ARTICLE pubs.acs.org/est
Abiotic Reductive Dechlorination of cis-Dichloroethylene by Fe Species Formed during Iron- or Sulfate-Reduction Hoon Y. Jeong,*,† Karthik Anantharaman,‡ Young-Soo Han,‡ and Kim F. Hayes‡ † ‡
Department of Geological Sciences, Pusan National University, Busan 609-735, Korea Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, Michigan 48109, United States.
bS Supporting Information ABSTRACT: This study investigated reductive dechlorination of cis-dichloroethylene (cis-DCE) by the reduced Fe phases obtained from in situ precipitation, which involved mixing of Fe(II), Fe(III), and S(-II) solutions. A range of redox conditions were simulated by varying the ratio of initial Fe(II) concentration ([Fe(II)]o) to initial Fe(III) concentration ([Fe(III)]o) for iron-reducing conditions (IRC) and the ratio of [Fe(II)]o to initial sulfide concentration ([S(-II)]o) for sulfate-reducing conditions (SRC). Significant dechlorination of cis-DCE occurred under highly reducing IRC and iron-rich SRC, suggesting that Fe (oxyhydr)oxides including green rusts are highly reactive with cis-DCE but that Fe sulfide as mackinawite (FeS) is nonreactive. Relative concentrations of sulfate to chloride were also varied to examine the anion impact on cis-DCE dechlorination. Generally, slower dechlorination occurred in the batches with higher sulfate concentrations. As indicated by higher dissolved Fe concentration, the slower dechlorination in the presence of sulfate was probably due to the decreased surface-complexed Fe(II). This study demonstrates that the chemical form of reduced Fe(II) is critical in determining the fate of cis-DCE under anoxic conditions.
’ INTRODUCTION Tetrachloroethylene (PCE) and trichloroethylene (TCE), the second and fourth most frequently detected volatile organic compounds in U.S. groundwater,1 are subject to reductive dechlorination under anoxic conditions.2 While complete dechlorination of both chlorinated pollutants to nonchlorinated products (e.g., acetylene and ethylene) occurs at some contaminated sites, their incomplete degradation is often reported at other sites.3,4 Partial reductive dechlorination of PCE and TCE results in significant accumulation of cis-dichloroethylene (cis-DCE) and vinyl chloride (VC) in contaminated groundwaters,35 posing a greater remediation challenge due to their higher toxicity and persistence to subsequent degradation compared with their parent compounds, PCE and TCE.6 The tendency toward reductive dechlorination decreases with the number of chlorines in chlorinated organic pollutants (PCE > TCE > cis-DCE > VC).7 Although a limited number of bacterial strains have been shown to catalyze reductive dechlorination of cis-DCE and VC,2,5,8 biologically mediated dechlorination can be hindered under certain environmental conditions such as high contaminant concentrations,9 low substrate concentrations,10 and toxic metabolites.11 Natural attenuation of cis-DCE at a contaminated groundwater site could not be explained by biotic reductive dechlorination alone,12 suggesting the abiotic dechlorination. Recently, both biotic and abiotic dechlorination mechanisms were invoked to explain the natural attenuation of TCE and cis-DCE under anoxic conditions.13 r 2011 American Chemical Society
Reduced Fe minerals, because of their ubiquitous presence and dechlorination reactivity,14,15 are probably the most significant natural reductants for chlorinated organic pollutants. Yet, the abiotic dechlorination by reduced Fe minerals has been relatively unexplored compared with a larger body of data for the biotic degradation. Several Fe phases including magnetite,14 pyrite,14 green rust,15 and phyllosilicates,16 and zerovalent iron 17 have been documented for their dechlorination capability with cis-DCE and VC. However, the reported reactivity of these Fe minerals is often not reproducible due to the difference in mineral preparation procedures and batch compositions as well as the susceptibility of such minerals to oxidation. For example, He et al. 18 found that freezedried mackinawite was less reactive toward TCE dechlorination than the nonfreeze-dried mackinawite because of the decreased surface area and mineralogical transformation by freeze-drying. The geochemical conditions favoring formation of single reactive Fe minerals were identified (see reviews in ref 19). However, given the potential for formation of multiple reactive Fe species, additional work is needed to better recognize the geochemical conditions where abiotic dechlorination of cis-DCE and VC prevails. The objectives of this study were to assess reductive dechlorination of cis-DCE by reduced Fe species and identify the geochemical Received: December 30, 2010 Accepted: April 28, 2011 Revised: April 26, 2011 Published: May 19, 2011 5186
dx.doi.org/10.1021/es104387w | Environ. Sci. Technol. 2011, 45, 5186–5194
5187
0.02
0.02 0.02
0.02
d1
d2 d3
d4
0.005
0.01
0.015
0.005
0.01
0.015
(M)
[Fe(III)]o
(Rohypo)
0.01
0.01 0.01
0.01
0.04
0.01 0.02
0.04
0.02
0.01
(M)
[S(-II)]o
0.045 0.015
0.06
0.015
0.045
0.06
0.06
0.06
0.06 0.06
0.06
0.06
0.06
(M)
[Cl]o
0.06 b
0.015 0.045
0.06
0.045
0.015
0.06
0.06 0.06
0.06
0.06
0.06
0.06
(M)
[SO42]o
7.8
7.5 7.7
7.8
7.7
7.8
7.8
7.6
7.4
7.8 7.6
7.7
7.7
7.7
7.8
7.4
7.5
7.6 7.8
7.6
7.7
7.8
pHo
1.9 102 2
3
5.0 10 8.2 10 2
2
3
3
4.8 10 2.2 103 1.5 103 6 8
2.7 10 2.4 10 3
3
2
564 571 586 582 581 628 452 565 568 583
o
1.5 10 1.8 102
3
1.2 10
3
2.0 102 2 2 2 2 2 2
1.5 10 1.5 102 1.4 102
8.2 106 3
3
3
2
3 3
4.8 10 5.2 103 6.5 103
602 586 587 582 583 582 584 581 586
2.2 10
1.0 10
4.3 10
3.4 10
1.8 10
1.0 10
1.6 10
1.7 10
1.5 10
1.4 10 1.7 102
6.5 10 3.1 103
586 571 4.8 10
2
1.0 10
1.1 10
2.0 10
3
1.0 10
4.1 10
1.8 10
2.6 10 2
2
8.2 10
1.2 105
430 1.8 10
5.0 103 3
(M)
Fe(II)solid
(M)
Fediss
(mV)
Eho
0.04
0.04 0.04
0.04
0.02
0.02
0.02
0.02
0.1
0.04 0.06
0.02
0.015
0.01
0.005
0.1
0.06
0.02 0.04
0.015
0.01
0.005
(M)
Rohypoa
3.6 104
