Catalytic Role of Palladium and Relative Reactivity of Substituted

Pegasus Technical Services, Inc. , #. Current address: Department of Civil Engineering, University of Texas at Arlington, 416 Yates Drive, Arlington, ...
0 downloads 0 Views 1MB Size
Download PDF
Environ. Sci. Technol. 2009, 43, 7510–7515

Catalytic Role of Palladium and Relative Reactivity of Substituted Chlorines during Adsorption and Treatment of PCBs on Reactive Activated Carbon H Y E O K C H O I , †,# S O U H A I L R . A L - A B E D , * ,† AND SHIRISH AGARWAL‡ National Risk Management Research Laboratory, U.S. Environmental Protection Agency, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268, and Pegasus Technical Services, Inc., 46 East Hollister Street, Cincinnati, Ohio 45221

Received May 1, 2009. Revised manuscript received August 14, 2009. Accepted August 17, 2009.

The adsorption-mediated dechlorination of polychlorinated biphenyls (PCBs) is a unique feature of reactive activated carbon (RAC). Here, we address the RAC system, containing a tunable amount of Fe as a primary electron donor coupled with Pd as an electrochemical catalyst to potentially respond to the characteristic of contaminated sites, effectively traps and treats various PCB congeners. A dramatic increase in RAC reactivity with Pd doping at as low as 0.01% suggests its critical role for accelerating hydrodechlorination of PCBs. Characteristic adsorption and dechlorination behavior and ensuing decomposition pathways of 13 selected PCB congeners are discussed with their surface interactions with RAC. Important findings include (i) inherent dechlorination susceptibility of chlorines in para > meta > ortho position, regardless of independent or competitive conditions as well as substrate effects, (ii) favorable reduction of more toxic coplanar PCB congeners, (iii) preferential electrophilic attack to chlorines in a less substituted phenyl ring and an isolated chlorine, regardless of the steric or inductive effect as a dominant limiting factor for the dechlorination of ortho or meta PCBs, respectively, (iv) prominent dechlorination inhibition for higher ortho congeners but negligible inhibition for higher meta congeners, and (v) eventual complete dechlorination of higher PCB congeners to biphenyl.

Introduction In situ capping employing a physical barrier of clean material amended with adsorptive activated carbon and reactive iron particles has been proposed to effectively isolate sites contaminated with organic chemicals from the surrounding environment (1-4). More recently, so-called reactive activated carbon (RAC), in which iron/palladium (Fe/Pd) bimetallic nanoparticles are embedded into the pores of granular activated carbon (GAC), has been developed to treat hydrophobic chlorinated organic compounds such as polychlorinated biphenyls (PCBs) (5). The ability of GAC to place * Corresponding author phone: (513) 569-7849; fax: (513) 5697879; e-mail: [email protected]. † U.S. Environmental Protection Agency. ‡ Pegasus Technical Services, Inc. # Current address: Department of Civil Engineering, University of Texas at Arlington, 416 Yates Drive, Arlington, Texas 76019-0308. 7510

