Reaction of 1,1,1-trichloroethane with zero-valent metals and bimetallic reductants

Article abstract of DOI:10.1021/es970784p

Information concerning the pathways and products of reaction of 1,1,1- trichloroethane (1,1,1-TCA) with zero-valent metals may be critical to the success of in situ treatment techniques. Many researchers assume that alkyl polyhalides undergo reduction via stepwise hydrogenolysis (replacement of halogen by hydrogen). Accordingly, 1,1,1-TCA should react to 1,1- dichloroethane (1,1-DCA), to chloroethane, and finally to ethane. Experiments conducted in laboratory-scale batch reactors indicate, however, that with zinc, iron, and two bimetallic reductants (nickel-plated iron and copper- plated iron) this simplistic stepwise scheme cannot explain observed results. 1,1,1-TCA was found to react rapidly with zinc to form ethane and 1,1-DCA. Independent experiments confirmed that 1,1-DCA reacts too slowly to represent an intermediate in the formation of ethane. In reactions with iron, nickel/iron, and copper/iron, cis-2-butene, ethylene, and 2-butene were also observed as minor products. Product ratios were dependent on the identity of the metal or bimetallic reductant, with zinc resulting in the lowest yield of chlorinated product. For reactions with iron and bimetallic reductants, a scheme involving successive one-electron reduction steps to form radicals and carbenoids can be invoked to explain the absence of observable intermediates, as well as the formation of products originating from radical or possibly from carbenoid coupling. Information concerning the pathways and products of reaction of 1,1,1-trichloroethane (1,1,1-TCA) with zero-valent metals may be critical to the success of in situ treatment techniques. Many researchers assume that alkyl polyhalides undergo reduction via stepwise hydrogenolysis (replacement of halogen by hydrogen). Accordingly, 1,1,1-TCA should react to 1,1-dichloroethane (1,1-DCA), to chloroethane, and finally to ethane. Experiments conducted in laboratory-scale batch reactors indicate, however, that with zinc, iron, and two bimetallic reductants (nickel-plated iron and copper-plated iron) this simplistic stepwise scheme cannot explain observed results. 1,1,1-TCA was found to react rapidly with zinc to form ethane and 1,1-DCA. Independent experiments confirmed that 1,1-DCA reacts too slowly to represent an intermediate in the formation of ethane. In reactions with iron, nickel/iron, and copper/iron, cis-2-butene, ethylene, and 2-butyne were also observed as minor products. Product ratios were dependent on the identity of the metal or bimetallic reductant, with zinc resulting in the lowest yield of chlorinated product. For reactions with iron and bimetallic reductants, a scheme involving successive one-electron reduction steps to form radicals and carbenoids can be invoked to explain the absence of observable intermediates, as well as the formation of products originating from radical or possibly from carbenoid coupling.

Full text of DOI:10.1021/es970784p

Environ. Sci. Technol. 1998, 32, 1980-1988
forming contaminants in situ through permeable “barriers”
Reaction of 1,1,1-Trichloroethane
with Zero-Valent Metals and
Bimetallic Reductants
employing zero-valent metals, in particular, has received
considerable scrutiny as a replacement for pump-and-treat
approaches (5-12). The success of this technique is,
however, contingent on the reaction products being relatively
innocuous.
At present, little information is available concerning the
pathways through which 1,1,1-TCA reacts with zero-valent
metals, including the distribution of the products that result.
Many researchers assume that organohalides react with zero-
valent metals via hydrogenolysis, involving stepwise replace-
ment of halogen by hydrogen (5, 6, 13). According to this
paradigm, 1,1,1-TCA should react to 1,1-dichloroethane (1,1-
DCA), to chloroethane, and finally to ethane. The data that
do exist concerning products of 1,1,1-TCA reaction with
metals are somewhat contradictory and do not always accord
with this sequence.
An early report by Archer (14) indicated that the reaction
of 1,1,1-TCA with Al(0) was complicated by a side reaction
of the solvent with AlCl3 or Al2O3 present at the metal surface,
which served as Lewis acids to catalyze dehydrohalogenation
to 1,1-dichloroethylene (1,1-DCE). Such a product would
be of considerable concern in a zero-valent metal treatment
scheme if it were to arise from Fe(0): 1,1-DCE tends to be
more persistent in the presence of Fe(0) than is 1,1,1-TCA
(5, 15), and drinking water standards for 1,1-DCE are 20 times
more stringent (on a molar scale) than for 1,1,1-TCA (16).
Archer further noted that 1,1,1-TCA reacts with Al(0), Sn(0),
and Fe(0) to form 2,2,3,3-tetrachlorobutane, which in turn
reacts with metals via reductive elimination to cis- and trans-
2,3-dichloro-2-butene.
More recently, Gillham and O’Hannesin (5) studied
reactions of Fe(0) with 14 halogenated organic contaminants
including 1,1,1-TCA. Products of 1,1,1-TCA reaction were
not however reported. Schreier and Reinhard (13) found
that 1,1,1-TCA reacts with both Fe(0) and Mn(0) in HEPES
buffer. No increase in 1,1-DCA concentration accompanying
1,1,1-TCA degradation was detected; rather, it was noted that
1,1-DCA (added as a cosubstrate) was stable under reaction
conditions. Finally, Wilson (17), citing work conducted by
the U.S. EPA National Exposure Research Laboratory,
reported that 1,1,1-TCA reacts with Fe(0) to form 1,1-DCA.
1,1-DCA is on the U.S. EPA 1991 Drinking Water Priority List
and must be monitored by all drinking water systems (18),
but at this time, no maximum contaminant levels have been
established by the EPA for this organohalide. Considerable
ambiguity therefore exists concerning the products likely to
be encountered in a zero-valent metal-based treatment
system, leading to uncertainties in the human health risk
that might be associated with the presence of partially
dehalogenated reaction products in effluent from a subsur-
face treatment wall.
This study was initiated to better characterize the
pathways through which 1,1,1-TCA reacts with two metals
(iron and zinc) and two bimetallic reductants (copper-plated
iron and nickel-plated iron), with a particular emphasis on
the identities of the organic products formed. Additional
studies also were carried out with 1,1-DCA and 2-butyne to
assess their potential roles as reaction intermediates. Finally,
a limited suite of experiments was conducted with isotopically
labeled 1,1,1-TCA-2,2,2-d3 in order to provide evidence in
support of the hypothesized reaction pathways.
J A Y P . F E N N E L L Y A N D
A . L YN N R O B E R T S *
Department of Geography and Environmental Engineering,
313 Ames Hall, The Johns Hopkins University, 3400 North
Charles Street, Baltimore, Maryland 21218-2686
Information concerning the pathways and products of
reaction of 1,1,1-trichloroethane (1,1,1-TCA) with zero-valent
metals may be critical to the success of in situ treatment
techniques. Many researchers assume that alkyl
polyhalides undergo reduction via stepwise hydrogenolysis
(replacement of halogen by hydrogen). Accordingly, 1,1,1-
TCA should react to 1,1-dichloroethane (1,1-DCA), to
chloroethane, and finally to ethane. Experiments conducted
in laboratory-scale batch reactors indicate, however, that
with zinc, iron, and two bimetallic reductants
(nickel-plated iron and copper-plated iron) this simplistic
stepwise scheme cannot explain observed results. 1,1,1-
TCA was found to react rapidly with zinc to form ethane
and 1,1-DCA. Independent experiments confirmed that
1,1-DCA reacts too slowly to represent an intermediate in
the formation of ethane. In reactions with iron, nickel/
iron, and copper/iron, cis-2-butene, ethylene, and 2-butyne
were also observed as minor products. Product ratios
were dependent on the identity of the metal or bimetallic
reductant, with zinc resulting in the lowest yield of
chlorinated product. For reactions with iron and bimetallic
reductants, a scheme involving successive one-electron
reduction steps to form radicals and carbenoids can be
invoked to explain the absence of observable intermediates,
as well as the formation of products originating from
radical or possibly from carbenoid coupling.
