Environ. Sci. Technol. 2006, 40, 6837-6843
Exploring the Influence of Granular Iron Additives on 1,1,1-Trichloroethane Reduction DAVID M. CWIERTNY,† STEPHEN J. BRANSFIELD,‡ KENNETH J. T. LIVI,§ D. HOWARD FAIRBROTHER,‡ AND A . L Y N N R O B E R T S * ,† Departments of Geography and Environmental Engineering, Chemistry, and Earth and Planetary Sciences, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218
Bimetallic reductants are frequently more reactive toward organohalides than unamended iron and can also alter product distributions, yet a molecular-level explanation for these phenomena remains elusive. In this study, surface characterization of six iron-based bimetallic reductants (Au/Fe, Co/Fe, Cu/Fe, Ni/Fe, Pd/Fe, and Pt/Fe) revealed that displacement plating produced a non-uniform overlayer of metallic additive on iron. Batch studies demonstrated that not all additives enhanced rates of 1,1,1-trichloroethane (1,1,1-TCA) reduction nor was there any clear periodic trend in the observed reactivity (Ni/Fe ≈ Pd/Fe > Cu/Fe > Co/ Fe > Au/Fe ≈ Fe > Pt/Fe). Pseudo-first-order rate constants for 1,1,1-TCA reduction (kobs values) did, however, correlate closely with the solubility of atomic hydrogen within each additive. This suggests absorbed atomic hydrogen, rather than galvanic corrosion, is responsible for the enhanced reactivity of bimetallic reductants. In addition, all additives shifted product distributions to favor the combined yield of ethylene plus ethane over 1,1-dichloroethane. In rateenhancing bimetallic systems, branching ratios between 1,1dichloroethane and the combination of ethylene and ethane were uniquely dependent on kobs values, indicating an intimate link between rate-determining and productdetermining steps. We propose that our results are best explained by an X-philic pathway involving atomic hydrogen with a hydride-like character.
Introduction Iron-based bimetallic reductants represent an attractive alternative for treatment of contaminated groundwater because of their ability to increase the rates of organohalide reduction (1) and to produce more fully dehalogenated reduction products (2). While concerns regarding the longevity of these materials may limit their use in permeable reactive barriers (3), subsurface injection of nanoscale reductants (4) has emerged as a promising option for source zone control. An improved understanding of the metal additive’s role in organohalide reduction might allow the * Corresponding author e-mail: lroberts@jhu.edu; phone: (410) 516-4387; fax: (410) 516-8996. † Department of Geography and Environmental Engineering. ‡ Department of Chemistry, § Department of Earth and Planetary Sciences. 10.1021/es060921v CCC: $33.50 Published on Web 10/04/2006
2006 American Chemical Society
performance of iron-based bimetallic reductants to be optimized. At present, the molecular-level phenomena responsible for the enhanced reactivity commonly associated with bimetallic reductants remain unclear. Competing theories exist, and most can be related to specific properties of the metal additives employed. Some researchers contend (5) that the increased reactivity is dependent upon the ability of the additives to enhance the rates of iron corrosion through the formation of a galvanic couple. In such a scenario, the reactivity of different bimetals might correlate with the additives’ corrosion potentials, a parameter that will influence the driving force for galvanic corrosion (6). Others postulate (7, 8) that faster rates of organohalide degradation result from an additive’s role as a hydrogenation catalyst. Reductant reactivity in this case might be expected to scale with a metal additive’s capacity to generate and store adsorbed (7-11) or absorbed (9, 12, 13) atomic hydrogen. One means of testing these hypotheses would be to compare the reactivity of a suite of iron-based bimetallic reductants. Unfortunately, prior studies considering multiple bimetals have been limited (1, 8, 13-15), and most (1, 14, 15) failed to quantify the additive loading on iron. As this parameter influences reductant reactivity (16), its omission complicates attempts to develop a reactivity trend that could test competing explanations. Additive loadings were quantified by Kim and Carraway (13) and by Lin et al. (8) in studies of trichloroethylene (TCE) reduction by iron-based bimetals. Although no attempt was made by Kim and Carraway to relate their observed reactivity trend to specific additive properties, the reactivity trend developed by Lin et al. (8) was attributed to an additive’s ability to adsorb atomic hydrogen. This assertion was based on a limited correlation between the reactivity of Ru/Fe and Au/Fe reductants and exchange current densities (i0 values) for hydrogen evolution taken from the literature for these additives. The authors contend that the i0 values, which represent the magnitude of current flow for an electrochemical reaction at equilibrium, should be proportional to the adsorbed atomic hydrogen concentrations in their systems. Previous investigations have also failed to clearly establish the metal additive’s role in organohalide reduction product partitioning. For instance, Kim and Carraway (13) report that TCE reduction by Ni/Fe, Cu/Fe, and Pd/Fe yields fully dehalogenated end products without the detection of halogenated intermediates. Lin et al. (8), on the other hand, observed the formation of partially halogenated, and in some cases persistent, products of TCE reduction in Pd/Fe, Pt/Fe, Ru/Fe, and Au/Fe systems. Moreover, these researchers noted that the yield of the halogenated products was independent of the metal additives considered. In contrast, Fennelly and Roberts (1) observed different 1,1,1-trichloroethane (1,1,1TCA) product distributions in Ni/Fe and Cu/Fe systems. A study that systematically examines trends in product partitioning for a broad suite of bimetallic reductants may help to resolve the influence of granular iron metal additives on the relative yields of organohalide reduction products. The foundation for the present study was established in our earlier work with Cu/Fe reductants (16). In that study, we observed a pronounced influence of Cu mass loading (determined from the loss of Cu in plating solutions) and two-dimensional Cu surface coverage (determined by Auger electron spectroscopy (AES)) on Cu/Fe reactivity toward 1,1,1TCA in batch systems. From the dependence of pseudofirst-order rate constants (kobs values) on these variables, we VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6837
