Graphite-Mediated Reduction of 2,4-Dinitrotoluene with Elemental Iron

toluene (DNT) is reduced with elemental iron were investigated through batch ... iron surface area, with high-purity iron powder and Master. Builders ...
0 downloads 0 Views 99KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 2178-2184

Graphite-Mediated Reduction of 2,4-Dinitrotoluene with Elemental Iron SEOK-YOUNG OH, DANIEL K. CHA, AND PEI C. CHIU* Department of Civil & Environmental Engineering, University of Delaware, Newark, Delaware 19716

The mechanism and pathway through which 2,4-dinitrotoluene (DNT) is reduced with elemental iron were investigated through batch experiments performed utilizing the same iron surface area, with high-purity iron powder and Master Builders scrap iron. In addition to different kinetics and adsorption patterns, the distribution of two intermediates, 4-amino-2-nitrotoluene (4A2NT) and 2-amino-4-nitrotoluene (2A4NT), contrasted sharply. This suggests that different mechanisms are involved in DNT reduction with pure iron and scrap iron. We hypothesized that exposed graphite in scrap iron transferred reductants from iron to adsorbed nitroaromatic molecules. This hypothesis was supported by an experiment conducted using two-compartment dialysis cells in which DNT and pure iron powder were separated by a graphite sheet. Results indicate that graphitemediated, indirect reduction of DNT occurred primarily through reduction of the ortho nitro group to form 2A4NT, whereas DNT reduction at the iron (hydr/oxide) surface occurred via para nitro reduction to give 4A2NT. Based on pH and product analysis, atomic hydrogen probably accounted for most of the reducing equivalents that passed through the graphite, reacting with adsorbed DNT mainly through ortho nitro reduction. In contrast, electron was a minor fraction of the reducing equivalents, reducing DNT mainly through para nitro reduction. The implications of graphite as a reaction site and conductor of electron and atomic hydrogen are discussed with respect to treatment processes involving iron.

Introduction Elemental iron has been increasingly used in permeable reactive barriers (PRBs) for site remediation since the mid1990s (1, 2). It was first used to remediate groundwater contaminated with chlorinated solvents (3, 4) and has since been evaluated for treating other halogenated compounds (5-9), nitroaromatic compounds (10, 11), nitramines (12, 13), azo dyes (14, 15), nitrate (16), metals (17-19), and radionuclides (20, 21). Recently, it was proposed that elemental iron may be applied to wastewater treatment to enhance the degradation of refractory organic compounds (22, 23). Much work has been performed on the kinetics of redox reactions with elemental iron. Many studies (1, 2, 5) showed that reduction of aliphatic halogenated compounds with elemental iron was pseudo-first-order with respect to the * Corresponding author phone: (302)831-3104; fax: (302)831-3640; e-mail: [email protected]. 2178

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

halogenated compounds. Among the variables that control the pseudo-first-order reaction rates, iron surface area appears to be an important factor (1, 10, 15). It has been proposed that surface area-normalized reaction rates be used for design of iron PRBs for reductive dechlorination (24, 25). In contrast to the kinetics, the pathway and mechanisms through which aqueous oxidants are reductively transformed with elemental iron are less understood (with the exception of chlorinated ethenes and acetylenes, which has been examined in detail (26)). The mechanism (and thus the pathway) of a redox reaction involving elemental iron is directly related to the reducing agent responsible and the sites where reduction occurs. Studies in recent years have provided important insights into the possible reducing agents involved and their modes of action in iron-water systems. It appears that redox reactions with elemental iron can occur through at least three mechanisms, each involving a different type of reducing agent. The first mechanism is direct transfer of electron from elemental iron to the adsorbed oxidant molecules. Such electron transfer most likely occurs through the (hydr)oxide layer at the iron surface. Odziemkowski et al. (27) showed that Fe(OH)2 was formed at the iron surface and gradually changed to magnetite (Fe3O4) during reduction of chlorinated compounds. Based on the electron-conducting property of magnetite, Bonin et al. (28) proposed a conceptual model for the reduction of chlorinated compounds for freshly cleaned and oxidized iron surfaces. Balko and Tratnyek (29) also concluded that the reduction rate of CCl4 with iron metal was controlled by electron transfer through iron oxides. Another possible reduction mechanism involves Fe(II) as a reducing agent. Although aqueous Fe(II) is not an effective reducing agent (with reduction of hexavalent chromium being an exception (30, 31)), surface-associated Fe(II) species have been shown to be much stronger reducing agents. Schwarzenbach and co-workers demonstrated that Fe(II) adsorbed to Fe-(hydr)oxides such as magnetite, goethite (R-FeOOH), and lepidocrocite (γ-FeOOH) can reduce nitroaromatic compounds to corresponding amines (32-34). In addition to adsorbed Fe(II), ferrous-containing minerals such as green rust and magnetite have also been shown to reduce pollutants including CCl4, nitrate, and Cr(VI) (35-39). The third mechanism of reduction involves atomic hydrogen as a reductant, which is produced during anaerobic iron corrosion by water (28, 40). By comparing the results of N-nitrosodimethylamine reduction in batch and column systems with elemental iron and with an electrochemical process, Odziemkowski et al. (41) concluded that the reduction reaction was effected by hydrogen atom rather than electron, since the potential required for the electrochemical reduction was too negative to achieve with elemental iron. These authors also observed different reduction products in the metallic and electrochemical systems, suggesting the mechanism involved in the scrap iron system was not direct electron transfer. Farrell et al. (42) observed faster reduction of trichloroethene (TCE) than tetrachloroethene (PCE) in a longterm column study, while current measurements indicated that direct reduction of PCE was faster. These authors attributed the difference to the higher rate of TCE reduction by atomic hydrogen than PCE. Li and Farrell (43) further showed that TCE reduction rate was limited by chemicaldependent factors rather than electron transfer, consistent with the hypothesis that TCE was reduced by atomic hydrogen. From these recent studies, it therefore appears that all three mechanisms are possible, although the importance of each mechanism may vary with the type of oxidants and medium conditions. 10.1021/es011474g CCC: $22.00

