Effects of Alcohols, Anionic and Nonionic Surfactants on the Reduction of Pce and Tce by Zero-Valent Iron

EFFECTS OF ALCOHOLS, ANIONIC AND NONIONIC SURFACTANTS ON THE REDUCTION OF PCE AND TCE BY ZERO-VALENT IRON

GREGORY A. LORAINE* Air Force Research Laboratory, AFRL/MLQR, 139 Barnes Dr. Tyndall AFB, FL 32403, USA

(First received 1 July 1999; accepted in revised form 1 July 2000)

Abstract}The effects of surfactants, sodium dodecyl sulfate (SDS) and Triton X-a00 (TX), and alcohols (methanol, ethanol, and propanol) on the dehalogenation of TCE and PCE by zero-valent iron were examined. Surface concentrations of PCE and TCE on the iron were dependent on aqueous surfactant concentrations. At concentrations above the CMC, sorbed halocarbon concentrations declined and concentrations associated with solution phase micelles increased. The anionic surfactant SDS ([SDS]5 CMC) did not affect reduction rates, until the CMC was exceeded after which reactivity decreased, possibly due to sequestering of the TCE and PCE in mobile micelles. The nonionic TX showed a mixed effect on reactivity, increasing the PCE reduction rate, but not affecting TCE removal. Production of TCE from PCE increased in the presence of TX. Similar experiments showed that methanol, ethanol, and propanol inhibited reduction of TCE and PCE by metallic iron. Zero-valent iron may be useful in recycling soil washing effluents contaminated with TCE and PCE. # 2001 Elsevier Science Ltd. All rights reserved

Key words}zero-valent iron, surfactants, TCE, PCE, cosolvents

INTRODUCTION two halogens are lost simultaneously. Both of these Alcohols and surfactants are increasingly used in the processes involve a net exchange of two electrons. remediation of subsurface contamination by spar- Hydrogenolysis of PCE produces TCE, while reduc- ingly soluble chlorinated organic compounds such tive b-elimination results in dichloroacetylene. Tri- as tetrachloroethene (PCE) and trichloroethene chloroethene itself can undergo b-elimination to (TCE) (Jawitz et al., 1998; Smith et al., 1997). If produce chloroacetylene, or hydrogenolysis to give the chlorinated organics could be removed the 1,1-dichloroethene (DCE), cis 1,2-DCE, or trans 1,2- recovered cosolvents or surfactants could be reused. DCE. Dichloroethene degrades to vinyl chloride or However, removal of the chlorinated contaminants is acetylene. The final products are ethene and ethane. problematic. One method of reducing halogenated Dichloroacetylene rapidly undergoes hydrogenolysis alkenes in aqueous solutions uses zero-valent iron to produce chloroacetylene. Like DCE, chloroacety- (Gillham and O’Hannesin, 1994; Campbell et al., lene will produce either vinyl chloride or acetylene. 1997). The feasibility of combining these two Arnold and Roberts (1999) examined the reduction technologies, using zero-valent iron to dechlorinate of PCE and the resulting product formations and PCE and TCE in typical enhanced solubilization reactions with pure iron granules in buffered water matrices, was examined. (pH=7.2). They determined the relative selectivity of In anaerobic solutions chlorinated organic com- each pathway by fitting a complex mechanism. The pounds such as PCE and TCE are dehalogenated by iron they used is different that used in this study, so metallic iron (Campbell et al., 1997). Reductive direct comparisons cannot be made. However, their dehalogenation occurs via two pathways hydrogeno- results are instructive. They found that of the PCE lysis (two sequential one e transfers removing a reduced 87% proceeded via b-elimination and 13% via hydrogenolysis and 97% of TCE reduced under- halogen and adding a hydrogen from H2O), and reductive b-elimination in which two e are added and went b-elimination and 3% hydrogenolysis. b-elimination was the dominant pathway for all of the above reactions. *Author to whom all correspondence should be addressed. It has been shown that surfactants can increase the Department of Civil and Environmental Engineering, surface concentration of sparingly soluble organics San Diego State University, 5500 Campanile Drive, San on non-reactive surfaces such as soil and sediment Diego, CA 92182, USA. E-mail: [email protected] particles in water (Ko et al., 1998; Sun et al., 1995;