3.6 10 3.6 104
4
6.5 10
4
5.8 10
4
8.0 10
4
1.1 10
3
6.6 10
4
3.6 10
4
5.8 10
4
6.6 10 6.5 104
4
1.1 10
3
4
5.8 10
(M)
Rob
6.0 101
1
1
1
2
6.4 10 6.4 101
8.1 10
6.7 10
5.2 10
4.9 10
1
7.2 101
NDd
6.0 10 NDd
1
6.7 101
NDd
NDd
NDd
NDd
NDd
1
7.2 10 8.1 101
8.6 10
2
6.5 101
NDd
(mM )
1
Kb
7.6 105
2.0 10 1.9 104
4
4.2 10
4
2.4 10
4
3.9 10
4
1.0 10
2
3.2 104
7.6 10
5
2.4 104
3.2 10 4.2 104
4
1.1 10
4
2.0 104
(h )
1
kLHb
7.2 105
1.4 104 1.3 104
4.1 104
2.2 104
3.5 104
9.0 103
2.9 104
7.2 105
2.2 104
2.9 104 4.1 104
1.1 104
2.0 104
(h1)
kobsc
c
Hypothetical reduction capacities were calculated from eq 7. Reduction capacities (R ), sorption coefficients (K), and pseudo-first-order rate constants (kLH) were determined using eq 4 and 5. Pseudo-first-order rate constants (kobs) were determined using eq 6. d ND means no dechlorination (indicated by both no appreciable decrease in aqueous cis-DCE concentrations compared with blanks and the absence of the degradation products) with the limit of kLH (kobs) at ∼1 105 h1.
a
0.02
c4
0.02
b4
0.02
0.015
b3
c3
0.01
b2
0.02
0.005
b1
c2
0.02
a7
0.02
0.02
a6
c1
0.02 0.02
a4 a5
0.02
0.015
a3
b7
0.01
a2
0.02 0.02
0.005
a1
b5 b6
(M)
ID
[Fe(II)]o
Table 1. Batch Compositions and Kinetic Parameters
Environmental Science & Technology ARTICLE
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology
ARTICLE
Figure 1. Time profiles of normalized aqueous concentrations of cis-DCE in the batches containing chloride under chemically simulated iron-reducing conditions (A) and sulfate-reducing conditions (B). The ID number of each batch in Table 1 is provided in parentheses inside the figures. Batch compositions are [Fe(II)]o þ [Fe(III)]o = 0.02 M, [Cl]o = 0.06 M, and pHo ∼7.7. In part (A), no sulfide is added; in part (B), no Fe(III) is added with varying [S(-II)]o. The initial aqueous concentration of cis-DCE is ∼6 μM. Blanks are 0.06 M NaCl solutions at pHo 7.7. Error bars indicate one standard deviation.
conditions leading to its complete dechlorination. For this, reduced Fe species were prepared by in situ precipitation, which involved mixing of Fe(III), Fe(II), and S(-II) solutions without washing. We believe that using in situ precipitates rather than pure synthetic Fe minerals better reflects the heterogeneous mineral compositions in subsurface environments. Also, considering the extensive dechlorination mediated by biologically produced Fe minerals,20 such precipitates should better mimic the dechlorination by biogenic Fe minerals compared with crystalline Fe minerals. Different redox conditions were chemically simulated by varying Fe(II), Fe(III), and S(-II) solutions to compare the reactivity among various Fe phases forming under iron- and sulfatereducing conditions. Chloride and sulfate concentrations were also evaluated for their impact on reductive dechlorination. This study is the first to report abiotic dechlorination of cis-DCE by reduced Fe phases prepared by in situ precipitation.
’ EXPERIMENTAL SECTION Reaction batches were prepared inside an anaerobic chamber with an atmospheric composition of ∼5% H2 in N2. All glassware was autoclaved, and all aqueous solutions were prepared using distilled, deionized, and deoxygenated water filtered through 0.22 μm membranes.21 Aqueous Fe(II), Fe(III), and S(-II) solutions were added to the solutions of electrolytes containing NaCl and/or NaHSO4 in ∼160 mL serum bottles to obtain the compositions in Table 1. While Fe(II) solutions were prepared by dissolving FeCl2 or FeSO4 in water, Fe(III) solutions were obtained by dissolving FeCl3 or Fe2(SO4)3 in water. Sulfide solutions were made using NaHS. Predetermined amounts of NaOH solutions were added to the batches to set the initial solution pH at ∼7.7. No buffer was added to avoid its potential impact on abiotic dechlorination.22 The solution volume was then adjusted to 150 mL by adding water, resulting in 11 mL of headspace. Subsequently, the batches were sealed with Tefloncoated butyl rubber septa and allowed to age by vigorously mixing on a reciprocating shaker for 7 days before dechlorination experiments. The aged batches were measured for pH, redox potential (Eh), and dissolved Fe concentration (Fediss) as described in ref 23. Since the batches were rapidly oxidized under the atmosphere,
pH was measured at the end of kinetic experiments inside the anaerobic chamber for the batches showing complete cis-DCE dechlorination (e.g., batch c2 in Table 1). In such batches, the pH change was less than ∼0.5 unit. Kinetic Experiments. Dechlorination experiments were initiated by introducing a methanolic cis-DCE solution, resulting in the methanol content less than 0.04% (v/v). Such content did not affect dechlorination rates.17 The reaction batches were removed from the anaerobic chamber and submerged in a reciprocating water bath at 170 rpm at 25 °C in the dark. In parallel, blank solutions containing NaCl and/or NaHSO4 were run to account for cis-DCE loss due to sampling, sorption to reaction vials, and volatilization though the septa. All batches including blanks were prepared in duplicate or triplicate. Initial aqueous cis-DCE concentration was ∼6 μM, and dechlorination was monitored for 70 days. At intervals, a 1.0 mL aliquot of the solution phase was withdrawn using a gastight glass syringe, and to maintain the solution-phase volume, a 1.0 mL of a blank solution having the same electrolyte composition as the reaction batch was added. The aliquot was transferred to headspace vials containing 5 mL water, with 5 μL of 1.03 mM trans-DCE in methanol added as an internal standard. cis-DCE and potential dechlorination products including VC, acetylene, ethylene, and ethane were analyzed using a gas chromatography method (see Supporting Information). Based on triplicate analyses of samples containing a range of concentrations of cis-DCE and its degradation products, analytical errors were typically ∼5%, with detection limits of ∼1 107 M.