 2002 American Chemical Society Published on Web 04/04/2002

Because iron is the ultimate source of reducing equivalents for all three mechanisms, it is commonly assumed that reactions occur at the iron (oxide/hydroxide) surface and many studies have been conducted using high-purity iron (10, 26, 44). For low-purity scrap iron, which is commonly used in field applications, the non-iron components are regarded as inert impurities. However, it has been suggested that adsorption of organic oxidants to scrap iron may play a role in controlling the overall reduction kinetics due to the high carbon contents of scrap iron (45). Burris and co-workers (46, 47) showed that PCE, TCE, and dichloroethenes (DCEs) were adsorbed to nonreactive sites in scrap iron filings. These authors proposed that the nonreactive sites to which chlorinated ethenes were adsorbed are exposed graphite inclusions in scrap iron and that the extent of adsorption was related to the hydrophobicity of the organic compound (45). Such nonreactive adsorption is believed to decrease the aqueous concentration of the pollutant and therefore decrease its apparent transformation rate. In this paper, we present evidence for a previously unrecognized mechanism that involves graphite as both an adsorption and reaction site by serving as a conduit for reducing agents such as electron and atomic hydrogen. We chose 2,4-dinitrotoluene (DNT) as a model compound for this study because its reductive transformation is regioselective and may permit distinction of different reaction mechanisms (48-50). Aqueous and surface concentrations of DNT and its reduction products were followed during the reduction with both scrap and high-purity iron powder in a batch system. Experiments using two-compartment dialysis cells were conducted to provide additional evidence for the proposed mechanism and to demonstrate the involvement of electron and atomic hydrogen in DNT reduction.

Experimental Section Chemicals. 2,4-Dinitrotoluene (DNT, 97%), 2,4-diaminotoluene (DAT, 98%), 4-amino-2-nitrotoluene (4A2NT, 97%), and 2-amino-4-nitrotoluene (2A4NT, 99%) were purchased from Aldrich (Milwaukee, MI). HEPES (N-[2-hydroxyethyl]piperazine-N′-[ethanesulfonic acid]) was obtained from Sigma (St. Louis, MO). Acetonitrile (HPLC grade) was purchased from Fisher Scientific (Pittsburgh, PA). All chemicals were used as received. Two types of iron were used in this study. Master Builder iron was chosen as a representative scrap iron because it has been used in field remediation and laboratory studies and its elemental composition has been characterized (45-47). High-purity iron powder (99.5%) was purchased from Alfa Aesar (Ward Hill, MA). These irons were used as received without pretreatment. Specific surface areas of Master Builder iron and the high-purity iron powder were 1.29 and 0.19 m2/g, respectively, as measured by BET adsorption method with N2. Graphite sheets (99.8%, 0.13 mm thickness) were obtained from Alfa Aesar (Ward Hill, MA). Equilibrium dialysis cells (total volume ) 10 mL) were purchased from Bell-Art Products (Pequannock, NJ). Batch Reduction Experiments. Batch reduction experiments were conducted in an anaerobic glovebox (95% N2 + 5% H2, Coy laboratory, Grass Lake, MI) using 8 mL borosilicate vials containing 5 mL of aqueous solution and either 1.0 g of scrap iron or 6.8 g of high-purity iron powder. The different iron masses were used to give approximately the same surface area of 1.29 m2. Replicate vials were set up for each experiment. The DNT solution contained 0.1 M HEPES buffer to maintain a constant pH of 7.4 throughout the experiment. The solution was purged in the glovebox to completely remove dissolved oxygen. Initial concentration of DNT was 0.250 ( 0.001 mM. After iron was added, the vials were shaken at 100 rpm in a horizontal position using an orbital shaker.