1453 1454 Gregory A. Loraine

Burris and Antworth, 1992). Ko et al. (1998) chlorinated organics have made the assumption that demonstrated that anionic and nonionic surfactants increasing surface concentrations would increase raised surface concentrations of aromatic compounds reaction rates. Sayles et al. (1997) found that the on kaolinite particles. The surface concentration was reduction of DDT (1,1,1-trichloro-2, 2-bis (p-chloro- dependent on the aqueous surfactant concentration. phenyl) ethane) by zero-valent iron increased 40% in Once the critical micelle concentration (CMC, the the presence of Triton X-114 (an octylphenoxy surfactant concentration at which monomers aggre- ethoxylate that differs from Triton X-100 in ethoxy- gate to form micelles) was reached, partitioning to ethanol chain length). The enhanced rate was the mobile phase micelles began to compete with attributed to increased aqueous solubility of the sorption into the surfactant layer on the clay hydrophobic DDT and its products leading to surfaces. Above the CMC the sparingly soluble increased mobility of the DDT. The concentrations organic concentration on the surface declined on the surface of the iron were not measured. Gu (Sun et al., 1995). This was especially true for anionic et al. (1997) found that a mixture of 2% Aerosol surfactants. While enhanced surface sorption has a MA-1 (an anionic surfactant) and 2% ethanol or negative effect on the mobilization of hydrophobic isopropanol did not affect the kinetics of TCE organics in groundwater, it may be possible to increase reduction by Fe0. Li (1998) studied the reduction of the concentration of chlorinated organics on reductive PCE by Fe0 and Fe0 coated with HDTMA (cationic iron surfaces using surfactants. The reduction rates of surfactant) or SDS. The reduction of PCE with halogenated compounds by zero-valent iron are slow HDTMA coated iron was three times faster than and long contact times are required. Surfactants could unmodified iron. The SDS coated iron was only be used to retard the flow of pollutants through slightly faster than unmodified iron. The surface permeable barriers thus increasing contact time and concentrations of PCE, TCE or SDS were not decreasing the required size of the barrier. determined. The aqueous concentrations of TCE Micelles of sodium dodecyl sulfate in water form produced from PCE were higher with the surfactant with the polar OSO3 head facing the aqueous phase modified irons. It was concluded that increasing the and the hydrophobic dodecyl tail in the interior. On surface concentration of PCE increased the reaction surfaces SDS monomers organized into patches of rates. These previous studies suggest that increased bilayers or hemimicelles with the anionic heads surface concentration increased reaction rate but did pointed out giving the surface a negative charge. not directly measure the surface concentrations. On surfaces with a net negative charge, like corroding In this study, the effects on reduction rates of TCE iron, electrostatic repulsion could have caused the and PCE with zero-valent iron by modifying the SDS bilayers to roll up into spherical micelles on surface concentration with surfactants and alcohols the surface (Rusling, 1997; Chen et al., 1999). The were determined. Surfactant solutions containing an nonionic octylphenol ethoxylate (TX) was not sub- anionic surfactant, sodium dodecyl sulfate or a ject to electrostatic repulsions and may have pro- nonionic surfactant, Triton X-100, at concentrations vided more complete coverage. above and below the CMCs were used. Reduction Electron transfer can be affected by non-electro- reactions in solutions of ethanol, methanol, and active surfactants (Rusling, 1997; Chen et al., 1999; 2-propanol at concentrations typical of field reme- Possidonio and El Seoud, 1999). For electron diation efforts were also examined. transfer to occur the substrate must be on or near the surface. If a surfactant monomer is on an active EXPERIMENTAL SECTION site the substrate must either displace it or approach within one headgroup distance of the surface (Rusl- All alcohol (ethanol, EtOH, Pharmco; methanol, MeOH, ing, 1997). Solutes can enter the bound surfactant Fisher, HPLC grade; and 2-propanol, IPA, Fisher, HPLC grade) and surfactant solutions (sodium dodecyl sulfate, layer either by diffusion from the bulk liquid or by Sigma, sodium salt 99%; Triton X-100, t-Octylphenoxypo- being transported with micelles merging with the lyethoxy ethanol, Sigma, SigmaUltra grade) were made in surface aggregates. Transfer into micelles of SDS by argon sparged Milli-Q water and degassed under vacuum small nonionic molecules such as PCE and TCE have for at least 20 min and stored in an anaerobic chamber (5% H2, 95% N2). The CMCs were found in the literature to be been shown to proceed at diffusion controlled rates 2.31 g/L (8.0 mM) for SDS and 0.15 g/L for Triton X-100 (Rusling, 1997). Diffusion out of micelles was (Liu and Roy, 1995; Mukerjee and Mysels, 1971). inversely related to the chain length of the hydro- Aqueous SDS concentrations were determined by TOC phobic tail (Possidonio and El Seoud, 1999). analyzer (Shimadzu, TOC-5000A) after dilution to Regardless of the transfer mechanism, once inside TOC51000 mg/L with Milli-Q water (detection limit 0.05 mg/L). Triton X-100 concentrations were measured the surfactant layer the solute orients itself with spectrophotometrically at 277.5 nm, e ¼ 1220 M/cm (Cary respect to the interface on a microsecond time scale 3E UV-Visible Spectrophotometer, detection limit 0.001 mg/ (Rusling, 1997). This orientation depends on solvent L). All samples were filtered through a 0.22 mm filter (Lida, interactions in the interior of the micelle and the size Pro-X 0.22 mm hydrophilic cellulose acetate membrane) before analysis. and charge of the headgroup. Sorption isotherm experiments for the surfactants were Recent work examining surfactants or surfactant/ run in zero headspace batch experiments with duplicate, alcohol mixtures in studies of Fe0 reduction of sacrificial samples. Initial aqueous concentrations for SDS Effects of PCE and TCE in alcohols and surfactants 1455 were 0.58, 1.73, 2.31, 4.61, and 23.07 g/L and for TX were 0.05, 0.1, 0.15, 0.3, 1.5, and 7.5 g/L. Acid washed iron filings (5 g, iron metal Fisher Scientific 40 mesh, SA=1.25 m2/g, 3.1% C w/w) were placed in 15 m serum vials (Wheaton). The vials were filled completely with liquid (14.2 mL) in an anaerobic chamber and crimp sealed with Teflon-faced rubber septa. Control samples contained no iron. The samples were mixed by axial rotation on a roller drum at 8 rpm, at 20 Æ 18C in the dark for 24 h. Surface concentrations were taken to be the difference between aqueous concentrations in the samples with iron and the controls. Reduction experiments of PCE and TCE were preformed in a similar manner with the following changes. Immediately after being sealed, the vials were removed from the chamber and spiked with methanol solutions of either tetrachloroethene (J. T. Baker, Photrex grade) or trichloroethene (Fisher Scientific, Certified ACS Grade) (500 mg/mL) to give a final concentration of 36 mg/L. Duplicate control and experimental samples were removed after varying periods of time and analyzed. Each vial was analyzed to determine aqueous and total PCE or TCE concentrations. Duplicate 20 mL aliquots of the aqueous phase were sampled from the vial and spiked into 1.0 mL of acetonitrile (Fisher Scientific, PHLC Grade) with 10 mg/L p-dichlorobenzene (Fisher Scientific, Reagent Grade) as an internal standard and vortexed. Total system concentration was determined by transferring the aqueous content of the vial (14.2 mL) into a 40 mL vial containing 10 mL of acetonitrile via cannule. The remaining solids were washed with two 2.5 mL portions of acetonitrile (spiked with internal standard) and vortexed for 1 min. The washing solvents were added to the 40 mL vial. A 100 mL aliquot was spiked into 0.90 mL of acetonitrile and analyzed via GC/ECD. Burris et al. (1995) reported extraction recoveries for PCE and TCE of approximately 100% for this method. Fig. 1. Sorption Isotherm of SDS (a) and Triton X-100 (b) Sorption isotherms for TCE and PCE were calculated on clean cast iron (Fisher, >40 mesh, SA=1.25 m2/g) at assuming the difference between the total and aqueous 208C after mixing at 8 rpm for 24 h in the dark. Iron/liquid concentrations was bound to the iron surface (control samples ratio was 5 g/14.2 mL. The data points are labeled with the with no iron present showed that sorption to the walls of the initial surfactant concentration. (a) [SDS]0=0.58, 1.7, 2.3 glass vial was negligible). Sorption isotherms (208C) were (CMCAqu), 4.6, and 23.1 g/L. (b) [Triton X-100]0=0.05, made by plotting the bound concentration (mg/g) versus the 0.10, 0.15 (CMCAqu), 0.30, 1.5, and 7.5 g/L. aqueous concentration (mg/ml) as the reaction progressed. Since both the aqueous and total concentrations were determined at the same time point the isotherm calculations reflect only the equilibrium and were independent of substrate of cast iron at 20 Æ 18C after 24 h. Sorption of SDS loss. Reaction rates were determined from the change in total onto the cast iron increased until the CMC in the concentration between time samples. aqueous phase was reached. Above the CMC, the Tetrachloroethene and trichloroethene concentrations were determined by GC on a 15 m 0.53 mm id HP-1 column concentration of bound surfactant reach a plateau with film thickness of 3 mm. The temperature program was and aqueous micelles formed. Surface sorption of TX 508C for 2 min, ramp at 258C/min to 2008C. Helium gas flow continued to increase beyond the published aqueous rate was 3.7 