’ RESULTS AND DISCUSSION Different redox conditions, the critical factors in determining the type of bulk-phase Fe precipitates, were created by varying the initial concentrations of Fe(II) ([Fe(II)]o), Fe(III) ([Fe(III)]o), and sulfide ([S(-II)]o) (see Table 1). The degree of iron-reduction under iron-reducing conditions (IRC) was simulated by changing the ratio of [Fe(II)]o/[Fe(III)]o at [S(-II)]o = 0. Similarly, different sulfate-reducing conditions (SRC) were obtained by varying the ratio of [Fe(II)]o/[S(-II)]o at [Fe(III)]o = 0 to mimic iron-rich or iron-deficient conditions. Total Fe concentration in 5188
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology
ARTICLE
Table 2. Results of Equilibrium Calculations for Batches in Table 1 ID
Fediss(M)
a1
1.16 105 3
a2 a3 a4 a5 a6 a7
1.75 10
3
2.58 10
3
4.83 10
3
2.16 10
3
1.45 10
6
2.67 10
8
stable Fe precipitatea
solubility (M)a
dissolved Fea,c
Magd
4.57 109
Fe2þ, FeOHþ
Cl-GRd
3.95 104
Fe2þ, FeOHþ
d
7
2þ
þ
d
3
Fe2þ, FeOHþ
2þ
þ
3
Fe2þ, FeOHþ
2þ
þ
3
Fe2þ, FeOHþ
2þ
þ
3
Fe2þ, FeOHþ
2þ
þ
6
Fe2þ, FeOHþ
2þ
þ
4
Fe(HS)32-, Fe(HS)2o
Mag
d
OH(1)-GR
d
OH(1)-GR, Py
d
OH(1)-GR, Py Py
Mag Magd
b3
4.10 103
Magd
b5
1.02 10
3
6.49 10
3
7
7.39 10
7
2.53 10
7
1.19 10
6
4.56 10
d
2.41 10 1.84 103
b4
5.06 10
d
b1 b2
2
2.83 10
Mag
7
8
7.92 10 4.35 106
d
4.70 106 d
OH(1)-GR
d
OH(1)-GR, Py
6.64 10
6
3.87 10
6
5
b6
3.05 10
OH(1)-GR, Py
1.14 10
b7
8.18 106
Pyd
4.97 108
c1
4.83 103
OH(1)-GRd
7.39 107
c2 c3
3.42 103 4.30 103
OH(1)-GRd OH(1)-GRd
c4
1.02 102
d1
2.16 103
d2
Fe , FeOH Fe , FeOH Fe , FeOH Fe , FeOH Fe , FeOH Fe , FeOH FeSO4o, FeSO4o, FeSO4o, FeSO4o, FeSO4o, FeSO4o, FeSO4o, 2þ
metastable Fe precipitateb
solubility (M)b 2.32 10
Cl-GR
3.64 10
d
Cl-GR
4.49 10
d
Cl-GR
1.80 10
d
Cl-GR, Mack
3.30 10
d
Mack
3.08 10
d
Mack
dissolved Feb,c
2þ
Fe Fe2þ
d
SO4-GR SO4-GRd
3
1.50 10 1.16 102
FeSO4o, Fe2þ FeSO4o, Fe2þ
Fe2þ
SO4-GRd
1.20 102
FeSO4o, Fe2þ
2þ
d
2
FeSO4o, Fe2þ
2
FeSO4o, Fe2þ
5
Fe
1.47 10
SO4-GR
1.00 10
d
2þ
Fe
Mack
Fe
Mack
1.13 10
FeSO4o, Fe2þ
Fe2þ
Greigd
1.63 105
Fe(HS)32-, Fe(HS)2o
Fe , FeOHþ
Cl-GRd
4.49 103
Fe2þ, FeOHþ
1.16 106 2.98 106
FeSO4o, Fe2þ FeSO4o, Fe2þ
SO4-GRd SO4-GRd
3.45 103 7.74 103
FeSO4o, Fe2þ FeSO4o, Fe2þ
OH(1)-GRd
6.64 106
FeSO4o, Fe2þ
SO4-GRd
1.47 102
FeSO4o, Fe2þ
OH(1)-GR, Pyd
2.53 107
Fe2þ, FeOHþ
Cl-GR, Mackd
1.80 103
Fe2þ, FeOHþ
4.80 103
OH(1)-GR, Pyd
5.84 106
FeSO4o, Fe2þ
Mackd
1.00 102
FeSO4o, Fe2þ
d3
5.16 103
OH(1)-GR, Pyd
5.12 106
FeSO4o, Fe2þ
Mackd
9.93 103
FeSO4o, Fe2þ
d4
6.49 103
OH(1)-GR, Pyd
3.87 106
FeSO4o, Fe2þ
Mackd
1.00 102
FeSO4o, Fe2þ
d
d
2þ
a
Stable Fe precipitates were obtained by including all Fe phases listed in Equilibrium Calculations, Supporting Information. b Metastable Fe precipitates were obtaining by excluding goethite, magnetite, all hydroxy green rusts (OH-GRs), and pyrite. c The dominant dissolved Fe species are listed, with the left one the most abundant. d Mag, Cl-GR, SO4-GR, OH(1)-GR, Py, Mack, and Greig are magnetite, chloride green rust, sulfate green rust, a hydroxy green rust (Fe3(OH)7), pyrite, mackinawite, and greigite, respectively.
