Congener-specific dechlorination of dissolved PCBs by microscale and nanoscale zerovalent iron in a water/methanol solution

Article abstract of DOI:10.1021/es049835q

Polychlorinated biphenyl (PCB)-contaminated sediments remain a significant threat to humans and aquatic ecosystems. Dredging and disposal is costly, so viable in situ technologies to dechlorinate PCBs are needed. This study demonstrates that nanoscale zerovalent iron (ZVI) dechlorinates PCBs to lower-chlorinated products under ambient conditions, provides insight into structure-activity relationships between PCB isomers, and compares the reactivity of nanoscale ZVI to that of palladized microscale ZVI. Six PCB congeners were studied (22′, 34′, 234, 22′35′, 22′45′, and 33′44′) to compare the initial rate of dechlorination of each and to monitor the order in which chlorines are removed. Using 200 g/L of nanoscale ZVI in a 30% MeOH/water mixture, observed surface-area-normalized pseudo-first-order PCB dechlorination rate constants ranged from 1 × 10^-6 to 5.5 × 10^-4 L yr ^-1 m^-2 depending on the PCB congener tested. Using 200 g/L of palladized (0.05 wt %) microscale ZVI, surface-area-normalized pseudo-first-order PCB dechlorination rate constants were significantly faster and ranged from 3.8 × 10^-2 to 1.7 × 10^-1 L yr^-1 m^-2, but these rates were not sustainable. For nanoscale ZVI, nonorthosubstituted congeners had faster initial dechlorination rates than orthosubstituted congeners in the same homologue group. Chlorines in the para and meta position were predominantly removed over chlorines in the ortho position, which suggests that more-toxic coplanar PCB congeners are not likely to form from less-toxic noncoplanar, orthosubstituted congeners. Complete dechlorination was not observed over the course of the experiments. PCB dechlorination is rapid enough that nanoscale ZVI may offer novel in situ remedial alternatives for PCB-contaminated sediments.

Full text of DOI:10.1021/es049835q

Environ. Sci. Technol. 2004, 38, 5208-5216
disposal of the dredged material in a landfill, or stabilization
Congener-Specific Dechlorination of
Dissolved PCBs by Microscale and
Nanoscale Zerovalent Iron in a
Water/Methanol Solution
(e.g., addition of silicate binders or cement) to reduce the
PCB mobility (1). Extraction and stabilization technologies
are costly and most dredged contaminated sediment is
disposed of in landfills. The cost to dredge and treat ∼2.7
million cubic yards of PCB-contaminated sediment from a
∼40 mile section of the Upper Hudson River is estimated at
nearly half a billion dollars (2). Lower-cost passive treatment
alternatives to remediate PCB-contaminated sediment in situ,
or to treat PCBs in the dredged material, are needed.
Few passive technologies to treat PCB-contaminated
sediments have been studied other than microbial degrada-
tion. Anaerobic bacteria may slowly dehalogenate higher
chlorinated congeners with unpredictable success, and some
congeners are resistant to anaerobic microbial dechlorination
(3, 4). The most common anaerobic dechlorination pathway
is meta, followed by para chlorine removal (5); however,
enrichment cultures derived from estuarine sediment have
recently been shown to dechlorinate ortho chlorines and
doubly flanked chlorines (6, 7). Ortho dechlorination is
generally limited though, and a build up of ortho-substituted
congeners such as 2, 22′ and 26-dichlorobiphenyl results (5,
8-12). Lesser-chlorinated congeners, typically mono-, di-,
or trichlorobiphenyls, can be microbially metabolized or
mineralized under oxic conditions (5, 8-10, 13, 14). This
suggests that complete microbial destruction of PCBs requires
that higher-chlorinated congeners (>3 chlorines) be partially
dechlorinated anaerobically to lower-chlorinated congeners
(<3 chlorines) before being metabolized by aerobes. Because
anaerobic microorganisms degrade higher-chlorinated con-
geners very slowly, identifying a nontoxic medium that
accelerates the rate of dechlorination of higher-chlorinated
PCBs congeners to lower-chlorinated congeners is desirable.
It is also desirable for the medium to preferentially dechlo-
rinate recalcitrant and toxic coplanar congeners, thereby
decreasing the exposure risk to biota.
Microscale zerovalent iron (ZVI) filings (µm to mm
particles) effectively dechlorinate many halogenated hydro-
carbon compounds (15-18) and are employed in permeable
reactive barriers at hundreds of pilot-scale and full-scale
remediation sites throughout the United States (19). Un-
fortunately, at ambient conditions, PCB dechlorination has
not been observed using microscale ZVI (e.g., 8-50 mesh
Peerless or <40 mesh electrolytic Fisher) (20, 21). Microscale
ZVI has only been shown to dechlorinate Aroclor mixtures
at elevated temperatures (>250 °C) (22-24), rendering this
impractical for sediment remediation. Palladized ZVI (Pd/
Fe0) has also been evaluated as a potential reactive medium
to treat PCB-contaminated sediments and process waters.
Small additions of palladium (0.05-0.25 wt %) to microscale
ZVI enables it to rapidly dechlorinate PCBs (20, 25), but this
dramatically increases the cost. It is therefore desirable to
find the lowest level of palladium that still effectively
dechlorinates PCBs. The longevity of the rate enhancement
provided by Pd is uncertain.
G R E G O R Y V . L O W R Y* , † A N D
† , ‡
K A T H L E E N M . J O H N S O N
Department of Civil & Environmental Engineering, Carnegie
Mellon University, Pittsburgh, Pennsylvania 15213-3890, and
SAIC, 3320 North Buffalo Drive, Suite 130,
Las Vegas, Nevada 89129
Polychlorinated biphenyl (PCB)-contaminated sediments
remain a significant threat to humans and aquatic ecosystems.