mL/min. Injector and detector temperatures CMC concentration. The apparent change in CMC were 150 and 2808C, respectively. Samples were injected splitless (1 mL). An anode purged Ni63 ECD detector was of non-homogeneous surfactants in water–particu- used, with a 5% CH4 in Ar make up gas flow of 20 mL/min. late systems has been observed previously (Hayworth Peaks were identified by co-elution with known standards, and Burris, 1997). Triton X-100 is a mixture of and detector responses were calibrated as the relative octylphenol ethoxylates of different ethoxyethanol responses of the analytes to the internal standard on a chain lengths. It is thought that these different multilevel standard curve. Limits of detection were 0.30 mg/ L(2.3mM) for TCE and 05.0 mg/L (3.1 mM) for PCE. monomers exhibit different solubilities and form Analyses of products of PCE and TCE reduction reactions micelles sequentially. The solid phase may contribute were based on the method of Campbell et al.(1997)withthe to the differentiation of micellization and the CMC following modifications: five grams of acid washed iron were observed may be greater than that seen in aqueous used instead of 20 g and pyrite was not added as a buffer; previous studies have shown that the pH increase is negligible solutions. The CMC in this system was not deter- in the Fisher iron/unbuffered DI system (Deng, 1998). mined. The actual CMC was assumed to be near 1.5 g/L Triton X-100. RESULTS AND DISCUSSION The sorption isotherms for SDS and TX were also determined after 2 weeks of exposure time (336 h). SDS matrices The isotherms were essentially identical to those Figure 1 shows the sorption isotherms of SDS obtained after 24 h for both surfactants (data not (Fig. 1(a)) and Triton X-100 (Fig. 1(b)) on the surface shown). After prolonged exposure to anaerobic water 1456 Gregory A. Loraine the net surface charge of the cast iron would be Although there was loss due to reduction on the expected to become increasingly negative and inhibit surface of cast iron, the reaction rates were relatively sorption of the anionic SDS and change the isotherm. slow and quasi-steady-state conditions should have A simplified conceptual model of the aqueous occurred (Burris et al., 1998). Plotting the mobile surfactant–iron system consists of five phases: the phase and bound phase concentrations yielded aqueous phase, mobile surfactant micelles, surfactant sorption isotherms. Figure 3 shows the sorption hemimicelles or double layers on the iron surface, isotherms for PCE on cast iron in deionized (DI) reactive binding sites on the iron surface, and non- water and four concentrations of SDS (initial reactive graphite inclusions on the iron (Fig. 2). concentrations 0.58, 2.31, 4.61, 23.07 g/L). The Burris et al. have shown that there is significant non- isotherms were nonlinear and were fit using the reactive, non-linear sorption of hydrophobic organics Langmuir isotherm, see Table 1 (Sposito, 1980). to graphite inclusions (3.1% w/w) in cast iron Burris et al. (1995) fit TCE and PCE sorption (Burris et al., 1998). isotherms using the generalized Langmuir equation. The concentrations of the chlorinated compounds The sorption of PCE on iron surfaces increased as were measured as mobile (solubilized and micellar surfactant concentration approached the CMC then concentrations) and bound (sorbed to iron active declined at higher surfactant concentrations. A sites, iron graphite, and iron-surfactant phase). similar profile was observed for TCE (Table 1). Partitioning of organic compounds from aqueous phase to surface-bound surfactants has been shown to be dependent on the hydrophobicity of the compound. As PCE is less soluble than TCE the change in sorption of PCE is more pronounced (solubility at 258C, PCE = 0.15 g/L, TCE =1.10 g/L, Verschueren, 1983). The pseudo-first-order rates of reduction of TCE and PCE in different concentrations of SDS are given in Table 2. Statistical analysis of the reduction rate constants showed no significant difference between DI water and SDS concentrations below 23.07 g/L (t-test, P ¼ 0:05). Above 23.07 g/L the rate constants decreased, possibly due to portioning of TCE and PCE from the iron surface to mobile micelles formed at this concentration (Fig. 1(a)). As can be seen from Fig. 2. In a simple model of an aqueous-surfactant-cast iron Table 2 and Fig. 3, addition of surfactant increased system a sparingly soluble organic (SSO) can partition into five phases. The mobile phases are the aqueous fraction PCE sorption but did not result in a corresponding (SSOAqu) and the micelles. The bound phases are the increase in reaction rate. There does not appear to be surfactant hemimicelles on the surface, graphitic inclusions a direct relationship between the dehalogenation rate in the cast iron (3% w/w), and the metallic iron surface. and net surface concentration.

