Environ. Sci. Technol. 2009, 43, 6983–6988
Graphite- and Soot-Mediated Reduction of 2,4-Dinitrotoluene and Hexahydro-1,3,5-trinitro-1,3,5-triazine S E O K - Y O U N G O H † A N D P E I C . C H I U * ,‡ Department of Civil and Environmental Engineering, University of Ulsan, Ulsan 680-749, South Korea, and Department of Civil and Environmental Engineering, University of Delaware, Newark, Delaware 19716
Received May 14, 2009. Revised manuscript received July 30, 2009. Accepted August 04, 2009.
Black carbon (BC) is an important class of geosorbents that influencethefateandtransportoforganicpollutants.Itiscommonly assumed that molecules sorbed to BC are chemically inert. Here we show that this is not true for redox-sensitive sorbates such as nitro-aromatic compounds. In the presence of graphite or n-hexane soot as a BC material, the reduction of 2,4dinitrotoluene (DNT) to 2,4-diaminotoluene by dithiothreitol was greatly accelerated. The para and ortho nitro groups of graphite or soot-sorbed DNT had an approximately equal probability of being reduced. This (1:1) regio-selectivity is different from that when DNT is reduced in homogeneous solution. That is, sorption to BC altered both the kinetics and pathway of DNT reduction. Transformation of hexahydro-1,3,5-trinitro-1,3,5triazine, a nonaromatic nitro compound, by dithiothreitol was also enhanced by graphite, with concurrent formation of formaldehyde. We propose that BC can catalyze the reduction of nitro compounds because it contains microscopic graphitic (graphene) domains, which are both sorption sites and electron conductors. The environmental significance and potential applications of these findings are discussed.
Introduction Sorption of organic chemicals to carbonaceous geosorbents is a pivotal process that controls the fate and transport of hydrophobic pollutants in aquatic and terrestrial systems (1). Black carbon (BC), such as soot, charcoal, coke, cenosphere, graphite, coal, and kerogen, has been recognized as an important class of geosorbents (2). Ubiquitous in soils and sediments, BC is derived from both natural processes and human activities. For example, incomplete combustion of fossil fuels and biomass are the predominant source of BC (2, 3), whereas rock weathering can produce graphite in significant quantities (4). Structurally, BC contains polycyclic aromatic (i.e., graphene) sheets arranged in a highly disordered manner, resulting in its microporous nature and high surface area (2, 3, 5). The current paradigm for organic sorption in soils and sediments, established by much research in recent decades, is a combination of absorption to natural organic matter and adsorption to BC (2, 6, 7). In contrast to natural organic matter, which is amorphous and resembles a solvent for * Corresponding author phone: 302-831-3104; fax: 302-831-3640; e-mail:
[email protected]. † University of Ulsan. ‡ University of Delaware. 10.1021/es901433m CCC: $40.75
Published on Web 08/14/2009
2009 American Chemical Society
nonpolar molecules to dissolve into, BC is hard and glassy and binds sorbate molecules through competitive, nonlinear adsorption onto its condensed surfaces and/or into the nanopores in its structures (8, 9). Because of these sorption mechanisms, BC is a particularly important sorbent for small molecules that have or can adopt a planar configuration, such as alkyl benzenes, small polycyclic aromatic hydrocarbons (PAHs), and polychlorinated biphenyls (PCBs) (10-13). These chemicals are among the most prevalent contaminants in groundwater, soil, and sediment and much research on organic sorption to BC has been carried out using these compounds. In 2005, Pignatello’s group reported that nitroaromatic compounds (NACs) were sorbed to BC even more strongly than substituted benzenes and PAHs (14, 15). These researchers showed that the extent of sorption was related to the electron-accepting or donating ability of the aromatic sorbate. They postulated that this