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

PCBs into its pores in close proximity to the Fe/Pd particles and role of Pd for enhancing Fe reactivity to dechlorinate PCBs make the RAC system attractive for its full scale application to contaminated sites (6, 7). The type, amount, and subsequent cost of the metals might be determined based on the characteristics of the contaminated site (e.g., presumably contamination level and water flow rate) and purpose of treatment. Fe, when coupled with a less active metal such as Pd, Cu, Co, and Ni, oxidizes faster, inducing increase in its reactivity but decrease in its longevity to dechlorinate PCBs (vice versa in coupling with more active metal). In general, even a small amount of Pd as a catalyst governs the dechlorination kinetics (8) while Fe as an electron donor determines the dechlorination longevity. A considerable amount of Pd on Fe might be needed for the prompt treatment of PCBs available in the liquid phase at high concentration, whereas a high content of Fe coupled with a trace amount of Pd might be preferred to respond to long-term and low level release of PCBs from a solid matrix such as aquatic sediments. In addition, the oxidation and particle growth of Fe over time affect the PCB permeability and adsorption capacity of RAC (7, 9). Two possible adsorption sites for PCBs might exist in RAC, i.e., Fe/Pd vs AC uncovered with the Fe/Pd. Considering complete dechlorination of PCBs once adsorbed to RAC (6) and the way RAC is synthesized (5), Fe particles seem to occupy most of AC surface (but RAC still has huge pore space to hold or adsorb PCBs) or at least to locate very close to any of the pristine AC sites enough to cause dechlorination of PCBs even adsorbed to the AC. This depends on Fe loading. As Fe and/ or Pd content increase, the RAC is expected to show slower kinetics for adsorption of PCBs but faster kinetics for their dechlorination if the two reactions are independent. However, availability of PCBs for the dechlorination in the RAC phase is largely limited by their decreased adsorption kinetics. Consequently, control of the metal content during RAC synthesis and its ensuing effect on PCB treatment are important. Meanwhile, treatability of metallic systems to highly chlorinated and positional PCBs is another important issue (10-13). In our previous studies (5-7), 2-chlorobiphenyl (2ClBP) was used as a model PCB due to the simplicity of system analysis. However, tri-, tetra-, penta-, and hexa-chlorinated biphenyls are the most problematic in real world treatment, one of the major concerns invoked by our research group and others (7). In general, ortho dechlorination may not match meta or para dechlorination, and dechlorination of a single chlorine may not match that of multiple chlorines. Even a large amount of nanoscale Fe was not effective to completely reduce highly chlorinated PCBs to BP (12). An apparent variation in relative resistance of substituted chlorines in PCBs to reductive dechlorination on metal surface was reported (11-13). Despite many theories proposed to explain the dechlorination susceptibility, conflicting results have been reported, depending on experimental conditions and metal particles used (10-17). Some reported faster dechlorination for higher PCB congeners (16) while some indicated the opposite tendency (17). In this RAC system, adsorption of PCBs mediated by GAC should proceed prior to their dechlorination on Fe/Pd (5, 6). The effect of substituted chlorines in PCBs on their reactivity with RAC is uncertain because various PCB congeners with considerably different nature and affinity for the GAC might exhibit different adsorption behavior and thus dechlorination susceptibility. RAC might strike a balance between strength of adsorption and rate of reaction. 10.1021/es901298b CCC: $40.75

 2009 American Chemical Society

Published on Web 08/25/2009

FIGURE 1. Fe and Pd effects on RAC performance: (a) relation between Fe added to GAC and that incorporated into RAC, (b) effect of Fe loading to RAC (Pd at 0.7%) on adsorption and dechlorination of 2-ClBP after 12 h reaction, (c) relation between Pd added to GAC and that incorporated into RAC, (d) effect of Pd loading to RAC (Fe at 14%) on adsorption and dechlorination of 2-ClBP after 4 h reaction, and (e) effect of low Pd loading to RAC (Fe at 14%) on 2-ClBP dechlorination kinetics. The slight decrease in the dechlorination efficiency over Fe loading in (b) is due to the decrease in the amount of 2-ClBP adsorbed to RAC, which is available for the surface reaction. The error bar is the standard deviation of triplicated results. Consequently, the main objective of this study is to investigate the ability of the RAC system with tunable amounts of metals to treat various PCB congeners. The effect of Fe and Pd content on the performance of RAC for PCB treatment was studied particularly to prove the catalytic effect of a small amount of Pd doping. In parallel, treatability of RAC toward various substituted chlorines in PCBs was tested to elucidate PCB structure-dechlorination relation and subsequent dechlorination pathways of PCBs.

Experimental Section Synthesis of RAC. Detailed descriptions for the procedure, chemistry, and route of RAC synthesis were reported elsewhere (5). The synthesis procedure of RAC containing around 14.4% Fe and 0.7% Pd (weight-base) is denoted as standard. To synthesize RAC composites with variable Fe and Pd content, the standard condition was slightly modified. For various Fe loading, desired amount of Fe(NO3)3 was embedded to GAC and the other steps were the same as those in the standard procedure except for the amount of NaBH4 added proportionally to the Fe added. This resulted in a set of RAC composites with various Fe content at up to 37% and fixed Pd content at 0.7%. Another set of RAC composites with various Pd content at up to 5% and fixed Fe content at 14.4% was prepared by adding certain amounts of Pd(CH3CO2)2 to GAC/Fe composite, where Pd2+ was reductively deposited to zerovalent iron (ZVI) surface. The Fe and Pd metal contents in RAC were measured using inductively coupled plasma-atomic emission spectroscopy (IRIS Intrepid, Thermo Electron Corporation). Batch Reaction. All experiments were based on sacrificial batch in anaerobic condition, as described elsewhere (6). For simple comparison purposes, 2-ClBP (Accustandard) was used to study the effect of metal content. A glass vial (Fisher) containing 10 mL of around 4 mg/L 2-ClBP dissolved in water was mixed with 1 g of RAC (standard batch setup). The solution was agitated using a gyroshaker at 60 rpm for up to 30 d. To study the reactivity of substituted chlorines in PCBs with the standard RAC containing around 14.4% Fe and 0.7% Pd, 13 PCB congeners were selected (Table S1 in Supporting Information). Individual mono PCB (2, 3, and 4-ClBP), mixture of the mono PCBs, and 234-triClBP were tested to elucidate dechlorination susceptibility of positional chlorines under independent or competitive conditions. Higher ortho congeners (2-, 22′-, 26-, 22′6-, and 22′66′-ClBP), meta congeners (3-, 33′-, 35-, 33′5-, and 33′55′-ClBP), and para congeners (4- and 44′-ClBP) prepared in ethanol/water solution (1:1) (18) were used to investigate the effect of chlorination degree of PCBs. Most of the experiments were triplicated, if statistically essential.