Dredging and disposal is costly, so viable in situ
technologies to dechlorinate PCBs are needed. This study
demonstrates that nanoscale zerovalent iron (ZVI)
dechlorinates PCBs to lower-chlorinated products under
ambient conditions, provides insight into structure-activity
relationships between PCB isomers, and compares the
reactivity of nanoscale ZVI to that of palladized microscale
ZVI. Six PCB congeners were studied (22′, 34′, 234,
22′35′, 22′45′, and 33′44′) to compare the initial rate of
dechlorination of each and to monitor the order in which
chlorines are removed. Using 200 g/L of nanoscale ZVI in a
30% MeOH/water mixture, observed surface-area-
normalized pseudo-first-order PCB dechlorination rate
constants ranged from 1 × 10-6 to 5.5 × 10-4 L yr-1 m-2
depending on the PCB congener tested. Using 200 g/L
of palladized (0.05 wt %) microscale ZVI, surface-area-
normalized pseudo-first-order PCB dechlorination rate
constants were significantly faster and ranged from 3.8 ×
10-2 to 1.7 × 10-1 L yr-1 m-2, but these rates were not
sustainable. For nanoscale ZVI, nonorthosubstituted
congeners had faster initial dechlorination rates than
orthosubstituted congeners in the same homologue group.
Chlorines in the para and meta position were predominantly
removed over chlorines in the ortho position, which suggests
that more-toxic coplanar PCB congeners are not likely
to form from less-toxic noncoplanar, orthosubstituted
congeners. Complete dechlorination was not observed
over the course of the experiments. PCB dechlorination is
rapid enough that nanoscale ZVI may offer novel in situ
remedial alternatives for PCB-contaminated sediments.
Introduction
Recently, nanoscale (<1 µm) ZVI and nanoscale bime-
tallics (e.g., Pd/ Fe0 and Ni/ Fe0) have been investigated as
remedial agents for chlorinated organic compounds and toxic
metals (21, 26-28). Nanoscale ZVI particles have a larger
specific surface area, and have higher surface reactivity than
microscale ZVI (27). There is just one published study on the
reactivity of PCBs with nanoscale ZVI. Wang and Zhang (21)
investigated the dechlorination of Aroclor 1254 in an ethanol/
water mixture at ambient conditions using unmodified
nanoscale ZVI and nanoscale Pd/ Fe0. The Pd/ Fe0 (∼50 g/ L)
resulted in complete dechlorination of Aroclor 1254 after 17
Historical and continued releases of polychlorinated bi-
phenyls (PCBs) to the environment have resulted in polluted
water, soil, and sediment. Several costly ex situ remediation
techniques are available to treat these media, such as dredging
and PCB extraction (e.g., soil washing, solvent extraction,
and thermal desorption) to minimize the waste volume,
* Corresponding author phone: (412)268-2948; fax: (412)268-7813;
e-mail: glowry@cmu.edu.
† Carnegie Mellon University.
‡ SAIC.
9
5 2 0 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004
10.1021/es049835q CCC: $27.50
2004 Am erican Chem ical Society
Published on Web 08/25/2004

h. In contrast, unpalladized nanoscale ZVI (∼50 g/ L) resulted
in <25% dechlorination over the same time period. Biphenyl
was detected in both cases, suggesting complete dechlori-
nation is possible using unmodified nanoscale ZVI. Because
Aroclors are mixtures of PCB congeners, little can be
determined from these experiments about the reactivity
trends of individual positional isomers or between different
PCB homologue groups. Because dioxinlike toxicity has been
reported in coplanar PCB congeners containing both para
and at least two meta substitutions (33′44′, 344′5, 33′44′5,
and 33′44′55′) (29, 30), it is important to predict if a buildup
of more-toxic coplanar congeners would result from the iron-
mediated reduction of PCBs under ambient conditions.
In this study, the PCB dechlorination rate and product
distributions afforded by unmodified nanoscale ZVI were
measured in water/ cosolvent batch reactors. No sediment
or soil was used in these studies to minimize confounding
factors such as PCB losses due to sorption or background
contamination. The research objectives were (i) to conclu-
sively demonstrate that nanoscale ZVI, without the addition
of a noble metal such as Ni, Pt, or Pd, can dechlorinate a
range of PCB congeners at ambient conditions, and (ii) to
determine the relative reactivity of positional isomers within
a homologue group and provide insight into structure-
activity relationships that may exist. Six strategically chosen
PCB congeners were used. Tetrachlorobiphenyl congener
22′45′ (BZ 49) has proven resistant to microbial dechlorination
in the Hudson River (3). Its more toxic coplanar isomer 33′44′
(BZ 77) is present in Aroclor mixtures at concentrations 10-
100 times greater than other toxic coplanar congeners (30,
31). Dechlorination of 22′35′ (BZ 44) by microscale ZVI and
microscale Pd/ Fe0 at ambient temperatures has been studied
(20) and provides a basis for comparing PCB reactivity in
this study. Anaerobic microbial PCB dechlorination has been
shown to result in a buildup of ortho-substituted dichloro-
biphenyl congeners such as 22′ (BZ 4) (32, 5). Comparing the
reactivity of ortho-substituted noncoplanar isomers with the
nonortho-substituted coplanar isomers will yield additional
information about their relative reactivity. The reactivity of
congener 234-trichlorobiphenyl (BZ 21), containing one
ortho, meta, and para chlorine was compared with the
reactivity of the di- and tetrachlorobiphenyls. Finally, the
feasibility of different schemes employing nanoscale
zerovalent iron for remediating PCB-contaminated sediments
is discussed.
washes of 0.37 M hydrochloric acid, acetone, and histological
grade methanol using 1 mL liquid for every 2 g ZVI and dried
in a 106-°C oven for 2 h under nitrogen gas. Palladized ZVI
(0.05 wt %) was prepared from acid-treated Peerless iron. A
0.5 mM solution of Pd(II)C4H6O4 was prepared in 1 L of
histological grade MeOH. Treated Peerless iron filings (∼100
g) were added and the mixture was stirred in a closed flask
for 6 h on an orbital shaker at 90 rpm. The liquid was decanted
and the palladized Peerless Fe(0) was oven-dried at 106 °C
for 3 h under nitrogen. It is assumed that all the palladium
was coated on the iron to give 0.05% Pd by weight.
