Photoeffects on the reduction of carbon tetrachloride by zero-valent iron

Article abstract of DOI:10.1021/jp973113m

While reduction of chlorinated hydrocarbons by zero-valent iron in water is strongly influenced by the oxide layer at the metal-water interface, the role of the oxide in the dechlorination mechanism has not been fully characterized. In this paper, we investigate the semiconducting properties of the oxide layer on granular iron and show how the electronic properties of the oxide affect electron transfer to aqueous CCl₄. Specifically, we determine whether conduction-band electrons contribute to the reduction of CCl₄ by using light to increase the number of conduction-band electrons at the oxide surface and measuring how this treatment affects the rate and products of CCl₄ degradation. We find that photogenerated conduction-band electrons do degrade CCl₄ and, more importantly, shift the product distribution to more completely dechlorinated products that are indicative of two-electron transfer with a dichlorocarbene intermediate. Since the photogenerated electrons give different reduction products than the dark reducers, we conclude that the latter must not be conduction-band electrons. Further investigation of the reduction with photogenerated electrons is carried out by adding hole scavengers to the system. Isopropyl alcohol reacts with photogenerated holes to yield the ?±-hydroxyalkyl radical, which is known to reduce CCl₄. With isopropyl alcohol present, we observe faster degradation of CCl₄ with higher light intensity. Since no such increase is seen without isopropyl alcohol, the rate of CCl₄ degradation by conduction-band electrons in water must not be limited by the number of photogenerated electron-hole pairs but rather by electron transfer from the oxide conduction band to CCl₄.

Full text of DOI:10.1021/jp973113m

J. Phys. Chem. B 1998, 102, 1459-1465
1459
Photoeffects on the Reduction of Carbon Tetrachloride by Zero-Valent Iron
Barbara A. Balko*
Department of Chemistry, Lewis & Clark College, 0615 SW Palatine Hill Road, Portland, Oregon 97219
Paul G. Tratnyek
Department of EnVironmental Science and Engineering, Oregon Graduate Institute of Science and Technology,
P.O. Box 91000, Portland, Oregon 97291-1000
ReceiVed: September 25, 1997
While reduction of chlorinated hydrocarbons by zero-valent iron in water is strongly influenced by the oxide
layer at the metal-water interface, the role of the oxide in the dechlorination mechanism has not been fully
characterized. In this paper, we investigate the semiconducting properties of the oxide layer on granular iron
and show how the electronic properties of the oxide affect electron transfer to aqueous CCl4. Specifically,
we determine whether conduction-band electrons contribute to the reduction of CCl4 by using light to increase
the number of conduction-band electrons at the oxide surface and measuring how this treatment affects the
rate and products of CCl4 degradation. We find that photogenerated conduction-band electrons do degrade
CCl4 and, more importantly, shift the product distribution to more completely dechlorinated products that are
indicative of two-electron transfer with a dichlorocarbene intermediate. Since the photogenerated electrons
give different reduction products than the dark reducers, we conclude that the latter must not be conduction-
band electrons. Further investigation of the reduction with photogenerated electrons is carried out by adding
hole scavengers to the system. Isopropyl alcohol reacts with photogenerated holes to yield the R-hydroxyalkyl
radical, which is known to reduce CCl4. With isopropyl alcohol present, we observe faster degradation of
CCl4 with higher light intensity. Since no such increase is seen without isopropyl alcohol, the rate of CCl4
degradation by conduction-band electrons in water must not be limited by the number of photogenerated
electron-hole pairs but rather by electron transfer from the oxide conduction band to CCl4.
‚CCl3 + e- + H+ f HCCl3
(3)
Introduction
Zero-valent iron is currently being used to remediate dilute,
aqueous, halogenated hydrocarbon contaminants in ground
water.1 This process is a large-scale, environmental application
of a dissolving metal reduction,2,3 where the oxidation of Fe0
which may undergo further hydrogenolysis to give dichlo-
romethane. The rate of chloroform reduction, however, is much
less than the rate of hydrogenolysis of CCl4.7,11 In reductive
elimination, the ‚CCl3 undergoes a second dissociative electron
transfer to form dichlorocarbene
(E0 ) -0.44 V4)
red
Fe0 h Fe2+ + 2e-
(1)
‚CCl3 + e- f :CCl2 + Cl-
(4)
drives the reduction of halogenated methanes, ethanes, ethenes,
and propanes. In this paper, we investigate the role that the
semiconducting properties of iron oxides play in these reme-
diation reactions, specifically degradation of CCl4. CCl4 was
chosen as our probe compound because it is the most widely
used model compound for studying degradation by zero-valent
iron, and detailed studies have been carried out on photoreduc-
tion of CCl4 using a colloidal semiconductor (TiO2).5,6
There are two main reaction pathways for the anaerobic
reduction of CCl4: hydrogenolysis (one-electron transfer) and
reductive elimination (two-electron transfer).7-9 Both of these
pathways are initiated by dissociative electron transfer10
which is hydrolyzed to give HCl and CO.12,13
:CCl2 + H2O f CO + 2HCl
(5)
The net two-electron transfer to form :CCl2 from CCl4, i.e.,
reductive elimination
CCl4 + 2e- f :CCl2 + 2Cl-
(6)
is thermodynamically favored over one-electron reduction
(reaction 2) and is spontaneous as long as the electron donor
has an E0′ < ∼0.2 V.9,14 From a remediation standpoint,
reductive elimination is preferred over hydrogenolysis because
the final products in the former pathway are more environmen-
tally benign.
