitrobenzene in this study probably
reflects, in part, differences in the properties of the two
organic reactants. However, reactivity per unit surface area
of iron is also likely to be a significant variable because the
Fe0 source and pretreatment were not the same in the two
studies. The specific reaction rate constants described
9
VOL. 30, NO. 1, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 5 7

above offer advantages similar to the half-lives normalized
to 1 m2/ mL that have been reported by Gillham and
O’Hannesin (5). One major difference however is that the
latter were based on single experiments, so they lack the
statistical power that derives from regression on a larger
quantity of data, as was done to obtain the fitted parameters
in eq 9.
Effect of pH. It was anticipated that changes in pH
might cause changes in the nitro reduction rate through (i)
direct involvement ofH+ in the contributing reactions (recall
eqs 6-8), (ii) mass transport limitations imposed by the
precipitation of a passive film on the metal surface, and/ or
(iii) mass transport limitations determined by the thickness
of the Nernst layer between the passive film and the bulk
electrolyte. Previous studies have shown that decreased
pH results in a modest increase in CCl4 dechlorination rate
under conditions that probably were borderline between
activation and diffusion control (4) and in a strong increase
in dechlorination rate where activation control may have
been predominant (9). In this study, the results indicate
no clear effect on k1 over the environmentally relevant pH
range of 6-8 and only a small decrease for pH < 6 (41).
The general lack of a pH effect, however, is consistent with
nitro reduction rates that are limited by mass transport to
reactive sites on a surface where pH-controlled precipitation
reactions are comparatively slow (and ionic strength is
roughly constant).
FIGURE5. Effectofmixingrate onthe pseudo-first-orderrate constant
for nitrobenzene degradation. All bottles contained 33.3 g/L 18-20
mesh acid-washed Fluka iron turnings, pretreated in 1.5 × 10-2
M
bicarbonate buffer (pCO ) 0.01 atm, initial pH ) 7.9) for 2 h, mixed
2
throughout by rotation at 15 °C.
A more dramatic effect of pH was observed on the
distribution of nitro reduction products. Following the
disappearance ofnitrobenzene and the intermediate nitroso
product, the final concentration of aniline provided nearly
complete mass balance at pH > 5. Values of k3, estimated
from the rate of aniline appearance, decreased slightly from
0.015 min-1 at pH 5 to 0.009 min-1 at pH 6.9. However,
below pH 5, no aniline production at all was detected. It
is unlikely that the lack of aniline appearance at pH < 5
indicates a real change in the product distribution since all
of the major transformation pathways produce aniline
(Figure 1). Instead, protonation of the aniline formed (pKa
) 4.6) probably prevents desorption of the product due to
electrostatic attractions that involve the anilinium cation
and specifically adsorbing counterions such as Cl- (40).
Effect of Mixing Rate. The usual method of mixing
employed in this study was 360° rotation at 10 rpm around
a fixed-length axis. Since the rate of rotation was rapid
relative to the rate ofreaction, rpm should be linearly related
to the actual velocity of mixing experienced by the metal
(at least until centrifugal forces become significant).
Numerous empirical determinations have shown that this
velocity is proportional to the square of the mass transfer
coefficient for diffusion across a stagnant boundary layer
(42). Therefore, nitro reduction rate constants should
exhibit a linear relationship with respect to (rpm)0.5, under
conditions where the reaction is mass transport limited.
Values of k1 measured in this study increased linearly with
(rpm)0.5 up to about 45 rpm (Figure 5), and regression on
these data gives
FIGURE6. EffectofextendedincubationofFluka Fe0 withbicarbonate
buffer on pseudo-first-order rate constants for nitrobenzene deg-
radation. Bottles containing 33.3 g/L 18-20 mesh acid-washed Fluka
iron turnings were pretreated variable amounts of time in 1.5 × 10-2
M bicarbonate buffer (pCO ) 1.0 atm, initial pH ) 5.9), with mixing
2
throughout by rotation at 10 rpm and 15 °C.