was due to the π-π electron donor-acceptor (EDA) interactions of NACs with edge sites of the graphene planes in BC. They also suggested the electron-rich moieties in BC might be involved in the sorption of nitrobenzene, 2,4-dinitrotoluene (DNT), and 2,4,6-trinitrotoluene (TNT), all of which are good π-acceptors due to the electron-withdrawing nature of the nitro function (15, 16). NACs are also prevalent pollutants due to their widespread uses as explosives (e.g., TNT), agrochemicals (e.g., trifluralin), personal care products (e.g., musk xylene and musk ketone), dyes, and chemical intermediates. Therefore, BC is expected to strongly influence the environmental fate and ecotoxicity of NACs, in addition to PCBs, PAHs, and other aromatic pollutants. It is commonly assumed that when an organic molecule is sorbed to a geosorbent such as BC, it becomes sequestered and inaccessible. Unlike its counterparts in aqueous solution, a BC-sorbed molecule is thought to be biologically and chemically inert. This assumption is common in fate modeling and risk assessment for contaminated soil and sediment (17). It is also the basis of a new remediation approach involving addition of sorbents such as activated carbon to sediments to reduce the bioavailable concentration of PCBs or PAHs (18, 19). However, recent findings by us and other authors suggest that BC-sorbed organic molecules may not be chemically inert, especially when the molecules are redoxsensitive (20-23). For example, we showed that NACs and heterocyclic nitramines could undergo reduction reactions while adsorbed to graphite, which facilitated the reactions by conducting electron from a reductant to the sorbed molecules (21, 22). We propose that this is due to the unique ability of graphite, a sorbent and semiconductor, to sorb organic compounds and to conduct electrons (22). Here, we further proffer the hypothesis that, because of its partial graphitic nature, BC such as soot may have the same ability as graphite to promote the reduction of nitro compounds. In this paper, we present experimental evidence for the dual roles of soot and graphite, as both a sorbent and a redox mediator, in the transformation of DNT and hexahydro-1,3,5trinitro-1,3,5-triazine (RDX) by a thiol reductant.
Materials and Methods Chemicals. We used 2,4-dinitrotoluene as a model NAC for this study. DNT is a structural analogue of the explosive TNT, the fragrances musk ketone and musk ambrette, and other NACs. DNT can also undergo reduction reactions in a regioselective manner (20, 24-26) and thus may offer insights into the reaction mechanism. RDX, a nonaromatic heterocyclic nitramine, was included in this work for comparison VOL. 43, NO. 18, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6983
because it has been examined in related studies (22, 23) and because it is a common explosive and known soil and groundwater contaminant. Dithiothreitol (EHo ) -0.33 V at pH 7) was chosen as a model reductant in this work. It is a common reductant in biochemistry and has been used as a bulk reducing agent in earlier studies (27, 28). DNT (97%), 2,4-diaminotoluene (DAT, 98%), 4-amino2-nitrotoluene (4A2NT, 97%), 2-amino-4-nitrotoluene (2A4NT, 99%), and dithiothreitol (>98%) were purchased from SigmaAldrich (Milwaukee, MI). RDX (>99%) was provided by Holston Army Ammunition Plant (Kingsport, TN). HEPES buffer (N-[2-hydroxyethyl]-piperazine-N’-[ethanesulfonic acid], >99%) was obtained from Sigma (St. Louis, MO). All chemicals were used as received. Graphite powder (20-84 mesh, Alfa Aesar, Ward Hill, MA) and n-hexane soot (kindly provided by Dwight M. Smith of the University of Denver) were the BC materials used in this study. The purity of the graphite powder was 99.9% and its specific surface area was 0.54 ( 0.09 m2/g, as determined using the Brunauer-Emmett-Teller (BET) method with nitrogen. The average BET-to-geometric surface area ratio of this graphite powder was approximately 30. The synthesis