Analytical Methods. A reasonable sampling schedule was set up to best accommodate comparison of experimental results, depending on the reaction kinetics in each set of experiments. One reactor was sacrificed each time for extraction and measurement of PCBs and BP in the liquid and RAC solid phases. A detailed description for the extraction (hexane extraction for liquid samples and automated Soxhlet for solid samples) and measurement (gas chromatograph/ mass spectrometer) of PCBs was reported elsewhere (6).

Results and Discussion Detailed information on many new definitions, concepts, and data handling and interpretation associated with this unique RAC system (e.g., extraction efficiency, mass recovery, overall and solid phase dechlorination, PCB congener ratio) was provided in our previous studies (6, 7). Impregnation of Fe and Catalytic Role of Pd. Control of the metal content in RAC is important for responding to site-specific treatment goals. As shown in Figure 1a, all Fe added in case of low loading was completely imbedded in GAC with high pore volume of 0.639 cm3/g. However, too high Fe loading did not result in linear increase in the amount of Fe incorporated. Rather, a level-off was observed at around 30%. This is a strong indication that Fe was primarily incorporated into the GAC pores rather than being stacked on the GAC grain boundary, which is beneficial to the mechanical stability of Fe/Pd and protection of its reactivity from instant oxidation (5, 7). As shown in Figure 1b, 2-ClBP adsorption efficiency of RAC was inversely proportional to its Fe content since surface area of GAC at 574 m2/g decreased to, for example, 358 m2/g after 14% Fe incorporation. Regardless of Fe content at 9-37%, RAC with the fixed amount of Pd at around 0.7% showed almost stable dechlorination ability at 60-70%. The driving force of Pd deposition onto Fe in GAC is completely different from that of physical placement of Fe into GAC pores. As shown in Figure 1c, Pd added to GAC/Fe was completely deposited (i.e., linear response). Fast reductive conversion of Pd2+ to elemental Pd on the surface of ZVI was observed as the brownish dark Pd(CH3CO2)2 solution mixed with GAC/ZVI turned colorless (5). The increase in Pd content up to 5% did not affect adsorption efficiency of RAC at around 60% (Figure 1d) since Pd deposition occurred onto Fe rather than GAC and thus Pd content in the low range negligibly affected the structural properties of RAC. However, Pd doping to Fe surface drastically increased the dechlorination reaction kinetics. Only GAC/Fe showed negligible reactivity within 4 h. A similar result was also observed in trichloroethylene dechlorination, where Fe alone exhibited VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7511