hypothesized that 1,1,1-TCA reduction occurred primarily at Cu sites on the reductant surface and not at the interface between Cu and Fe. We also proposed that the enhanced reactivity of Cu/Fe might result from the production and storage of reactive atomic hydrogen within the developing Cu lattice. In the current study, we examine five additional bimetals (Au/Fe, Co/Fe, Ni/Fe, Pd/Fe, and Pt/Fe) and investigate their reactivity toward 1,1,1-TCA as a means of testing hypotheses presented in our earlier work. Prior to reaction, reductants were characterized by an array of analytical techniques (AES, X-ray photoelectron spectroscopy (XPS), and scanning electron microscopy with energy dispersive X-ray analysis (SEM-EDS)). Batch studies yielded rate constants for 1,1,1TCA reduction, while modeling of reaction time course data enabled quantification of product branching ratios. The examination of these parameters as a function of additive loading enabled direct comparison of each additive’s influence on 1,1,1-TCA dehalogenation. Insights into the possible roles played by granular iron metal additives were obtained through comparing experimental reactivity trends to various metal additive properties that might be anticipated to influence bimetal reactivity. In fact, the additives studied were specifically selected in hopes of observing clear periodic trends in reactivity, analogous to those encountered for many chemical reactions catalyzed by transition metal surfaces (17). Of particular interest was the comparison of our experimental trend to those predicted on the basis of each additive’s ability to form a galvanic couple with iron and to accumulate reduced hydrogen species.
Experimental Section Chemicals. All reagents are described in the Supporting Information (hereafter referred to as SI). Bimetallic Reductant Preparation. Bimetallic reductants were generated via displacement plating of Fisher electrolytic iron with metal chloride salts of each additive. Details are provided in Table S-1 and the accompanying text in the SI. Quantification of Deposited Metal Additive Mass. The mass of metal additive deposited during displacement plating was determined from measured losses of metal ion concentration in plating solutions after reductive precipitation onto the granular iron. Metal ion concentrations were measured via atomic absorption spectroscopy (AAS). Details are included in the SI. Deposition of Pd and Pt required special consideration. Results from AAS suggested that all Pd and Pt initially present in plating solutions was deposited on the iron. Upon exposure of iron to the most highly concentrated Pd and Pt plating solutions (>1 mM), however, small amounts of black precipitates were observed. As the mass of these precipitates was not quantified, it represents an unknown amount of Pd or Pt that was not deposited on the iron. Reported mass loadings for the reductants with the highest Pd (50 µmoles Pd/g Fe) and Pt (63 µmoles Pt/g Fe) contents should therefore be regarded as upper limits. Reductant Characterization. All characterization techniques were applied to samples of dried bimetallic particles prior to reaction with 1,1,1-TCA. Details pertaining to reductant characterization with XPS, AES, and SEM-EDS are provided elsewhere (16). XPS binding energies used to identify metal oxidation states were taken from Muilenburg (18). Twodimensional additive surface coverages were determined from AES using a convention developed in our earlier work with Cu/Fe reductants (16). Surface coverages for Au and Pt are not reported because the characteristic AES peaks of these metals were obscured by spectral features associated with other elements on the reductant surface (e.g., adventitious oxygen and carbon, residual chlorine, or the underlying iron). 6838
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006
Surface Area Measurements. The specific surface area of reductants was measured with N2-BET adsorption using a Micromeritics ASAP 2010 surface area analyzer. Because ∼10 g of reductant was required for each analysis, the surface area measurements were limited to acid-washed granular iron and select Ni/Fe bimetals. In agreement with our results for Cu/Fe (16), the Ni/Fe reductants investigated possessed a specific surface area similar to that of unamended iron (∼0.2 m2/g). Batch Experimental Procedure. Batch systems investigating the reduction of 1,1,1-TCA were prepared and sampled in a manner identical to that described by Cwiertny and Roberts (19). Reactors were assembled within an anaerobic chamber and were initially free of headspace. Each reactor contained 0.1 M NaCl/50 mM Tris (pH 7.20) and a reductant loading of 1.6 g/L. Control experiments with unamended iron and with powders of each metal additive in the absence of iron were conducted according to the procedures outlined in the SI. As in our prior studies (19), headspace developed in our reactors over the course of most experiments, presumably from water reduction to H2. Minimal headspace evolved prior to the addition of 1,1,1-TCA to the reactors, while the volume of headspace produced after 1,1,1-TCA addition was estimated as described in ref 19. For most bimetals, this volume was similar to that observed in unamended iron systems, although Pd/Fe and Pt/Fe generated considerably more headspace than the other reductants. Gas Chromatography. Aqueous concentrations of 1,1,1TCA and its reduction products were determined by static headspace gas chromatography with flame ionization detection (GC/FID). The analytical methods and GC/FID system employed for this study are identical to those described in ref 19. All aqueous concentrations obtained from GC/FID have been corrected to account both for headspace partitioning and for dilution resulting from the addition of fresh buffer solution during reactor sampling, as described elsewhere (19). Kinetic Modeling. Product formation rate constants were determined from experimental time course data using nonlinear regression methods. Analyses were performed with Scientist for Windows (version 2.01, Micromath) assuming that 1,1,1-TCA reduction occurs via parallel reaction pathways to four groups of products (see Scheme S-1 in the SI). One product was 1,1-dichloroethane (1,1-DCA). Because ethylene was reduced to ethane by certain bimetals, the summed concentration of ethylene and ethane was considered to be a single product (referred to as EE). Similarly, the sum of cisand trans-2-butene and 2-butyne, presumably produced from radical coupling reactions, was also considered as a single reduction product (referred to as C4). An additional term, UN, was included to maintain system mass balance. This term accounted for any lost system mass, as well as for trace reaction products that were observed but not quantified (such as C5 and C6 species). Measured carbon mass balances were typically better than 80% in all reductant systems, such that UN represented at most 20% of the initial 1,1,1-TCA mass.