Triton X-100 matrices Figure 4 shows the sorption isotherms of PCE in DI water and Triton X-100 matrices. The TCE isotherms were similar (Table 1). The surface layer of TX continued to form until the aqueous concentration reaches 1.5 g/L (Fig. 1(b)), however, the [PCE]bound decreases with increasing nonionic surfactant concentration. This differed from what was observed with SDS where [PCE]bound increased until CMC level of SDS was reached. The reduction rate constants of TCE and PCE in the presence of Triton X-100 are listed in Table 3. The rate constants for TCE reduction were statistically identical (P ¼ 0:05) at all nonionic surfactant concentrations. Fig. 3. PCE in SDS sorption isotherm on cast iron (>40 On the other hand, the PCE reduction rates increased mesh cast iron, SA=1.25 m2/g) at 208C in the dark at a with surfactant concentration (rates at the two lowest rotation speed of 8 rpm. (^) DI Water; ( ) 0.58 g/L SDS; TX concentrations were not significantly different than ( ) 2.31 g/L SDS; ( ) 4.61 g/L SDS; (*) 23.1 g/L SDS. Curves fit using Langmuir equation (outliers were deter- DI, P ¼ 0:05) and reached a maximum value at mined by residual analysis and omitted from fitting), [TX]=1.5 g/L, the concentration where maximum samples were taken after 2, 24, 48, 72, 96, 168, and 240 h. [TX]bound is reached. At 1.5 g/L of Triton X-100 the Effects of PCE and TCE in alcohols and surfactants 1457