Introduction
Until its use in applications such as vapor degreasing, cold
metal cleaning, printed circuit board manufacture, aerosols
and as a solvent for inks, coatings, and adhesives was curtailed
under provisions of the Montreal Protocol (and subsequent
amendments), 1,1,1-trichloroethane (1,1,1-TCA) was the
single most extensively employed chlorinated solvent, with
a U.S. demand in 1989 of 338 000 t (1, 2). In keeping with
its widespread former use, this solvent has been detected at
about 20% of the sites on the National Priorities (Superfund)
List as identified by the U.S. EPA (3).
Groundwater contaminated with 1,1,1-TCA (as with other
dense chlorinated solvents) is difficult to remediate using
conventional pump-and-treat methods, which are inefficient
at extracting organic contaminants from heterogeneous
subsurface environments (4). This provides a substantial
incentive for development of innovative and cost-effective
alternative methods for minimizing the risks associated with
contaminants in the subsurface. The possibility of trans-
Experimental Section
Reagents. The following chemicals were used as received:
1,1,1-trichloroethane (99%; ChemService); 1,1-dichloro-
* Corresponding author e-mail address: lroberts@jhu.edu; phone:
(410)516-4387; fax: (410)516-8996.
9
1 9 8 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 13, 1998
S0013-936X(97)00784-0 CCC: $15.00
1998 Am erican Chem ical Society
Published on Web 05/21/1998

ethane (95+%; TCI America); vinylidene chloride (1,1-di-
chloroethylene, 99%; Aldrich); 2-butyne (99%; Aldrich);
chloroethane (in methanol; Supelco); ethane, ethylene, and
cis-2-butene (Scott Specialty Gases); and 1,1,1-trichloroeth-
ane-2,2,2-d3 (98%; Cambridge Isotope Laboratories). Spiking
solutions of 1,1,1-TCA, 1,1-DCA, 2-butyne, and trideuterated
1,1,1-TCA were prepared in methanol (HPLC grade, J. T.
Baker). Samples for additional analyses were extracted into
hexane (95% n-hexane, ultra-resi analyzed grade, J. T. Baker).
Acid solutions for cleaning metals were prepared within
an anaerobic chamber (containing an atmosphere of 10% H2
and 90% N2) using deoxygenated (argon-sparged) deionized
water (Milli-Q Plus UV, Millipore). The argon stream was
purified using an in-line molecular sieve and oxygen traps.
Copper and nickel solutions for synthesis of bimetallic
reductants were prepared by dissolving CuCl2 (97%; Aldrich)
or NiCl2 (98%, Aldrich) in deoxygenated, deionized Milli-Q
water under N2/ H2. Reaction bottles were filled with an Ar-
sparged solution consisting of 0.1 M NaCl (99%; J. T. Baker)
and 50 mM Tris buffer (tris(hydroxymethyl)aminomethane,
reagent grade; Sigma), pH 7.5.
Metal Preparation. The surfaces of the zero-valent metals
(iron and zinc) used in these experiments were cleaned with
acid according to methods recommended by ref 19 to remove
any surface oxides present. All cleaning steps were carried
out within an anaerobic chamber. Zinc metal (Baker, 30
mesh) was washed with 0.4% H2SO4 for 10 min, and iron
metal (Fisher, 100 mesh electrolytic) was washed with 1 M
HCl for 2 min. The acid-washed metals were rinsed 3 times
with deoxygenated Milli-Q water, rinsed with acetone,
removed from the anaerobic chamber, dried under argon at
100 °C for 30 min, and used within 24 h. Surface area analyses
conducted via Kr and N2 BET adsorption using a Micromer-
itics Flowsorb II 2300 device indicated a surface area of 0.16
m2 g-1 for iron and 0.035 m2 g-1 for zinc.
the reaction bottle using a syringe equipped with a long
needle while simultaneously adding 1 mL of deoxygenated
buffer solution to the top of the bottle with a second syringe.
Samples were sealed in 2.6 mL crimp-cap autosampler vials
for headspace analysis by gas chromatography (GC) with
flame ionization detection (FID), as described below. A 10
µL sample aliquot was extracted into 1 mL of hexane for
additional analysis of 1,1,1-TCA by GC with electron capture
detection (GC/ ECD). Efforts were made to follow the reaction
of 1,1,1-TCA for at least 3 half-lives.
Reactions of 1,1-DCA. Reactions of 1,1-DCA with metals
and bimetallic reductants were carried out in 160 mL serum
bottles sealed with Teflon septa. Serum bottles were used
in preference to the 125 mL reactors employed with 1,1,1-
TCA because the pressure buildup caused by hydrogen gas
evolution at high metal loadings or over long periods causes
the glass stopcock adapters on the reactors to leak. Reactions
were run with 2.5 g of zinc, 5.0 g of iron, 5.0 g of nickel/ iron,
or 5.0 g of copper/ iron, i.e., at a 5-fold higher metal loading
than employed with either 1,1,1-TCA or 2-butyne. The bottles
were filled with deoxygenated Tris/ NaCl buffer and sealed
without headspace. Less than ∼5 mL of headpace evolved
during these experiments. The bottles were spiked with 150
µL of a 0.2 M 1,1-DCA methanolic solution and were mixed
and sampled as described for 1,1,1-TCA. Because of the
dangers posed by the high pressures built up in the serum
bottles, time courses could only be monitored for relatively
low conversions (7-15%) of 1,1-DCA.
Reactions of Deuterated 1,1,1-TCA. To confirm hypoth-
esized reaction pathways, additional experiments were
conducted using Cl3C-CD3 as a starting material. Reactions
were carried out in 25 mL serum bottles sealed with Teflon
septa. The bottles contained 2 g of iron, 2 g of nickel/ iron,
or 2 g of copper/ iron and were filled with 15 mL of
deoxygenated Tris/ NaCl buffer leaving approximately 10 mL
of headspace in the bottle consisting of 10% H2 and 90% N2.
The sealed bottles were spiked with 5 µL of a 0.7 M solution
of trideuterated 1,1,1-TCA in methanol to give an initial
concentration of approximately 200 µM. The bottles were
mixed overnight to allow reaction products to accumulate
before 50 µL headspace samples were taken directly from
the reaction bottle for immediate analysis of reaction products
by gas chromatography/ mass spectrometry (GC/ MS) as
described below. Deuterium label retention in the parent
compound was verified in separate experiments by monitor-
ing the mass spectrum of the 1,1,1-TCA-2,2,2-d3 over several
hours during the course of its reaction with iron.