Results and Discussion Bimetallic Reductant Characterization. Surface spectroscopy revealed qualitative similarities among the bimetals investigated. For instance, XPS analyses indicated that all additives were deposited in their zerovalent oxidation states (representative XP spectra are shown in Figure S-1). In addition, results from AES (Figure S-2) demonstrated that increasing the mass of each additive deposited on iron produced an initial increase in the additive’s two-dimensional surface coverage, followed by an apparent plateau in surface coverage at higher mass loadings. An iron AES signal was observed for all additives and additive loadings considered. This was
FIGURE 1. Values of kobs for the reduction of 1,1,1-TCA as a function of deposited additive mass on both (a) linear and (b) logarithmic scales. The dashed horizontal lines in a and b represent kobs for 1,1,1-TCA reduction by unamended granular iron (kobs(Fe) ) 0.53 (( 0.05) × 10-4 s-1). The dashed vertical line in b indicates an additive loading of 13 µmol/g Fe at which the relative bimetallic reactivities toward 1,1,1-TCA were determined. Uncertainties represent 95% confidence intervals. interpreted (as with our previous results for Cu/Fe bimetals, ref 16) to signify that some iron remains exposed at, or very near to, the particle/solution interface for all bimetals investigated. Bimetallic reductants were further characterized via elemental mapping of grain cross sections using SEM-EDS. Elemental mapping of Cu/Fe (16), Pd/Fe (Figure S-3), Pt/Fe (Figure S-4), and Au/Fe (Figure S-4) revealed a non-uniform additive distribution across the reductant surface. Localized regions were observed with relatively thick additive overlayers while other regions contained no discernible additive. In contrast, SEM-EDS elemental mapping of Co/Fe and Ni/Fe reductants with comparable additive loadings (as determined by AAS) failed to reveal Co or Ni on the particle surface, perhaps indicating thinner deposits. We note that XPS and AES analyses confirmed the presence of Co and Ni in the near-surface region of these reductants. Moreover, AES elemental mapping of Ni/Fe bimetals (Figure S-5) illustrated that Ni was unevenly distributed on the iron, as SEM-EDS analyses indicated had been the case for Pd, Pt, Co, and Cu on iron. Thus, although the additive overlayers on our reductants share many similarities, some differences exist. As discussed in greater detail in the SI, these may stem from differences in the thermodynamic ease with which additives are reductively deposited on granular iron. Influence of Additive Loading on kobs Values for 1,1,1TCA Reduction. Pseudo-first-order rate constants (kobs values) for the reduction of 1,1,1-TCA by each of our bimetals are shown in Figure 1. Data are plotted as a function of deposited additive mass (from AAS) on both a linear (Figure 1a) and a logarithmic (Figure 1b) scale. For Ni/Fe, Pd/Fe, Cu/Fe, and
Co/Fe, an increase of the mass of deposited additive generally increased kobs for 1,1,1-TCA reduction. Moreover, all kobs values in these systems were greater than that measured with unamended iron (shown as the dashed horizontal line). In contrast, Au failed to exert a discernible influence on kobs at any loading tested, while increasing Pt loading progressively inhibited 1,1,1-TCA reduction rates. We note that in the absence of iron, powders of each metal additive were essentially unreactive toward 1,1,1-TCA, consistent with iron serving as the primary source of reducing equivalents in our systems. To assess relative reactivity, rate constants for 1,1,1-TCA reduction by each bimetal were compared at an additive loading of approximately 13 µmol/g Fe (shown as the dashed vertical line in Figure 1b; plots of 1,1,1-TCA concentration versus time for each bimetal with this loading are shown in Figure S-6a). As discussed in our earlier work (16) this value approximates the loading at which one monolayer of each additive is anticipated on the iron surface. At this loading, Ni/Fe was most reactive, producing nearly an 8-fold increase in kobs for 1,1,1-TCA reduction. In contrast, Pt/Fe depressed kobs for 1,1,1-TCA reduction by ∼40% (Figure S-6b). One experimental variable that could potentially influence the reactivity of our bimetals is system pH. As in our study with unamended iron (19), the pH in all batch reactors increased over time, a behavior that might influence the magnitude of rate enhancement (or inhibition) reported for each bimetal. For most bimetallic systems, however, the