CCl4 + e- f ‚CCl3 + Cl
(2)
The immediate source of the electrons that are transferred to
CCl4 is difficult to determine due to the complex chemistry of
the interfacial region between iron and water. In principle,
reduction might occur by electron transfer directly from Fe0 or
to give trichloromethyl radical and chloride ion. In hydro-
genolysis, the ‚CCl3 is reduced to chloroform
* Corresponding author. Email: balko@lclark.edu.
S1089-5647(97)03113-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 02/03/1998

1460 J. Phys. Chem. B, Vol. 102, No. 8, 1998
Balko and Tratnyek
from either of the products formed during anaerobic corrosion
of the metal (Fe2+ or H2).7
Fe0 + H2O f Fe2+ + H2 + 2OH-
(7)
Electron transfer from Fe0 to the aqueous CCl4 may occur
directly from the metal or indirectly through the oxide film.
Direct electron transfer between the metal and the organic could
occur by tunneling (direct or resonance)15-19 or through defects
such as pits or pinholes that allow the solution to directly contact
the iron.15,20,21 For indirect electron transfer through the oxide
film to be significant, there must be a facile mechanism for
transporting electrons across the oxide layer, presumably through
the conduction band. Exploring the role of conduction-band
electrons in reduction of organic adsorbants is the primary
objective of this study.
The mechanism of electron transport through the oxide layer
(or passive film) will depend, in part, on the composition of
this material. There is general agreement that the typical passive
film on Fe0 is composed of Fe3O4 (magnetite), which is a mixed-
valent, highly conductive oxide,22-26 and γ-Fe2O3 (maghemite),
which is an n-type semiconductor.22,23,27-31 The arrangement
of these oxides in the passive film is undoubtedly variable but
is thought to conform to one of two models.30 In one model,
the film has an inner layer of Fe3O4 and an outer layer of
γ-Fe2O3.23,27-30 Alternatively, the film may have a continuous
Fe2+ concentration gradient, with the concentration equal to that
found in Fe3O4 at the oxide-iron interface and decreasing to
that of γ-Fe2O3 at the oxide-solution interface.22,30,31
Raman studies suggest that the air-formed oxide film present
on the granular iron used in our studies is similar in composition
to the passive film.32 These studies show that the outer layer
of the air-formed film is primarily comprised of γ-Fe2O3 with
some R-Fe2O3 present (the percentage depends on the original
composition and subsequent handling of the iron particles) while
the inner layer is Fe3O4. Exposure of the iron particles for
several weeks to ground water contaminated with chlorinated
solvents results in disappearance of the Fe2O3, presumably due
to autoreduction of the Fe(III) oxide to Fe3O4 and formation of
green rust.32 However, since our experiments were completed
within 3 h, our films presumably will retain the composition of
the original air-formed oxide.
Figure 1. Band diagram of the iron/oxide/aqueous CCl4 system at pH
7, constructed using electron energies versus the normal hydrogen
electrode (NHE) from the literature. The diagram shows (i) The Fermi
energies, EF, for Fe0 (calculated from the work function42) and the oxide
layer (literature values cluster between -0.1835 and -0.33 V;34 the
average, -0.25 V, is used in the diagram); (ii) the lower edge of the
oxide conduction band, ECB, (EF - ECB ) ∼0.06 eV, assuming an
effective density of energy levels at the conduction band edge of ∼1019
cm-3 41 and a doping density of 1020 cm-3 34,36,39
(iii) the upper edge
;
of the oxide valence band, EVB, calculated from the value of ECB and
assuming a band gap of 1.9 eV;34,36,38 (iv) the E0’s for relevant redox
couples: CCl4 + e- f ‚CCl3 + Cl-,14 ‚CCl3 + H+ + e- f HCCl3,14
CCl4 + H+ + 2e- f HCCl3 + Cl-,14 CCl4 + H+ + 2e- f CHCl3 +
Cl-,40 and ‚CCl3 + e- f :CCl2 + Cl- (calculated using the method
outlined in ref 45 with thermodynamic values from refs 43 and 44)
and assuming [Cl-] ) 10-3 M).
from a semiconductor to the electrolyte is greater if there is
good overlap between the conduction-band edge and the
unoccupied electron energy levels of the redox couple.15,46 Third,
because EF of the oxide is more negative than E0(CCl4/‚CCl3,
Cl-), the oxide film may be in depletion; i.e., there may be a
space charge region in the oxide film that acts as a barrier to
electron transport from the conduction band to the solution. It
should be noted that this barrier arises as equilibrium is
established between the semiconductor and the redox couple
in solution, and that electron transfer occurs from the semicon-
ductor to solution until E0(oxide) ) EF(redox).41 This barrier
may not be fully established, however, because the rate at which
equilibration occurs depends on the rate of charge transfer
between the oxide and the redox couple,41 and this is likely to
be slow for the iron/oxide/aqueous CCl4 system.11,47
The details of the electronic structure of the passive film differ
from that of the single-crystal semiconductor portrayed in Figure
1, primarily because passive films lack long-range order.
Passive films are typically only a few monolayers thick, so long-
range order normal to the surface cannot develop. Thicker films
tend to be amorphous because a more coherent film would be
strained and break up.15,17,18,38,48 While the electronic structure
will vary with film thickness and the conditions under which
the film is grown (potential, electrolyte, etc.), several generaliza-
tions can be made regarding the air-formed oxide film involved
in this study. First, while the film will have bands equivalent
to the conduction and valence bands found in single-crystal
semiconductors, additional bands, which result from a large
density of impurities or ions in a different oxidation state, may
also be found.15,22 In addition, there will be a much greater
density of localized states than found in single-crystal semi-
Under illumination, the typical passive film behaves like an
n-type semiconductor with a band gap of ∼1.9 eV, a flatband
potential of -0.3-0.0 V (vs NHE at pH 7), and a doping density
of ∼1020 cm-3 33-36
.