all batches was fixed at 0.02 M to produce a well-defined sequence of Fe(II)Fe(III) redox changes. In addition, the concentrations of chloride and sulfate were varied to evaluate their potential impact on cis-DCE dechlorination. Data Analysis. In this study, cis-DCE dechlorination exhibits an initially rapid disappearance followed by a slower removal (see Figures 14), which is attributable to the consumption of reactive sites as the reaction proceeds.1416 To describe such behavior, Lee and Batchelor 14 modified the Langmuir Hinshelwood rate law as follows: dC kRC ¼ dt 1=K þ C
ð1Þ
where k is the decay rate constant of a target organic compound at reactive sites; K is the sorption coefficient of the organic compound at reactive sites; R is the reductive capacity for the organic compound at time t; and C is aqueous concentration of the organic compound at time t. In applying the above modification, it is assumed (1) a target organic compound adsorbs to a finite number of reactive sites; (2) its transformation occurs at reactive sites by a first-order reaction; and (3) reactive sites are the source of reduction capacity.14 By relating the decrease in the reduction capacity of a target organic compound to that in its aqueous concentration, the following form can be derived from eq 1 (see ref 14 for details): dC ðk=pÞðR o pðCo CÞÞC ¼ dt 1=K þ C
is the initial aqueous concentration. Neglecting sorption of the organic compound onto the solid-phase,21,24 p is given by p ¼ 1þH
ð3Þ
where H is the dimensionless Henry’s constant for cis-DCE (0.167 from ref 24); Vg is the headspace volume (11 mL); and Vaq is the solution-phase volume (150 mL). The analytical solution to eq 2 is in an implicit form: C 1=K ln t ¼ o ð Co þ R =pÞk Co ð1=K þ Co R o =pÞ C Co þ R o =p ð4Þ þ ln ð Co þ R o =pÞk R o =p Initial aqueous concentration of cis-DCE (Co) after partitioning to the headspace was directly measured from blanks. The kinetic parameters (k, K, and Ro) were then determined by a nonlinear regression of aqueous cis-DCE concentrations with eq 4 using KaleidaGraph 3.51 (Synergy Software 2000). Using the kinetic parameters, pseudo-first-order rate constants (kLH) were determined by kLH ¼
ð2Þ
where p is the partitioning coefficient for a target organic compound; Ro is the initial reductive capacity of the organic compound; and Co
Vg Vaq
ðk=pÞR o 1=K þ Co
ð5Þ
For several data, the regression with eq 4 gave a poor fit at longer reaction times (see Supporting Information Figure S1) likely due to the presence of multiple reactive sites with significantly 5189
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology
ARTICLE
Figure 2. Time profiles of normalized aqueous concentrations of cis-DCE in the batches containing sulfate under chemically simulated iron-reducing conditions (A) and sulfate-reducing conditions (B). The ID number of each batch in Table 1 is provided in parentheses inside the figures. Batch compositions are [Fe(II)]o þ [Fe(III)]o = 0.02 M, [SO42]o = 0.06 M, and pHo ∼7.7. In part (A), no sulfide is added; in part (B), no Fe(III) is added with varying [S(-II)]o. The initial aqueous concentration of cis-DCE is ∼6 μM. Blanks are 0.06 M Na2SO4 solutions at pHo 7.7. Error bars indicate one standard deviation.
different reactivities and competitive adsorption/transformation for reactive sites between cis-DCE and its degradation products. Highly reactive surface-complexed Fe(II) may preferentially react compared with bulk-phase Fe minerals. The dechlorination products (e.g., acetylene) may accumulate and compete with cisDCE for reactive sites as the reaction proceeds.17 Both effects would retard cis-DCE dechlorination, and become more evident at longer reaction times. Such effects are not taken into account in the modified LangmuirHinshelwood rate law. Thus, the fit was limited to the data at t < ∼1000 h. When the initial reduction capacity remains nearly constant (i.e., Ro . p(Co C)) and the adsorption of a target organic compound to reactive sites is energetically weak (i.e., 1/K . C), the dechlorination kinetics given by eq 2 simplifies to a first-order rate law: dC ¼ kobs C ð6Þ dt Pseudo-first-order rate constants (kobs) were also determined by a nonlinear regression of aqueous cis-DCE concentrations with the analytical solution to eq 6 (C = Co 3 exp(kobs 3 t)) over the data at t < ∼1000 h. Pseudo-first-order rate constants determined by both approaches are presented in Table 1, where the rate constants estimated by the modified LangmuirHinshelwood rate law (kLH) are slightly larger than those by a first-order rate law (kobs). Dechlorination in Chloride-Containing Batches. For chloridecontaining batches, cis-DCE dechlorination under chemically simulated IRC and SRC is shown in Figure 1A and 1B, respectively. Under IRC, the greatest dechlorination occurs at the highest [Fe(II)]o/[Fe(III)]o (i.e., in the [Fe(III)]o-free batch), but the ratio, indicative of the degree of iron-reduction, does not exactly match the extent of cis-DCE dechlorination. Under SRC, the batch with excess [Fe(II)]o relative to [S(-II)]o exhibits significant cisDCE dechlorination, but the batch with the ratio of [Fe(II)]o/ [S(-II)]o e 1 shows no detectable dechlorination. Although strong oxidants including molecular oxygen can oxidize S(-II) to sulfate,25 relatively weak oxidants such as cisDCE may oxidize S(-II) to elemental sulfur.26 Thus, based on the redox pairs of Fe(II)Fe(III) and S(-II)S(0), a hypothetical
reduction capacity (Rohypo) was estimated by o Rhypo ¼ ½FeðIIÞo þ 2½Sð-IIÞo
ð7Þ
Although Rohypo increases from batch a1 to a7, the observed rate constant does not show a similar trend (see Table 1). Furthermore, the reduction capacity (Ro) estimated from the kinetic analysis is far smaller than Rohypo, indicating most of the initially added Fe(II) and S(-II) is unavailable for cis-DCE dechlorination. While dissolved sulfide effectively dechlorinated the readily degradable chlorinated organic compounds such as hexachloroethane 27 and carbon tetrachloride,28 it caused less than 10% of PCE transformation during a 33-day reaction period.29 In our previous study,21 the degradation of PCE and TCE by dissolved sulfide was also minimal. Mackinawite (FeS), the initial Fe sulfide in sulfidic environments,30 did not show the ability to reductively dechlorinate cis-DCE.31 Consistent with the nonreactivity of both dissolved sulfide and Fe sulfide, no appreciable dechlorination of cis-DCE was observed in the sulfide-rich batch under SRC (see batch a7 in Table 1). Dissolved Fe(II) is also a less effective reductant than solid-phase Fe(II).32,33 Accordingly, greater dechlorination is expected at higher solid-phase Fe(II) concentration (Fe(II)solid), which is given by the difference between [Fe(II)]o and Fediss. However, batch a7 exhibits no dechlorination despite the highest Fe(II)solid. Furthermore, Fe(II)solid correlates poorly with kobs (see Supporting Information Figure S2). Thus, the chemical form of solid-phase Fe(II), not the quantity, is critical for cis-DCE dechlorination. Solid-phase Fe(II) is either surface-complexed or in the form of bulk-phase precipitates. Equilibrium calculations were performed using MINEQLþ4.5 (Environmental Research Software) to predict bulk-phase Fe precipitates (see Supporting Information for details and Table 2 for the results). Under IRC, magnetite (Fe3O4) is thermodynamically stable in the presence of the initially added Fe(III) (Fe(III)o) (batches a1a3), and a hydroxy green rust (Fe3(OH)7) becomes stable in the absence of Fe(III)o (batch a4). Formation of green rusts even without Fe(III)o would be possible due to anoxic corrosion by which oxidation of Fe(II) to Fe(III) is coupled with reduction of Hþ to H2(g). Although this work was performed under anoxic conditions to 5190