pH increase (∼0.2 pH units over the reactor lifetime (∼30 h)) was comparable to that measured with unamended iron (Figure S-7). Pd/Fe and Pt/Fe reductants were exceptions with measured increases as great as ∼0.5 pH units. We note that at high Pt loadings (>1.25 µmol/g Fe), changes in the reactor pH became independent of the deposited Pt mass (see Figure S-7 and accompanying discussion). As a 2-fold decrease in kobs values coincided with this same range of Pt loadings (see Figure 1), increases in solution pH cannot be solely responsible for the depressed reactivity of Pt/Fe reductants. Kim and Carraway (15) also encountered slower rates of pentachlorophenol reduction by iron-based bimetals than they measured with unamended iron. An important distinction, however, is that pentachlorophenol reduction was inhibited by all metal additives investigated (Cu, Pd, Ni, and Pt) in their study. In contrast, we found that certain additives (Ni, Pd, Cu, and Co) functioned as promoters of 1,1,1-TCA reductive dehalogenation, while others (Pt) acted as inhibitors. For additives that promoted 1,1,1-TCA reduction, the rate of increase in kobs was greatest at loadings of less than ∼10 µmol/g Fe (see Figure 1a). This behavior was most evident for Ni/Fe, which yielded statistically equivalent kobs values for all loadings greater than this value. It is unlikely that this plateau resulted from reaction rates limited by mass transfer because we have measured kobs values for the reduction of cis-dichloroethylene (which possesses a diffusivity similar to that of 1,1,1-TCA) that are nearly three times greater in otherwise identical bimetallic batch systems (data not shown). We suggest that the dependence of kobs on additive mass loading results from changes in the additive overlayer structure similar to those postulated in our earlier work with Cu/Fe reductants (16). While all additives are unlikely to deposit on iron in an identical fashion, AES and SEM-EDS data support the hypothesis that, at low-additive loadings, deposition occurs directly onto iron; this scenario would provide the greatest increase in additive surface area. At higher loadings, deposition onto preexisting metal additive predominates, producing only a minimal increase in additive surface area. We propose, therefore, that the dependence of VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6839
kobs on Ni, Pd, Cu, and Co loadings is consistent with 1,1,1TCA reduction occurring at the surface of the deposited additive rather than at the additive-iron interface. As discussed in ref 16, reduction at the additive-iron interface would produce a different relationship between kobs and additive loading than that reported herein. Generalized Reactivity Trend. Although Pd/Fe was more reactive than Ni/Fe at additive loadings of less than ∼1 µmol/g Fe, the relative reactivity of our bimetals is generally consistent over the entire range of additive loadings investigated (Figure 1). This generalized reactivity trend for 1,1,1TCA reduction is as follows: Ni/Fe ≈ Pd/Fe > Cu/Fe > Co/ Fe > Au/Fe ≈ Fe alone > Pt/Fe. While Kim and Carraway (13) report a reactivity trend for TCE reduction (Pd/Fe > Ni/Fe > Cu/Fe > Fe) that tends to agree with our results, the magnitudes of rate enhancement they report for Pd/Fe (∼1000×) and Ni/Fe (∼100×) reductants with comparable additive loadings are far greater than those encountered herein. On the other hand, the trend for TCE reduction (Pd/Fe . Ru/Fe > Pt/Fe > Au/Fe > Fe) reported by Lin et al. (8) suggests that Pt/Fe may in certain instances be an effective promoter of organohalide reductive dehalogenation. The dissimilarities between our reactivity trend and those presented in these prior studies likely reflect the differences in the target organohalides. For instance, vinyl polyhalides (e.g., TCE) may react by a different mechanism or through different rate-determining steps in bimetallic systems than saturated alkyl polyhalides (e.g., 1,1,1-TCA). Differences in the extent of vinyl polyhalide adsorption to reactive surface sites relative to alkyl polyhalide adsorption could also account for the different reactivities displayed by these pollutant classes in bimetallic systems. Relating Bimetallic Reductant Reactivity to the Properties of Metal Additives. Although SEM-EDS elemental maps suggest that the additive overlayers on Pd/Fe, Pt/Fe, Au/Fe, and Cu/Fe reductants are similar, their reactivity toward 1,1,1TCA varies considerably. Thus, while overlayer characteristics may contribute to