Preparation of Bim etallic Reductants. Iron was acid-
washed and rinsed with deoxygenated water as described
above. A dilute solution (50 µM) of CuCl2 or NiCl2 was added
slowly (within an anaerobic chamber) to 2 g of iron that was
suspended in water by agitation. Once the metal chloride
solution had been added, the metal was agitated for an
additional minute, rinsed with deoxygenated Milli-Q water,
rinsed with acetone, and dried as described above. Assuming
that all of the catalytic metal was reductively precipitated
onto the iron base metal, the content of the Cu or Ni in the
bimetallic reductant was calculated as 0.035 mol %.
Reactions of 1,1,1-TCA and 2-Butyne. Reactions of 1,1,1-
TCA and 2-butyne with metals and bimetallic reductants
were carried out in 125 mL (nominal volume; actual volume
= 150 mL) glass bottles with glass stopcock adapters. The
stopcocks were fitted with an NMR septum through which
samples could be taken by syringe while maintaining anoxic
reaction conditions; the stopcocks served to isolate the rubber
septa from the flask contents except during the brief intervals
required for sampling. For 1,1,1-TCA, reactions were run
with 0.5 g of zinc, 1.0 g of iron, 1.0 g of nickel/ iron, or 1.0 g
of copper/ iron. Metal loadings were the same for 2-butyne
(except that reaction with zinc was not investigated). The
bottles were filled with deoxygenated Tris/ NaCl buffer under
an anaerobic atmosphere. Initially, the reaction bottles
contained no headspace; less than 2 mL of headspace
(presumably resulting from reduction of protons to H2 by
the metal) evolved during the course of a typical experiment.
The bottles were spiked with a 0.2 M methanolic solution of
1,1,1-TCA for an initial concentration of approximately 200
µM or with a 0.2 M solution of 2-butyne for an initial
concentration of ∼10 µM. The bottles were rotated about
their longitudinal axes on a rotator (Cole-Parmer) at 40 rpm
throughout the course of the experiments. At regular
intervals, a 1-mL sample was removed from the bottom of
Sam ple Analysis. Headspace samples equilibrated with
the 1-mL aqueous reaction aliquots were analyzed on a Carlo
Erba GC 8000 gas chromatograph equipped with a Carlo
Erba HS850 headspace autosampler, a J&W Scientific GS-Q
PLOT column (30 m ×0.53 mm i.d.), and an FID. The samples
were equilibrated in the autosampler at 60 °C for 30 min
prior to injection of 200 µL of headspace in splitless mode.
Data were acquired by a PC-based data acquisition system
(XChrom v. 2.1; LabSystems, Beverly, MA).
Peak areas were converted to aqueous concentrations by
the external standard method, using calibration curves
prepared from aqueous standards or gas standards, as
appropriate. For 1,1,1-TCA, 1,1-DCA, 1,1-DCE, and 2-butyne,
aqueous standards were prepared in 20-mL glass syringes,
which were analyzed as described for samples. For ethane,
ethylene, and cis-2-butene, gas standards (Scott Specialty
Gas) were employed. A 2-mL glass syringe with a wetted
barrel and three-way stopcock (one end of which was fitted
with a septum) was used to mix and dilute gas standards. A
200 µL aliquot of gas was removed and was manually injected
in splitless mode. Results for samples were converted to
aqueous concentrations using the appropriate Henry’s law
constant. The dimensionless Henry’s law constant (ex-
9
VOL. 32, NO. 13, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 9 8 1

pressed as [(mol/ Lair)/ (mol/ Lwater)] at 60 °C), measured in
our laboratory by a modified EPICS method (20) in Tris/
NaCl buffer, is 15.8 for ethane, 8.7 for ethylene, and 5.4 for
cis-2-butene (15).
two isotopomers cannot be differentiated by mass spec-
trometry. In the second ethane fit, the previous program
was supplemented with contributions from ethane-1,1,1,2-
d4 and ethane-1,1,2,2-d4. The following compounds were
included in trying to fit observed spectra of ethylene:
unlabeled ethylene, ethylene-d1, ethylene-d2, and ethylene-
d3. Because reported mass spectra for the three dideuterated
ethylenes are virtually identical, these isomers were lumped
together as a single species in our analysis. For ethylene
isotopomers, the most abundant ion is always the parent
ion. For ethanes, the molecular ion is not the most abundant
ion; relative intensities for the parent ion range from 27% to
42%. The most abundant ions for these species are as
follows: unlabeled ethane, m/ e 28; ethane-d1, m/ e 29; ethane-
1,1-d2, m/ e 29; ethane-1,2-d2, m/ e 30; ethane-1,1,1-d3, m/ e
30; ethane-1,1,1,2-d4, m/ e 31; ethane-1,1,2,2-d4, m/ e 31.
A best fit distribution was found by minimizing the
objective function:
The existence of a small amount of headspace could, in
principle, result in a partitioning of the volatile compounds
under investigation, thereby influencing reaction mass
balances. Based on 25 °C Henry’s law constants (21) for
1,1,1-TCA (0.71 [(mol/ Lair)/ (mol/ Lwater)]), 1,1-DCA (0.26), and
1,1-DCE (1.1), less than 1.5% of the mass of these halocarbons
would have been lost through partitioning to the volumes of
headspace that accumulated. For more volatile hydrocarbon
constituents such as ethane (KH′ ) 20.4 [(mol/ Lair)/ (mol/
Lwater)]; 22) and ethylene (KH′ ) 8.7; 22), however, losses could
have been more significant, amounting to approximately 20%
of the ethane mass and 10% of that of ethylene. Smaller
losses would have been anticipated for the other hydrocar-
bons observed, namely, cis-2-butene (KH′ ) 5.8 at 24 °C;
determined in our lab according to ref 23) and 2-butyne (KH′
estimated as 0.44; 24). Aqueous concentrations reported for
all species therefore include a correction for calculated
volatilization losses, based on our estimates of the headspace
volume that had accumulated at each sampling point.
Detection limits for selected species were calculated from
confidence bands around linear calibration curves using the
method of Hubaux and Vos (25). For purposes of assessing
detection limits, five-point calibration curves were based on
low concentration ranges (generally within a factor of 5 of
the estimated detection limit) rather than on the full
concentration range that was used to quantitate the results
of the kinetic experiments.
36
Z )
(Ui + V ) + L
(1)
i
∑
i)25
where the index i denotes the m/ e values, subject to the
following constraints:
aij xj + Ui - V ) ei L
i, i ) 25, ..., 36 (2)
i
∑
j
aij xj e L
i
(3)
∑
j
Due to the coelution of 1,1,1-TCA with one possible
reaction product, 1-hexene, verification of 1,1,1-TCA con-
centrations was made by analyzing hexane extracts via cold
on-column injection onto an Rtx-1 GC column (30 m × 0.32
mm i.d. ×5.0 µm film thickness; Restek) followed by detection
with a Carlo Erba 63Ni ECD. The GC/ ECD results agreed well
with results obtained from the headspace GC/ FID analysis
technique. Identities of all major reaction products observed
were confirmed by headspace injections onto a Hewlett-
Packard (HP) 5890 GC equipped with an HP 5970 mass
spectrometer detector and a GS-Q PLOT column (30 m ×
0.32 mm i.d.) or the Rtx-1 column previously described.