FIGURE 2. Dechlorination efficiencies after 24 h reaction of RAC with individual mono PCB congener (left side) and mixture of the mono PCB congeners (right side). The error bar is the standard deviation of triplicated results. around 80 times lower first order reaction rate than Fe with 1-5% Pd (8). In case of Fe, direct reduction of 2-ClBP occurs via formation of an organic complex adsorbed to the surface of Fe which serves as an electron donor. Meanwhile, indirect but faster reduction of 2-ClBP at the Pd surface also occurs in the Fe/Pd system, involving atomic hydrogen as a powerful secondary reducing agent, which is produced via electrolysis of water at the Fe surface and transferred to Pd surface (8, 19). In the Fe/Pd system, the surface of Pd with the low cathodic hydrogen overpotential is covered by atomic hydrogen due to low activation barrier for H2 dissociation, while such an indirect reduction is not considerable on Fe surface with high hydrogen overpotential (20, 21). The linear increase in the reactivity over Pd doping in Figure 1d strongly supports the critical role of Pd. Pd serves as an electrocatalyst to control the reaction kinetics. Once the amount of Fe in RAC is fixed, its dechlorination capacity is almost determined regardless of Pd content. As shown in Figure 1e, the dechlorination occurred on the Fe surface even without Pd doping, but significantly below batch reactions with Pd loading. A small amount of Pd worked in such a bimetallic system (22). After 7 d reaction, half of the 2-ClBP added was dechlorinated through Pd-mediated hydrodechlorination in the presence of even 0.01% Pd. As reported previously (7), Pd dissolution is limited due to the spontaneous redeposition process of dissolved Pd species (if any in solution) to Fe surface (5, 22). It was also confirmed that there is no apparent dechlorination reaction in the absence of Fe and/or Pd in GAC. Dechlorination of Mono PCB Congener under Independent or Competitive Conditions. The dechlorination efficiency after 24 h treatment of individual mono PCB congener is summarized in Figure 2 (left side) (also note Table S2). The dechlorination increased in order of 4-ClBP > 3-ClBP > 2-ClBP, which is in a good agreement with a general observation on relative dechlorination resistance of the positional isomers (11-13). The reduction of para and meta isomers is slightly easier than that of the ortho isomers (note Figure S1 for kinetics). The same tendency was also observed even in case of the dechlorination of a mixture of the positional mono PCBs, as shown in Figure 2 (right side) (also note Table S3). The dechlorination was at 86% for para > 81% for meta > 62% for ortho. Comparing independent (Figure 2 left side) and competitive (Figure 2 right side) conditions, the dechlorination of meta or para mono PCB was not affected while that of ortho PCB significantly decreased from 77.0 ( 3.5% to 62.1 ( 4.4% in the competition environment. These results imply preferential dechlorination of meta and para PCBs. 7512

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

FIGURE 3. Stepwise dechlorination of 234-trichlorinated biphenyl (234-ClBP) on RAC. Relative fraction of 234-ClBP, reaction intermediates, and BP is reported. The error bar is the standard deviation of triplicated results. Reactivity of Substituted Chlorines and Its Implications. To explain the relative dechlorination resistance of positional PCBs, Greaves et al. (14) introduced lowest unoccupied molecular orbital (LUMO) energies of PCBs and Arbon et al. (15) correlated the absorbed dose of γ-radiation of various PCBs with their LUMO energies. They found the highest dose constants occur for the para/meta-substituted congeners within a homologue, which have the lowest LUMO energies (vice versa for ortho substituents). On the basis of their observations, Yak et al. (11) proposed a possibility of predicting dechlorination resistance of PCB congeners. According to the calculation by Greaves et al. (14), the twist angle between the two phenyl rings increases from 41° for BP to 56° for 2-ClBP. The chlorine in ortho position causes increase in electron density at the biphenyl bond and thus the two phenyl rings are twisted to minimize the steric repulsion. On the other hand, the meta and para chlorines are far from the phenyl rings to induce this effect (but not negligible). As a result, the presence of ortho chlorines results in the nonplanarity of the phenyl rings and restricts their free spinning along the biphenyl bridge. The first step of the reaction of RAC with PCBs is surface interaction of chlorines in PCBs with the anion vacancy sites in the metal (23). The steric hindrance of chlorines affects their reactivity with Fe/Pd. Parallel configuration of the phenyl ring of PCBs to the active metallic surface was more favorable for their adsorption (24). Such a favorable adsorption was also found between more planar PCBs (4-ClBP > 3-ClBP > 2-ClBP) and the GAC used as a reaction mediator in this study (note Table S2). In a previous study using PCB congeners of similar hydrophobicity, considerably higher affinities toward soot-like sorbent were observed for the nonortho substituted PCBs (24). According to their study, ortho chlorine substitution has a more pronounced effect on the congener hydrophobicity than on their soot binding strength. As a result, it is likely that the inhibition of the intra molecular rotation around the 1,1′ bond, exerted by chlorines close to the bond (ortho > meta > para position in order), restricts surface interaction of PCBs with the planar GAC. Similar observation of preferential coplanar sorption was reported for PCB interaction with humic substrates (25). A configuration-specific sorption strength was also observed with different artificial analogues to naturally occurring forms of soot such as activated carbon (26). In addition, the conformation of ortho PCBs forces the electron cloud of the ortho chorine on one phenyl ring to hang on the top of the other phenyl ring. The electronic effect of chlorine (mostly inductive effect) is also known to cause partial deactivation to electrophilic substitution (27). Dechlorination of triClBP with Ortho, Meta, and Para Chlorines. The highest dechlorination susceptibility of the para positional chlorine was also confirmed during dechlo-