FIGURE 1. Schematic of a dialysis cell with graphite sheet. At different elapsed times, one of the vials was sacrificed, and 4.5 mL of the solution was filtered through a 0.22 µm mixed cellulose membrane filter (Millipore, MA) before analysis using a high performance liquid chromatograph (HPLC). Acetonitrile extraction of used filters indicated that DNT did not adsorb to the filters. Extraction of Adsorbed Compounds. After 4.5 mL of supernatant was removed, the iron and 0.5 mL of solution remaining in the vial were extracted once for high-purity iron powder and twice for scrap iron, each using 2 mL of acetonitrile, to recover adsorbed molecules. Further extraction did not yield meaningful increases in recovery. For the high-purity iron powder, after 2 mL of acetonitrile was added, the vial was shaken vigorously for 1.5 min using a vortex mixer. Two milliliters of supernatant was withdrawn and passed through a glass fiber filter, and the filtrate was analyzed by HPLC. Masses of the sorbed compounds were calculated using eq 1

MS ) Ce‚Ve - Ca‚Va

(1)

where MS is the adsorbed mass (µmol), Ce is the concentration in extract (mM), obtained from LC analysis, Ve is the extract volume ) 2.5 mL, Ca is the aqueous concentration (mM), from LC analysis, and Va is the volume of residual solution extracted ) 0.5 mL. For the scrap iron, the extraction step was repeated with another 2 mL of acetonitrile. The total adsorbed mass was taken to be the sum of the two adsorbed masses calculated based on the two extract concentrations, using eqs 2-4. The fractions of the adsorbed DNT mass extracted in the first and second extractions, defined as MS1/MS and MS2/MS, were 0.766 ( 0.006 and 0.234 ( 0.006, respectively,

MS1 ) Ce1‚Ve - Ca‚Va

(2)

MS2 ) Ce2‚Ve - Ce1‚Va

(3)

MS ) MS1 + MS2

(4)

where MS1 and MS2 are the adsorbed masses from first and second extractions (µmol), respectively, and Ce1 and Ce2 are concentrations in the first and second extracts (mM), respectively. Dialysis Cell Experiments with Graphite Sheet. The dialysis cell experiments were also performed in the anaerobic glovebox. The two 5-mL compartments were separated by a graphite sheet and tightened using double-sided tape and stainless steel screw (Figure 1). An opening at the top of each compartment allowed sampling and transfer of solutions. One compartment (“compartment 1”) was filled with highpurity iron powder (approximately 32 g) in deoxygenated 0.1 M HEPES buffer solution. The other compartment (“comVOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2179

partment 2”) was filled with 4 mL of DNT solution (0.308 ( 0.003 mM) in deoxygenated deionized water. Replicate dialysis cells were prepared and shaken using an orbital shaker at 60 rpm in an upright position. At predetermined times, one of the cells was sacrificed for analysis. DNT solution was withdrawn from compartment 2 using a needled syringe (Micro-mate, New Hyde Park, NY) and analyzed for DNT, its reduction products, and pH. Molecules adsorbed to the graphite sheet were extracted twice, each with 4 mL of acetonitrile. After addition of 4 mL of acetonitrile, the opening of the cell was plugged, and the cell was shaken for 1.5 min using a vortex mixer. The extract was withdrawn and analyzed by HPLC. The total mass of each analyte adsorbed to the graphite sheet was taken to be the sum of the adsorbed masses obtained from the two extractions. The fractions of the adsorbed DNT mass extracted in the first and second extractions were 0.747 ( 0.014 and 0.253 ( 0.014, respectively. Analytical Methods. 2,4-Dinitrotoluene (DNT), 2,4-diaminotoluene (DAT), 4-amino-2-nitrotoluene (4A2NT), and 2-amino-4-nitrotoluene (2A4NT) were analyzed using a Varian HPLC (Walnut Creek, CA) equipped with a Supelguard guard column (20 × 4.6 mm, Supelco, Bellefonte, PA), a SUPELCO LC-18 column (250 × 4.6 mm, 5 µm, Supelco, Bellefonte, PA), a UV detector (2510 Varian, Walnut Creek, CA), and an isocratic pump (2550 Varian, Walnut Creek, CA). Methanol-water mixture (70/30, v/v) was used as the mobile phase at a flow rate of 1.0 mL/min for DNT, 4A2NT, and 2A4NT. The wavelength for the UV detector was set to 254 nm. For DAT, acetonitrile-phosphate buffer (20 mM, pH 7.0, 30/70, v/v) was used as an eluent at 1.0 mL/min. The wavelength for the UV detector was 224 nm. The injection volume for all samples was 10 µL. pH was measured using a pH-30 pH sensor (Corning, Big Flats, NY). X-ray powder diffraction (-200 mesh) analysis of Master Builder iron was performed with Philips/Norelco diffractometer using CuKR radiation to determine the crystalline species at the iron surface.