Nanoscale ZVI preparation. Nanoscale ZVI was synthe-
sized using a slightly modified method as previously described
(21, 26). The primary differences between this method and
previous methods are that synthesis was performed in a hood
rather than in an N2 atmosphere (except for drying), a fivefold
lower FeSO4 concentration was used (0.07 M instead of 0.36
M), a NaBH4 solution was used instead of adding solid NaBH4,
and the particles were centrifuged and over-dried in N2 rather
than filtered from solution and vacuum-dried. The synthesis
is briefly described here. Complete synthesis details are
provided in the Supporting Information. Into 1 L of a 30%
(volume) MeOH/ deionized water solution was dissolved 20
g of FeSO4‚7H2O. While stirring, 10 mL of 5 N NaOH aqueous
solution was added dropwise to the dissolved iron solution,
yielding a pH of 6.1. Next, 50 mL of a 2.1 M NaBH4 aqueous
solution was added at a rate ∼0.5 mL/ sec. This is ap-
proximately 2.9 times the stoichiometric amount of NaBH4
required to reduce all the dissolved Fe2+ in solution to
zerovalent iron. A fine black precipitate formed instantly upon
addition of NaBH4 and remained in solution throughout the
process. The mixture was stirred for 20 min before being
centrifuged in plastic centrifuge tubes for 5 min at 3500 rpm
to separate the solids. The supernatant was discarded and
the remaining solids were rinsed with MeOH twice to remove
any excess salt. The nanoscale ZVI particles were then dried
in a 106-°C oven for 4 h under nitrogen gas. Once dry, the
particles remained in the oven for a minimum of 8 h to cool
and allow oxygen to slowly bleed into the oven. This
passivated the iron surface slightly and prevented the highly
reactive iron particles from spontaneously igniting. The dried
particles were finely ground and stored in sealed serum
bottles until use. A second batch of iron was synthesized
similarly, but dried in a glovebox containing 99.5% N2/ 0.5%
O2 rather than pure N2.
Nanoscale Iron Characterization. Particle size and
electron diffraction patterns for the nanoscale iron were
collected using a Philips EM420 transmission electron
microscope (TEM) at 120 kV. The morphology of the
nanoscale ZVI particles was determined using the JEM-ARM
1000 Atomic Resolution Microscope (JEOL, Ltd.) at the
National Center for Electron Microscopy at Lawrence Ber-
keley National Laboratory. The ARM was operated at 800 kV,
which allowed magnification levels of up to 1 000 000 ×. ZVI
samples were dispersed (sonicated) in acetone, then dripped
onto a 300-mesh Cu TEM grid with Formvar substrate or
lacey carbon. The N2-BET surface area of the nanoscale ZVI
was measured using a NOVA 2200 BET-surface area analyzer
(Quantachrome, Boynton Beach, FL). A five-point BET
isotherm was used to determine the total available surface
area of the iron. The elemental composition of nanoscale
ZVI was determined by Inductively Coupled Plasma-Atomic
Emission Spectrometry (ICP-AES) according to EPA Method
6010B.
Materials and Methods
Chem icals. All PCB congeners studied (22′, 34′, 234, 22′35′,
22′45′, and 33′44′), their dechlorination byproducts, and the
internal standard (22′33′) were 97+% pure and obtained from
Ultra Scientific (N. Kingstown, RI). Standard solutions in
hexane were used for gas chromatograph (GC) calibration
and neat congeners were used in batch experiments without
further purification. The microscale ZVI materials tested were
Fisher Scientific electrolytic powder (<100 mesh) and Peerless
iron filings (8-50 mesh). Hydrochloric acid (37.6%), acetone
(99.5+%), and methanol (histological grade) were used to
pretreat the Peerless iron filings obtained from Fisher. Some
filings were palladized using Pd(II)(CH3CO2)2 (47.5 wt % Pd,
Acros). Nanoscale ZVI particles were synthesized from FeSO4‚
7H2O (99.4%, Fisher) and NaBH4 (98+%, Acros). Methanol
(histological grade), sodium azide (99%), and 1,1,1,3,3,3-
hexamethyldisilazane (HMDS) (98%) to silanize the serum
bottles were obtained from Acros. Hexane (99.9+%), metha-
nol (pesticide grade), and anhydrous Na2SO4 (10-60 mesh)
used to extract PCBs from the reactors for analysis were
obtained from Fisher. The Na2SO4 was oven-dried at 106 °C
for a minimum of 4 h prior to use.
Sacrificial Batch Experim ents. PCB spike solutions were
prepared by dissolving a known mass of neat PCB congener
in acetone or methanol. The concentration of the parent
compound and any unwanted PCB congeners present (i.e.,
impurities) in each spike solution were determined using a
Hewlett-Packard 6890 gas chromatograph with an electron
Microscale Iron Preparation. Fisher electrolytic iron was
used as received. Peerless iron was treated with separate
9
VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 2 0 9

TABLE 1. GC Operating Parameters
parameter
GC/µECD
GC/FID
30 m HP-5MS
GC/MSD
colum n
60 m HP-5
30 m PTE-5
d ) 250 µm , film ) 0.25 µm
d ) 250 µm , film ) 0.25 µm
d ) 250 µm , film ) 0.25 µm
He (1 psi)
N/A
carrier gas (pressure/rate)
detector gases (rate)
He (20 psi)
He (2.1 m L/m in)
Ar/(5%) CH4 (60 m L/m in)
H2 (45.0 m L/m in)
air (320.0 m L/m in)
He (10.0 m L/m in)
detector tem p
325 °C
300 °C
280 °C
inlet m ode (tem p/press)
oven tem perature
program (run tim e)
splitless (250 °C/20 psi)
115 °C for 3 m in, 5 °C/m in
to 230 °C, 15 °C/m in to
290 °C, 30°C/m in to
300 °C, hold 300 °C for
10 m in (40 m in)
splitless (250 °C/21 psi)
60 °C for 1 m in, 20 °C/ m in
to 220 °C, 30 °C/m in to
280 °C, hold 280 °C
for 4 m in (15 m in)
splitless (250 °C/1 psi)
105 °C for 3 m in, 5 °C/ m in
to 230°C, 15 °C/m in to 290°C,
30 °C/m in to 300 °C,
hold 300 °C for 5 m in (38 m in)
capture detector (GC/ ECD). An aliquot of the spike solution
(∼ 0.1 to 1 mL) was added to 500 mL of a 30 vol % methanol/
water mixture to provide the desired initial PCB concentration
for each experiment without exceeding the solubility limit
in the water/ cosolvent mixture (33). Sodium azide (∼1 mM)
was added to inhibit microbial activity and the solution was
stirred on an orbital shaker for a minimum of 1 h prior to
spiking the reactors. For each PCB congener tested, 8 reactors
(12-mL amber serum bottles) were prepared by placing 2 g
of iron into each bottle, adding 10 mL of solution containing
the desired PCB congener, and capping with Teflon-coated
rubber septa and aluminum seals. Control experiments were
run identically but without addition of iron. The reactors
were placed on an orbital shaker (90 rpm) or an end-over-
end rotator at 20 rpm and kept at laboratory temperature (22
( 3 °C) until sacrificed for analysis. To minimize PCB
adsorption to glass, the serum bottles were silanized in a
hexane solution with 5% HMDS by volume for 1 h, rinsed
with DI water, and air-dried prior to use. Approximately 2
mL of headspace remained in each bottle.