carbonate speciation. In the presence of Fe0, H2CO3 and
-
HCO3 can be reduced and thereby accelerate corrosion
(17, 18) or precipitate as siderite, which eventually will
inhibit corrosion by limiting mass transfer to the metal
surface. The net impact of these phenomena on the
application of Fe0 to remediation of contaminated ground-
waters can be predicted from results obtained in this study
using CO2-buffered model systems. In early experiments,
it was observed that k1 was notably greater in the presence
of moderate concentrations of carbonate buffer relative to
unbuffered systems at comparable pH. This increase in k1
appears to be due to carbonate-enhanced corrosion in the
absence of significant precipitation. In subsequent experi-
ments, it was discovered that k1 decreased at high total
carbonate concentrations [from 0.035 ( 0.001 min-1 at 1.5
× 10-2 M to 0.029 ( 0.001 at 6.0 × 10-2 M (41)] and with
extended exposure to a particular bicarbonate buffer (Figure
6). Thus, nitro reduction rates decline as circumstances
that favor precipitation of iron carbonates improve. Pre-
cipitates such as siderite that are much less redox active
than the underlying zero-valent metal generally will inhibit
k1 ) (0.015 ( 0.001) × (rpm)0.5 - (0.015 ( 0.002) (10)
where r2 ) 0.998 for n ) 6. The result provides further
evidence for the conclusion that nitro reduction rates are
mass transport limited under the conditions of this study.
Effect of Bicarbonate: In most natural waters, ambient
levels of pCO2 and alkalinity determine the pH and
9
1 5 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 1, 1996

consistent with the kinetics of nitrosobenzene reduction
observed in this study (Figures 2 and 3) and may have
general significance to the application of Fe0 as a reducing
agent in groundwater remediation.
A likely reduction mechanism for the adsorbed NAC
involves the planar donor-acceptor complex as a precursor
to electron transfer (23, 44). The adsorbed molecule must
undergo a series of electron transfers, proton transfers, and
dehydrations to achieve complete reduction, and the order
of these steps has been the subject of many detailed
electrochemical studies. For example, nitro reduction may
be initiated by electron transfer to form an anion radical
(43, 45) or by protonation to form a cation which then
accepts an electron (46). The precise sequence will vary as
a function of pH, solvent properties, and composition of
the surface where adsorption and reaction occur. Schwarz-
enbach et al. (38, 47, 48) have concluded that electron
transfer is initiating and rate limiting in several homoge-
neous model systems based on correlations of nitro
reduction rates to one-electron reduction potentials. In
this study, reduction rates reflect mass transport to the
metal surface, so there is no correlation to the energetics
of electron transfer. Mass transport also influences the
kinetics of further degradation and, therefore, the distribu-
tion of detectable reduction products. For this reason, the
sequence of electron and proton transfer is unlikely to have
a discernible effect on the reaction of NACs by Fe0 under
environmentalconditions where the metalsurface develops
a passivating diffusion boundary of iron oxides or carbon-
ates.
FIGURE 7. Scheme showing competing sequences of substrate
adsorption and reduction at a metal surface. The dotted arrows
represent the transformation process that was observed by analysis
of the solution phase, but reaction took place primarily through
adsorptionandreductionatthe surface (representedbysolidarrows).
contaminant reduction by creating a barrier for mass
transport to active sites on the metal surface. Therefore,
these conditions may prove to be important determinants
of field performance in contaminant remediation.
Mechanism of Nitro Reduction. Although mass trans-
port of NACs to the iron surface is rate-limiting and
determines the observed first-order rate of disappearance,
competition between desorption and further reduction
determines the degree to which intermediate transforma-
tion products are observed in the solution. This relationship
is illustrated in Figure 7 for reduction of nitrobenzene.
Appearance of nitrosobenzene as an intermediate in this
study implies that its desorption is rapid relative to further
reduction to the hydroxylamine (i.e., the step corresponding
to k-NO is rapid in comparison to that corresponding to
Acknowledgments
The authors acknowledge Mark A. Williamson, Atlantic
Geoscience Centre, GeologicalSurvey ofCanada for support
during manuscript preparation. This study was supported
in part by the University Consortium Solvents-In-Ground-
water Research Program, the U.S. EnvironmentalProtection
Agencythrough Cooperative Agreement CR82078-01-0, and
the National Science Foundation through Award BCS-
9212059. This report has not been subject to review by the
U.S. Environmental Protection Agency and therefore does
not necessarilyreflect the views ofthe Agency, and no official
endorsement should be inferred.
k
2S). It is important to note that ArNHOH(ads) must still be
formed even though ArNHOH(aq) was not identified, just as
ArNO(ads) is believed to be an intermediate in electro-
chemical nitro reduction even though ArNO(aq) is rarely
detected (24, 43). An analogous situation probably exists
with dechlorination reactions caused by iron metal. In the
latter case, sequential dechlorination is likely to be the
dominant reduction pathway at the metal surface (4), even
though the distribution of products in solution generally
does not show the appropriate amounts of intermediate
species. The major reason for this is that the distribution
of products observed in the aqueous phase is determined
by the rates at which various species undergo desorption
and transport away from the metal as well as the intrinsic
rates of the reductive dechlorination at the surface.