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology
ARTICLE
Figure 3. Time profiles of normalized aqueous concentrations of cis-DCE in the batches prepared under chemically simulated iron-reducing conditions (A) and sulfate-reducing conditions (B). The ID number of each batch in Table 1 is provided in parentheses inside the figures. Batch compositions in part (A) are [Fe(II)]o = 0.02 M, [S(-II)]o = 0 M, and pHo ∼7.7. Batch compositions in part (B) are [Fe(II)]o = 0.02 M, [S(-II)]o = 0.01 M, and pHo ∼7.7. The initial aqueous concentration of cis-DCE is ∼6 μM. Blanks are 0.06 M NaCl or Na2SO4 solutions at pHo ∼7.7. Error bars indicate one standard deviation.
Figure 4. Time profiles of the number of moles (Mt) of cis-DCE, acetylene, ethylene, and ethane in the batches prepared under chemically simulated iron-reducing conditions (A) and sulfate-reducing conditions (B). Batch compositions in part (A) are [Fe(II)]o = 0.02 M, [S(-II)]o = 0 M, [Cl]o = 0.045 M, [SO42]o = 0.015 M, and pHo ∼7.7 (see batch c2 in Table 1). Batch compositions in part (B) are [Fe(II)]o = 0.02 M, [S(-II)]o = 0.01 M, [Cl]o = 0.06 M, and pHo ∼7.7 (see batch d1 in Table 1).
avoid exposure to air, oxygen contamination could also cause Fe(II) oxidation. Previously, Doelsch et al.34 observed formation of green rusts by raising the pH of ferrous chloride solutions to neutral pH. Under SRC, an assemblage of pyrite (FeS2) and OH-GR is thermodynamically stable in batches a5 and a6, and only pyrite is stable in the sulfide-rich batch (batch a7). Nevertheless, chloride green rust (Fe4(OH)8Cl), sulfate green rust (Fe6(OH)12SO4), and mackinawite (FeS) may possibly form as metastable phases.30,35 Recently, mackinawite precipitates were reported to form under chemically simulated SRC that were similar to our experimental conditions.36 Although most batches prepared under IRC (batches a2a4) resulted in significant cis-DCE dechlorination, only the iron-rich batch under SRC (batch a5) showed dechlorination. Thus, some Fe (oxyhydr)oxide precipitates are reactive with cisDCE, whereas Fe sulfide precipitates are nonreactive. In our previous studies,21,31 green rust-like precipitates reductively dechlorinated cis-DCE, but mackinawite did not transform cis-DCE
despite its reactivity toward PCE and TCE. Since precipitates like green rusts are highly susceptible to oxidation by dehydration and exposure to the atmosphere,15,36 no attempt was made to identify bulk-phase precipitates using X-ray diffraction. Dechlorination in Sulfate-Containing Batches. For sulfatecontaining batches, cis-DCE dechlorination under IRC and SRC is shown in Figure 2A and 2B, respectively. As in case of chloridecontaining batches, cis-DCE dechlorination is most evident in the batch at [Fe(III)]o = 0 under IRC and the one with [Fe(II)]o in excess of [S(-II)]o under SRC. In sulfate-containing batches, Ro accounts for only a small fraction of Rohypo, and Fe(II)solid does not correlate with the rate constants. This finding is consistent with that observed in chloride-containing batches. Under SRC, a range of redox conditions were created by varying the ratio of [S(-II)]o/[SO42]o, with the lower ratio corresponding to a less reducing condition. In Table 1, the batch with the lowest [S(-II)]o/[SO42]o ratio shows the highest 5191
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology
ARTICLE
cis-DCE dechlorination despite representing the highest redox condition. Instead, as discussed in chloride-containing batches, the relative abundance of Fe(II) to S(-II) more likely controls cisDCE dechlorination under SRC. Anion Impact. In Figures 1 and 2, cis-DCE dechlorination is found to proceed at higher rates in the batches containing chloride than sulfate. To further assess the impact of these anions on cis-DCE dechlorination, the [SO42]o/[Cl]o ratio was varied while other compositions (i.e., the ratios of [Fe(II)]o/[Fe(III)]o and [Fe(II)]o/[S(-II)]o, and pHo) were maintained. Timeprofiles of the normalized aqueous cis-DCE concentrations are shown for the batches prepared under IRC (Figure 3A) and SRC (Figure 3B). In both Figure 3 and Figure S3, the anion composition significantly affects cis-DCE dechlorination. Under IRC, no clear relationship is noted between the dechlorination rate and the [SO42]o/[Cl]o ratio. Under SRC, the dechlorination rate decreases monotonically with the [SO42]o/[Cl]o ratio. As shown in Table 2, the thermodynamically stable Fe precipitate is independent of the anion composition. In most cases, the measured Fediss exceeds the solubility of the stable Fe precipitates by several orders of magnitude. In contrast, equilibrium calculations without involving magnetite, hydroxy green rusts, and pyrite as potential precipitates reveal that the measured Fediss is generally closer to the solubility of the resultant metastable phases (e.g., chloride green rust, sulfate green rusts, and mackinawite) (see Table 2). This is particularly true of batches c1c4 in Figure 3A and d1d4 in Figure 3B. Green rusts, the corrosion products of iron and steel, occur in natural environments,35 and mackinawite is ubiquitous in sulfidic sediments.30 Thus, formation of different metastable precipitates in the presence of chloride or sulfate may account for differences in cis-DCE dechlorination rates. Previously, a range of cis-DCE dechlorination rates were reported among mackinawite and green rusts.14,15,37 Batch c2 exhibits far greater dechlorination than batch c3 although the same metastable phase (sulfate green rust) may precipitate at similar amounts (see Tables 1 and 2). Also, the observed cis-DCE dechlorination in batches d2d4 is inconsistent with the presence of metastable mackinawite predicted by equilibrium calculations considering the nonreactivity of mackinawite with cisDCE (see Tables 1 and 2). Thus, besides formation of different metastable Fe precipitates, other factors need to be considered to explain such features. In Table 1, Fediss generally increases with higher [SO42]o. Once bulk-phase Fe precipitates form in reaction batches, these precipitates can provide surface sites where the excess dissolved Fe(II) can form surface complexes. Thus, the sulfate initially present may affect the stability of not only bulk-phase Fe precipitates but also surface-complexed Fe(II). In Table 2, Fe2þ is generally the most dominant dissolved species in sulfate-free batches, whereas FeSO4o becomes dominant in sulfate-containing batches. Consequently, the solubility of Fe precipitates is expected to increase by forming dissolved FeSO4o in the presence of sulfate. Similarly, the surface complexation of Fe(II) becomes less favorable in the presence of sulfate by forming dissolved complexes: O-Feþ þ SO4 2 T O þ FeSO4 o