Results and Discussion DNT Reduction with High-Purity Iron Powder. Figure 2 shows the aqueous, surface, and total masses of DNT, 4A2NT, 2A4NT, and DAT during reduction with high-purity iron powder. DNT was completely removed from the solution in 60 min. Two intermediates, 4A2NT and 2A4NT, were detected, and the amounts at 30 min were 0.315 and 0.085 µmol, respectively (Figure 2(a)). DAT, the end product of DNT reduction, appeared within 10 min, and its concentration in solution after 3 h accounted for 90.4% of the initial DNT. A portion of the DNT removed from the aqueous phase was due to adsorption to the iron surface (Figure 2(b)). In contrast to the aqueous phase, the two intermediates were not found at the surface. DAT was detected at the surface after 10 min, and the final mass at the surface was approximately 0.12 µmol (ca. 10% of the initial DNT). Figure 2(c) shows the total masses of DNT and its reduction products. These profiles, which were obtained by adding the aqueous and surface masses, represent mass changes due only to transformation. The mass recovery during the course of the reaction ranged from 80.0 to 99.2%. The lower recovery in the early stage of the reaction might be accounted for by reduction intermediates we did not measure, such as hydroxyamino and nitroso compounds (10). The observation that 4A2NT was the major intermediate suggests that either DNT reduction in the high-purity iron system occurred primarily through reduction of the nitro group in the para position or 2A4NT was reduced faster than 4A2NT with high-purity iron. The first explanation is more likely since similar regioselectivity for the reduction of the para nitro group has been observed in other DNT reduction reactions (48-50). Reduction of nitroaromatic compounds 2180

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

FIGURE 2. Aqueous (a), surface (b), and total (c) masses of DNT, 4A2NT, 2A4NT, and DAT during reduction of DNT with high-purity iron powder. The error bars were calculated based on analyses of replicate reactors. such as DNT and trinitrotoluene (TNT), which do not contain a polar functional group in the 1-position, is known to proceed primarily or exclusively through reduction of the para nitro group (48). Barrows et al. (49) suggested that the preferential reduction of the para nitro group of DNT and TNT was due to the shielding effect of the methyl group on the ortho nitro group, therefore favoring the one-electron reduction of the fully exposed and solvated para nitro group. DNT Reduction with Scrap Iron. Figure 3 shows the aqueous, surface, and total masses of DNT, 4A2NT, 2A4NT, and DAT during reduction with scrap iron. Compared with the results with high-purity iron (Figure 2), there are several marked differences. First, adsorption was more significant with scrap iron. Greater amounts of DNT (up to 0.375 µmol) and DAT (ca. 0.220 µmol) were adsorbed to scrap iron than to high-purity iron. In addition, 4A2NT and 2A4NT, which

FIGURE 4. X-ray powder diffraction pattern of Master Builder iron.

FIGURE 3. Aqueous (a), surface (b), and total (c) masses of DNT, 4A2NT, 2A4NT, and DAT during reduction of DNT with Master Builder scrap iron. The error bars are based on replicate reactors. were not detected at high-purity iron surface, were largely associated with the surface of scrap iron. Second, reduction appeared to occur more slowly in the scrap iron system. The reduction of DNT to DAT required more than 2 h with scrap iron, whereas complete transformation was achieved within an hour with high-purity iron powder (Figure 2). The slower reduction was also evident from the greater quantities of 4A2NT (0.35 µmol) and 2A4NT (0.48 µmol) detected. Third, in contrast to the result with high-purity iron, 2A4NT, rather than 4A2NT, was the dominant intermediate in both the solution and solid phases. This indicates that either the ortho nitro group of DNT was preferentially reduced or 4A2NT reacted more rapidly than 2A4NT in the scrap iron system. One might attribute the more pronounced adsorption and slower reduction observed with scrap iron to adsorption