Literature Cited
(1) O’Hannesin, S. F. M.S. Thesis, University of Waterloo, Ontario,
1993.
(2) Abstr. Pap.sAm. Chem. Soc. 1995, No. 209th, pp 689-835.
(3) Gillham, R. W.; Blowes, D. W.; Ptacek, C. J.; O’Hannesin, S.
Proceedings of the 33rd Hanford Symposium on Health & the
EnvironmentsIn Situ Remediation: Scientific Basis for Current
and Future Technologies; Battelle Pacific Northwest Laborato-
ries: Columbus, OH, 1994; Vol. 2, pp 913-930.
(4) Matheson, L. J.; Tratnyek, P. G. Environ. Sci. Technol. 1994, 28,
2045-2053.
(5) Gillham, R. W.; O’Hannesin, S. F. Ground Water 1994, 32, 958-
967.
(6) Johnson, T. L.; Tratnyek, P. G. Proceedings of the 33rd Hanford
Symposium on Health & the EnvironmentsIn Situ Remediation:
Scientific Basis for Current and Future Technologies; Battelle
Pacific Northwest Laboratories: Columbus, OH, 1994; Vol. 2, pp
931-947.
(7) Schreier, C. G.; Reinhard, M. Chemosphere 1994, 29, 1743-1753.
(8) Lipczynska-Kochany, E.; Harms, S.; Milburn, R.; Sprah, G.;
Nadarajah, N. Chemosphere 1994, 29, 1477-1489.
(9) Warren, K. D.; Arnold, R. G.; Bishop, T. L.; Lindholm, L. C.;
Betterton, E. A. J. Hazard. Mater. 1995, 41, 217-227.
(10) Hartter, D. R. In Toxicity of Nitroaromatic Compounds; Rickert,
D. E., Ed.; Hemisphere: Washington, DC, 1985; pp 1-13.
Aniline cannot be further reduced by iron and, in fact,
is a well-known corrosion inhibitor. The mechanism of
inhibition is believed to be simple interference with mass
transport of oxidant to the metal surface, which in turn is
strongly influenced by the orientation of aniline adsorption
(40). Corrosion inhibition by this mechanism implies that
contaminant reduction rates may also be limited by
competition for reactive sites on the metal. This would be
9
VOL. 30, NO. 1, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 5 9

(11) Kagan, J.; Wang, T. P.; Benight, A. S.; Tuveson, R. W.; Wang, G.-R.
Chemosphere 1990, 20, 453-466.
(12) Leuenberger, C.; Czuczwa, J.; Tremp, J.; Giger, W. Chemosphere
1988, 17, 511-515.
(13) Spanggord, R. J.; Mabey, W. R.; Chou, T. W.; Smith, J. H. In
Toxicity of Nitroaromatic Compounds; Rickert, D. E., Ed.; Hemi-
sphere: Washington, DC, 1985; pp 15-33.
(14) Dickel, O.; Haug, W.; Knackmuss, H.-J. Biodegradation 1993, 4,
187-194.
(15) Bollag, J.-M. Environ. Sci. Technol. 1992, 26, 1876-1881.
(16) Hardy, L. I. M.S. Thesis, University of Waterloo, Ontario, 1994.
(17) Wieckowski, A.;Ghali, E.;Szlarczyk, M.;Sobkowski, J. Electrochim.
Acta 1983, 28, 1619-1626.
(18) Wieckowski, A.;Ghali, E.;Szlarczyk, M.;Sobkowski, J. Electrochim.
Acta 1983, 28, 1627-1633.
(36) Parris, G. E. Environ. Sci. Technol. 1980, 14, 1099-1106.
(37) Haderlein, S. B.; Schwarzenbach, R. P. Environ. Sci. Technol.
1993, 27, 316-326.
(38) Dunnivant, F. M.; Schwarzenbach, R. P.; Macalady, D. L. Environ.
Sci. Technol. 1992, 26, 2133-2141.
(39) Wardman, P. Curr. Top. Radiat. Res. Q. 1977, 11, 347-398.
(40) Banerjee, G.; Malhotra, S. N. Corrosion 1992, 48, 10-15.
(41) Agrawal, A. M.S. Thesis, Oregon Graduate Institute, 1995.
(42) Sherwood, T. K.;Pigford, R. L.;Wilke, C. R. MassTransfer;McGraw
Hill: New York, 1975.
(43) Laviron, E.; Vallat, A.; Meunier-Prest, R. J. Electroanal. Chem.