bimetal reactivity, they cannot entirely explain our reactivity trend. Rather, some other metal additive property is likely responsible for the enhanced rates of 1,1,1TCA reduction, although a cursory analysis fails to reveal any clear periodic trend in reactivity (select properties of the metal additives investigated are provided in Table S-2). Our reactivity trend is inconsistent with that expected if galvanic couple formation were to influence rates of 1,1,1TCA reduction. Specifically, as Au, Pt, and Pd typically possess similar corrosion potentials in aqueous systems (6), Au/Fe and Pt/Fe would be expected to exhibit comparable reactivity to Pd/Fe. We caution, however, that such arguments are based solely upon thermodynamic considerations and do not account for the possibility that iron corrosion in Au/Fe and Pt/Fe systems is kinetically limited. We can also consider the possible involvement of adsorbed atomic hydrogen (Hads) in our bimetallic systems. Lin et al. (8) proposed that exchange current densities for hydrogen evolution (i0 values) represent a metric for the concentration of Hads on a metal additive surface. If true, then the inhibitory effect of Pt is inconsistent with a prominent role for Hads in 1,1,1-TCA reduction because the i0 values cited in ref 8 for Pd and Pt are nearly equivalent (10-3 and 10-3.1 A/cm2, respectively). We further note that Raman spectroscopy (11) has demonstrated the existence of Hads on polycrystalline Pt in aqueous solutions. If we assume that Hads is also produced on the surface of our Pt/Fe reductants, then our results indicate that this species must not be particularly reactive toward 1,1,1-TCA. In contrast, we found a strong correlation between kobs for 1,1,1-TCA reduction and the solubility of atomic hydrogen within each additive (Figure 2). In Figure 2 the relative partial molar enthalpy for an infinitely dilute solution of hydrogen 6840
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006
FIGURE 2. Correlations between log(kobs) for 1,1,1-TCA reduction and ∆HH∞. Relationships were developed with kobs values measured at additive loadings of (a) ∼1 and (b) ∼13 µmol of deposited additive/g of Fe. Data for the reduction of 1,1,1-TCA by unamended Fisher electrolytic iron are also included using the ∆HH∞ value reported for r-Fe. The dashed lines represent linear regression fits to the experimental data. The data for r-Fe were not included in the regression analyses. in each additive (∆HH∞) (20) was chosen as a metric for atomic hydrogen solubility. This term represents a metal’s ability to absorb atomic hydrogen, with large, negative ∆HH∞ values corresponding to greater solubilities (21). The values of ∆HH∞ for each additive (taken from ref 20) are provided in Table S-3. The correlations in Figure 2 suggest that hydrogen absorbed in the metal additive’s lattice participates in 1,1,1TCA reduction, as previously proposed by Cheng et al. (12) for the reduction of 4-chlorophenol by Pd and Pt electrodes. Electronic structure calculations (22) of H absorbed in metals suggest that such atoms possess excess electron density and can be considered anionic in character. On the basis of the correlations in Figure 2, we therefore hypothesize that atomic hydrogen generated via the reduction of water is absorbed into the additive lattice, thereby producing a hydride-like species (H-) that represents the primary redox-active entity in our bimetallic systems. Involvement of such species would likely require its presence in the first few subsurface layers of the additive, which would facilitate its diffusion to the additive-solution interface where 1,1,1-TCA dehalogenation primarily occurs. Consistent with our hypothesis, hydrides (23) and subsurface hydrogen (9) have previously been proposed as reactive species toward vinyl polyhalides in ironbased bimetallic systems. In Figure 2, the quality of the correlation between kobs and ∆HH∞, as well as the trend in bimetal reactivity toward 1,1,1-TCA, was somewhat dependent upon the additive loading at which the rate constants were measured. For instance, a larger R2 value was obtained with kobs values determined at low (∼1 µmol/g Fe, Figure 2a) rather than high (∼13 µmol/g Fe, Figure 2b) additive loadings. This may reflect differences
FIGURE 4. Relationship between kDCA/kEE and kobs for metal additives found to enhance the rate of 1,1,1-TCA reduction (Ni, Pd, Cu, and Co) over the range of additive loadings tested. Data from an unamended iron system are included for purposes of comparison.