Reactors were sacrificed at predetermined times for
analysis. PCB sorption to iron was significant, so both the
aqueous and solid phases in each reactor were extracted to
effectively search for dechlorination byproducts. Extractions
were conducted in the reactor bottles without removing the
septa. Experiments using microscale ZVI or Pd/ Fe0 were
extracted using the following procedure. The serum bottle
was centrifuged for 5 min at 3500 rpm to separate the solid
iron from the water/ suspended iron. A 7-mL aliquot of the
aqueous phase was withdrawn from the reactor and extracted
with 0.5 mL of hexane containing 1 mg/ L of the internal
standard 22′33′ (extraction solution) by mixing on a vortex
orbital mixer. Approximately 100 µL of the hexane extract
was drawn off and added to an autosampler vial containing
a 100-µL glass insert and ∼6 mg of oven-dried anhydrous
sodium sulfate, and analyzed for PCBs using GC/ µECD. To
the wet iron remaining in the bottle, approximately 1 mL of
pesticide-grade MeOH was added and shaken on the vortex
mixer. Next, 0.5 mL of the extraction solution (hexane/ internal
standard) was added and the bottle, vortexed, and analyzed
for PCBs as above. Reactors containing nanoscale iron were
analyzed in a similar fashion but the iron particles were not
separated from the aqueous phase. Instead, 1 mL of pesticide
grade methanol was added to the reactor, mixed on the vortex
mixer, followed by addition of 0.5 mL of the extraction
solution, shaking, and transfer to a GC vial for analysis.
Experiments using microscale ZVI were conducted once.
Experiments using nanoscale ZVI were repeated to verify
results and assess the relative reactivity of different batches
of nanoscale ZVI. Control reactors (without iron) were
extracted using 0.5 mL of hexane only (no methanol).
PCB Analysis and Calibration. Analysis was by GC/ µECD
(for PCBs), GC/ FID (for biphenyl), and GC/ MSD (Model
5972). The GC operating parameters for each are provided
(Table 1). In all analyses, a 1-µL aliquot of sample was injected
using a Hewlett-Packard liquid autosampler (Model 18593B).
These operating parameters provided excellent separation
of the PCB congeners used as well as of the dechlorination
byproducts, except where noted. When necessary, the
identities of PCB dechlorination byproducts were verified
with GC/ MSD. The detector was tuned using the standard
spectra autotune to compare the mass spectra with the NIST/
EPA/ NIH Chemical Structures Database (library NBS75K.L)
with >90% quality.
The GC/ µECD was calibrated for all PCB congeners tested
and all potential dechlorination products using 22′33′ as an
internal standard (IS). Response factors were determined for
each congener and the internal standard. Regressions (R2 >
0.99) were determined from a three-point calibration of
biphenyl, a three-point calibration of monochlorobiphenyls,
a four-point calibration of dichlorobiphenyls, and either a
four or five-point calibration for tri- and tetrachlorobiphenyls.
The approximate detection limits were 1 nmol for biphenyl
and 3-chlorobiphenyl, 0.1 nmol for 2, 4, 22′, 33′, and 34′, and
0.01 nmol for all other congeners.
In experiments using 22′45′, the potential byproducts 22′5
and 34′ (resulting from loss of the ortho chlorines) coelute.
It was assumed that no 34′ was formed due to loss of the
ortho chlorines and only 22′5 constituted the peak at 27.9
min. In experiments with Pd/ Peerless Fe0, it was assumed
congeners 24 and 25 coeluted and had the same response
(i.e., each congener made up half of the peak area).
System Mass Balance. The mass balance at each sampling
event was calculated as the ratio of the total PCB congeners
measured to the initial PCB mass measured in the system
Mi
∑
(
)
MB (%) )
× 100
(1)
Mo, t ) 0
where Mi is the mass of all PCBs in solution at time t including
all dechlorination byproducts and the parent compound,
and Mo, t)0 is the mass of the parent congener at the initial
sampling event in µmol. In experiments where aqueous and
iron phases could be separated, PCB masses in the aqueous
and iron extracts were summed to calculate the total mass
in the reactor.
PCB Dechlorination Rate Constants. The pseudo-first-
order PCB dechlorination rate constants were determined
from the rate of product formation rather than loss of the
parent congener. This is primarily because (i) conversion of
the parent PCB congener over the 1.5-to-6-month reaction
time was low (40% or less), (ii) nondestructive loss of mass
of the parent PCB compound due to irreversible sorption is
likely, and (iii) no products were observed in control reactors
without iron. The observed pseudo-first-order rate constants
9
5 2 1 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004

TABLE 2. Iron Physical and Chemical Properties
particle size (nm)
N2 BET-surface area (m2/g)
elemental composition (wt %)
nanoscale ZVI
30-50 (∼1 µm aggregates)
36.5
Fe 60
B 4
O 36a
Fe > 99
Fe 98
C ∼2c
fisher electrolytic iron
peerless iron
<150 000 (<100 m esh)
0.18b
1.23b
2 300 000 < dp < 300 000 (8 to 50 m esh)
a
Only Fe and B were m easured. The balance of the com position is assum ed to be O in the Fe-oxide shell. The presence of O has been
qualitatively determ ined using EDX and no other elem ents were detected in the particles. Alowitz and Scherer (44). From Peerless product
inform ation sheet.
b
c
were calculated from fits of the products formed in each
reactor using eq 2
d
Products
dMI
dt
∑
-
)
) kobsMi
(2)
dt
where Mi is the mass of the parent PCB congener, ∑products
is the sum of all observed products, and kobs is the observed
pseudo-first-order rate constant. Fits were performed using
a kinetic modeling software package, Scientist, v. 2.01
(Micromath, St. Louis, MO). Reported errors are 95%
confidence intervals for these fits. All product data was used
for the nanoiron experiments. Products data collected prior
to deactivation (t <3 days) was used to calculate PCB
dechlorination rates using Pd/ Fe.