The passive film on Fe0 has been
successfully modeled as semiconducting Fe2O3 that is highly
doped (5 × 1020 cm-3) with Fe2+ 37
.
(However, if the passive
film has a continuous Fe2+ concentration gradient, the doping
density cannot be uniform through the thickness of the film.35)
Electron energy levels for the iron/oxide/aqueous CCl4 system
compiled from previously published values14,34-36,38-45 are
summarized in Figure 1. The energy levels of the oxide are
from photoelectrochemical experiments in which the film was
treated as a one-component, single-crystal semiconductor so the
oxide layer is drawn as such.
Figure 1 reveals several constraints on electron transfer
through the oxide film. First, the Fermi energy (EF) of the iron
oxide is more negative than that of the iron metal, so once
equilibrium is established there will be a potential barrier for
injecting electrons from the iron metal to the conduction band
of the iron oxide.42 Second, E0(CCl4/‚CCl3, Cl-) is more positive
than the conduction-band energy, which means that the unoc-
cupied electron energy levels of the redox couple can overlap
with the conduction band. The likelihood of electron transfer

Reduction of CCl4 by Zero-Valent Fe
J. Phys. Chem. B, Vol. 102, No. 8, 1998 1461
conductors. These localized states result from the lack of
identical lattice sites throughout the film.15,17,18,22,38
Electron transfer between the oxide layer and adsorbed
species can occur via the oxide conduction band, impurity bands,
or localized states.15,35,39,49 For example, several studies have
shown that electrons are transferred from the conduction band
3- 50,51
of the passive film on iron to Fe(CN)6
.
Yet Fe2+ provides
localized states, which can facilitate resonance tunneling through
iron oxide films.52 A specific goal of this study was to
determine if conduction-band electrons are responsible for the
reduction of CCl4 by iron metal. The approach taken was to
manipulate the number of conduction-band electrons that reach
the surface of the oxide film by the addition of light.
There are two main mechanisms whereby light can increase
the number of conduction-band electrons at the oxide-water
interface. First, assuming that the oxide film is in depletion,
photogenerated electrons and holes cause the bands to flatten,
thereby reducing the barrier for tunneling of conduction-band
electrons through the space charge region to the oxide sur-
face.41,53 The second effect that the light may have is generation
of conduction-band electrons in the oxide layer. If the contact
potential barrier between the iron metal and the oxide layer is
large enough to impede electron injection from the iron metal
into the oxide layer, then there will be a greater number of
conduction-band electrons available at the oxide surface.
In the experiments described below, batch studies were done
with CCl4 and zero-valent iron in the presence and absence of
light with energy greater than the oxide layer band gap. The
results show that while photogenerated conduction-band elec-
trons are formed on iron particles and contribute to the
degradation of CCl4, conduction-band electrons are not respon-
sible for the dechlorination of CCl4 by Fe0 in the dark.
Figure 2. Typical kinetics of CCl4 degradation by Fe0 with and without
the addition of light (AM0). Squares and circles are for experiments
run in the dark and light, respectively. The lines show the best fit to a
first-order kinetics model.
sphere. We estimated the power supplied to be 720 W/m2 in
the 280-700-nm range and 170 W/m2 in the 320-450-nm
range, based on data supplied by Oriel. The projected area of
the iron particles in the vials was ∼3 cm2. In some experiments,
the intensity of the solar simulator was decreased by inserting
an Air Mass 1 (AM1) filter (Oriel #81074), which gives an
output equivalent to the solar spectrum at ground level. With
this filter installed, we estimated the output power to be 575
W/m2 in the 280-700-nm range and 115 W/m2 in the 320-
450-nm range.
In all experiments, we included dark controls by wrapping
vials in aluminum foil and adding them to the rotator. The
degradation rates obtained in the light (kL+D) represent the sum
of the rate of CCl4 degradation in the dark (kD) and CCl4
degradation due to the addition of light (kL). To quantify the
rate of degradation due to the light alone, the rate constant for
dark degradation was subtracted from that measured in the light
(i.e., kL ) kL+D - kD). Direct homogeneous photolysis should
be negligible since CCl4 has very little absorbance at wave-
lengths greater than 270 nm,55 the borosilicate glass vials used
transmit little light below 290 nm,56 and the solar simulator puts
out only a small percentage of its power at wavelengths below
320 nm (0.8% for AM0 and 0.2% for AM1). However, because
the effect of light on CCl4 degradation was also small, we felt
it was important to quantify the rates of homogeneous CCl4
photolysis due to the addition of light (kno Fe,L) as well as CCl4
loss in the dark in the absence of iron (kno Fe,D), which is likely
due to processes such as adsorption on the septa, volatilization,
and hydrolysis. We found kno Fe,D ) 0.07 ( 0.01 h-1 and kno
Experimental Section
In the typical batch experiment, vials were filled under an
N2/H2 atmosphere with 12 mL of deoxygenated/deionized water,
1.0 g of 20-32 mesh Fe0 turnings (Fluka, puriss. grade, specific
surface area ) 0.019 m2 g-1), and CCl4 at an initial concentra-
tion of 85 µM. The vials were crimp-sealed without headspace
and rotated at 30 rpm. Periodically, vials were sacrificed and
aqueous samples extracted with hexane. The extract was then
analyzed by gas chromatography with electron-capture detection.