FIGURE 3. Values of kDCA/kEE as a function of deposited additive mass on a logarithmic scale for the (a) Ni/Fe, Pd/Fe, Cu/Fe, and Co/Fe reductant systems (which enhanced rates of 1,1,1-TCA reduction), as well as for the (b) Pt/Fe and Au/Fe bimetals (which did not enhance rates of 1,1,1-TCA reduction). The dashed horizontal line represents the kDCA/kEE value measured in an unamended iron system (kDCA/kEE(Fe) ) 3.2 ( 0.4). among our reductants in the point at which displacement plating onto previously deposited additive begins to predominate. Such differences would cause relative trends in additive surface area to change as loading increases. We note that the reactivity of unamended iron appears to be less than that anticipated from the correlations developed for the bimetallic reductants. This does not rule out the possibility that absorbed atomic hydrogen may play a role in unamended iron systems, as the surface of our iron particles is likely to be coated by Fe(II) and Fe(III) (hydr)oxides; the solubility of atomic hydrogen in such phases may not be identical to that reported for R-Fe. Influence of Metal Additives on 1,1,1-TCA Product Partitioning. Nonlinear regression analyses (model fits shown as dashed lines in Figure S-8) revealed that the branching ratio between 1,1-DCA formation and ethane plus ethylene formation (defined as kDCA/kEE) was less in all bimetallic systems than that obtained with unamended iron (Figure 3). This ratio also tended to decrease with increasing additive loading. Thus, even though certain reductants (Au/ Fe and Pt/Fe) failed to increase the rate of 1,1,1-TCA reduction, all bimetals altered 1,1,1-TCA product partitioning to favor the formation of fully dehalogenated products. Independent experiments revealed 1,1-DCA to be nearly unreactive toward all bimetals. 1,1-DCA cannot, therefore, represent an intermediate to ethylene or ethane formation in our systems. Instead, observed changes in product partitioning must result from a shift in the pathway through which 1,1,1-TCA undergoes reduction in the presence of the metal additives. Shifts in the 1,1,1-TCA reduction pathway are best discerned by the examination of trends in kDCA and kEE as a function of additive loading (Figure S-9). For reductants that enhanced the rate of 1,1,1-TCA reduction, values of both
kDCA and kEE were generally greater than those observed in an unamended iron system. These values also increased with additive loading, although increases in kEE were usually considerably greater than those observed for kDCA. As a result, the yield of ethane and ethylene (defined as kEE/kobs) was enhanced relative to unamended iron, while the yield of 1,1DCA (kDCA/kobs) was depressed (product yields for each reductant are shown in Figure S-10). Although similar trends in kEE were observed with Au/Fe and Pt/Fe bimetals, values of kDCA tended to decrease with additive loading in these systems. As with the other bimetals, these trends increased the yield of ethane and ethylene relative to unamended iron, while decreasing the yield of 1,1-DCA. Interestingly, our results revealed a close relationship between kDCA/kEE and kobs in Pd/Fe, Ni/Fe, Cu/Fe, and Co/ Fe reductant systems (Figure 4). Specifically, large values of kobs produced small values of kDCA/kEE. Furthermore, the dependence of kDCA/kEE on kobs appears to be independent of the metal additive identity or additive loading used to achieve the kobs value in question. Thus, the predominant factor governing 1,1,1-TCA product partitioning is not so much the identity or loading of the metal additive, but rather the reactivity of the bimetal toward 1,1,1-TCA reduction. This suggests that Pd/Fe, Ni/Fe, Cu/Fe, and Co/Fe enhance the rate of 1,1,1-TCA reduction by a common mechanism. Data from Pt/Fe and Au/Fe reductant systems did not adhere to the correlation in Figure 4, as these bimetals reduced values of kDCA/kEE without increasing kobs for 1,1,1TCA reduction. It is likely, therefore, that the altered branching ratios observed in these systems result from different product formation pathways than those in rate-enhancing bimetallic systems. Proposed Role of Metal Additives in 1,1,1-TCA Reduction. Fennelly and Roberts (1) proposed that the reduction of 1,1,1-TCA by granular iron proceeds via a dissociative oneelectron transfer (eq 1). This reaction yields a dichloroethyl radical that others (24) suggest is associated with the reductant surface.
H3C-CCl3 + e- f H3C-C˙ Cl2 + Cl-
(1)
Such radicals could couple (giving rise to C4 products) or else react with H• to generate the hydrogenolysis product 1,1-DCA (eq 2). One potential source of H• in our batch system may be surface-adsorbed atomic hydrogen, which is generated via water reduction as an intermediate in H2 evolution.
M-Hads H3C-C˙ Cl2 98 H3C-CHCl2
(2)
The dichloroethyl radical could also undergo a second oneelectron reduction to generate a dichloroethyl carbanion (eq 3), whose subsequent protonation represents another VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6841
possible route for 1,1-DCA formation (eq 4). Alternatively
H3C-C˙ Cl2 + e- f [H3C-C ¨ Cl2]-
(3)
[H3C-C ¨ Cl2]- + H2O f H3C-CHCl2 + OH-
(4)
(and potentially in concert with the electron transfer depicted in eq 3), this carbanion could undergo R-elimination of Clto yield chloromethylcarbene, which may also be surfaceassociated (eq 5). Subsequent reduction and elimination steps
[H3C-C ¨ Cl2]- f H3C-C ¨ Cl + Cl-
(5)
could convert chloromethylcarbene to ethylene and ethane (1). Prior researchers (25, 26) have suggested that surface stabilization of carbon-containing radicals generated during organohalide reduction may influence product branching ratios. Accordingly, a possible outcome in our systems is that the metal additives stabilize the dichloroethyl radical intermediate generated via eq 1, thereby influencing its fate. Although our experimental product distributions might at first seem consistent with such a scenario, we argue that the ability of metal additives to stabilize carbon-containing radicals (presumably via an organometallic surface complex) should produce an obvious periodic trend in reduction product partitioning. Because no such trend could be discerned, alternative explanations require consideration. The unique dependence of kDCA/kEE on kobs in Figure 4 suggests an intimate link between the rate-limiting step and the product-forming step in 1,1,1-TCA reduction. Moreover, the data in Figure 2 suggest that absorbed atomic hydrogen plays a role in this reaction. If, as we proposed earlier, this species behaves as hydride (H-), an X-philic (halophilic) reaction might occur (a review of X-philic reactions is provided in ref 27). This would involve the nucleophilic attack of H- at a Cl on 1,1,1-TCA to produce a dichloroethyl carbanion, which would subsequently eliminate Cl- to yield chloromethylcarbene (eq 6).