Products were identified on the basis of retention time (using
authentic compounds as standards) and mass spectra. The
mass spectra were recorded under electron impact (EI)
ionization by conducting scans in the range of m/ e 15-140;
a narrower range (m/ e 15-40) was used to maximize
sensitivity in measuring the spectra of low molecular weight
deuterated products (ethane and ethylene) in the experiment
with isotopically labeled 1,1,1-TCA.
xj ) 1
(4)
(5)
∑
j
0 e xj e 1
j
where the indexj refers to the specific isotopomers considered
in the mixture. The constant aij represents the theoretical
abundance of m/ e i for isotopomer j, and ei is the experi-
mentally determined abundances for the mixture at each
m/ e value. The decision variables xj are the fractions of
isotopomer j in the mixture, while Ui and V are so-called
i
target-hitting variables. Their function in the linear program
is such that minimizing the objective function is equivalent
to minimizing the absolute value of the difference between
the experimental and calculated mixtures. Through con-
straints (3), the decision variable L equals the largest
calculated m/ e abundance and is used to normalize the
calculated values to obtain relative intensities.
Results
Mass Spectral Modeling. One of the objectives in the
experiments conducted with Cl3C-CD3 was to assess the
deuterium content in the reaction products. GC/ MS spectra
indicated that ethylene and ethane products resulting from
reactions of trideuterated 1,1,1-TCA consisted of a mixture
of coeluting isotopomers with varying numbers of deuterium
atoms. To estimate the retention of the deuterium label
among the products, a set of linear programs was developed.
Two such programs were developed for ethane and one for
ethylene modeling. For unlabeled compounds, model fits
were based on mass spectra of gas standards determined on
our instrument, while for deuterated compounds, model fits
were based on the eight major peaks reported in ref 26. In
the first ethane fit, the observed mass spectra were assumed
to include contributions from unlabeled ethane; ethane-d1;
ethane-1,1-d2; ethane-1,2-d2; and ethane-1,1,1-d3. We were
unable to locate mass spectral information for ethane-1,1,2-
d3, although we expect its mass spectrum is quite similar to
that of ethane-1,1,1-d3. Ponec and Bond (27) note that these
Reaction of 1,1,1-TCA and 1,1-DCA with Zn. 1,1,1-TCA
reacts rapidly with zinc to form ethane and 1,1-DCA (Figure
1). A model fit to the first five data points for 1,1,1-TCA
shows a reasonable adherence to exponential decay, with
some evidence that the reaction slows down toward the end
of the experiment (Figure 1). Pseudo-first-order kobs values
obtained from initial rates and rate constants normalized to
metal surface area loading (kSA; 9) are shown in Table 1.
Ethane is the more abundant product observed, with a
lesser amount of 1,1-DCA formed. The yield of 1,1-DCA was
calculated as
(1,1-DCA)t
(1,1,1-TCA)o - (1,1,1-TCA)t
where the subscript t denotes concentrations measured at
all but the initial time point and (1,1,1-TCA)o is the model-fit
initial TCA concentration. Results indicated a 1,1-DCA yield
9
1 9 8 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 13, 1998

FIGURE 1. Reduction of 1,1,1-TCA in 0.1 M NaCl/0.05 M Tris buffer
(pH 7.5) by Zn(0). Solid line reflects exponential decay fit to initial
1,1,1-TCA data.
TABLE 1. Summary of Rate Data Obtained for Reactions of
1,1,1-TCA with Zero-Valent Metals and Bimetallic Reductants
in NaCl/Tris Buffer, pH 7.5a
FIGURE 2. Reduction of 1,1,1-TCA in 0.1 M NaCl/0.05 M Tris buffer
(pH 7.5) by Fe(0). Solid line reflects exponential decay fit to initial
1,1,1-TCA data. Inset shows minor products not observed in reaction
with Zn.
reductant
metal mass (g)
kobs (h-1
)
kSA (L m-2 h-1
)
zinc
iron
nickel/iron
copper/iron
0.5
1.0
1.0
1.0
1.15 ( 0.18
0.48 ( 0.06
1.86 ( 0.20
1.35 ( 0.04
10.0
0.46
1.77
1.29
1,1-DCA represents a two-electron reduction product, and
ethane is a six-electron reduction product. We do not,
however, believe that 1,1-DCA is an intermediate in ethane
formation. Both reaction products appeared simultaneously,
and both persisted throughout the 3.5 h course of the
experiment. Independent experiments demonstrated that
1,1-DCA reacts slowly with zinc to produce ethane, with a
half-life of approximately 14 days at a 5-fold higher loading
of zinc. This slow rate of reaction precludes 1,1-DCA as an
intermediate in the formation of ethane from 1,1,1-TCA.
Reaction of 1,1,1-TCA and 1,1-DCA with Fe. Somewhat
different products were obtained with iron (Figure 2). With
this reductant, 1,1,1-TCA reacted to form mainly 1,1-DCA as
a major product (67 ( 2% yield) along with lesser amounts
of ethane, cis-2-butene, ethylene, and a trace of 2-butyne.
Final mass balances on carbon were approximately 84%
relative to the calculated spiked concentration. We do not
believe that sorption to the iron metal or impurities therein
(as observed by Burris et al.; 28) is responsible for the
imperfect mass balance. Unlike the cast iron employed by
Burris, the Fisher electrolytic iron used in the present study
has a very low carbon content (29); furthermore, we have
not observed any losses in aqueous concentration that could
be attributed to sorption of chlorinated ethenes to either
Fisher electrolytic iron or to zinc (15). Note that the total
mass balance throughout the time course coincides closely
with the initial 1,1,1-TCA measurement; if reaction to
relatively polar products were significant, mass balances
would have decayed over time. As with zinc, reactions
initially followed pseudo-first-order decay with an apparent
slowing at later times. Initial rate constants are shown in
Table 1.
a
Metal loading represents m ass of m etal per 150 m L of buffer solution
in reactor. Uncertainties on kobs values represent 95% confidence lim its
obtained from an exponential fit to the first five data points.
of 25.8 (( 0.5)%, where the stated uncertainty reflects 95%
confidence limits. Replicate experiments indicated that
product distributions determined for reaction of 1,1,1-TCA
with zinc (as well as with iron; see below) were reproducible
within 5-10%. No organic products were observed in blank
experiments conducted with zinc (or iron or the bimetallic
reductants employed) plus NaCl/ Tris buffer in the absence
of 1,1,1-TCA either in the presence or absence of methanol
at the concentration employed as a carrier in the 1,1,1-TCA
spike.
The carbon mass balance at the end of the experiment
(calculated as the sum of all organic species measured) was
approximately 93% of the calculated spiked 1,1,1-TCA
concentration (based on the volume of water in the reactor
and the concentration of the 1,1,1-TCA in the methanol stock).
We suspect that minor losses may have occurred during
spiking through the stopcocks, as no other organic products
were observed, and mass balances throughout the time
course agree well with initial 1,1,1-TCA measurements.