of DNT and its daughter products to graphite inclusions, which were proposed by Burris and co-workers to be nonreactive adsorption sites for chlorinated ethenes (45-47). Such adsorption would retain a portion of the DNT and the intermediates at nonreactive sites and decrease the amounts of “accessible” reactants, thus decreasing their apparent reduction rates. However, adsorption to nonreactive sites would not change the pathway or mechanism of the reaction, nor would it change the relative reactivity of 2A4NT and 4A2NT. Therefore, nonreactive adsorption should not alter the distribution of the intermediates, as all reaction would take place only at iron (oxide) surface in both pure and scrap iron systems. This is clearly not the case. Comparing Figures 2(c) and 3(c), the reduction of DNT with high-purity iron occurred mostly through reduction of the para nitro group to 4A2NT, whereas more DNT was reduced through reduction of the ortho nitro group to form 2A4NT with scrap iron. This difference cannot be explained by nonreactive adsorption to carbon. Instead, the data suggest that, in the scrap iron system, there existed a mechanism that favored reduction of the ortho nitro group of DNT. In addition, the marked disparity in the distribution of the intermediates further suggests that this mechanism was probably important in the scrap iron system but was absent or less important in the pure iron system. To explain the different dominant intermediates, we hypothesized that the electron is transferred from the Fe(0) core through exposed graphite to the DNT and intermediates adsorbed to graphite. Master Builder iron is known to contain 2-4% carbon in the form of graphite (45, 51), which may account for the more significant adsorption we observed. We also identified graphite in the Master Builder iron samples used in our experiment by XRD analysis (Figure 4). In addition to its ability to adsorb organic compounds through hydrophobic interactions, graphite is also a good conductor of electricity (52). According to our hypothesis, DNT and its daughter products were reduced while they were adsorbed to graphite; that is, exposed graphite in scrap iron represents another type of reactive sites. To test this hypothesis, a DNT reduction experiment was conducted using dialysis cells in which aqueous DNT was physically separated from elemental iron by a graphite sheet. While aqueous DNT concentration may decrease due to adsorption to the graphite sheet, no reduction products would be detected unless a reductant generated in the iron compartment is transported through the graphite sheet to the DNT compartment. DNT Reduction Mediated by Graphite. Figure 5 shows the aqueous, surface, and total masses of DNT, 4A2NT, 2A4NT, and DAT in the DNT compartment of dialysis cells at different elapsed times. The DNT in deionized water was adsorbed to the graphite and reduced to DAT through 2A4NT and 4A2NT over 2 days. The slower transformation in the VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2181

The result in Figure 5 clearly shows that DNT and its daughter products can be reduced even when they are physically separated from elemental iron by graphite. In control cells where iron powder was omitted, no DNT reduction (only adsorption) was observed in the presence of H2 (ca. 5% in the glovebox). This indicates that the observed DNT reduction was dependent on the iron in the other compartment and was not effected by H2. Electron or other reducing agents generated from the iron compartment must have permeated the graphite sheet to reduce the adsorbed nitro compounds. The amount of 2A4NT relative to 4A2NT (Figure 5(c)) was even greater than that observed with scrap iron (Figure 3(c)), suggesting that reduction of the ortho nitro group was the dominant reaction that occurred at the graphite surface. The proposed mechanism of graphite-mediated nitroaromatic reduction can explain the product distribution observed in the scrap iron experiment (Figure 3). In contrast to the pure iron reactors where all DNT reduction occurred at the iron oxide surface to form mostly 4A2NT, reduction of DNT in the scrap iron reactors took place at the surfaces of both iron oxide (to mainly 4A2NT) and graphite (to primarily 2A4NT), resulting in a shift in the distribution of the intermediates. The data therefore support our hypotheses that exposed graphite in the scrap iron can serve as both adsorption and reaction sites and that electron or other reducing agents from iron powder can be transferred through graphite to reduce the nitroaromatic compounds adsorbed to graphite. According to these hypotheses, the observed reduction was slower with scrap iron than with high-purity iron not because nonreactive adsorption to graphite decreased the effective reactant concentrations but because reduction of adsorbed nitroaromatic molecules on graphite was intrinsically slower than that on iron oxides. Alternatively, the lower reaction rate with scrap iron might be due to its thicker oxide coating. Reducing Agent for Reduction of DNT through Graphite. In the DNT compartment of the dialysis cells, DNT and water (or proton) were the only two potential oxidizing agents (i.e., electron acceptors). If the reducing agent that permeated the graphite sheet was electron, then the pH of the DNT solution should increase since hydroxide ion would be generated stoichiometrically regardless of whether DNT (eq 5) or water (eq 6) was reduced. The amount of electron transported through graphite can therefore be measured simply by measuring the pH increase in the DNT compartment. Since DNT is neutral and its daughter products are unprotonated at pH 7 or greater (pKas of 2A4NT, 4A2NT, and DAT are 2.29, 3.07, and 4.46 and 5.24, respectively, as calculated using the Hammett equation in reference (53)), they would not affect the pH of the DNT solution. FIGURE 5. Aqueous (a), surface (b), and total (c) masses of DNT, 4A2NT, 2A4NT, and DAT during reduction of DNT with high-purity iron powder and graphite sheet using dialysis cells. The error bars are based on replicate reactors. dialysis cells might be due to factors such as different reactor geometry, shaking speeds, iron corrosion rate, and surface area-to-solution volume ratios between the dialysis cell and the pure and scrap iron systems. Alternatively, the reduction of nitroaromatic compounds adsorbed to graphite may be intrinsically slower than that adsorbed to iron hydr/oxides. It is also possible that the graphite sheet used in this experiment was thicker and thus more resistant to the transfer of electron or other reducing agents than the graphite inclusions in scrap iron, which presumably exist in the form of flakes (45). Figure 5(c) also showed that the mass balance first decreased (to 50.1%) and then improved as the reduction progressed (93.5% after 2 days). The lower mass balance in early reaction might be due to accumulation of the nitroso and other intermediates not measured in this study. 2182