Some fraction of this dichloroethyl carbanion could also protonate to generate 1,1-DCA, consistent with the modest increase in kDCA generally observed with increased additive loading in our rate-enhancing bimetallic systems (see Figure S-9). We therefore propose that the role of granular iron metal additives is to generate and store H-, thereby increasing the rate of an X-philic reaction. For this scenario, the values of both kobs and kDCA/kEE will depend upon how much H- has accumulated in a bimetallic system, which in turn is a function of the identity and the loading of the metal additive. While we cannot rule out the existence of M-Hads (or H•) in our bimetallic systems, our trends in product formation and the relationship between kDCA/kEE and kobs (Figure 4) do not support its role in 1,1,1-TCA reduction. We maintain that such a species with radical character would react with 1,1,1-TCA in a manner that would ultimately increase the production of 1,1-DCA (via eq 2) relative to ethane and ethylene. This change in product partitioning was not observed in any bimetallic system. We also note that values of kDCA/kEE in Figure 4 appear bounded by an upper limit near the branching ratio observed in unamended iron systems and a lower threshold at which values of kDCA/kEE become independent of kobs values. Such behavior suggests that experimental branching ratios in bimetallic systems may represent a weighted sum of those kDCA/kEE values characteristic of 1,1,1-TCA reduction by 6842
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 21, 2006
unamended iron (the upper limit of ∼3) or H- (the lower limit of ∼0.2-0.4). At low H- concentrations, 1,1,1-TCA reduction would occur primarily on iron, and the kDCA/kEE values would reflect that of those observed in an unamended iron system. On the other hand, smaller values of kDCA/kEE would be obtained when 1,1,1-TCA reduction occurs primarily at additive surfaces possessing elevated concentrations of H-. We further hypothesize that the lower limit in the kDCA/ kEE values reflects the branching between R-elimination (to chloromethylcarbene) and protonation (to 1,1-DCA) of the proposed dichloroethyl carbanion intermediate generated in eq 6. The similarity of this lower limit for the additives shown in Figure 4 may imply that the dichloroethyl carbanion is at best weakly associated with most bimetallic surfaces, so that its tendency to undergo R-elimination or protonation would be constant. We cannot, however, rule out the possibility that additional factors influence product formation in Pd/Fe systems because it was the only rate-enhancing bimetal for which the values of kDCA did not increase with additive loading (see Figure S-9). Finally, because solution-phase hydrides are frequently regarded as participating in dehalogenation via a bimolecular nucleophilic substitution (SN2) reaction at a carbon center (28), it is worth considering if such a reaction could occur in our bimetallic systems. Our data do not support this possibility; 1,1-DCA, which should be more susceptible to nucleophilic attack at carbon than 1,1,1-TCA, proved to be unreactive. We contend, therefore, that if surface-absorbed H- participates in 1,1,1-TCA reduction, its behavior must be fundamentally different from that commonly associated with solution-phase hydrides (e.g., metal hydride complexes). The possibility of such differences illustrates the need for additional experimental investigations before the exact role of reactive atomic hydrogen species in iron-based bimetallic systems can be fully elucidated.
Acknowledgments This research was funded by the NSF through grant number BES0086755 and a Collaborative Research Activities in the Environmental Molecular Sciences (CRAEMS) grant (CHE0089168) in Environmental Redox-Mediated Dehalogenation Chemistry at Johns Hopkins University. Additional funding for DMC was provided by NSF and U.S. EPA Graduate Research Fellowships. The contents of this publication are solely the responsibility of the authors and do not necessarily reflect the official view of the NSF or the U.S. EPA. We are especially grateful for the efforts of four anonymous reviewers, whose insightful comments greatly improved this manuscript.
Supporting Information Available XPS spectra, results from AES analyses, SEM-EDS elemental maps, detailed experimental protocols, additional results pertaining to the reactor pH, the relative rate of 1,1,1-TCA reduction by each bimetal, and reduction product partitioning, as well as property tables for the additives investigated. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Fennelly, J. P.; Roberts, A. L. Reaction of 1,1,1-trichloroethane with zero-valent metals and bimetallic reductants. Environ. Sci. Technol. 1998, 32, 1980-1988. (2) Muftikian, R.; Fernando, Q.; Korte, N. A method for the rapid dechlorination of low molecular weight chlorinated hydrocarbons in water. Water Res. 1995, 29, 2434-2439. (3) Grenier, A.; McGuire, M. M.; Fairbrother, D. H.; Roberts, A. L. Treatment of vapor-phase organohalides with zero-valent iron and Ni/Fe reductants. Environ. Eng. Sci. 2004, 21, 421-435.