of n-hexane soot has been described in detail elsewhere (29). In contrast to the graphite powder, the surface area of n-hexane soot was 89 ( 2 m2/g as measured by the same (BET) method, and its elemental composition has been reported to be 89.5 ( 2.5% C, 1.4 ( 0.2% H, and 8.5 ( 2.5% O (30). A scanning electron microscopy (SEM) image of the n-hexane soot shows that it consisted of agglomerates of spherical particles 50-100 nm in diameter (Supporting Information Figure S1a). An energy dispersive X-ray (EDX) analysis indicates that the carbon and oxygen contents of the soot were 90.1% and 9.9%, respectively (data not shown), similar to the reported values (29, 30). X-ray diffraction (XRD) result of the n-hexane soot shows a broad peak of moderate intensity, suggesting that the structure was poorly crystalline (Supporting Information Figure S1b). Based on the XRD data, the spacing between the graphene layers in the soot was calculated to be 0.346 nm, similar to that of graphite (0.335 nm). Batch Experiments. Sorption and reduction experiments were conducted using sealed 250 mL borosilicate bottles in an anaerobic glovebox (Bel-Art, Pequannock, NJ) under N2. Each bottle contained 200 mL of aqueous solution and either 10 g of graphite powder or 0.05 g of n-hexane soot. The BC masses were chosen to give approximately the same BET surface area per bottle and desirable extent of sorption of DNT in the reductant-free control. The initial concentration of DNT was 0.227 or 0.178 mM. For RDX reaction, 2 g of graphite was added to 200 mL of buffer solution and the initial RDX concentration was 0.182 mM. Solution was deoxygenated by purging with N2 for at least 30 min in the glovebox. The DNT and RDX solutions contained 0.1 M HEPES to maintain a constant pH of 7.4 throughout the experiment. Two hundred mg of dithiothreitol (1.30 mmol) was added to each bottle as a bulk reductant. Each bottle was sealed with a Mininert valve (VICI, Baton Rouge, LA) and low-permeability vinyl tape (Scotch 471, 3M, St. Paul, MN). Two sets of control bottles were set up in parallel under identical conditions except either dithiothreitol or BC was omitted. All bottles were placed horizontally in an orbital shaker at 150 rpm throughout the experiment except during sampling. At different elapsed times, a 1 mL aliquot was withdrawn using a glass syringe and immediately passed through a 25 nm cellulose syringe filter (Millipore, MA) prior to analysis. Only dissolved analytes were quantified in this study. Recovery of sorbed mass was hindered by the strong binding affinity of the BC and was particularly difficult from soot due to its small particle size, low bulk density, and the small mass (50 mg) used. 6984
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 18, 2009
FIGURE 1. Reduction of 2,4-DNT by dithiothreitol in the presence of graphite. Error bars represent one standard deviation calculated from replicates. Chemical Analysis. DNT, DAT, 4A2NT, and 2A4NT were analyzed using a Dionex Summit HPLC (Sunnyvale, CA) equipped with a Dionex Acclaim 120 guard column (4.3 × 10 mm) and an Acclaim 120 C-18 column (250 × 4.6 mm, 5 µm). A methanol-water mixture (50/50, v/v) was used as the mobile phase at a flow rate of 1.0 mL/min for DNT, 4A2NT, and 2A4NT. The wavelength of the UV detector was set at 254 nm. The retention times for DNT, 4A2NT, and 2A4NT were 21.60, 10.12, and 11.26 min, respectively. For DAT, an acetonitrile-phosphate buffer (20 mM, pH 7.0, 30/70, v/v) was used as eluent at 1.0 mL/min. The wavelength of the UV detector was 224 nm and the retention time was 4.73 min for DAT. RDX was analyzed using a Dionex HPLC equipped with a Supelguard guard column (20 × 4.6 mm, Supelco) and a Supelco LC-18 column (250 × 4.6 mm, 5 µm). A methanol-water mixture (70/30, v/v) was used as the mobile phase at a flow rate of 1.0 mL/min. Formaldehyde, a reduction product of RDX, was quantified using the same procedures as described in our previous work (21).