Lacking authentic standards of 2,2,3,3-tetrachlorobutane and
cis- and trans-2,3-dichloro-2-butene, however, we cannot
rule out the possibility that these could have been missed by
the analytical techniques employed. Also, the headspace
analytical technique is not optimal for products such as
acetaldehyde or ethanol, which have low Henry’s law
constants. The detection limit for acetaldehyde was approxi-
mately 9.2 µM (more than an order of magnitude higher
than for the volatile organic products reported herein), and
that for ethanol was greater than 200 µM. The high detection
limit for the latter compound may be in part attributable to
the GC column employed, which is not well-suited to
quantitative analysis of alcohols.
Complementary experiments with 1,1-DCA again indicate
that even though this chlorocarbon reacts to form ethylene
and ethane as major products, reaction with iron is too slow
(with a half-life of approximately 25 days at a loading of 5.0
g of Fe) for 1,1-DCA to represent an intermediate in the
9
VOL. 32, NO. 13, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 9 8 3

FIGURE 3. Reduction of 1,1,1-TCA by bimetallic (Ni-plated Fe)
reductant in 0.1 M NaCl/0.05 M Tris buffer (pH 7.5). Solid line reflects
exponential decay fit to initial 1,1,1-TCA data. Inset shows minor
products not observed in reaction with Zn.
FIGURE 4. Reaction of 1,1,1-TCA with bimetallic (Cu-plated Fe)
reductant in 0.1 M NaCl/0.05 M Tris buffer (pH 7.5). Solid line reflects
exponential decay fit to initial 1,1,1-TCA data. Inset shows minor
products not observed in reaction with Zn.
formation of ethane or ethylene. Nor is ethylene (a four-
electron reduction product of 1,1,1-TCA) an intermediate in
ethane formation, as independent experiments conducted
in our laboratory demonstrate that ethylene reacts too slowly
for such a reaction to account for the amount of ethane
observed in reduction of 1,1,1-TCA (15). Further evidence
that suggests neither 1,1-DCA nor ethylene is an intermediate
in ethane formation comes from the observation that yields
of 1,1-DCA, ethylene, and ethane were independent of time.
These three products must therefore originate from parallel
and not consecutive reactions.
Reaction of 1,1,1-TCA and 1,1-DCA with Bim etallic
Reductants. 1,1,1-TCA reacted with nickel/ iron (Figure 3)
at a significantly faster rate than with iron (Table 1). This
enhanced reactivity of Ni/ Fe accords with observations by
other researchers for reactions of chlorinated ethylenes (30,
31). The increased rate of reaction was accompanied by a
small change in the product distribution: a slightly higher
yield of cis-2-butene and ethylene was obtained and a lower
yield (43 ( 4%) of 1,1-DCA. As with iron, only a trace of
2-butyne was detected. The carbon mass balance at the end
of the experiment was approximately 83% of the calculated
spiked concentration. 1,1-DCA again reacts slowly with
nickel/ iron, with a half-life of approximately 9 days at a
loading of 5.0 g of Ni/ Fe.
Reduction of 2-Butyne. Reactions of 2-butyne with iron
and bimetallic reductants also were examined to assess
whether this species could represent an intermediate in cis-
2-butene formation. With iron and the two bimetallic
reductants studied, the major reaction product observed was
cis-2-butene, with only traces oftrans-2-butene and 1-butene
formed. Preliminary modeling studies based on a simple
pseudo-first-order approach suggested, however, that the
measured rate of 2-butyne reaction was insufficient to
account for all of the cis-2-butene observed in 1,1,1-TCA
experiments. This may indicate limitations of a pseudo-
first-order approach; alternatively, some additional pathway
may be responsible for cis-2-butene formation.
In summary, these experiments with 1,1,1-TCA show no
evidence that the dehydrohalogenation reported with Al(0)
(14) takes place with zero-valent zinc or iron. The expected
dehydrohalogenation product (1,1-DCE) is observed only
with copper/ iron and only in trace concentrations (less than
2% yield). The major reaction products (1,1-DCA and ethane)
are consistent with reductive dehalogenation. A simple
stepwise reduction from 1,1,1-TCA to 1,1-DCA to chloro-
ethane to ethane does not however appear to be involved.
Discussion
A more dramatic change in product distribution was found
in the copper/ iron system (Figure 4). A substantial increase
in yield was observed for ethylene, while a decrease in 1,1-
DCA (30 (1% yield) and cis-2-butene formation was obtained
relative to nickel/ iron. A significant concentration of 2-bu-
tyne was also measured; this product exhibited accumulation
and subsequent disappearance indicative of a reaction inter-
mediate. In addition, a trace of 1,1-dichloroethylene (1,1-
DCE; detection limit 0.03 µM) was observed, a product
confirmed by GC/ MS. Final mass balances on carbon were
approximately 76% relative to calculated spiked concentra-
tion. Once more, kinetic arguments indicate that 1,1-DCA
cannot represent an intermediate in ethane or ethylene
formation.
Reaction Pathways. One potential scheme that could
account for all of the products observed in our experiments
with Fe(0) or reported in the literature is indicated in Figure
5. According to this scheme, ethane is formed via a sequence
that involves successive one-electron reduction steps to form
radicals and carbenoids, analogous to the pathway involved
in reduction of 1,1,1-TCA by Cr(II) in aqueous solution (32).
The first step in the sequence involves a one-electron
dissociative electron-transfer step to result in the 1,1-
dichloroethyl radical:
H3C–CCl3 + e–
[H3C–CCl3]• –
H3C–CCl2 + Cl–
9
1 9 8 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 13, 1998

FIGURE 5. Proposed scheme for reaction of 1,1,1-TCA with Fe(0). Boxes indicate reaction products identified in this study; dashed boxes
indicate products previously reported in the literature (14). Dashed arrows reflect potential pathways that are too slow to explain the
observed distribution of reaction products, as shown by this study or by unpublished data from our laboratory (15). Note that intermediates
shown for simplicity as free carbenes are in fact more likely to occur as carbenoids. Not shown: potential coupling of chloroethyl radicals
or of carbenoid species to C4 compounds.
This radical could couple with other dichloroethyl radicals
to form 2,2,3,3-tetrachlorobutane (which could undergo two
successive reductive â-elimination reactions to form 2,3-
dichloro-2-butenes and hence 2-butyne, which in turn was
observed to undergo reduction to cis-2-butene). The im-
portance of such coupling reactions would be expected to
be dependent on the concentrations of the parent compound,
as higher initial 1,1,1-TCA concentrations should give rise to
higher steady-state concentrations of radicals.
Alternatively, the dichloroethyl radical could undergo a
second one-electron reduction to form the dichloroethyl
carbanion, either free or complexed as an organometallic
species. This carbanion could undergo protonation (to yield
1,1-dichloroethane) or else could undergo R-elimination to
form the H3C-C¨ -Cl carbene (or carbenoid). Carbanions
are known to be stabilized by halogen substituents (33),
which could favor the competition between protonation
detectable accumulation of this product.