(4) Elliott, D. W.; Zhang, W. Field assessment of nanoscale bimetallic particles for groundwater treatment. Environ. Sci. Technol. 2001, 35, 4922-4926. (5) Xu, Y.; Zhang, W. Subcolloidal Fe/Ag particles for reductive dehalogenation of chlorinated benzenes. Ind. Eng. Chem. Res. 2000, 39, 2238-2244. (6) Jones, D. A. Principles and Prevention of Corrosion, 2nd ed.; Prentice Hall: Upper Saddle River, NJ, 1996. (7) Schrick, B.; Blough, J. L.; Jones, A. D.; Mallouk, T. E. Hydrodechlorination of trichloroethylene to hydrocarbons using bimetallic nickel-iron nanoparticles. Chem. Mater. 2002, 14, 5140-5147. (8) Lin, C. J.; Lo, S. L.; Liou, Y. H. Dechlorination of trichloroethylene in aqueous solution by noble metal-modified iron. J. Hazard. Mater. 2004, 116, 219-228. (9) Odziemkowski, M. S.; Gillham, R. W. Electroless Hydrogenation of Trichloroethylene by Fe-Ni(P) Galvanic Couples. In Environmental Issues in the Electronics/Semiconductor Industries and Electrochemical/Photochemical Methods for Pollution Abatement; Simpson, C. R., Mendicino, L., Rajeshwar, K., Fenton, J. M., Eds.; The Electrochemical Society, Inc.: Pennington, NJ, 1998; Vol. 98-5, pp 91-97. (10) Odziemkowski, M. S.; Gui, L.; Gillham, R. W. Reduction of N-nitrosodimethylamine with granular iron and nickelenhanced iron. 2. Mechanistic studies. Environ. Sci. Technol. 2000, 34, 3495-3500. (11) Odziemkowski, M. S.; Simpraga, R. P. Distribution of oxides on iron materials used for remediation of organic groundwater contaminantssImplications for hydrogen evolution reactions. Can. J. Chem. 2004, 82, 1495-1506. (12) Cheng, I. F.; Fernando, Q.; Korte, N. Electrochemical dechlorination of 4-chlorophenol to phenol. Environ. Sci. Technol. 1997, 31, 1074-1078. (13) Kim, Y.-H.; Carraway, E. R. Reductive dechlorination of TCE by zero valent bimetals. Environ. Technol. 2003, 24, 69-75. (14) Zhang, W.; Wang, C.-B.; Lien, H.-L. Treatment of chlorinated organic contaminants with nanoscale bimetallic particles. Catal. Today 1998, 40, 387-395. (15) Kim, Y.-H.; Carraway, E. R. Dechlorination of pentachlorophenol by zero valent iron and modified zero valent irons. Environ. Sci. Technol. 2000, 34, 2014-2017. (16) Bransfield, S. J.; Cwiertny, D. M.; Roberts, A. L.; Fairbrother, D. H. Influence of Cu loading and surface coverage on the reactivity
(17) (18) (19) (20) (21) (22) (23) (24) (25)
(26)
(27) (28)
of granular iron toward 1,1,1-trichloroethane. Environ. Sci. Technol. 2006, 40, 1485-1490. Somorjai, G. A. Introduction to Surface Chemistry and Catalysis; Wiley and Sons: New York, 1994. Muilenberg, G. E., Ed. The Handbook of X-ray Photoelectron Spectroscopy; Perkin Elmer Corporation: Eden Prairie, MN, 1979. Cwiertny, D. M.; Roberts, A. L. On the nonlinear relationship between kobs and reductant mass loading in iron batch systems. Environ. Sci. Technol. 2005, 39, 8948-8957. Gallagher, P. T.; Oates, W. A. Partial excess entropies of hydrogen in metals. Trans. Metall. Soc. AIME. 1969, 245, 179-182. McLellan, R. B.; Oates, W. A. The solubility of hydrogen in rhodium, ruthenium, iridium and nickel. Acta Metall. 1973, 21, 181-185. Jena, P.; Fradin, F. Y.; Ellis, D. E. Models of electronic structure of hydrogen in metals: Pd-H. Phys. Rev. B. 1979, 20, 35433551. Kim, Y.-H.; Carraway, E. R. Dechlorination of chlorinated ethenes and acetylenes by palladized iron. Environ. Technol. 2003, 24, 809-819. Arnold, W. A.; Winget, P.; Cramer, C. J. Reductive dechlorination of 1,1,2,2-tetrachloroethane. Environ. Sci. Technol. 2002, 36, 3536-3541. Ta´mara, M. L.; Butler, E. B. Effects of iron purity and groundwater characteristics on rates and products in the degradation of carbon tetrachloride by iron metal. Environ. Sci. Technol. 2004, 38, 1866-1876. Elsner, M.; Schwarzenbach, R. P.; Kellerhals, T.; Luzi, S.; Zwank, L.; Angst, W.; Haderlein, S. B. Mechanisms and products of surface-mediated reductive dehalogenation of carbon tetrachloride by Fe(II) on goethite. Environ. Sci. Technol. 2004, 38, 2058-2066. Zefirov, N. S.; Makhon’kov, D. I. X-philic reactions. Chem. Rev. 1982, 82, 615-624. Smith, M. B.; March, J. March’s Advanced Organic Chemistry: Reactions, Mechanisms and Structure, 5th ed.; Wiley and Sons: New York, 2001.
Received for review April 17, 2006. Revised manuscript received July 21, 2006. Accepted August 4, 2006. ES060921V
VOL. 40, NO. 21, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6843