FIGURE 4. Carbon mass balance (mole fraction) during adsorption and dechlorination reaction of (a) meta PCB congeners and (b) ortho PCB congeners. The mole fraction was based on the initial concentration of each congener reported in Table S1 and its final and intermediate concentrations after 24 h reaction which is considered a reasonable time frame to best accommodate comparison of experimental results. Note that BP in the mass balance is not an actual PCB. * “Theoretical” corresponds to PCBs adsorption efficiency or mass fraction of PCBs theoretically present in the RAC phase (counterpart of total mass fraction in the liquid phase), and “extracted” stands for actual mass fraction of PCBs extracted from the RAC solid phase. rination of 234-triClBP. As shown in Figure 3 (also note Figure S2), in the first round dechlorination of 234-triClBP, 23-diClBP was a more dominant intermediate than any other diClBPs found, whereas in the second round dechlorination of 234triClBP (or first round dechlorination of diClBPs formed), 2-ClBP was present much more than 3-ClBP and 4-ClBP. Noma et al. (28) found that the reactivity of a chlorine in aromatics decreases upon the presence of adjacent chlorines. On the basis of the discussion so far and the conformation of 234-triClBP, the para chlorine (Cl-‘Cl’-H) > meta chlorine (Cl-‘Cl’-Cl: inductive effect) > ortho chlorine (phenyl-‘Cl’-Cl: steric and inductive effects) is the preferential target to remove from 234-triClBP. Since the similar result was also observed in other metallic systems including Fe (11, 12), Mg/Pd (13), and Ni/SiO2 (23) which might exhibit different affinity for PCBs (e.g., substrate effect), the dechlorination susceptibility of positional chlorines in order of para > meta > ortho position, seems to be inherent regardless of the substrate effect. Promisingly, this indicates that more toxic coplanar PCB congeners are preferentially reduced. All positional chlorines were completely removed after 30 d reaction, resulting in BP formation as a sole reaction product. Treatment of Higher Meta and Ortho PCB Congeners. Instead of tracing the dechlorination pathway of higher PCB congeners, which occurs simply stepwise with a chlorine atom being replaced by a hydrogen atom in each single reduction step as shown in Figure 3 (11, 13), carbon mass balance (mole fraction) was monitored after 24 h reaction, which might be more proper for comparing and explaining the complicated results. Figure 4 shows partitioning of mother PCB, reaction intermediates, and final product BP between the liquid and solid phases. Take 33′55′-PCB for an example of data interpretation. After 24 h reaction, 21.2% of PCBs were partitioned to the liquid phase, consisting of 33′55′: 7.4%, 33′5: 3.1%, 35: 2.6%, 33′: 2.5%, 3: 2.2%, and BP: 3.3%. As a result, 78.8% of PCBs should be theoretically present in the RAC solid phase (i.e., PCBs adsorption efficiency). However, due to incomplete extraction of PCBs from the solid phase, only 60.9% out of 78.8% was extracted, yielding solid phase extraction efficiency 77.3%. Complete (or reasonably high) extraction of PCBs from adsorptive materials such as activated carbon used here is difficult to achieve and

their extraction efficiencies (using conventional solvent extraction, Soxhlet, and automated Soxhlet) significantly decrease over their aging on the RAC (6, 7). We have reported importance of the ratio (i.e., relative mass fraction or distribution) of PCB congeners actually recovered from the system (6). Consequently, the ratio (distribution) of PCB congeners is particularly important for indirectly estimating the dechlorination reaction of PCBs, no matter how much of PCBs are extracted. To reflect this, total mass fraction of PCB congeners in RAC solid at 60.9% was reset as 100%, giving the ratio among PCB congeners in RAC at 33′55′: 48.3%, 33′5: 12.8%, 35: 9.1%, 33′: 4.1%, 3: 4.1%, and BP: 21.3%. However, the distribution of PCBs remaining in the RAC after extraction might be different from the extracted PCBs distribution (Fortunately, our results and conclusions deduced from the approach are generally consistent with findings elsewhere. This will be noted later). As shown in Figure 4a (note Table S4) demonstrating meta PCB treatment, adsorption efficiency of RAC was consistently decreased from 96.9% for the smallest size monoClBP to 78.8% for the largest size tetraClBP. In the RAC solid phase, lesser chlorinated PCBs showed more dominant BP fraction than their mother fraction while higher chlorinated PCBs exhibited more dominant mother PCB fraction than BP fraction. BP was detected in all cases, indicating that it is possible to remove all chlorines from the higher PCBs. The same results were also deduced for the ortho PCBs, as shown in Figure 4b. In addition, the results on the comparative adsorption tendency and dechlorination susceptibility between positional mono PCB congeners (i.e., 3-ClBP and 2-ClBP) presented in Figure 2 could be also applicable to the higher PCB congeners here. Preferential Removal of Chlorines in a Certain Environment. Comparing 33′-ClBP with 35-ClBP as a mother PCB, 33′-ClBP in RAC survived much less than 35-ClBP. A similar trend was also observed during the treatment of higher meta PCBs. If no selective dechlorination among the chlorines within the meta position exists in the first round dechlorination of 33′5-ClBP and the second round dechlorination of 33′55′-ClBP, the ratio of 35-ClBP to 33′-ClBP should be 0.5 on the basis of 33′5-ClBP conformation. However, the ratio of 35-ClBP/33′-ClBP was rather reversed at 3.9 for 33′5-ClBP VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7513