We suspect that H3C-C¨ -Cl exists as a carbenoid (e.g., as
a carbene-metal complex or R-haloorganometallic com-
pound; 33) rather than as a free carbene. A free carbene
might be expected to undergo a significant amount of trap-
ping by the solvent, which would be expected to result in the
formation of acetaldehyde in the case of H3C-C¨ -Cl. There
is precedent in the chemical literature for carbenoids as
intermediates of polyhaloalkane reduction by zero-valent
metals in organic solvents. For example, in the Simmons-
Smith synthesis of cyclopropanes through reaction of olefins
with CH2 (34), it is believed that the methylene (generated
through the reduction of CH2I2 by the zinc-copper couple
in ether) exists in the form of iodomethylzinc iodide, which
is in equilibrium with the dialkylzinc species (ICH2)2Zn‚ZnI2
(35, 36). Although the iodomethylzinc iodide intermediate
is sufficiently stable in ether to be isolable (35), the presence
of water could have an enormous impact on the stability of
organometallic intermediates, which may only possess only
a fleeting existence as reactive intermediates in the present
system.
vs R-elimination to
a carbene; the extra stabilization
energy provided by two halogens may explain why 1,1-
dichloroethane was observed but not chloroethane (which
would be expected to result from protonation of a chloro-
ethyl carbanion). We note that our detection limit for
chloroethane is approximately 0.24 µM. Assuming that
(as with 1,1-dichloroethane) chloroethane is reduced
slowly by the zero-valent metals investigated, even a chlo-
roethane yield less than 0.2% should have resulted in
Further reduction of H3C-C¨ -Cl, accompanied by proton
transfer, would result in formation of the chloroethyl radical.
This radical could undergo reduction to the H3C-C¨ -H
carbenoid. One of the characteristic reactions of alkylcar-
benes is rearrangement (37), with migration of hydrogen:
9
VOL. 32, NO. 13, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 9 8 5

H
C
H
H
H
H
H
H
C
H
C
C
Such a rearrangement could explain the ethylene observed.
A similar rearrangement of H3C-C¨ -Cl does not appear to
be taking place to any significant extent, as no vinyl chloride
(detection limit 0.19 µM) was observed. This may serve as
further indication that H3C-C¨ -Cl occurs as a carbenoid
rather than as a free carbene.
Finally, H3C-C¨ -H could undergo two successive one-
electron reduction steps (accompanied by protonation) to
generate the ethyl radical and finally ethane. The latter
product also results from reduction of ethylene, albeit at rate
that is too slow to account for its formation from 1,1,1-TCA
reduction (15).
FIGURE 6. Comparison of experimental with calculated GC/MS
spectra for reaction of isotopically labeled 1,1,1-TCA-2,2,2-d3 with
iron: (a) experimental ethane mixture (solid bars); best-fit mass
spectrum assuming a mixture of unlabeled, mono-, di-, and
trideuterated ethane isotopomers (fit 1, patterned bars); and best-fit
mass spectrum based on a mixture of unlabeled through tetra-
deuterated ethanes (fit 2, open bars); (b) experimental ethylene
mixture (solid bars) and best-fit mass spectrum based on a mixture
of unlabeled through trideuterated ethylene isotopomers (patterned
bars).
As with the dichloroethyl radical, the chloroethyl radical
and the ethyl radical could potentially undergo self-coupling,
forming meso- and D,L-2,3-dichlorobutane (in the case of
the chloroethyl radical) and n-butane (from the ethyl radical).
We have previously found that erythro- and threo-2,3-
dibromopentane display a high degree of stereospecificity
in their reductive elimination by zero-valent metals, reacting
almost exclusively to trans- and cis-2-pentene, respectively
(38, 39). Even if the reaction of 2,3-dichlorobutanes were to
be less stereospecific than reactions of the brominated
analogues, we would expect if anything that an excess of
trans-2-butene (as the more stable isomer) over the cis isomer
should result from reduction of a racemic mixture of 2,3-
dichlorobutanes. Our data indicate that trans-2-butene is
not formed in appreciable amounts; it did not accumulate
above detection limits (0.07 µM), nor was its likely reduction
product (butane) ever observed. Unless the coupling of
radicals at the metal surface were to result somehow in a
substantial excess of D,L-2,3-dichlorobutane over the meso
isomer, the apparent lack oftrans-2-butene production would
argue against the coupling of chloroethyl radicals. The
absence of butane formation is evidence that coupling of
ethyl radicals is not a significant process. The lack of apparent
coupling by the chloroethyl and ethyl radicals may reflect
the decreased stability of these less highly substituted species
(33), resulting in lower steady-state concentrations and thus
diminished probability of an encounter with one another.
The possibility that carbenes (or carbenoids) could couple
should also be considered. Some authorities (40) have
suggested that carbenes can couple to give rise to olefins.
Other researchers (33) argue that because of their very high
reactivity, it is unlikely that a free carbene could encounter
another before reacting in some other manner, such as by
rapid intramolecular rearrangement or reaction with the
solvent; alternative explanations should therefore be sought
for such apparent “dimerization” reactions. Although cou-
pling of free carbenes in homogeneous solution may indeed
be an extremely remote occurrence, the formation of
organometallic carbenoids, either as dialkyl species or as
monoalkyl species at locally elevated concentrations at the
solid-water interface, might increase the probability of
coupling. Blanchard and Simmons (35) have studied the
formation of ethylene as a side product of methylene iodide
reaction with the zinc-copper couple. Their evidence
suggests that ethylene formation occurs through two simul-
taneous reactions of an organozinc carbenoid intermediate:
(i) reaction of (ICH2)2Zn‚ZnI2 with unreacted CH2I2 to form
diiodoethane, which subsequently undergoes reductive
elimination to ethylene;(ii)reaction of(ICH2)2Zn‚ZnI2 to ICH2-
CH2ZnI, which in turn reacts to ethylene.
dergo reductive elimination to 2-butyne and then be reduced
to cis-2-butene); from H3C-C¨ -H, cis- and/ or trans-2-butene
would be expected. Once again, unless carbenoid coupling
occurs in a highly stereospecific manner, the apparent lack
of trans-2-butene formation would seem to rule out the
possibility that H3C-C¨ -H species couple. Further assess-
ment of the likelihood of carbenoid coupling must await a
more detailed examination of the natures of these interme-
diates.
Experim ents with Trideuterated 1,1,1-TCA. Additional
evidence concerning reaction pathways was obtained by
conducting experiments using trideuterated 1,1,1-TCA. GC/
MS studies of products obtained from reaction of this
compound with zero-valent iron, nickel/ iron, and copper/
iron showed that ethylene and ethane reaction products are
partially deuterated. According to the scheme shown in
Figure 5, we would predict that the ethane should consist of
ethane-1,1,1-d3 and that the ethylene should also be tri-
deuterated. The observed spectra did not match library
spectra of pure ethane-1,1,1-d3 or trideuterated ethylene (26)
but rather appeared to represent mixtures of isotopomers
exhibiting varying degrees of deuterium label retention. A
set of linear programs, described in the Experimental Section,
provided estimates of the distribution of isotopomers in each
case.
Experimental and optimized results are shown in Figure
6a for ethane and Figure 6b for ethylene. The optimal
distribution for the first ethane fit (from zero to three
deuterium atoms) consisted of 16% unlabeled ethane, 34%
ethane-d1, and 50% ethane-1,1,1-d3. The calculated spectrum
does not match the measured spectra well for the higher
molecular weight ions. Improved fit is obtained by allowing
ethane-1,1,1,2-d4 and ethane-1,1,2,2-d4 to be present (Figure
6a, fit 2), but the abundance of the m/ e 33 value is still
underestimated. The resulting optimized distribution con-
sisted of 6% unlabeled ethane, 32% ethane-d1, 31% ethane-
1,1,1-d3, and 31% ethane-1,1,2,2-d4. Incorporation of higher
molecular weight isotopomers (pentadeuterated or hexa-
deuterated ethane) into the model did not result in any
substantial improvement in fit between predicted and
measured spectra.