Results
Iron Characterization. TEM images of the N2-dried nanoscale
ZVI particles (before reaction) are shown in Figure 1a and
1b. The N2-BET surface area and elemental composition of
all particles used in this study are provided in Table 2. The
nanoscale ZVI exists as aggregates of spherical particles
averaging 30-50 nm in diameter. The specific surface area
of nanoscale ZVI particles is ∼2 orders of magnitude greater
than for the microscale iron particles. Assuming spherical
particles and identical densities for each type of iron, the
40-nm-diameter nanoscale ZVI particles should have ap-
proximately a 4 order of magnitude larger surface area-to-
volume ratio (3/ R) than the other particle types, so the
measured surface area of 36.5 m2/ g is not unreasonable. It
is also consistent with previously reported measurements
(21, 26, 27). The nanoscale ZVI particles appear to have a
core-shell morphology (Figure 1b). The particles contain
∼60 wt % iron and ∼4 wt % boron, also consistent with other
reported studies (26). The remaining 36 wt % of the particle
mass is presumed to be atomic oxygen in Fe-oxides. The d
spacings for electron diffraction patterns [0.201 nm (110),
0.119 nm (211), and 0.145 nm (200)] are consistent with d
spacings for the three strongest reflections of bcc Fe0 [0.203
nm (110), 0.117 nm (211), and 0.143 (200)] indicating that
R-Fe(0) is present in the particles. The presence of oxygen
atoms was verified qualitatively using EDX and no other
elements were present in substantial quantities. It is therefore
likely that the core of these particles is R-Fe(0) while the shell
consists mostly of Fe-oxide. Nanoscale ZVI synthesized in a
similar fashion had an enrichment of boron (oxidized) in the
surface layers (∼11%) (26). It is assumed that the boron
distribution and speciation in these particles is similar. The
nature and extent that boron influences particle reactivity is
unclear and is under investigation.
FIGURE 1. (a) Aggregate spherical particles of nanoscale ZVI (50
000×). (b) Core-shell morphology of nanoscale ZVI particles (250
000×).
were formed in reactors containing Pd/ Peerless Fe0 and
nanoscale ZVI. Figure 2 shows the byproduct distributions
from dechlorination of 22′35′ with Pd/ Peerless Fe0. Figure 3
shows the total mass of the products formed during dechlo-
rination of each of the congeners evaluated by unmodified
nanoscale ZVI, and the corresponding data fits using eq 2.
Figure 4 shows the loss of 33′44′ and byproducts formed
during dechlorination using unmodified nanoscale ZVI, along
with the corresponding data fits using eq 2. The surface-
area-normalized PCB dechlorination rate constants and
corresponding reactor-specific halflives for all PCB congeners
tested are provided in Table 3. The dechlorination rate
constants in Table 3 were all determined using the same ZVI
(glovebox-dried) so the reactivity of different congeners is
directly comparable.
PCB Dechlorination Rates Using Different form s of ZVI.
No products formed in control reactors without iron. No
PCB dechlorination products were observed in batch reactors
with microscale Fisher electrolytic iron powder or Peerless
Fe0 filings after 180 days, suggesting that these forms of iron
are not reactive. Quantifiable PCB dechlorination products
System Mass Balance. Mass balances for PCBs were
calculated for all experiments. It is assumed that the extraction
efficiencies were the same for all PCB congeners. PCB mass
9
VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 2 1 1

FIGURE 2. Byproduct distribution during dechlorination of PCB congener 22′35′ using Pd/Peerless Fe0 (0.05 wt % Pd). Lines on larger plot
are data fits using eq 2. Lines on inset are interpolated (not data fits) and only meant to guide the eye.
FIGURE 3. Byproduct distribution during dechlorination of PCB congeners 22′35′, 22′45′, 234, and 2,2′ using glovebox-dried (99.5% N2/0.5%
O ) nanoscale ZVI. Lines represent data fits using eq 2.
2
balances in experiments with Pd/ Peerless Fe0 ranged from
36 to 70%. PCB mass balances with Fisher electrolytic iron,
which lacks carbon inclusions, were higher at 50-90%. PCB
mass balances of 56-104% were obtained in nanoscale ZVI
experiments. With Peerless Fe0 that is 2% carbon, PCBs may
not have been efficiently extracted from these carbon
inclusions, resulting in poor mass balances after 4 days.
Because PCBs are likely to undergo stepwise reductive
dechlorination with ZVI (23, 24), the formation of soluble
reaction products that would not be measured with GC/
µECD is doubtful. During the experiment, a small fraction
of PCBs would have volatilized into the headspace. Using
estimates of Henry’s constant (34), this fraction is , 1% and
does not account for the total mass deficit. Microbial
degradation is doubtful owing to the presence of NaN3. By
using amber glass, the likelihood of PCB photodegradation
is small. Preliminary studies showed less PCB sorption to
glass reactors that were silanized, but sorption was not
eliminated as has been documented in other studies (35).
Control reactors that lacked iron also showed some loss of
the spiked PCB congener (up to 50%), but no products were
formed. The less than 100% mass balance observed in these
experiments is therefore likely due to irreversible sorption
of the parent congener to glassware and the Teflon-lined
septa. It should be noted that nondestructive loss of the
dechlorination products would bias the observed dechlo-
rination rate constants reported here toward underestimating
the actual PCB dechlorination rate constants, however, the
order of magnitude differences in the observed rates (e.g.,
22′ vs 34′ and 22′35′/ 2,2′4,5′ and 33′44′) exceed the potential
error due to nondestructive mass loss of PCB dechlorination
products.
Discussion
Microscale ZVI. The lack of reactivity of microscale ZVI (Fisher
and Peerless) after 180 days is reasonable considering that
its surface area is ∼2 orders of magnitude lower than that
of nanoscale ZVI. It has been demonstrated that the ZVI-
promoted reaction rate is first order with respect to the iron
surface area (36). The surface-area-normalized PCB dechlo-
rination rate constant for 33′44′ using nanoscale ZVI is 5.5
( 1.4 × 10-4 yr-1 m-2 L. Based on this rate constant, the
9
5 2 1 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004

FIGURE 4. Loss of parent compound and byproduct distribution during dechlorination of toxic, coplanar PCB congener 33′44′ using
glovebox-dried (99.5% N2/0.5% O ) nanoscale ZVI. PCB mass balance is 104%. Lines represent data fits using eq 2.