TABLE 1. Apparent Dechlorination Efficiencies of Highly Chlorinated PCB Congeners on RACa dechlorinationb Cl no. 1 2 2 3 4

meta PCBs 3 33′ 35 33′5 33′55′

T% 97.9 85.5 74.3 63.8 51.7

C% 97.9 77.3 69.2 33.5 21.3

dechlorination CL

ortho PCBs c

N.A. 157.4 137.2 135.3 140.6

2 22′ 26 22′6 22′66′

T% 88.2 82.9 33.8 45.3 6.0

C% 88.2 74.5 31.0 28.5 2.4

dechlorination CL c

N.A. 157.4 64.8 52.9 8.6

para PCBs

T%

C%

CL

4 44′ n.a. n.a. n.a.

99.0 92.5

99.0 85.3

N.A.c 178.1

a Based on PCBs observed in the RAC solid phase after 24 h reaction. b Categorized to (i) T% (target %) based on disappearance of mother PCB, (ii) C% (complete %) based on complete transformation of mother PCB to BP, and (iii) CL, arbitrary value based on relative chlorine removal (CL is not % but a relatively calculated value for indirect comparison of chlorine removal efficiency. Note Table S5 for detailed calculation example). c CL for mono-ClBPs is not comparable since their dechlorination reaction was almost completed after 24 h while that with other higher PCBs still progressed.

and 2.2 for 33′55-ClBP. This means that the 3′ positional chlorine was preferentially removed from 33′5-ClBP than the 3 or 5 positional chlorine. In addition, 35-ClBP, once formed, was more resistant to dechlorination than 33′-ClBP. The result strongly suggests that the isolated chlorine in a phenyl ring is more susceptible to attack than one of the coexisting chlorines in the other phenyl ring, consistent with finding by Noma et al. (28) that the less chlorinated phenyl ring should be prone to electrophilic attack. This trend was also found in the treatment of ortho PCBs (Figure 4b) exhibiting high steric hindrance effect. As observed in all 6 comparison cases in Figure 4, the electrophilic attack in higher PCB congeners occurred preferentially to the chlorines in a less substituted phenyl ring and isolated chlorine. This specific electrophilic attack seems independent of the steric or inductive effect as a dominant limiting factor for the dechlorination of ortho or meta PCBs, respectively, since the preferential removal of chlorines in such an environment was observed in many cases (28-30). However, we can not exclude substrate effect since Agarwal et al. (13) reported the same result for meta PCBs on Pd/Mg but the opposite result for ortho PCBs where the phenyl ring with the higher chlorination degree was more susceptible to attack. Chlorine Removal Inhibition over PCB Chlorination Degree. Dechlorination efficiencies for the highly chlorinated PCB congeners are summarized in Table 1 (note Table S5). With increasing chlorination degree of meta PCBs, complete and target dechlorination efficiencies almost linearly decreased due to the linearly increased number of chlorines to remove. As discussed on the preferential removal of isolated chlorine, the dechlorination efficiency for 33′-ClBP was slightly higher than that for 35-ClBP. Interestingly, chlorine removal efficiencies of meta PCBs were similar at 140 (excluding slightly easier removal of isolated chlorine in 33′ClBP at 157.4) regardless of their chlorination degree. This strongly suggests that the complete dechlorination of higher meta PCBs, which exhibit negligible steric effect, takes more time proportionally due to the increased number of chlorines. The dechlorination trend of ortho PCBs was generally the same as that of meta PCBs, except chlorine removal efficiency. The preferential removal of an isolated chlorine in ortho position was more obvious, when comparing Cl removal at 157.4 for 22′-ClBP and only 64.8 for 26-ClBP. In case of ortho positional PCBs showing significant steric effect, prominent dechlorination inhibition for higher PCB congeners was found. Chlorine removal efficiency obviously decreased from 64.8 for 26-ClBP to only 8.6 for 22′66′-ClBP. From one to four ortho chlorine substitution, the twist angles between two phenyl rings were reported to increase at around 56, 76, 86, and 87°, respectively (14). At three and four chlorine, the phenyl rings are almost orthogonal to each other, making PCBs more nonplanar, which are more difficult to dechlorinate (more steric hindrance). However, after prolonged reaction up to 30 d, the ortho and meta PCB congeners and 7514