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

DNT + 12e - + 8H2O f DAT + 12OH-

(5)

1 H2O + e- f H2 + OH2

(6)

As shown in Figure 6(a), the pH in the DNT compartment increased from 7.66 to 9.12 in the first 12 h but remained constant thereafter, indicating that electron transfer through graphite occurred only in the early stage of the reaction. In contrast, the pH of the deionized water did not increase in the blank cells, which were identical to the experimental cells except DNT was omitted. This shows that the amount of electron transferred to water was negligible. Therefore, electron was most likely transferred through graphite to adsorbed DNT (eq 5) rather than to water (eq 6). As shown in Figure 6(b), hydroxide formation, or electron transferred to DNT, did not occur immediately but started only after a large fraction of the aqueous DNT had been adsorbed to the graphite sheet (Figure 5(a)).

that graphite can mediate the reductive dechlorination of polychlorinated biphenyls by serving as a hydrogen atom transfer agent (54).

DNT + 12H‚ f DAT + 4H2O

FIGURE 6. pH (a) and OH- concentration (b) in the DNT compartment of the dialysis cell during reduction of DNT with high-purity iron powder. Reduction of DNT to DAT requires 12 mol of electron per mole of DNT (eq 5). Based on the initial DNT concentration of 0.308 mM and assuming complete reduction of DNT to DAT in 48 h (Figure 5(c)), 3.7 mM of OH- would have been produced if electron was the only reducing agent that permeated the graphite sheet. This would have resulted in a final pH of 11.56 in the DNT compartment. However, the observed final pH of 9.12 ( 0.12, which corresponded to a hydroxide concentration of 0.014 ( 0.004 mM (Figure 6(b)), indicated that electron only accounted for 0.35% of the reducing equivalents involved in the observed reduction. Therefore, another reducing agent formed in the iron compartment must have been conducted through the graphite sheet to reduce DNT. In addition, this reducing agent must be uncharged so that its addition to the DNT solution would not have affected the pH. The only reducing agent that fulfills these criteria is atomic hydrogen. As stated earlier, H2 gas was unable to reduce DNT. Ferrous ion, the other reducing agent formed during iron corrosion, was also ruled out because graphite is not permeable to Fe2+ and because we observed no reduction in control cells where iron powder was replaced with 1 mM Fe2+ (in 0.1 M HEPES buffer; data not shown). We therefore propose that, in addition to electron, atomic hydrogen produced during anaerobic iron corrosion was transferred to the graphite in contact with iron and then permeated graphite to react with adsorbed nitroaromatic molecules. The reaction does not change the pH, as shown in eq 7. This hypothesis is supported by other studies. Hydrogen atom is believed to be involved in the reduction of N-nitrosodimethylamine (40, 41) and TCE (42, 43) with commercial scrap iron. It has also been proposed

(7)

In a separate study, we observed that TNT was reduced with scrap iron primarily via ortho nitro reduction (55), as was observed for DNT. Barrows et al. (49) suggested that the regioselectivity in nitroaromatic reduction via electron addition is controlled by the distribution of negative charge between the para and ortho nitro groups of the radical anion formed after the first electron transfer to the nitroaromatic compound. For DNT and TNT, reduction of the more highly charged and better solvated para nitro group following electron addition is strongly favored. Consistent with their explanation, the small amounts of 4A2NT formed in the dialysis cells (up to 0.09 µmol, Figure 5(c)) can be largely accounted for by the electron transferred through graphite (0.056 ( 0.016 µmol, Figure 6(b)). If most DNT in the dialysis cells was reduced by atomic hydrogen as we propose, it is possible that reduction of DNT (and TNT (55)) adsorbed to graphite by hydrogen atom occurs preferentially through reduction of the ortho nitro function. However, the reason for this regioselection is unclear. It should be noted that the very high hydrogen atomto-electron ratio observed in the graphite dialysis cells should not be extrapolated to the scrap iron system. Because the two compartments were isolated, transport of electron from one compartment to the other would result in charge accumulation and therefore buildup of a counter-potential. This would decrease the rate of electron transport across graphite but would have little effect on the conduction of neutral species such as atomic hydrogen. The relative importance of electron and hydrogen atom and the regioselectivity in pure and scrap iron systems are currently under investigation. In summary, our results show that graphite can serve as a conduit for electron and atomic hydrogen generated during iron corrosion. These reducing agents can then reduce nitroaromatic compounds adsorbed to graphite in a regioselective manner. The findings may have important implications for remediation and treatment processes involving scrap iron. First, in addition to being adsorption sites, graphite inclusions in scrap iron represent a previously unrecognized type of reaction sites. This means that our understanding of how hydrophobic oxidants are reduced with scrap iron and our approach to mathematically model these processes in iron PRBs need to be modified. For example, hydrophobic adsorption of reduction end products and nonreducible cocontaminants (e.g., toluene) to graphitic carbonsa process often considered independent of the reduction reactions may decrease the overall reaction rate through depletion of available reactive (graphite) sites. In terms of modeling the process, scrap iron should be regarded as a system containing at least two distinct types of reactive sites having different adsorption and reaction rate constants. Second, graphitemediated reduction may be a dominant mechanism for highly hydrophobic oxidants, since they would preferentially accumulate at graphite (as opposed to iron oxide) sites. Although the mass fraction of carbon in scrap iron is only 2-4% (45), if most of the carbon exists as graphite, graphite would account for approximately 10% of the volume due to its relatively low density (2.2 g/cm3). Graphite may cover an even greater percentage of the total surface area because of its planar (sheet) structure. Therefore, the kinetics, mechanism, and pathway through which an organic oxidant is reduced with scrap iron may depend on its hydrophobicity. Third, carbon content may be a factor to consider in selecting scrap iron for treatment purposes, as it may affect reactivity VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2183