1994, 379, 427-435.
(44) Kastening, B.; Holleck, L. Electroanal. Chem. Interfacial Elec-
trochem. 1979, 27, 355-368.
(45) Laviron, E.; Meunier-Prest, R.; Lacasse, R. J. Electroanal. Chem.
1994, 375, 263-274.
(46) Zuman, P.; Fijalek, Z.; Dumanovic, D.; Suznjevic, D. Electroanal.
1992, 4, 783-794.
(47) Heijman, C. G.; Grieder, E.; Holliger, C.; Schwarzenbach, R. P.
Environ. Sci. Technol. 1995, 29, 775-783.
(19) Archer, W. L.; Simpson, E. L. Ind. Eng. Chem. Prod. Res. Dev.
1977, 16, 158-162.
(20) Fry, A. J. In The Chemistry of Amino, Nitroso, and Nitro
Compounds and Their Derivatives; Patai, S., Ed.; Wiley-
Interscience: Chichester, 1982; Vol. 1, pp 319-337.
(21) Zuman, P.; Shah, B. Chem. Rev. 1994, 94, 1621-1641.
(22) Lund, H. In Organic Electrochemistry; Baizer, M. M., Ed.; Marcel
Dekker: New York, 1973; pp 315-345.
(48) Schwarzenbach, R. P.; Stierli, R.; Lanz, K.; Zeyer, J. Environ. Sci.
Technol. 1990, 24, 1566-1574.
(49) Encyclopedia ofElectrochemistryofthe Elements;Bard, A. J., Lund,
H., Eds.; Marcel Dekker: New York, 1979; Vol. 13.
(50) Kemula, W.; Krygowski, T. M. In Encyclopedia of Electrochemistry
of the Elements; Bard, A. J., Lund, H., Eds.; Marcel Dekker: New
York, 1979; Vol. 13, pp 131-161.
(51) Meites, L.; Zuman, P. CRC Handbook Series in Organic Elec-
trochemistry; CRC: Cleveland, OH, 1979.
(52) Kemula, W.; Krygowski, T. M. In Encyclopedia of Electrochemistry
of the Elements; Bard, A. J., Lund, H., Eds.; Marcel Dekker: New
York, 1979; Vol. 13, pp 77-130.
(23) Holleck, L.; Kastening, B. Rev. Polarogr. 1963, 11, 129-140.
(24) Shindo, H.; Nishihara, C. J. Electroanal. Chem. 1989, 263, 415-
420.
(25) Martigny, P.; Simonet, J. J. Electroanal. Chem. 1983, 148, 51-60.
(26) Pizzolatti, M. G.; Yunes, R. A. J. Chem. Soc. Perkin Trans. 2 1990,
759-764.
(27) Dalmagro, J.; Yunes, R. A.; Simionatto, E. L. J. Phys. Org. Chem.
1994, 7, 399-402.
(28) Laviron, E.; Vallat, A. Electroanal. Chem. Interfacial Electrochem.
1973, 46, 421-426.
(29) Gibian, M. J.; Baumstark, A. L. J. Org. Chem. 1971, 36, 1389-
1393.
(53) Smyth, M. R.; Osteryoung, J. G. Anal. Chim. Acta 1978, 96, 335-
344.
(30) Bunton, C. A.; Rubin, R. J. Tetrahedron Lett. 1975, 55-58.
(31) Hudlicky´, M. Reductions in Organic Chemistry; Ellis Horwood:
Chichester, 1984.
(32) Lyons, R. E.; Smith, L. T. Chem. Ber. 1927, 60, 173-182.
(33) Kamm, O.; Marvel, C. S. In Organic Syntheses; Gillman, H., Ed.;
Wiley: New York, 1932; Vol. 1, pp 445-447.
(34) Haderlein, S. B.; Schwarzenbach, R. P. In Biodegradation of
Nitroaromatic Compounds; Spain, J., Ed.; Plenum: New York,
1995; pp 199-225.
(54) Tucker, W. A.; Nelken, L. H. In Handbook of Chemical Property
Estimation Methods; Lyman, W. J., Reehl, W. F., Rosenblatt, D.
H., Eds.; McGraw-Hill: New York, 1982; pp 17.1-17.25.
Received for review March 27, 1995. Revised manuscript
received July 31, 1995. Accepted August 7, 1995.X
ES950211H
(35) Weber, E. J.; Wolfe, N. L. Environ. Toxicol. Chem. 1987, 6, 911-
919.
X Abstract published in Advance ACS Abstracts, November 1, 1995.
9
1 6 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 1, 1996