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

reaction intermediates were almost completely transformed to BP, which was also sequestered in RAC. The dechlorination behavior of para PCBs was very similar to that of meta. Even though we have not tested all 209 PCB congeners, the results so far with 13 selected congeners strongly support the promising ability of the RAC system to effectively treat various PCB congeners and give some insights on reductive dechlorination behavior of PCBs on RAC. For large scale field applications of RAC system, however, its reactivity under various and practical conditions relevant in natural systems, where the technology is ultimately intended to be implemented, should be eventually confirmed. These include pH, coexisting natural organic matter and ionic species, and microbial activity.

Acknowledgments This research was funded and conducted by the National Risk Management Research Laboratory of U.S. Environmental Protection Agency (EPA), Cincinnati, Ohio. This paper has not been subjected to internal policy review of the U.S. EPA. Therefore, the research results do not necessarily reflect the views of the agency or its policy. Mention of trade names and commercial products does not constitute endorsement or recommendation for use. We recognize the support of Mr. Eric Graybill of the EPA for GC/MS analysis. Donation of the GAC by Norit Americas Inc. is appreciated.

Supporting Information Available PCB concentration (Table S1), mono PCB treatment (Table S2), mono PCB mixture treatment (Table S3), meta PCB treatment (Table S4), chlorine removal efficiency (Table S5), mono PCB congener dechlorination (Figure S1), and 234ClBP treatment (Figure S2). This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Zimmerman, J. R.; Ghosh, U.; Millward, R. N.; Bridges, T. S.; Luthy, R. G. Addition of carbon sorbents to reduce PCB and PAH bioavailability in marine sediments: Physicochemical tests. Environ. Sci. Technol. 2004, 38, 5458–5464. (2) Shimotori, T.; Cussler, E. L.; Arnold, W. A. High-density polyethylene membrane containing Fe(0) as a contaminant barrier. J. Environ. Eng. 2006, 7, 803–809. (3) Murphy, P.; Marquette, A.; Reible, D. D.; Lowry, G. V. Predicting the performance of activated carbon-, coke-, and soil-amended thin layer sediment caps. J. Environ. Eng. 2006, 132, 787–794. (4) McDonough, K. M.; Murphy, P.; Olsta, J.; Zhu, Y.; Reible, D. D.; Lowry, G. V. Development and placement of a sorbent-amended thin layer sediment cap in the Anacostia River. Soil Sed. Contam. 2007, 16, 313–322. (5) Choi, H.; Al-Abed, S. R.; Agarwal, S.; Dionysiou, D. D. Synthesis of reactive nano Fe/Pd bimetallic system-impregnated activated carbon for the simultaneous adsorption and dechlorination of PCBs. Chem. Mater. 2008, 20, 3649–3655.