2
likely to be economical given the cost and the current inability
TABLE 3. Measured Surface Area normalized PCB
to provide long-term dechlorination rate enhancement.
Dechlorination Rate Constants and Reactor-Specific Half-Life
Nanoscale ZVI. Nanoscale ZVI had a greater specific
Times
surface area than microscale iron and was more reactive.
0.05 wt % Pd/Fe0
nanoscale ZVIa
Additionally, gas bubbles (likely H2) were observed in these
reactors but not with the microscale iron. The high surface-
to-volume ratio and the rapid formation of hydrogen gas
may have enabled nanoscale ZVI to dechlorinate PCBs during
the 45-day experiment. The presence of B (4 wt %, ∼18 at.
%) may also play a role if an Fe-B alloy or Fe-B-oxide is
formed that significantly alters the electronic properties of
the oxide shell. It is also possible that the nanoiron reactivity
(compared to microscale iron) is due to the small particle
size. The observed primary particle size (∼50 nm) of Fe/ B is
probably not small enough to increase the Fermi potential
and lower the potential of the Fe2+/ Fe0 redox couple, as has
been reported for silver nanoparticles with low aggregation
numbers (ca. < Ag100) (38), but size effects may be possible
if the 50-nm particles are in fact aggregates of very fine clusters
of Fe0 atoms with similarly low aggregation numbers.
While the data set collected here is not extensive enough
to support strong conclusions regarding structure-activity
relationships, some general trends are apparent. Considering
PCB congeners containing two ortho chlorines (22′, 22′35′,
and 22′45′), the dechlorination rate increased with increasing
chlorine substitution. Much less variability in the PCB
dechlorination rate was observed with the Pd/ Peerless Fe0.
Furthermore, nonortho-substituted congeners are signifi-
cantly more reactive than their ortho-substituted isomers
(34′ vs 22′ and 33′44′ vs 22′35′/ 22′45′). The highest chlorinated,
nonortho-substituted congener (33′44′) was dechlorinated
faster and to a greater extent than any other congener studied.
The reactivity of the trichloro-PCB congener (234) was similar
to a dichloro nonortho-substituted congener (34′) and a
tetrachloro diortho-substituted congener (22′35′). This con-
gener contains one ortho chlorine rather than two, so its
relative reactivity may be affected by both the number of
chlorine substitutions and the number of ortho chlorines.
The number of orthochlorines present on congeners may
affect reactivity because the torsion angles between the
biphenyl ring increases with increasing ortho substitution,
38° (nonortho), 58° (mono-ortho), and 73° (di-ortho) when
they are sorbed (39). Nonortho-substituted congeners may
be able to adsorb in a closer planar position with iron, which
kobs 102
t1/2
(yr)
kobs 105
t1/2
(yr)
congener
(L yr-1 m-2
)
(L yr-1 m-2
)
22′
34′
234
22′35′
22′45′
33′44′
3.8 ( 1.4
17 ( 10
5 ( 2
0.07
0.02
0.06
0.02
0.02
0.1 (est.)c
1.8 ( 1.8
3.7 ( 0.4
2.6 ( 0.1
9.3 ( 0.1
55 ( 14
77
5.3
2.6
3.6
1.0
0.2
14 ( 5
12.6 ( 6
NDb
ND
Glovebox-dried Nanoscale ZVI. Not determ ined. c Estim ated from
trace of 2-CBP observed. Reported errors are 95% confidence intervals
for pseudo-first-order fit of product data (eq 2).
a
b
expected pseudo-first-order rate constant for 33′44′ in a
reactor containing 200 g L-1 of Peerless Fe0 (N2-BET surface
area ) 1.23 m2/ g) is 0.14 ( 0.03 yr-1. Given this rate constant,
some PCB dechlorination products (∼10% of the initial PCB
mass) should have been measurable after 180 days had
Peerless Fe0 been capable of dechlorinating PCBs. No
byproducts were observed in reactors containing 33′44′ and
unmodified Peerless Fe0, suggesting that factors other than
iron surface area prevented PCB dechlorination from oc-
curring. The fastest observed PCB dechlorination rates were
achieved using Pd/ Peerless Fe0. These rates are similar to
previously reported rates by palladized iron (20). Palladium
likely serves as a catalyst, lowering the activation energy of
the reaction (25). The rapid dechlorination rate, however, is
only sustained for 2 days (Figure 2). After this initial period,
the reactivity appears to cease in all experiments and product
levels remain unchanged for a month. This trend was
observed for all experiments using Pd/ Peerless Fe0. A similar
trend was noted in a previous study using 0.25 wt % Pd/
Fisher electrolytic (<40 mesh) iron but was not discussed
(20). Loss of reactivity could result from passivation of the
palladium surface or poisoning (37). Assuming a historical
bulk iron cost of $350/ ton and a bulk Pd market price of
$14.75/ gram, the additional materials cost of palladized ZVI
filings (at 0.05 wt %) to unpalladized ZVI is approximately
20:1. The use of palladized iron for PCB remediation is not
9
VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 2 1 3

FIGURE 5. (a) Chlorine removal patterns after 2 days with 0.05% Pd/Peerless Fe(0). (b) Chlorine removal patterns after 45 days with
glovebox-dried nanoscale ZVI.
may be more favorable for reductive dechlorination. The
increased reactivity of higher-chlorinated PCB congeners and
nonortho-substituted congeners correlates with the more
positive reduction potentials measured in previous studies
for such congeners (40-43). These observations suggest that
(1) the measured reduction potentials may be useful in
predicting trends in the dechlorination rate constants of PCBs
using nanoscale ZVI, (2) higher-chlorinated congeners should
dechlorinate faster than lower-chlorinated congeners, and
(3) nonortho-substituted congeners should dechlorinate
faster than their ortho-substituted isomers.
Effect of Iron Drying Atm osphere. The dechlorination
rate constant for 22′35′ using N2-dried nanoscale ZVI was 2.3
yr-1 with a halflife of 0.3 years (data not shown). For glovebox-
dried nanoscale ZVI, the dechlorination rate constant for
22′35′ was only 0.19 yr-1 with a halflife of 3.6 years (Table 2).