For H3C-C¨ -Cl, the resulting coupling products would
be cis- and trans-2,3-dichloro-2-butenes (which could un-
The imperfect agreement between measured and pre-
dicted ethane spectra may reflect limitations of our model
9
1 9 8 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 13, 1998

and/ or the resolution of our mass spectrometric analysis
(e.g., limits to instrumental accuracy; variability in back-
ground noise leading to imperfect background corrections;
or, perhaps most likely, reliance on published spectra rather
than on spectra for partially deuterated ethane isotopomers
determined on our instrument). We feel that other pos-
sibilities for the imperfect match between predicted and
measured spectra, such as that some other product coelutes
with the ethane, are unlikely. When experiments were run
using unlabeled 1,1,1-TCA, the peak obtained for ethane was
readily resolved from all other nearby peaks on the GC column
employed, and the mass spectrum was identical to that of
an authentic ethane standard. Mass spectrometric analysis
of the isotopically labeled 1,1,1-TCA failed to reveal significant
levels of impurities (such as 1,1,1-trichloroethane-2,2-d2) that
might have led to production of dideuterated or lower
molecular weight ethane or ethylene isotopomers. No ethane
or ethylene accumulation was detected in the absence of
1,1,1-TCA, as previously discussed.
The best-fit distribution of labeled and unlabeled products
in the ethylene mixture was 6% ethylene, 11% ethylene-d1,
70% ethylene-d2 (mixture of all three possible dideuterated
isomers), and 14% ethylene-1,1,2-d3. In this case, the
simulated spectrum matched the experimental spectrum
well, as shown in Figure 6b.
The deviations from the expected deuterium label reten-
tion in the ethane and ethylene products suggest that the
actual reaction mechanism is more complex than the scheme
illustrated in Figure 5. It is possible that H-D exchange
could occur either in the reactive intermediates or in the
final products ethane or ethylene. Some of the deuterium
lost could persist at the surface of the metal, thereby
accounting for the variable deuterium retention. Such metal-
catalyzed H-D exchange of alkanes and olefins has received
intensive scrutiny in the field of metal catalysis (27).
According to the generally accepted model, the reaction
involves a sequence of steps: (a) adsorption of the hydro-
carbon to the metal surface, followed by activation of the
C-H (or C-D) bond; (b) dissociation of the carbon-hydrogen
bond to yield a metal-bound radical; (c) coupling of the radical
with metal-bound atomic H (or D). Most of the prior
research, however, pertains to gas-phase reactions, and it is
difficult to assess without additional experiments whether
such exchange would also occur under the conditions
employed in the present study. Certainly no evidence of
D-H exchange could be detected in the mass spectrum of
1,1,1-TCA-d3 over several half-lives of reaction with iron.
Furthermore, no significant deuterium label loss was ap-
parent in the 2-butyne-1,1,1,4,4,4-d6 or the 1,1-dichloro-
ethane-2,2,2-d3 products observed (see below). This indicates
that H-D exchange, if it occurs under the present reaction
conditions, is not invariably a facile process.
The other reaction products (1,1-dichloroethane, 2-bu-
tyne, and cis-2-butene) were found to consist of 1,1-
dichloroethane-2,2,2-d3 (Figure 7a), 2-butyne-1,1,1,4,4,4-d6
(Figure 7b), and cis-2-butene-d6, shown in Figure 7c. Note
that because of the tendency of hydrocarbons to exhibit a
scrambling of hydrogen atoms under electron impact
ionization (41, 42), the positions of the deuterium atoms
within the cis-2-butene cannot be determined from the mass
spectrum. The presence of six deuterium atoms in the C4
products is consistent with their origin via coupling of radicals
(or possibly D3C-C¨ -Cl carbenoids) that then undergo
reductive elimination and reduction to 2-butyne and cis-2-
butene. Retention of six deuterium atoms would not have
been anticipated if C4 products were to originate from
reduction of aqueous carbonate (43) or from reaction of
carbide impurities within the metal (29).
FIGURE 7. GC/MS spectra of (a) 1,1-dichloroethane-2,2,2-d3; (b)
2-butyne-1,1,1,4,4,4-d6; and (c) cis-2-butene-d6 obtained in reaction
of 1,1,1-TCA-2,2,2-d3 with iron.
reacts rapidly with Fe(0). Kinetic analyses and product
studies indicate that the reaction does not, however, proceed
exclusively through stepwise replacement of halogen by
hydrogen, as has been previously proposed as a paradigm
for reaction of organohalides with metals. Rather, the data
for iron and iron-based bimetallic reductants are consistent
with a sequence involving reduction to radicals and car-
benoids, resulting in products that reflect reductive R-elim-
ination and coupling in competition with hydrogenolysis.
To determine if reaction with zinc also involves free radical
intermediates requires additional study. Certainly, none of
the coupling products that are often characteristic of free
radical reactions and that were observed for reaction with
iron, Ni/ Fe, and Cu/ Fe were detected with zinc.
The data reveal interesting differences in the distribution
of products obtained with the various reductants employed.
A lower yield of 1,1-DCA, for example, was obtained with
zinc than with iron or with either of the two bimetallic
reductants. In all cases, the reaction of 1,1-DCA was shown
by independent experiments to be sufficiently slow as to
preclude major differences between the reactivity of the
various reductants as an explanation. It seems that the metal
serves a role that goes beyond that of a simple source of
electrons: specific chemical interactions between (potentially
partially oxidized) metal species present at the solid-water
interface and the organohalide (or reactive intermediates
resulting therefrom) are likely to be involved in dictating
reactivity and reaction product distribution. Boronina et al.
(7) have also found that the identity of the metal plays a part
in the distribution of products obtained in the reduction of
organohalides: reaction of CCl4
with zinc led to dechlori-
nation to methane, while reaction with Sn(0) led to CO2 and
chloroform as major products. A complete explanation for
such differences in reactivity and reaction product distribu-
tions displayed by various metals and bimetallic reductants
is bound to involve a multitude of factors, many of which are
still imperfectly understood. Even for reactions as well-
studied as metal-promoted H-D exchange in alkanes, Ponec
Relevance of Results. In summary, our results confirm
Gillham’s (5) and Schreier’s (13) observations that 1,1,1-TCA
9
VOL. 32, NO. 13, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1 9 8 7

(9) Johnson, T. L.; Scherer, M. M.; Tratnyek, P. G. Environ. Sci.
Technol. 1996, 30, 2634-3640.
(10) Johnson, T. L.; Scherer, M. M.; Tratnyek, P. G. Environ. Sci.
Technol. 1996, 30, 2634-3640.
(11) Senzaki, T.; Kumagai, Y. Kogyo Yosui 1988, 357, 2-7.
(12) Senzaki, T.; Kumagai, Y. Kogyo Yosui 1989, 369, 19-25.