ð8Þ
where O and O-Feþ are deprotonated hydroxyl surface groups and Fe(II)-complexed surface groups, respectively. As discussed above, the far higher dechlorination in batch c2 than batch c3 cannot be explained, if the small difference in Fediss between them is considered to reflect only the amount of Fe
precipitates. Instead, given the much smaller quantity of surfacecomplexed Fe(II) than structural Fe(II) in bulk-phase precipitates, even small differences in Fediss with the [SO42]o/[Cl]o ratio may translate into significant differences in the amount of surface-complexed Fe(II). Furthermore, considering the lower stability of surface-complexed Fe(II) than bulk-phase Fe precipitates as evidenced by the high extractability of the former,38 sulfate may have a greater effect on surface complexation of Fe(II) than on bulk-phase precipitation. Notably, no apparent impact of sulfate was observed for dechlorination of hexachloroethane by a preparation of pure mackinawite, which contained little surfacecomplexed Fe(II) due to intensive rinsing during the synthesis.39 In contrast to the poor correlation between kobs and Fe(II)solid for the whole data set in Figure S2, kobs is shown to increase with higher Fe(II)solid within a group of batches having the same batch compositions (thus similar mineralogy) except the anion composition (see Supporting Information Figure S3). Lee and Batchelor 14 observed that addition of Fe(II) to magnetite suspensions substantially enhanced cis-DCE dechlorination because of formation of surface-complexed Fe(II). Thus, changes in the amount of surface-complexed Fe(II) may be in part responsible for the observed anion impact. Another possible cause for the anion impact may include coagulation of bulk-phase Fe precipitates by anions. The divalent sulfate should aggregate Fe precipitates more strongly than the monovalent chloride,40 thus decreasing their reactivity by lowering the surface area. This is consistent with the dechlorination trend in Figure 3B, but not in Figure 3A. Therefore, multiple factors, not a single cause, may be responsible for the observed anion impact.41 Dechlorination Products and Pathways. The number of moles (Mt) of an organic compound within a batch is given by Mt ¼ ðHVg þ Vaq ÞC
ð9Þ
where H = 0.167 for cis-DCE,24 0.887 for acetylene,24 8.71 for ethylene,42 and 20.42 for ethane.42 Figure 4 shows time-profiles of Mt for cis-DCE and its degradation products. At the end of kinetic experiments (70 days), the mass recoveries (the sums of Mt for all organic compounds) are 80100%, with low recoveries more evident in highly reactive batches. Thus, incomplete mass recoveries result from formation of unidentified products rather than cis-DCE losses. For example, acetylene subsequently transforms to C4-hydrocarbons via coupling reactions.17 No additional attempt was made to clarify unidentified products. cis-DCE reductively transforms to vinyl chloride (VC) via hydrogenolysis or acetylene via β-elimination.17,31 In this study, acetylene was observed, but VC was not detected even at trace quantities. The absence of VC as a dechlorination product suggests the potential benefit of abiotic dechlorination over biotic dechlorination by which the highly toxic VC is produced.2 Ethylene and ethane were also present as the dechlorination products. Ethylene generally forms via subsequent hydrogenation of acetylene.17 However, formation of ethylene in some batches was dominant from the beginning (see Figure 4A). Similar results were observed in our previous study,31 where ethylene was postulated to form via a concerted β-elimination and hydrogenation without having acetylene as the reaction intermediate. In this study, thus, ethylene can form via a successive path of hydrogenation following β-elimination (cis-DCE f acetylene f ethylene) or a concerted path (cisDCE f ethylene). Ethane production in this study results from subsequent hydrogenation of ethylene. 5192
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology Environmental Implications. Given the pronounced dechlorination of cis-DCE under highly reducing IRC and ironrich SRC, metastable Fe (oxyhydr)oxides including green rusts are proposed to control the fate of cis-DCE under anoxic conditions. Consistent with this, green rusts have been shown to effectively dechlorinate cis-DCE.15,31 This study also suggests that surface-complexed Fe(II) may contribute to reductive transformation of cis-DCE. Previously, the dominant reductant in iron- and sulfate-reducing sediments has been attributed to surface-complexed Fe(II).38,43 Even when the aquatic systems are not saturated with respect to green rusts, cis-DCE dechlorination can be sustained by adsorbed Fe(II) on soil minerals.16,33 Magnetite, another potential precipitate formed under IRC, was reported to dechlorinate cis-DCE at the rate with kLH = 8.4 105 h1.14 In our work, however, no such dechlorination was observed within the measurement limit of kLH ∼1 105 h1. Moreover, cis-DCE degradation by magnetite 14 occurred at a 2 orders of magnitude slower rate than by sulfate green rust.15 Except under iron-rich conditions, no appreciable cis-DCE dechlorination was observed under SRC, indicating Fe sulfide(s) are not likely to dechlorinate cis-DCE. Consistent with this result, mackinawite, often the initial Fe sulfide phase to form,30,36 was found to be incapable of reductively transforming cis-DCE.31 Mackinawite persists for long periods of time before transforming to the thermodynamically stable pyrite.44 Thus, pyritemediated dechlorination would be expected to be insignificant in less diagenetically transformed sulfidic sediments despite its reported reactivity with cis-DCE.14 Taken together, abiotic transformation of cis-DCE under SRC is likely limited to iron-rich conditions, with bulk-phase Fe (oxyhydr)oxides or surface-complexed Fe(II) the most likely facilitators of cis-DCE dechlorination. While only a few bacterial strains associated with the Dehalococcoides group can directly catalyze reductive dechlorination of cis-DCE and VC,2,5,8 other anaerobes may also contribute to the abiotic degradation by producing Fe(II), which subsequently forms surface complexes with soil minerals or reactive Fe precipitates. Furthermore, Fe(III) formed during reductive dechlorination may be recycled back into Fe(II) by dissimilatory iron-reducing bacteria. Thus, even anaerobes incapable of dechlorinating cis-DCE and VC can play significant roles in controlling their reductive dechlorination in anoxic sediments.