The decrease in reactivity of glovebox-dried nanoscale ZVI
is likely due to the formation of a less-active or thicker oxide
layer during drying (Figure 1b). Thus, the synthesis method
of the nanoscale ZVI particles, and exposure to oxygen, can
affect the particle reactivity. The dechlorination product
distribution, however, was the same regardless of the iron
synthesis/ drying technique so conclusions regarding po-
tential dechlorination products are unaffected.
each PCB congener with Pd/ Peerless Fe0 are given in Figure
5a. Based on byproduct formation from dechlorination of
tri- and tetrachlorobiphenyls, meta and para chlorine removal
was dominant but some ortho dechlorination did occur.
Furthermore, para chlorines may be preferentially removed
over meta chlorines for 234 and 22′45′. For the dichloro-
biphenyls, some ortho dechlorination was observed with 22′.
Nearly equal fractions of meta and para chlorine removal
occurred with 34′. For Pd/ Peerless Fe0, nonortho chlorines
(para or meta) are preferentially removed over ortho chlorines
when both are present, but ortho dechlorination is possible.
No biphenyl was observed in these systems, suggesting that
complete dechlorination is not likely at the low Pd loading
used (0.05 wt % Pd). This is consistent with previous
observations (20).
Similar data for PCB dechlorination after 45 days using
glovebox-dried nanoscale ZVI are shown in Figure 5b. In the
ortho-substituted tri- and tetrachlorobiphenyls, meta and
para chlorine removal was dominant as no ortho dechlo-
rination occurred after 45 days. Para chlorines may be
preferentially removed over meta chlorines. Ortho chlorines
were removed from 22′, but only trace amounts of products
were identified.
These results suggest that Pd/ Peerless Fe0 and nanoscale
ZVI tend to dechlorinate PCBs in a predictable manner with
chlorine removal from each position following the general
Positional Isom er Reactivity. The relative masses of ortho,
meta, and para chlorines that were removed after 2 days for
9
5 2 1 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004

trend para g meta . ortho chlorines. This is consistent with
results from previous studies. The dechlorination products
measured by West et al. (20) using 22′35′ and 0.25%-Pd 40-
mesh iron also suggest that ortho chlorines are more resistant
to removal than meta chlorines. Yak et al. (24), who
investigated PCB dechlorination by microscale ZVI in sub-
critical water at 250 °C, found ortho-substituted congeners
were less likely to undergo reductive dechlorination than
meta or para-substituted congeners. Furthermore, they found
meta chlorines more resistant to removal than para chlorines
(24). For example, they estimated the ratio of 33′4/ 344′ from
the dechlorination of 33′44′ to be 3/ 2. For nanoscale ZVI
under ambient temperatures in this study, this ratio is 7/ 2.
Thus abiotic, nanoscale ZVI-mediated PCB dechlorination
should result in byproducts that have lower fractions of para
and meta chlorines but a higher fraction of ortho chlorines.
This is favorable because ortho-substituted PCB congeners
are less toxic than non ortho-substituted, coplanar PCB
congeners.
Engineering Im plications. It is encouraging that reactivity
was noted with all the congeners studied and does offer some
promise that nanoscale ZVI could dechlorinate all 209
congeners. Coating microscale iron with palladium makes
microscale iron reactive, but the rate enhancement is short-
lived and does not justify using the expensive catalyst unless
the catalytic activity can be sustained or if the Pd/ Fe0 media
can be regenerated in situ. This will be difficult and cost
prohibitive in large-scale field applications. The slow rates
of dechlorination with nanoscale ZVI require it to be in
prolonged contact with the PCBs. Thus, ex situ treatment
reactors for PCB-contaminated waste streams (e.g., effluent
from soil/ sediment washing processes) requiring rapid
dechlorination rates are not likely to be practical. Dechlo-
rination of PCBs with nanoscale ZVI may, however, be
possible in situ (e.g., mixing nanoscale ZVI into PCB-
contaminated sediments stored in confined disposal facilities
(CDF) or in diffusion-dominated systems such as a reactive
barrier covering contaminated sediments left in place). PCBs
in natural media are predominantly sorbed to solids rather
than in solution, so the PCB dechlorination rates achievable
in in situ applications will ultimately be limited by the PCB
desorption rate and subsequent transport to a reactive
nanoiron particle. The presence of competing substrates,
dissolved solids, dissolved oxygen, etc. will also affect the
PCB dechlorination rate. The physical processes limiting the
availability of PCBs to iron will have to be considered in any
remedial design. Currently, the cost of producing nanoscale
ZVI by methods described here (∼ $200-300 per kg) is too
high for widespread application in sediments, but these costs
are expected to decrease as the market for nanoscale ZVI
increases and alternative synthesis processes become avail-
able. Methods to cost-effectively synthesize nanoscale ZVI
must become available, and the long-term reactivity of
nanoscale ZVI under natural reaction conditions must be
assessed, before this technology becomes a viable approach
for remediating PCBs in situ.
Literature Cited
(1) A Risk Management Strategy for PCB-Contaminated Sediments;
National Research Council, National Academy Press: Wash-
ington DC, 2001.
(2) Superfund Proposed Plan: Hudson River PCBs Superfund Site,
New York; U.S. EPA Region 2; Dec 2000, p 26.
(3) Rhee, G.-Y.; Sokol, R. C.; Bush, B.; Bethoney, C. M. Environ. Sci.
Technol. 1993, 27 (4), 714-719.
(4) Wiegel, J.; Wu, Q. FEMS Microbiol. Ecol. 2000, 32, 1-15.
(5) Zwiernik, M. J.; Quensen, J. F., III; Boyd, S. A. Environ. Sci.
Technol. 1998, 32 (21), 3360-3365.
(6) Cutter, L.; Watts, J.; Sowers, K.; May, H. D. Environ. Microbiol.
2001, 3 (11), 699-709.
(7) Wu, Q.; Watts, J.; Sowers, K.; May, H. D. Appl. Environ. Microbiol.
2002, 68 (2), 807-812.
(8) Fish, K. M.; Principe, J. M. Appl. Environ. Microbiol. 1994, 60
(12), 4289-4296.