both directly and through adsorption. Fourth, care should be exercised to avoid incorrect conclusions when extrapolating results from laboratory studies using high-purity iron to interpret or predict field data involving scrap iron. Fifth, the regioselectivity in nitroaromatic reduction by electron and hydrogen atom merits further studies, as the reduction products may have different mobility, degradability, and toxicity. Finally, it remains to be determined whether reduction of organic oxidants other than nitroaromatics can be mediated by graphite and how important the reaction is relative to those occurring at iron oxide surfaces.

Acknowledgments This study was supported by the Water Environment Research Foundation (Project 99-CTS-3-UR) and the National Science Foundation (Award #9984669).

Literature Cited (1) Gillham, R. W.; O’Hannesin, S. F. Ground Water 1994, 32, 958967. (2) Matheson, L. J.; Tratnyek, P. G. Environ. Sci. Technol. 1994, 28, 2045-2053. (3) Puls, R. W.; Paul, C. J.; Powell, R. M. Appl. Geochem. 1999, 14, 989-1000. (4) United States Environmental Protection Agency. Field application of in situ remediation technologies: Permeable reactive barriers; EPA 542-R-99-002; U.S. EPA: Washington, DC, 1999. (5) Orth, W. S.; Gillham, R. W. Environ. Sci. Technol. 1996, 30, 6671. (6) Roberts, A. L.; Totten, L. A.; Arnold, W. A.; Burris, D. R.; Campbell, T. J. Environ. Sci. Technol. 1996, 30, 2654-2659. (7) Kim, Y. H.; Carraway, E. R. Environ. Sci. Technol. 2000, 34, 20142017. (8) Sayles, G. D.; You, G.; Wang, M.; Kupferle, M. J. Environ. Sci. Technol. 1997, 31, 3448-3454. (9) Eykhort, G. R.; Davenport, D. T. Environ. Sci. Technol. 1998, 32, 1482-1487. (10) Agrawal, A.; Tratnyek, P. G. Environ. Sci. Technol. 1996, 30, 153160. (11) Devlin, J. F.; Klausen, J.; Schwarzenbach, R. P. Environ. Sci. Technol. 1998, 32, 1941-1947. (12) Singh, J.; Comfort, S. D.; Shea, P. J. J. Environ. Qual. 1998, 27, 1240-1245. (13) Singh, J.; Comfort, S. D.; Shea, P. J. Environ. Sci. Technol. 1999, 33, 1488-1494. (14) Webber, E. J. Environ. Sci. Technol. 1996, 30, 716-719. (15) Nam, S.; Tratnyek, P. G. Water Res. 2000, 34, 1837-1845. (16) Huang, C. P.; Wang, H. W.; Chiu P. C. Water Res. 1998, 32, 22572264. (17) Blowes, D. W.; Ptacek, C. J.; Jambor, J. L. Environ. Sci. Technol. 1997, 31, 3348-3357. (18) Su, C.; Puls, R. W. Environ. Sci. Technol. 2001, 35, 1487-1492. (19) Shokes, T. E.; Mo¨ller, G. Environ. Sci. Technol. 1999, 33, 282287. (20) Fiedor, J. N.; Bostick, W. D.; Jarabek, R. J.; Farrell, J. Environ. Sci. Technol. 1998, 32, 1466-1473. (21) Gu, B.; Liang, L.; Dickey, M. J.; Yin, X.; Dai, S. Environ. Sci. Technol. 1998, 32, 3366-3373. (22) Mantha, R.; Taylor, K. E.; Biswas, N.; Bewtra, J. K.; Environ. Sci. Technol. 2001, 35, 3231-3236. (23) Perey, J. R.; Oh, S.-Y.; Lubenow, B. L.; Cha, D. K.; Huang, C. P.; Chiu, P. C. In Ex Situ Biological Treatment Technologies, the Sixth International In Situ and On-Site Bioremediation Symposium; Battelle, 2001; Vol. 6, pp 149-155. (24) Johnson, T. L.; Scherer, M. M.; Tratnyek, P. G. Environ. Sci. Technol. 1996, 30, 2634-2640.