(6) Choi, H.; Agarwal, S.; Al-Abed, S. R. Adsorption and simultaneous dechlorination of PCBs on GAC/Fe/Pd: Mechanistic aspects and reactive capping barrier concept. Environ. Sci. Technol. 2009, 43, 488–493. (7) Choi, H.; Al-Abed, S. R.; Agarwal, S. Effects of ageing and oxidation of palladized iron embedded in activated carbon on the dechlorination of 2-chlorobiphenyl. Environ. Sci. Technol. 2009, 43, 4137–4142. (8) Lien, H. L.; Zhang, W. X. Nanoscale Pd/Fe bimetallic particles: Catalytic effects of palladium on hydrodechlorination. Appl. Catal., B 2007, 77, 110–116. (9) Furukawa, Y.; Kim, J.-W.; Watkins, J.; Wilkin, R. T. Formation of ferrihydrite and associated iron corrosion products in permeable reactive barriers of zero-valent iron. Environ. Sci. Technol. 2002, 36, 5469–5475. (10) Yak, H. K.; Wenclawiak, B. W.; Cheng, I. F.; Doyle, J. G.; Wai, C. M. Reductive dechlorination of polychlorinated biphenyls by zerovalent iron in subcritical water. Environ. Sci. Technol. 1999, 33, 1307–1310. (11) Yak, H. K.; Lang, Q.; Wai, C. M. Relative resistance of positional isomers of polychlorinated biphenyls toward reductive dechlorination by zerovalent iron in subcritical water. Environ. Sci. Technol. 2000, 34, 2792–2798. (12) Lowry, G. V.; Johnson, K. M. Congener-specific dechlorination of dissolved PCBs by microscale and nanoscale iron in a water/ methanol solution. Environ. Sci. Technol. 2004, 38, 5208–5216. (13) Agarwal, S.; Al-Abed, S. R.; Dionysiou, D. D.; Graybill, E. Reactivity of substituted chlorines and ensuing dechlorination pathways of select PCB congeners with Pd/Mg bimetallics. Environ. Sci. Technol. 2009, 43, 915–921. (14) Greaves, J.; Harvey, E.; MacIntyre, G. Correlation between electron capture negative chemical ionization mass spectrometric fragmentation and calculated internal energies for polychlorinated biphenyls. J. Am. Soc. Mass. Spectrom. 1994, 5, 44–52. (15) Arbon, R. E.; Mincher, B. J.; Knighton, W. B. γ-ray destruction of individual PCB congeners in neutral 2-propanol. Environ. Sci. Technol. 1994, 28, 2191–2196. (16) Liu, Y.; Schwartz, J.; Cavallaro, C. L. Catalytic dechlorination of polychlorinated biphenyls. Environ. Sci. Technol. 1995, 29, 836– 840. (17) Kim, Y.-H.; Shin, W.-S.; Ko, S.-O. Reductive dechlorination of chlorinated biphenyls by palladized zero-valent metals. J. Environ. Sci. Health, Part A 2004, 39, 1177–1188.

(18) Korte, N. E.; West, O. R.; Liang, L.; Gu, B.; Zutman, J. L.; Fernando, Q. The effect of solvent concentration on the use of palladized-iron for the step-wise dechlorination of polychlorinated biphenyls in soil extracts. Waste Manage. 2002, 22, 343–349. (19) Zhang, W. X.; Wang, C. B.; Lien, H. L. Treatment of chlorinated organic contaminants with nanoscale bimetallic particles. Catal. Today 1998, 40, 387–395. (20) Brewster, J. H. Mechanisms of reductions at metal surfaces. I. A general working hypothesis. J. Am. Chem. Soc. 1954, 76, 6361– 6363. (21) Masel, R. I. Principles of Adsorption and Reaction on Solid Surfaces; John Wiley and Sons, Inc.: New York, 1996. (22) Agarwal, S.; Al-Abed, R. S.; Dionysiou, D. D. Enhanced corrosionbased Pd/Mg bimetallic systems for dechlorination of PCBs. Environ. Sci. Technol. 2007, 41, 3722–3727. (23) Shin, E.; Keane, M. A. Detoxification of dichlorophenols by catalytic hydrodechlorination using a nickel/silica catalyst. Chem. Eng. Sci. 1999, 54, 1109–1120. (24) Bucheli, T. D.; Gustafsson, O. Soot sorption of non-ortho and ortho substituted PCBs. Chemosphere 2003, 53, 515–522. (25) Uhle, M. E.; Chin, Y.-P.; Aiken, G. R. Binding of polychlorinated biphenyls to aquatic humic substrates: The role of substrate and sorbate properties on partitioning. Environ. Sci. Technol. 1999, 33, 2715–2718. (26) Feltz, K. P.; Tillitt, D. E.; Gale, R. W.; Peterman, P. H. Automated HPLC fractionation of PCDDs and FCDFs and planar and nonplanar PCBs on C18-dispersed PX-21 carbon. Environ. Sci. Technol. 1995, 29, 709–718. (27) Hagh, B. F.; Allen, D. T. Catalytic hydroprocessing of chlorinated benzens. Chem. Eng. Sci. 1990, 45, 2695–2701. (28) Noma, Y.; Ohno, M.; Sakai, S. Pathways for the degradation of PCBs by palladium-catalyzed dechlorination. Fresenius Environ. Bull. 2003, 12, 302–308. (29) Noma, Y.; Muramatsu, T.; Nishizawa, K.; Ohno, M.; Sakai, S. Dechlorination of pathways of PCBs by photochemical reaction and catalytic hydro-dechlorination. Organohalogen Comp. 2002, 56, 413–416. (30) Miyoshi, K.; Nishio, T.; Yasuhara, A.; Morita, M. Dechlorination of hexachlorobiphenyl by using potassium-sodium alloy. Chemosphere 2000, 41, 819–824.

ES901298B

VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7515