’ ASSOCIATED CONTENT
bS
Supporting Information. Gas chromatography analysis, MINEQLþ4.5 calculations, the model fit of data, and the relationship between solid-phase Fe(II) concentrations and pseudo-first-order rate constants. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Phone: 82-51-510-2249; fax: 82-51-517-6389; e-mail: hjeong@ pusan.ac.kr.
’ ACKNOWLEDGMENT We thank Tom Yavaraski for his help with GC analysis. This work was supported by NSF No. CBET-0730064, the Korean Ministry of Environment as ‘The GAIA Project’, and Pusan National University Research Grant, 2010.
ARTICLE
’ REFERENCES (1) The Quality of Our Nation’s Waters: Volatile Organic Compounds in the Nation’s Ground Water and Drinking-Water Supply Wells; U.S. Geological Survey Circular 1292; U.S. Geological Survey: Reston, Virginia, 2006; http://www.usgs.gov/. (2) Freedman, D. L.; Gossett, J. M. Biological reductive dechlorination of tetrachloroethylene and trichloroethylene to ethylene under methanogenic conditions. Appl. Environ. Microbiol. 1989, 55, 2144–2151. (3) Parson, F; Wood, P. R.; DeMarco, J. Transformations of tetrachloroethylene and trichloroethylene in microcosms and groundwater. J. Am. Water Works Assoc. 1984, 76, 56–59. (4) Milde, G.; Nerger, M.; Mergler, R. Biological degradation of volatile chlorinated hydrocarbons in groundwater. Water Sci. Technol. 1998, 20, 67–73. (5) Maymo-Gatell, X.; Nijenhuis, I.; Zinder, S. H. Reductive dechlorination of cis-dichloroethene and vinyl chloride by Dehalococcoides ethenogenes. Environ. Sci. Technol. 2001, 35, 516–521. (6) Case study of natural attenuation of trichloroethene at St. Joseph, Michigan; EPA/540/R-97/504; U.S. EPA Office of Research and Development: Washington, DC, 1997; http://www.epa.gov/ nrmrl/pubs/540r97504/540r97504.pdf. (7) Vogel, T. M.; Criddle, C. S.; McCarty, P. L. Transformations of halogenated aliphatic compounds. Environ. Sci. Technol. 1987, 21, 722–736. (8) Bradley, P. M. History and ecology of chloroethene biodegradation: A review. Biorem. J. 2003, 7, 81–109. (9) Sabalowsky, A. R.; Semprini, L. Trichloroethene and cis-1,2dichloroethene concentration-dependent toxicity model simulates anaerobic dechlorination at high concentrations: I. Batch-fed reactors. Biotechnol. Bioeng. 2010, 107, 529–539. (10) Cupples, A. M.; Spormann, A. M.; McCarty, P. L. Vinyl chloride and cis-dichloroethene dechlorination kinetics and microorganism growth under substrate limiting conditions. Environ. Sci. Technol. 2004, 38, 1102–1107. (11) McCartney, D. M.; Oleszkiewicz, J. A. Sulfide inhibition of anaerobic degradation of lactate and acetate. Water Res. 1991, 25, 203–209. (12) Ferrey, M. L.; Wilkin, R. T.; Ford, R. G.; Wilson, J. T. Nonbiological removal of cis-dichloroethylene and 1,1-dichloroethylene in aquifer sediment containing magnetite. Environ. Sci. Technol. 2004, 38, 1746–1752. (13) Darlington, R.; Lehmicke, L.; Andrachek, R. G.; Freedman, D. L. Biotic and abiotic anaerobic transformations of trichloroethene and cis-1,2-dichlorothene in fractured sandstone. Environ. Sci. Technol. 2008, 42, 4323–4330. (14) Lee, W.; Batchelor, B. Abiotic reductive dechlorination of chlorinated ethylenes by iron-bearing soil minerals. 1. Pyrite and magnetite. Environ. Sci. Technol. 2002, 36, 5147–5154. (15) Lee, W.; Batchelor, B. Abiotic reductive dechlorination of chlorinated ethylenes by iron-bearing soil minerals. 2. Green rust. Environ. Sci. Technol. 2002, 36, 5348–5354. (16) Lee, W.; Batchelor, B. Abiotic reductive dechlorination of chlorinated ethylenes by iron-bearing phyllosilicates. Chemosphere 2004, 56, 999–1009. (17) Arnold, W. A.; Roberts, A. L. Pathways and kinetics of chlorinated ethylene and chlorinated acetylene reaction with Fe(0) particles. Environ. Sci. Technol. 2000, 34, 1794–1805. (18) He, Y. T.; Wilson, J. T.; Wilkin, R. T. Impact of iron sulfide transformation on trichloroethylene degradation. Geochim. Cosmochim. Acta 2010, 74, 2025–2039. (19) Identification and Characterization Methods for Reactive Minerals Responsible for Natural Attenuation of Chlorinated Organic Compounds in Ground Water; EPA 600/R-09/115; U.S. EPA Office of Research and Development: Ada, Oklahoma, 2009; http://www.epa. gov/ada. (20) McCormick, M. L.; Bouwer, E. J.; Adriaens, P. Carbon tetrachloride transformation in a model iron-reducing culture: Relative kinetics of biotic and abiotic reactions. Environ. Sci. Technol. 2002, 36, 403–410. 5193
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