2184

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002

(25) Tratnyek, P. G.; Johnson, T. L.; Scherer, M. M.; Eykholt, G. R. Ground Water Monit. Rem. 1997, 17, 108-114. (26) Arnold, W. A.; Roberts, A. L. Environ. Sci. Technol. 2000, 34, 1794-1805. (27) Odziemkowski, M. S.; Schuhmacher, T. T.; Gillham, R. W.; Reardon, E. J. Corro. Sci. 1998, 40, 371-389. (28) Bonin, P. M. L.; Odziemkowski, M. S.; Gillham, R. W. Corro. Sci. 1998, 40, 1391-1409. (29) Balko, B. A.; Tratnyek, P. G. J. Phys. Chem. 1998, 102, 14591465. (30) Fendorf, S. E.; Li, G. Environ. Sci. Technol. 1996, 30, 1614-1617. (31) Buerge, I. J.; Hug, S. J. Environ. Sci. Technol. 1997, 31, 14261432. (32) Heijman, C. G.; Grieder, E.; Holliger, C.; Schwarzenbach, R. P. Environ. Sci. Technol. 1995, 29, 775-783. (33) Klausen, J.; Tro¨ber, S. P.; Haderlein, S. B.; Schwarzenbach, R. P. Environ. Sci. Technol. 1995, 29, 2396-2404. (34) Hofstetter, T. B.; Heijman, C. G.; Haderlein, S. B.; Holliger, C.; Schwarzenbach, R. P. Environ. Sci. Technol. 1999, 33, 14791487. (35) Hansen, H. C. B.; Koch, C. B.; Nancke-Krogh, H.; Borggaard, O. K.; Sorensen, J. Environ. Sci. Technol. 1996, 30, 2053-2056. (36) Erbs, M.; Hansen, H. C. B.; Olsen, C. E. Environ. Sci. Technol. 1999, 33, 307-311. (37) Loyaux-Lawniczak, S.; Refait, P.; Ehrhardt, J.-J.; Lecomte, P.; Ge´nin, J.-M. R. Environ. Sci. Technol. 2000, 34, 438-443. (38) Williams, A. G. B.; Scherer, M. M. Environ. Sci. Technol. 2001, 35, 3488-3494. (39) Sivavec, T. M.; Horney, D. P. Natl. Mtg.-Am. Chem. Soc., Div. Environ. Chem. 1997, 37, 115-117 (Abstr.). (40) Gui, L.; Gillham, R. W.; Odziemkowski, M. S. Environ. Sci. Technol. 2000, 34, 3489-3494. (41) Odziemkowski, M. S.; Gui, L.; Gillham, R. W. Environ. Sci. Technol. 2000, 34, 3495-3500. (42) Farrell, J.; Melitas, N.; Kason, M.; Li, T. Environ. Sci. Technol. 2000, 34, 2549-2556. (43) Li, T.; Farrell, J. Environ. Sci. Technol. 2001, 35, 3560-3565. (44) Scherer, M. M.; Westall, J. C.; Ziomek-Moroz, M.; Tratnyek, P. G. Environ. Sci. Technol. 1997, 31, 2385-2391. (45) Burris, D. R.; Allen-King, R. M.; Manoranjan, V. S.; Campbell, T. J.; Loraine, G. A.; Deng, B. J. Environ. Eng. 1998, 124, 10121019. (46) Burris, D. R.; Campbell, T. J.; Manoranjan, V. S. Environ. Sci. Technol. 1995, 29, 2850-2855. (47) Allen-King, R. M.; Halket, R. M.; Burris, D. R. Environ. Toxicol. Chem. 1997, 16, 424-429. (48) Larson, R. A.; Weber, E. J. Reaction Mechanisms in Environmental Organic Chemistry; Lewis/CRC: Boca Raton, FL, 1994. (49) Barrows, S. E.; Cramer, C. J.; Truhlar, D. G.; Elovitz, M. S.; Weber, E. J. Environ. Sci. Technol. 1996, 30, 3028-3038. (50) Jolas, J. L.; Pehkonen, S. O.; Maloney, S. W. Water Environ. Res. 2000, 72, 179-188. (51) Deng, B.; Campbell, T. J.; Burris, D. R. Environ. Sci. Technol. 1997, 31, 1185-1190. (52) Halliday, D.; Resnick, R. Fundamentals of Physics; John Wiley & Sons Inc.: New York, 1988. (53) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; Wiley: New York, 1993. (54) Kawahara, F. K.; Michalakos, P. M. Ind. Eng. Chem. Res. 1997, 36, 1580-1585. (55) Oh, S. Y.; Cha, D. K.; Kim B. J.; Chiu, P. C. Environ. Toxicol. Chem., in press.

Received for review December 12, 2001. Revised manuscript received March 5, 2002. Accepted March 6, 2002. ES011474G