In experiments where 0.1 M isopropyl alcohol was used, the
concentration of acetone formed was monitored by derivatization
with 2,4-dinitrophenylhydrazine and analysis of the product by
high-performance liquid chromatography.54 First-order kinetics
for the disappearance of CCl4 was observed, as expected, and
rate constants were obtained by linear regression of ln[CCl4]
versus time data. Reported uncertainties are 1 standard deviation
about the regression line.11
To create conduction-band electrons, it was necessary to
supply light with energy greater than the band gap of the oxide
film. Photocurrent has been detected from an iron passive film
in borate buffer with light up to ∼730 nm, with the peak
photocurrent occurring at ∼400 nm.38 The wavelength maxi-
mum, however, undoubtedly depends on the conditions under
which the film was grown. To ensure that the photons supplied
have sufficient energy, we chose to use a 1000-W solar simulator
(Oriel #81193), which emits light from ∼250 to ∼2500 nm.
Since the solar simulator has a 6 × 6 in. beam, this had the
added advantage of allowing us to illuminate all vials simul-
taneously. For most of the experiments, we used an Air Mass
0 (AM0) filter (Oriel #81011) which corrects the xenon arc lamp
output to match the solar spectrum outside the earth’s atmo-
) 0.07 ( 0.02 h-1
. These were subtracted from the
Fe,L
measured rates of CCl4 degradation in the presence of iron
(kmeas,D and kmeas,L+D) for all the kinetic data reported below
(i.e., kD ) kmeas,D - kno Fe,D and kL+D ) kmeas,L+D - kno Fe,L
no Fe,D).
-
k
Results and Discussion
Effect of Light on the Rate of CCl4 Degradation. In Water.
The addition of light to batch systems of iron and water
consistently increased the rate of CCl4 degradation; this increase
in rate shows that conduction-band electrons do indeed reduce
aqueous CCl4. Figure 2 shows typical kinetics for the degrada-
tion of CCl4 with and without the addition of light. In this

1462 J. Phys. Chem. B, Vol. 102, No. 8, 1998
Balko and Tratnyek
Figure 3. Effect that the addition of 0.1 M tert-butyl alcohol has on
CCl4 degradation by Fe0 in the light (natural sunlight). Squares are for
0.1 M tert-butyl alcohol and circles are for deionized water. The lines
show the best fit to a first-order kinetics model.
example, kL ) 0.20 ( 0.05 h-1 and kD ) 0.68 ( 0.01 h-1, so
the degradation rate due to photogenerated electrons is ap-
proximately one-third of the rate observed in the dark (based
on kL/kD ) 0.33 ( 0.1). The photoeffect, the increase in the
rate of CCl4 degradation due to the addition of light to the
experiment, may be small because degradation is slower when
photogenerated electrons are responsible for reduction or
because fewer photogenerated electrons are available than dark
reducers. (By dark reducers we mean the species responsible
for reduction in the dark, whether it be surface Fe(II), electrons
that are transferred directly from the iron via tunneling or
defects, or electrons that are transported through the conduction
band of the oxide film.) It is not known, however, how many
conduction-band electrons are produced for each photon sup-
plied, or the number of active dark reducers, so it is not yet
possible to compare the rates for light and dark degradation on
a microscopic level.
Figure 4. Effect of light on major products of CCl4 degradation by
Fe0. Concentrations of CCl4 (circles/solid lines), HCCl3 (squares/dashed
lines), and Cl- (triangles/dotted lines) were measured in the dark (D,
filled symbols) and in the light (L+D, open symbols). The lines connect
average concentrations.
degradation by photogenerated electrons follows a different
reaction pathway than that involving the dark reducers. As
shown in Figure 4, the addition of light did not increase the
amount of chloroform formed, even though the rate of CCl4
degradation and Cl- accumulation both increase. Reduction
of HCCl3 by iron is much slower than reduction of CCl4 in the
dark (kSA ≈ 10-3 L m-2 h-1 for HCCl3 versus ≈10-1 L m-2
11
h-1 for CCl4 where kSA is the first-order degradation rate
constant adjusted for surface area) as well as in the light (data
not shown). Therefore, it is likely that the more complete
dechlorination observed in the presence of light is due to two-
electron transfer to give dichlorocarbene, which then undergoes
hydrolysis to give HCl and CO (according to reaction 5), rather
than further reduction of chloroform. The change in product
distribution that occurs with the addition of light cannot be due
to direct, homogeneous photolysis for two reasons. First, direct
photolysis of CCl4 in aqueous solution is quite slow, as
evidenced by the fact that there is no significant production of
Cl- during control experiments without iron. Second, the
homogeneous photolysis pathway most likely involves only the
production of ‚CCl3, which would lead to HCCl3.
In 0.1 M tert-Butyl Alcohol. Although we expected that the
photoeffect is due to production of conduction-band electrons,
it is also possible that the photoeffect is due to formation of
surface Fe(II) sites via ligand-to-metal charge transfer (LMCT).