Environmental Science & Technology (21) Jeong, H. Y.; Hayes, K. F. Reductive dechlorination of tetrachloroethylene and trichloroethylene by mackinawite (FeS) in the presence of metals: Reaction rates. Environ. Sci. Technol. 2007, 41, 6390–6396. (22) Danielsen, K. M.; Gland, J. L.; Hayes, K. F. Influence of amine buffers on carbon tetrachloride reductive dechlorination by the iron oxide magnetite. Environ. Sci. Technol. 2005, 39, 756–763. (23) Jeong, H. Y.; Han, Y.-S.; Park, S. W.; Hayes, K. F. Aerobic oxidation of mackinawite (FeS) and its environmental implication for arsenic mobilization. Geochim. Cosmochim. Acta 2010, 74, 3182–3198. (24) Butler, E. C.; Hayes, K. F. Kinetics of the transformation of trichloroethylene and tetrachloroethylene by iron sulfide. Environ. Sci. Technol. 1999, 33, 2021–2027. (25) Hyland, M. M.; Jean, G. E.; Bancroft, G. M. XPS and AES studies of Hg(II) sorption and desorption reactions on sulphide minerals. Geochim. Cosmochim. Acta 1990, 54, 1957–1967. (26) Aller, R. C.; Rude, P. D. Complete oxidation of solid phase sulfides by manganese and bacteria in anoxic marine sediments. Geochim. Cosmochim. Acta 1988, 52, 751–765. (27) Miller, P. L.; Vasudevan, D.; Gschwend, P. M.; Roberts, A. L. Transformation of hexachloroethane in a sulfidic natural water. Environ. Sci. Technol. 1998, 32, 1269–1275. (28) Kriegman-King, M. R.; Reinhard, M. Transformation of carbon tetrachloride in the presence of sulfide, biotite, and vermiculite. Environ. Sci. Technol. 1992, 26, 2198–2206. (29) Doong, R.-A.; Wu, S.-C. Reductive dechlorination of chlorinated hydrocarbons in aqueous solutions containing ferrous and sulfide ions. Chemosphere 1992, 24, 1063–1075. (30) Wolthers, M.; van der Gaast, S. J.; Rickard, D. The structure of disordered mackinawite. Am. Mineral. 2003, 88, 2007–2015. (31) Jeong, H. Y.; Kim, H.; Hayes, K. F. Reductive dechlorination pathways of tetrachloroethylene and trichloroethylene and subsequent transformation of their dechlorination products by mackinawite (FeS) in the presence of metals. Environ. Sci. Technol. 2007, 41, 7736–7743. (32) Cui, D.; Eriksen, T. E. Reduction of pertechnetate by ferrous iron in solution: Influence of sorbed and precipitated Fe(II). Environ. Sci. Technol. 1996, 30, 2259–2262. (33) Amonette, J. E.; Workman, D. J.; Kennedy, D. W.; Fruchter, J. S.; Gorby, Y. A. Dechlorination of carbon tetrachloride by Fe(II) associated with goethite. Environ. Sci. Technol. 2000, 34, 4606–4613. (34) Doelsch, E.; Rose, J.; Masion, A.; Bottero, J. Y.; Nahon, D.; Bertsch, P. M. Hydrolysis of iron(II) chloride under anoxic conditions and influence of SiO4 ligands. Langmuir 2002, 18, 4292–4299. (35) Genin, J.-M. R.; Bourrie, G.; Trolard, F.; Abdelmoula, M.; Jaffrezic, A.; Refait, P.; Maitre, V.; Humbert, B.; Herbillon, A. Thermodynamic equilibra in aqueous suspensions of synthetic and natural Fe(II)Fe(III) green rusts: Occurrences of the mineral in hydromorphic soils. Environ. Sci. Technol. 1998, 32, 1058–1068. (36) Hyun, S. P.; Hayes, K. F. Feasibility of using in-situ FeS precipitation for TCE degradation. J. Environ. Eng. 2009, 135, 1009–1014. (37) Choi, J.; Batchelor, B.; Chung, J. Reductive dechlorination of tetrachloroethylene by green rusts modified by copper. Water Air Soil Pollut. 2010, 212, 407–417. (38) Shao, H.; Butler, E. C. The influence of iron and sulfur mineral fractions on carbon tetrachloride transformation in model anaerobic soils and sediments. Chemosphere 2007, 68, 1807–1813. (39) Butler, E. C.; Hayes, K. F. Effects of solution composition and pH on the reductive dechlorination of hexachloroethane by iron sulfide. Environ. Sci. Technol. 1998, 32, 1276–1284. (40) Zhang, Y; Cremer, P. S. Interactions between macromolecules and ions: The Hofmeister series. Curr. Opin. Chem. Biol. 2006, 10, 658–63. (41) Vikesland, P. J.; Heathcock, A. M.; Rebodos, R. L.; Makus, K. E. Particle size and aggregation effects on magnetite reactivity toward carbon tetrachloride. Environ. Sci. Technol. 2007, 41, 5277–5283. (42) Nirmalakhandan, N. N.; Speece, R. E. QSAR model for predicting Henry’s constant. Environ. Sci. Technol. 1988, 22, 1349–1357.
ARTICLE
(43) Kenneke, J. F.; Weber, E. J. Reductive dehalogenation of halomethanes in iron- and sulfate-reducing sediments. 1. Reactivity pattern analysis. Environ. Sci. Technol. 2003, 37, 713–720. (44) Berner, R. A. A new geochemical classification of sedimentary environments. J. Sediment. Petrol. 1981, 51, 359–365.
5194
dx.doi.org/10.1021/es104387w |Environ. Sci. Technol. 2011, 45, 5186–5194