(9) Abramowicz, D. A.; Brennan, M. J.; Van Dort, H. M.; Gallagher,
E. L. Environ. Sci. Technol. 1993, 27 (6), 1125-1131.
(10) Master, E. R.; Lai, V. W.; Kuipers, B.; Cullen, W. R.; Mohn, W.
M. Environ. Sci. Technol. 2002, 36 (1), 100-103.
(11) Dingyi, Y.; Quensen, J. F., III; Tiedje, J. M.; Boyd, S. A. Appl.
Environ. Microbiol. 1992, 58 (4), 1110-1114.
(12) Natarajan, M. R.; Nye, J.; Wu, W.-M.; Wang, H.; Jain, M. K. Biotech.
Bioeng. 1997, 55 (1), 182-190.
(13) Williams, W. A.; May, R., J. Environ. Sci. Technol. 1997, 31 (12),
3491-3496.
(14) Han, S.; Eltis, L. D.; Timmis, K. N.; Muchmore, S. W.; Bolin, J.
T. Science 1995, 270 (5238), 976-980.
(15) Matheson, L. J.; Tratnyek, P. G. Environ. Sci. Technol. 1994, 28
(12), 2045-2053.
(16) Vogan, J. L.; Focht, R. M.; Clark, D. K.; Graham, S. L. J. Hazard.
Mater. 1999, 68 (1), 97-108.
(17) Gillham, R. W.; O’Hannesin, S. F. Ground Water 1994, 32, 958-
967.
(18) Eykholt, G.; Davenport, D. Environ. Sci. Technol. 1998, 32 (10),
1482-1487.
(19) Powell, R. M.; Powell, P. D.; Puls, R. W. Economic Analysis of the
Implementation of Permeable Reactive Barriers for Remediation
of Contaminated Groundwater EPA/ 600/ R-02/ 034; Environ-
mental Protection Agency, U.S. Office or Research and Devel-
opment: Cincinnati, June 2002.
(20) West, O. R.; Liang, L.; Holden, W. L.; Korte, N. E.; Fernando, Q.;
Clausen, J. L. Degradation of Polychlorinated Biphenyls (PCBs)
Using Palladized Iron; ORNL/ TM-13217; Oak Ridge National
Laboratory: Oak Ridge, TN, 1996; 37831.
(21) Wang, C.; Zhang, W. Environ. Sci. Technol. 1997, 31 (7), 2154-
2156.
(22) Chuang, F.; Larson, R. A.; Wessman, M. S. Environ. Sci. Technol.
1995, 29 (9), 2460-2463.
(23) Yak, H. K.; Wenclawiak, B. W.; Cheng, I. F.; Doyle, J. G.; Wai, C.
M. Environ. Sci. Technol. 1999, 33 (8), 1307-1310.
(24) Yak, H. K.; Lang, Q.; Wai, C. M. Environ. Sci. Technol. 2000, 34
(13), 2792-2798.
(25) Grittini, C.; Malcomson, M.; Fernando, Q.; Korte, N. Environ.
Sci. Technol. 1995, 29 (11), 2898-2900.
(26) Schrick, B.; Blough, J. L.; Jones, D.; Mallouk, T. E. Chem. Mater.
2002, 14, 5140-5147.
(27) Zhang, W. X. J. Nanopart. Res. 2003, 5, 323-332.
(28) Ponder, S. M.; Darab, J. G.; Mallouk, T. E. Environ. Sci. Technol.
2000, 34 (12), 2564-2569.
(29) Safe, S. Crit. Rev. Toxicol. 1990, 21 (1), 51-88.
(30) Safe, S. Crit. Rev. Toxicol. 1994, 24 (2), 87-149.
(31) Frame, G. M.; Cochran, J. W.; Boewadt, S. S. J. High Resolut.
Chromatogr. 1996, 19, 657-668.
Acknowledgments
This research was supported in part by the Hazardous
Substance Research Center (HSRC) South and Southwest
through a research Grant to Dr. Lowry (R139634), the National
Science Foundation through a Graduate Student Fellowship
to K.M.J., and Alcoa, Inc. The authors also thank Dr.
Christopher S. Kim and the National Center for Electron
Microscopy (NCEM) for providing the ARM images and Dr.
Dorothy Farrell for the TEM and electron diffraction images.
(32) Alder, A. C.; Ha¨ggblom, M. M.; Oppenhelmer, S. R.; Young, L.
Y. Environ. Sci. Technol. 1993, 27 (3), 530-538.
(33) Morris, K. R.; Abramowitz, R.; Pinal, R.; Davis, P.; Yalkowsky, S.
H. Chemosphere 1988, 17 (2), 285-298.
(34) Shiu, W. Y.; Mackay, D. J. Phys. Chem. 1986, 15 (2), 911-929.
(35) Poerschmann, J.; Go´ recki, T.; Kopinke, F.-D. Environ. Sci.
Technol. 2000, 34 (17), 3824-3830.
(36) Johnson, T. L.; Scherer, M. M.; Tratnyek, P. G. Environ. Sci.
Technol. 1996, 30 (8), 2634-2640.
(37) Lowry, G. V.; Reinhard, M. Environ. Sci. Technol. 2000, 34 (15),
3217-3223.
(38) Mostafavi, M.; Marignier, J.; Amblard, J.; Belloni, J. Radiat. Phys.
Chem. 1989, 34 (4), 605-617.
Supporting Information Available
Detailed synthesis steps for the nanoscale iron used in this
study. This material is available free of charge via the Internet
at http:/ / pubs.acs.org.
(39) Jakusˇ, V.; Miertusˇ, S. Chem. Pap. 1990, 44 (5), 589-601.
9
VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 5 2 1 5

(40) Huang, Q.; Rusling, J. F. Environ. Sci. Technol. 1995, 29 (1),
98-103.
(44) Alowitz, M. J.; Scherer, M. M. Environ. Sci. Technol. 2002, 36 (3),
299-306.
(41) Farwell, S. O.; Beland, F. A.; Geer, R. D. Electroanal. Chem.
Interface Electrochem. 1975, 61, 315-324.
(42) Connors, T. F.; Rusling, J. F.; Owlia, A. Anal. Chem. 1985, 57,
170-174.
Received for review February 1, 2004. Revised manuscript
received June 18, 2004. Accepted July 7, 2004.
(43) Rusling, J. F.; Miaw, C. L. Environ. Sci. Technol. 1989, 23 (4),
476-479.
ES049835Q
9
5 2 1 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004