Light may drive LMCT between adsorbed hydroxide ions and
excited surface Fe(III) lattice atoms57,58
tFe(III)-OH- + hν f tFe(III)*-OH- f
tFe(II) + ‚OH (8)
Previous reports have shown that the two-electron-transfer
pathway is favored when the potential of the reducer is made
more negative. For example, Pt deposits, which are known to
shift the Fermi energy level in TiO2 to more negative poten-
tials,59 increase the amount of two-electron-transfer relative to
one-electron transfer in the photoreduction of halothane by
n-TiO2.60 While the second electron transfer step is thermo-
dynamically more favorable than the first step, the Pt deposits
are necessary to ensure efficient two-electron transfer as there
are many other species present that compete with the photoge-
nerated conduction-band electrons for the intermediate radical,
‚CHClCF3. Further evidence that a more negative potential
favors two electron transfer can be found in the photodegradation
of CCl4 using n-TiO2 particles, where high pH was found to
strongly favor two-electron transfer via dichlorocarbene. While
part of the explanation for this pH effect is that hydrolysis of
Studies on the photochemistry of aqueous Fe(III), however,
show that the yield of aqueous Fe(II) due to the LMCT
mechanism is quite small in the absence of ‚OH scavengers
[quantum yield ∼10-4 58]. It is unlikely, then, that LMCT is
responsible for our photoeffect. Moreover, addition of the ‚OH
scavenger tert-butyl alcohol, which should increase the yield
of Fe(II) via the LMCT mechanism,58 did not lead to a
significant increase in the rate of CCl4 degradation in the
presence of light (Figure 3). Thus, we conclude that the increase
in degradation rate that occurs with the addition of light is not
due to the formation of Fe(II) sites via LMCT but is due to the
creation of conduction-band electrons.
Effect of Light on the Distribution of CCl4 Degradation
Products. Comparison of the product distributions obtained
from CCl4 degradation in the light and the dark suggests that

Reduction of CCl4 by Zero-Valent Fe
J. Phys. Chem. B, Vol. 102, No. 8, 1998 1463
TABLE 1: Light Intensity Effects on CCl4 Degradation in
Water
kL/h-1
kD/h-1
kL/kD
AM1
AM0
0.12 ( 0.09
0.20 ( 0.05
0.31 ( 0.04
0.61 ( 0.02
0.38 ( 0.3
0.33 ( 0.1
a All rate constants have been corrected for degradation in the absence
of iron. In correcting kL(AM1), we assumed that the homogeneous
photolysis rate constant (kno Fe,L) is the same as with the AM0 filter,
making kL and kL/kD upper estimates.
However, results described in the following section indicate that
increasing the light intensity in the presence of hole scavengers
increases the number of holes produced so it must also increase
the number of conduction-band electrons. This leaves us with
the conclusion that reaction of conduction-band electrons with
CCl4 is a minor pathway and therefore negligibly effected by
modest changes in light intensity.
In 0.1 M Isopropyl Alcohol. Isopropyl alcohol was used to
investigate the weak dependence of the CCl4 reduction rate on
light intensity in water. Isopropyl alcohol reacts with photo-
generated holes or ‚OH (from h+ + OH-) to generate the
R-hydroxyalkyl radical, (CH3)2(‚C)OH, which can reduce CCl4
to ‚CCl3.5,6,64,65 The ‚CCl3 can then abstract H atoms from the
alcohol to give HCCl3.
Figure 5. Typical kinetics showing the effect of lower intensity light
(AM1). The squares and circles are data from experiments run in the
dark and light, respectively. The lines show the best fit to a first-order
kinetics model.
:CCl2 to form CO and HCl is base-catalyzed, the Nernstian shift
in the conduction band electron reduction potential that occurs
with higher pH is also important because it favors electron
transfer.5,6
h+/‚OH + (CH3)2CHOH f (CH3)2(‚C)OH + H+/H2O (9)
The change in distribution of products that occurs with the
addition of light implies that the degradation mechanisms
involving photogenerated electrons and dark reducers are
different. Thus, the dark reducers cannot be conduction-band
electrons but must be either electrons transferred directly from
the iron or surface Fe(II). In principle, comparison of the
reduction potential of the two possible dark reducers with the
reduction potential of conduction-band electrons (-0.3 V)
allows us to determine which is likely to be more important
(recall Figure 1). Electrons transferred directly from the iron
are a good candidate for the dark reducer because they are at
the Fermi energy of the metal, which has a more positive
reduction potential than the photogenerated electrons. The
reduction potential for aqueous Fe2+/Fe3+ is 0.77 V,61 but the
reduction potential of Fe(II) complexes are considerably more
negative: for example, E0 at pH 7 for Fe2+/R-Fe2O3(s) is -0.28
V and for Fe2SiO4(s)/Fe3O4(s) is -0.43 V.62 Until the reduction
potential of Fe(II) sites in the oxide can be determined, the role
of surface Fe(II) sites in the dark reaction cannot be fully
resolved from this study.
Effect of Light Intensity on CCl4 Degradation. In Water.
Since conduction-band electrons appear to be responsible for
reduction of CCl4 in the light, we expected that the rate of
degradation would vary with the irradiation intensity. Figure
5 shows the kinetics of CCl4 degradation at the low light
intensity (AM1). Comparison of rate constants obtained from
this experiment with those obtained under high light intensity
(AM0), shows that light intensity does not have a significant
effect on the ratio kL/kD (Table 1). One possible explanation
for the apparent lack of an intensity effect is that the reaction
of conduction-band electrons with CCl4 is slow relative to other
reactions involving these reducers. For example, the photoge-
nerated conduction-band electrons may react predominantly with
trapped holes, reduce Fe(III) lattice sites,63 or reduce H2O to
H2. It is also possible that the effect of light intensity is not
detectable because the range of light intensity used is too narrow;
e.g., production of electron-hole pairs may be saturated at low
light intensities because the oxide film is not a strong absorber.