(13) Schreier, C. G.; Reinhard, M. Chemosphere 1994, 29, 1743-
1753.
and Bond (27) note “It is still quite unclear why adjacent
metals can behave so differently as catalysts... Unfortunately
the chemical rules governing metals’ behavior have not yet
been elucidated... Much work remains to be performed in
this field.” These statements can be echoed for metal-
promoted reductive dehalogenation reactions. Nevertheless,
the strong influence that the metal identity exerts on reaction
product distribution suggests that it may be feasible to
develop a reductant through which the generation of
undesirable byproducts could be minimized.
Although the pathways outlined herein for iron are
complex, they do make it possible to formulate a reaction
scheme that can account for all of the observed reaction
products (including the inability of 1,1-DCA and ethylene to
represent reaction intermediates). This should facilitate the
development of process models that could be used for the
design of subsurface iron-based treatment walls for 1,1,1-
TCA. Additional studies will be required to assess whether
reaction with zinc (which does not appear to give rise to
coupling products) also proceeds through similar pathways.
(14) Archer, W. L. Ind. Eng. Chem. Prod. Res. Dev. 1982, 21, 670-672.
(15) Arnold, W. A. Unpublished data.
(16) Pontius, F. W. J. Am. Water Works Assoc. 1995, 87, 48-58.
(17) Wilson, E. K. Chem. Eng. News 1995, 19-22.
(18) Pontius, F. W. J. Am. Water Works Assoc. 1992, 84, 36-50.
(19) American Society for Metals. Metals Handbook, 9th ed.; Vol. 5,
Surface Cleaning Finishing and Coating; ASM: Metals Park,
OH, 1982; Vol. 5.
(20) Gossett, J. M. Environ. Sci. Technol. 1987, 21, 202-208.
(21) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook of
Physical-Chemical Properties and Environmental Fate for
Organic Chemicals; Lewis Publishers: Chelsea, MI, 1993; Vol.
3.
(22) Hine, J.; Mookerjee, P. K. J. Org. Chem. 1975, 40, 292-297.
(23) McAuliffe, C. Chem. Technol. 1971, 1, 46-51.
(24) Meylan, W. M.; Howard, P. H. Environ. Toxicol. Chem. 1991, 10,
1283-1293.
(25) Hubaux, A.; Vos, G. Anal. Chem. 1970, 42, 849-855.
(26) Mass Spectrometry Data Centre. Eight Peak Index of Mass
Spectra, 2nd ed.; Pendragon House: Palo Alto, CA, 1974; Vol.
I-1.
(27) Ponec, V.; Bond, G. C. Catalysis by Metals and Alloys; Elsevier:
Amsterdam, 1995.
(28) Burris, D. R.; Campbell, T. J.; Manoranjan, V. S. Environ. Sci.
Technol. 1995, 29, 2850-2855.
(29) Deng, B.; Campbell, T. J.; Burris, D. R. Environ. Sci. Technol.
1997, 31, 1185-1190.
(30) Appleton, E. L. Environ. Sci. Technol. 1996, 30, 536A-539A.
(31) Sivavec, T. M.; Mackenzie, P. D.; Horney, D. P. Effect of site
groundwater on reactivity of bimetallic media: deactivation of
nickel-plated iron. Natl. Meet.sAm. Chem. Soc., Div. Environ.
Chem. 1997, 37 (1), 83-85 (Abstr.).
Acknowledgments
This work was partially funded through an undergraduate
research grant awarded to J.P.F. by the G. W. C. Whiting
School of Engineering at Johns Hopkins University. Additional
support was obtained through Contract F08637 95 C6037
awarded to A.L.R. by the Department of the Air Force as well
as through an N.S.F. Young Investigator Award (Grant BES-
9457260) to A.L.R. Bill Arnold conducted some followup
determinations and generously shared unpublished data
concerning ethylene reaction rates and Henry’s law constants
for ethylene, ethane, and cis-2-butene. Hugh Ellis provided
helpful advice with modeling GC/ MS spectra through linear
programming. We are also grateful for the comments of four
anonymous reviewers, whose suggestions greatly improved
this manuscript.
(32) Castro, C. E.; Kray, W. C. J. Am. Chem. Soc. 1966, 88, 4447-4455.
(33) March, J. Advanced Organic Chemistry: Reactions, Mechanisms,
and Structures, 3rd ed.; John Wiley & Sons: New York, 1985.
(34) Simmons, H. E.; Smith, R. D. J. Am. Chem. Soc. 1959, 81, 4256-
4264.
(35) Blanchard, E. P.; Simmons, H. E. J. Am. Chem. Soc. 1964, 86,
1337-1347.
Literature Cited
(1) Reed, D. J. Chlorocarbons and Chlorohydrocarbons. In Kirk-
Othmer Encyclopedia of Chemical Technology; Kroschwitz, M.
H.-G., Ed.; John Wiley & Sons: New York, 1993; Vol. 5.
(2) Snedecor, G. Other Chloroethanes. In Kirk-Othmer Encyclo-
pedia of Chemical Technology; Kroschwitz, M. H.-G., Ed.; John
Wiley & Sons: New York, 1993; Vol. 6.
(3) U.S. Environmental Protection Agency. 1,1,1-Trichloro-
ethanesATSDR Public Health Statement; Agency for Toxic
Substances and Disease Registry, EPA: Washington, DC, 1990.
(4) National Research Council. Alternatives for Ground Water
Cleanup; National Academy Press: Washington, DC, 1994.
(5) Gillham, R. W.; O’Hannesin, S. F. Ground Water 1994, 32, 958-
967.
(36) Simmons, H. E.; Blanchard, E. P.; Smith, R. D. J. Am. Chem. Soc.
1964, 86, 1347-1356.
(37) Kirmse, W. Carbene Chemistry; Academic Press: New York, 1964.
(38) Totten, L. A.; Roberts, A. L. Investigating electron-transfer
pathways during reductive dehalogenation reactions promoted
by zero-valent metals. Natl. Meet.sAm. Chem. Soc., Div. Environ.
Chem. 1995, 35 (1), 706-710 (Abstr.).
(39) Totten, L. A. Unpublished data.
(40) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic
Chemistry, 3rd ed.; Harper & Row: New York, 1987.
(41) Meisels, G. G.; Park, J. Y.; Giessner, B. G. J. Am. Chem. Soc. 1969,
91, 1555-1556.
(42) McFadden, W. H. J. Phys. Chem. 1963, 67, 1074-1077.
(43) Hardy, L. I.; Gillham, R. W. Environ. Sci. Technol. 1996, 30, 57-
65.
(6) Matheson, L. J.; Tratnyek, P. G. Environ. Sci. Technol. 1994, 28,
2045-2053.
(7) Boronina, T.; Klabunde, K. J.; Sergeev, G. Environ. Sci. Technol.
1995, 29, 1511-1517.
(8) Agrawal, A.; Tratnyek, P. G.; Stoffyn-Egli, P.; Liang, L. Processes
Affecting Nitro Reduction by Iron Metal: Mineralogical Con-
sequences of Precipitation in Aqueous Carbonate Environments.
Natl. Meet.sAm. Chem. Soc., Div. Environ. Chem. 1995, 35, 720-
723 (Abstr.).
Received for review September 3, 1997. Revised manuscript
received February 18, 1998. Accepted March 2, 1998.
ES970784P
9
1 9 8 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 13, 1998