Results and Discussion Graphite-Mediated DNT Reduction. Result of DNT sorption and reduction in the presence of graphite and/or dithiothreitol is shown in Figure 1. In the reductant-free control containing graphite, the aqueous concentration of DNT decreased from 0.227 to 0.184 mM within a day, presumably due to sorption to graphite. Sorption of DNT continued after one day but at a much lower rate; its concentration in solution decreased to 0.153 mM after 21 days. None of the three DNT reduction products were detected during the 21-day period with graphite alone. In the graphite-free control, dithiothreitol was able to reduce DNT slowly in solution. The DNT concentration decreased by about 18% over 21 days (Figure 1). 4A2NT was the only daughter product detected, although its concentration never exceeded 1 µM (data not shown). This shows that the para nitro group of DNT was preferentially reduced by dithiothreitol, a regio-selectivity that is consistent with that observed for DNT reduction initiated by transfer of one electron in homogeneous aqueous solution (24). The preferential reduction of the para nitro group of freely dissolved DNT has been explained by the fact that the para nitro group
is more open and better solvated than the ortho nitro group, and hence can absorb a larger portion of the negative charge following acceptance of an electron by DNT (24). The low reduction rate of DNT and the small quantity of 4A2NT formed suggest that the missing mass was likely the precursors of 4A2NT, such as 4-nitroso- and 4-hydroxyamino-2nitrotoluenes, which were not measured in this study. In contrast to the two controls, the disappearance of aqueous DNT was much faster in reactors containing both dithiothreitol and graphite (Figure 1). The half-life was approximately 2 days, a 40-fold decrease relative to the graphite-free control. DNT reduction in the presence of graphite appeared to follow the well-established pathway of sequential reduction of the two nitro groups to amino groups, each of which involves transfer of six electrons (1, 26, 31). As aqueous DNT disappeared, 4A2NT and 2A4NT were concomitantly produced, reaching 28.0 and 28.8% of the initial DNT mass, respectively. Both intermediates were further reduced to DAT, albeit at lower rates. The final mass recovery was 56.6%, comparable to that for the reductant-free control (Figure 1). Interestingly, the concentrations of 4A2NT and 2A4NT were essentially the same throughout the experiment, in contrast to the (para) selectivity observed in homogeneous solutions, such as our graphite-free control and Barrows et al.’s systems containing either H2S or H2S plus juglone (24). This difference indicates that reduction of graphite-sorbed DNT had a different regio-selectivity than aqueous DNT. In previous work using elemental iron as a reductant, we showed that the ortho nitro group of graphite-sorbed DNT was preferentially reduced (20), and that the reduction was initiated mostly by atomic hydrogen and, to a lesser extent, by electron (20, 32). At that time it was unclear whether the difference in regio-selection between para (for DNT reduction by electron in solution) and ortho (reduction by hydrogen atom on graphite) was due to different reductants or due to sorption to graphite. Because dithiothreitol can donate electron only, the equimolar formation of 4A2NT and 2A4NT in Figure 1 suggests that both sorption to graphite and atomic hydrogen as a reductant would favor reduction of the ortho nitro function of DNT. Figure 1 also suggests that, unlike in aqueous solution, when a DNT molecule was adsorbed to graphite its para and ortho nitro groups had an approximately equal probability of being reduced by electron. Given the slow reduction of aqueous DNT by dithiothreitol and its rapid sorption to graphite, the data in Figure 1 show that, in the presence of both graphite and dithiothreitol, most of the DNT was reduced while being sorbed to graphite. That is, graphite-sorbed DNT molecules were more reductionprone than freely dissolved DNT in water. An adsorbent and semiconductor, graphite played dual roles in this experiment: It adsorbed DNT molecules from water, and facilitated their reduction by shuttling electron to them. The data suggest transfer of electron from dithiothreitol through graphite to sorbed DNT was faster than direct electron transfer without graphite. Combined with our previous finding with elemental iron (20), the results show that, as reductants, Fe(0) and thiols can both support graphite-mediated redox transformation. A conceptual scheme illustrating the proposed graphitemediated reduction of NACs by a thiol is shown in Figure 2. According to this mechanism, sorption and redox reactions are not mutually exclusive but are in fact linked; i.e., graphene-sorbed NAC molecules are more reactive, and extensive sorption to graphene is necessary for fast transformation. Furthermore, given that NAC reduction rates in solution are often controlled predominantly or partly by transfer of the first electron (refs 1, 31, and references therein), the rate enhancement by graphite suggests that the oneelectron reduction potential of graphene-sorbed DNT, or “DNT-graphene adduct,” may be less negative than that of
FIGURE 2. A conceptual scheme for the proposed graphite- and soot-mediated reduction of NAC by a thiol. 1. sorption of aqueous NAC to graphite/soot, 2. electron transfer from thiol through graphite to sorbed NAC, and 3. reductive transformation of sorbed NAC.
FIGURE 3. Reduction of 2,4-DNT by dithiothreitol in the presence of n-hexane soot. Error bars represent one standard deviation calculated from replicates. aqueous DNT, which was reported to be -0.360 (33) and -0.397 V (34). Soot-Mediated DNT Reduction. When n-hexane soot was used in place of graphite, DNT was adsorbed and reduced in a parallel manner (Figure 3). The data variations between duplicate reactors were greater than those in Figure 1, in part due to the larger error associated with measuring 50 mg of soot. Nonetheless, the similarity between the soot and graphite results is clear. In the dithiothreitol-free control, 45% of the DNT was sorbed to soot within a day and 62% in three weeks. Interestingly, traces (