(CH3)2(‚C)OH + CCl4 f (CH3)2CO + H+ + Cl- + ‚CCl3
(10)
‚CCl3 + (CH3)2CHOH f HCCl3 + (CH3)2(‚C)OH (11)
The product distribution we observed confirms that reduction
of CCl4 via (CH3)2(‚C)OH occurs with the addition of isopropyl
alcohol. Figure 6 shows CCl4, HCCl3, and Cl- concentrations
versus time in the light (AM0) and the dark with isopropyl
alcohol present. The observed effect of light on the product
distribution differs from that without alcohol (Figure 4) in that
there is significantly more HCCl3 formed as a result of H atom
abstraction by ‚CCl3. Acetone was also detected, and Figure 7
shows the concentration of acetone and HCCl3 produced due
solely to the addition of light and isopropyl alcohol. Throughout
most of the experiment, the amounts of (CH3)2CO and HCCl3
produced are approximately equal, as expected if eqs 9-11
apply. Toward the end of the experiment, the concentrations
of the two species begin to differ, which may be due to the
lower CCl4 concentration or changes in the oxide layer.
Light intensity has a significant effect on the CCl4 degradation
rate in the presence of isopropyl alcohol. Figure 8 and Table
2 show that this is entirely due to changes in kL. The effect
can be explained by the production of more h+/‚OH pairs with
higher intensity light, which leads to a greater number of (CH3)2-
(‚C)OH radicals. That we generate more h+/‚OH with the
additional light implies that more conduction-band electrons are
produced: i.e., we did not saturate the number of electron-
hole pairs. It should be noted that the hole-scavenging ability
of isopropyl alcohol5,58 is not responsible for the increase in
CCl4 degradation rate since a depletion region likely exists in
the oxide film, which already serves to separate the photoge-
nerated electron-hole pairs. If hole scavenging were critical,
addition of tert-butyl alcohol would have increased the CCl4
degradation rate. Thus, since light intensity had no significant
effect on the CCl4 degradation rate in the absence of isopropyl

1464 J. Phys. Chem. B, Vol. 102, No. 8, 1998
Balko and Tratnyek
Figure 8. Effect that light intensity has on the CCl4 degradation rate
in iron/0.1 M isopropyl alcohol/CCl4 systems. In both (a) and (b), the
filled circles are data from experiments run in the dark. The open circles
show data taken from the experiment run in the light. In (a), AM1
light intensity was used while in (b), AM0 was used. Note that the
same axis scale applies to (a) and (b). The solid lines show the best fit
assuming first-order kinetics.
Figure 6. Typical CCl4 degradation and formation of HCCl3 and Cl-
products showing the effect of light in a batch experiment with the
iron/0.1 M isopropyl alcohol/CCl4 system. The concentrations of CCl4
(circles/solid lines), HCCl3 (squares/dashed lines), and Cl- (triangles/
dotted lines) were measured in the dark (D, filled symbols) and in the
light (L+D, open symbols). The lines connect average concentrations.
TABLE 2: Light Intensity Effects on CCl4 Degradation in
0.1 M Isopropyl Alcohol
kL/h-1
kD/h-1
kL/kD
AM1
AM0
0.18 ( 0.06
0.64 ( 0.09
0.47 ( 0.03
0.52 ( 0.02
0.38 ( 0.14
1.3 ( 0.3
a All rate constants have been corrected for degradation in the absence
of iron. In correcting kL(AM1), we assumed that the homogeneous
photolysis rate constant (kno Fe,L) is the same as with the AM0 filter,
making kL and kL/kD upper estimates.
tunneling or defects such as pits) and reduction by Fe(II) sites
on the oxide surface.
If electron transport across the oxide film via tunneling is
how electrons are transferred from the iron to CCl4, the
degradation reaction might be enhanced by increasing conduc-
tivity across the layer. The addition of resonance sites into the
oxide film by ion-implantation,66 increasing the Fe2+ bulk
concentration by passivation of the iron in the light after removal
of the air-formed layer,33,52 or the use of iron alloys may increase
the probability of resonance tunneling and, as a result, the rate
of electron transfer. Increasing the number of defects in the
film, for example, by inducing pitting, may allow direct contact
between the iron metal and solution, leading to an increase in
the electron-transfer rate. Alternatively, if Fe(II) sites on the
oxide surface are responsible for the reduction of CCl4 in the
dark,67,68 increasing their concentration by adsorption of Fe2+
or electron injection by highly reducing species should increase
the rate of CCl4 degradation.
Figure 7. Concentrations of (CH3)2CO (triangles/solid line) and HCCl3
(squares/dashed line) produced solely due to the photogenerated
electron-hole pairs versus time in batch experiments run in 0.1 M
isopropyl alcohol. These concentrations ([ ]L) were calculated by
subtracting the concentrations measured in the dark([ ]D) from those
measured in the light ([ ]L+D). The lines connect average concentrations.
alcohol, reduction of CCl4 by conduction-band electrons must
become “saturated” at high light intensities.
Implications for Environmental Applications
CCl4 reduction by photogenerated conduction-band electrons
leads to greater dechlorination, presumably because the conduc-
tion-band electrons have a more negative reduction potential
than the dark reducers. Since HCCl3 is an undesirable product
in remediation applications, the results of this work suggest that
formation of the unwanted product might be minimized by
designing the system to favor two-electron transfer: e.g., by
lowering the reduction potential of the dark reducer.
The results of this work narrow the possible mechanisms for
how electron transfer occurs in the reduction of CCl4 by zero-
valent iron in the dark. The different product distributions that
we observe for CCl4 degradation in the light and dark suggest
that dark reduction does not involve transport of electrons from
the iron metal to the CCl4 via the oxide conduction band.
Alternative mechanisms that deserve further investigation
include direct electron transfer from the metal to CCl4 (via

Reduction of CCl4 by Zero-Valent Fe
J. Phys. Chem. B, Vol. 102, No. 8, 1998 1465
(29) Ryan, M. P.; Newman, R. C.; Thompson, G. E. J. Electrochem.
Soc. 1995, 142, L177.
The addition of isopropyl alcohol dramatically increased the
rate of reduction of CCl4 by zero-valent iron in the light. This
is due to the generation of the R-hydroxyalkyl radical (CH3)2-
(‚C)OH, which reduced the CCl4 to ‚CCl3. Under these
conditions, HCCl3 was a significant product because of H atom
abstraction from (CH3)2CHOH. This result suggests that the
presence of H atom donors (including dissolved organic matter
or organic buffers) may shift product distribution toward HCCl3,
whenever ‚CCl3 is an intermediate. In this way, secondary
components (in laboratory solutions as well as ground water)
may affect product distributions to a greater degree than is
generally recognized.
(30) Oblonsky, L. J.; Davenport, A. J.; Ryan, M. P.; Isaacs, H. S.;
Newman, R. C. J. Electrochem. Soc. 1997, 144, 2398.
(31) Cahan, B. D.; Chen, C.-T. J. Electrochem. Soc. 1982, 129, 921.
(32) Odziemkowski, M. S.; Gillham, R. W. DiVision of EnVironmental
Chemistry, Extended Abstracts; 213th American Chemical Society National
Meeting, San Francisco, CA, 1997, Vol. 37, p 177.
(33) Schmuki, P.; Bo¨hni, H. Electrochim. Acta 1995, 40, 775.
(34) Abrantes, L. M.; Peter, L. M. J. Electroanal. Chem. 1983, 150,
593.
(35) Stimming, U.; Schultze, J. W. Ber. Bunsen-Ges. Phys. Chem. 1976,
80, 1297.
(36) Wilhelm, S. M.; Yun, K. S.; Ballenger, L. W.; Hackerman, N. J.
Electrochem. Soc. 1979, 126, 419.
(37) Schmuki, P.; Bu¨chler, M.; Virtanen, S.; Bo¨hni, H.; Muller, R.;
Gauckler, L. J. J. Electrochem. Soc. 1995, 142, 3336.
(38) Searson, P. C.; Latanision, R. M.; Stimming, U. J. Electrochem.
Soc. 1988, 135, 1358.
Acknowledgment. M. M. Scherer made many helpful
contributions to the experimental design and data analysis. D.
Romanowski contributed preliminary data as a summer student
in the Apprenticeships in Science and Engineering Program. J.
Higa and D. Severson made helpful contributions to final
experiments as part of the Lewis & Clark College Summer
Science Research Program, with support from the Murdock
Trust. Acknowledgment is made to the donors of the Petroleum
Research Fund, administered by the ACS, for the primary
support of this research (29995-AC5).
(39) Wilhelm, S. M.; Hackerman, N. J. Electrochem. Soc. 1981, 128,
1668.
(40) Kolthoff, I. M.; Lee, T. S.; Stocesova, D.; Parry, E. P. Anal. Chem.
1950, 22, 521.
(41) Finklea, H. O. In Semiconductor Electrodes; Finklea, H. O., Ed;
Elsevier: New York, 1988; pp 1.
(42) Hummel, R. E. Electronic Properties of Materials, 2nd ed.;
Springer-Verlag: New York, 1993.
(43) Benson, S. W. Thermochemical Kinetics: Methods for the Estima-
tion of Thermochemical Data and Rate Parameters, 2nd ed.; Wiley: New
York, 1976.
(44) Woodbury, G. Physical Chemistry; Brooks/Cole: Pacific Grove,
CA, 1997.
(45) Vogel, T. M.; Criddle, C. S.; McCarty, P. L. EnViron. Sci. Technol.
1987, 21, 722.
(46) Gerischer, H. In Physical Chemistry: An AdVanced Treatise;
Eyring, H., Ed; Academic: New York, 1970; Vol. IXA, p 463.
(47) Scherer, M. M.; Westall, J. C.; Ziomek-Moroz, M.; Tratnyek, P.
G. EnViron. Sci. Technol. 1997, 31, 2385.
(48) Stimming, U. Electrochim. Acta 1986, 31, 415.
(49) Schmickler, W. Ber. Bunsen-Ges. Phys. Chem. 1978, 82, 477.
(50) Schultze, J. W.; Stimming, U. Z. Phys. Chem., Neue Folge 1975,
98, 285.
(51) Meisterjahn, P.; Schultze, J. W.; Siemensmeyer, B.; Stimming, U.;
Dean, M. H. Chem. Phys. 1990, 141, 131.
(52) Schreyer, A.; Eng, L.; Bo¨hni, H. J. Vac. Sci. Technol. B 1996, 14,
1162.
(53) Gerischer, H. In Photoelectrochemistry, Photocatalysis, and Pho-
toreactors; Schiavello, M., Ed; D. Reidel: Holland, 1985; p 39.
(54) Kieber, R. J.; Mopper, K. EnViron. Sci. Technol. 1990, 24, 1477.
(55) Loraine, G. A. Hazard. Waste Hazard. Mater. 1993, 10, 185.
(56) Moore, J. H.; Davis, C. C.; Coplan, M. A. Building Scientific
Apparatus; Addison-Wesley: Reading, MA, 1983.
(57) Cunningham, K. M.; Goldberg, M. C.; Weiner, E. R. EnViron. Sci.
Technol. 1988, 22, 1090.
(58) Langford, C. H.; Carey, J. H. Can. J. Chem. 1975, 53, 2430.
(59) Henglein, A.; Lindig, B.; Westerhausen, J. J. Phys. Chem. 1981,
85, 1627.
(60) Bahnemann, D. W.; Mo¨nig, J.; Chapman, R. J. Phys. Chem. 1987,
91, 3782.
References and Notes
(1) Tratnyek, P. G. Chem. Ind. 1996, 1 July, 499.
(2) Hudlicky, M. Reductions in Organic Chemistry; Halsted: New
York, 1984.
(3) House, H. O. In Modern Synthetic Reactions, 2nd ed.; Benjamin:
Menlo Park, CA, 1972; p 145.
(4) Bratsch, S. G. J. Phys. Chem. Ref. Data 1984, 18, 1.
(5) Choi, W.; Hoffmann, M. R. EnViron. Sci. Technol. 1995, 29, 1646.
(6) Choi, W.; Hoffmann, M. R. J. Phys. Chem. 1996, 100, 2161.
(7) Matheson, L. J.; Tratnyek, P. G. EnViron. Sci. Technol. 1994, 28,
2045.
(8) Roberts, A. L.; Totten, L. A.; Arnold, W. A.; Burris, D. R.;
Campbell, T. J. EnViron. Sci. Technol. 1996, 30, 2654.
(9) Criddle, C. S.; McCarty, P. L. EnViron. Sci. Technol. 1991, 25,
973.
(10) Save´ant, J.-M. In AdVances in Physical Organic Chemistry; Bethell,
D., Ed; Academic: London, 1990; Vol. 26, pp 1.
(11) Johnson, T. L.; Scherer, M. M.; Tratnyek, P. G. EnViron. Sci.
Technol. 1996, 30, 2634.
(12) Pliego, J. R., Jr.; De Almeida, W. B. J. Phys. Chem. 1996, 100,
12410.
(13) Robinson, E. A. J. Chem. Soc. 1961, 1663.
(14) Totten, L. A.; Roberts, A. L. DiVision of EnVironmental Chemistry,
Preprints of Papers; 209th American Chemical Society National Meeting.
Anaheim, CA, 1995; Vol. 35, p 702.
(15) Morrison, S. R. Electrochemistry at Semiconductor and Oxidized
Metal Electrodes; Plenum: New York, 1980.
(16) Schmickler, W. J. Electroanal. Chem. 1977, 82, 65.
(17) Stimming, U. Langmuir 1987, 3, 423.
(18) Newmark, A. R.; Stimming, U. Electrochim. Acta 1987, 32, 1217.
(19) Leiva, E.; Meyer, P.; Schmickler, W. Corros. Sci. 1989, 29, 225.
(20) Miller, C.; Cuendet, P.; Gra¨tzel, M. J. Phys. Chem. 1991, 95, 877.
(21) Gu, Y.; Waldeck, D. H. J. Phys. Chem. 1996, 100, 9573.
(22) Gerischer, H. Corros. Sci. 1989, 29, 191.
(23) Stimming, U.; Schultze, J. W. Electrochim. Acta 1979, 24, 858.
(24) Castro, P. A.; R.Vago, E.; Calvo, E. J. J. Chem. Soc., Faraday
Trans. 1996, 92, 3371.
(61) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Funda-
mentals and Applications; Wiley: New York, 1980.
(62) Stumm, W. Chemistry of the Solid-Water Interface: Processes at
the Mineral-Water and Particle-Water Interface in Natural Systems;
Wiley: New York, 1992.
(63) Waite, T. D. In Mineral-Water Interface Geochemistry; Hochella,
M. F., Jr.; White, A. F., Eds.; Mineralogical Society of America:
Washington, DC, 1990; Vol. 23, p 559.
(64) Packer, J. E.; Wilson, R. L.; Bahnemann, D.; Asmus, K.-D. J. Chem.
Soc., Perkin Trans. 2 1980, 296.
(65) Huston, P. L.; Pignatello, J. J. EnViron. Sci. Technol. 1996, 30,
3457.
(25) Cox, P. A. Transition Metal Oxides: An Introduction to their
Electronic Structure and Properties; Oxford: New York, 1995.
(26) Anderman, M.; Kennedy, J. H. In Semiconductor electrodes;
Finklea, H. O., Ed; Elsevier: New York, 1988; pp 147.
(27) Davenport, A. J.; Sansone, M. J. Electrochem. Soc. 1995, 142, 725.
(28) Bockris, J. O. M. Corros. Sci. 1989, 29, 291.
(66) Schultze, J. W.; Danzfuss, B.; Meyer, O.; Stimming, U. Mater.
Sci. Eng. 1985, 69, 273.
(67) Klausen, J.; Trober, S. P.; Haderlein, S. B.; Schwarzenbach, R. P.
EnViron. Sci. Technol. 1995, 29, 2396.
(68) Johnson, T. L.; Fish, W.; Gorby, Y. A.; Tratnyek, P. G. J. Contam.
Hydrol. 1998, 29, 377.