Table 1. Langmuir coeﬃcients for TCE and PCE sorption isotherms in SDS and TX, b¼ absorption capacity (mg/g) and K¼ equilibrium constant (mL/mg)

TCE PCE

(TX) g/L b (mg/g) K (mL/mg) r2 b (mg/g) K (mL/mg) r2

0 28.5 0.33 0.90 46.3 0.49 0.94 0.05 36.4 0.15 0.91 33.1 0.93 0.89 0.1 28.4 0.13 0.97 29.3 1.03 0.86 0.15 38.6 0.11 0.97 27.3 1.34 0.84 0.3 26.8 0.42 0.81 25.7 1.40 0.84 1.5 24.3 0.29 0.71 18.9 2.36 0.88 7.5 ND NDa ND 23.9 0.50 0.6 15.0 ND ND ND 16.0 0.14 0.56 (SDS) g/L b (mg/g) K (mL/mg) r2 b (mg/g) K (mL/mg) r2 0 28.5 0.33 0.95 46.3 0.49 0.96 0.58 30.4 0.32 0.90 76.4 0.11 0.97 1.73 33.6 0.27 0.92 91.9 0.11 0.93 2.31 31.7 0.32 0.81 123.1 0.07 0.97 4.61 23.0 0.28 0.91 66.5 0.07 0.98 23.1 12.3 0.20 0.86 0.3 0.04 0.20 a ND – not determined.

Table 2. Effect of SDS anionic surfactant on pseudo-first order rate Table 3. Effect of triton X-100 nonionic surfactant on pseudo-first constants for reduction of TCE and PCE (total concentrations) with order rate constants for reduction of TCE and PCE with zero-valent 0 0 zero-valent Iron. [TCE]0=36 mg/L, [PCE]0=36 mg/L, [Fe ]0=5.00 iron. [TCE]0=36 mg/L, [PCE]0=36 mg/L, [Fe ]0=5.00 Æ 0.005 g, Æ 0.005 g, SA=1.25 m2/g, in 14 mL DI water or surfactant solution, SA=1.25 m2/g in 14 mL DI water or surfactant solution, well mixed well mixed TCE PCE TCE PCE Triton X-100 g/L Rate ( Â 107/s m2) Rate ( Â 107/s m2) SDS g/L Rate ( Â 107/s m2) Rate ( Â 107/s m2) 0 16.7 Æ 0.16 4.35 Æ 0.42 0 16.7 Æ 0.2 4.35 Æ 0.42 0.05 15.6 Æ 0.24 4.18 Æ 0.11 0.58 12.4 Æ 0.8 3.62 Æ 0.80 0.1 14.1 Æ 1.4 4.24 Æ 0.27 1.73 11.4 Æ 0.04 3.32 Æ 0.13 0.15 14.2 Æ 0.4 5.06 Æ 0.46 2.31 12.1 Æ 0.6 3.81 Æ 0.78 0.3 12.8 Æ 0.16 5.29 Æ 0.36 4.61 12.6 Æ 0.96 4.00 Æ 0.96 1.5 13.0 Æ 0.8 6.05 Æ 0.38 23.1 8.8 Æ 0.3 1.55 Æ 0.9 7.5 NDa 5.06 Æ 0.14 15.0 ND 3.25 Æ 0.13

a ND – not determined.

resulting effluent would be a surfactant matrix with PCE concentrations much higher than those possible in water. Reduction rates of PCE by Fe0 in 1.5 g/L Triton X-100 were determined at five initial concentrations (Fig. 5). Heterogeneous reactions can be though of as three step processes, (1) adsorption- desorption of the substrate on the reactive surface, (2) reaction of the surface-bound substrate, (3) the desorption of the products (Carberry, 1976). The products of PCE and TCE reduction of Fe0 also absorbed to the surface and could have competed Fig. 4. PCE in Triton X-100, sorption isotherm on cast iron with the parents for active sites (Arnold and Roberts, 2 (>40 mesh cast iron, SA=1.25 m /g) at 208C in the dark at 1999). However, at high initial PCE concentrations ^ a rotation speed of 8 rpm. ( ) Water; ( ) 0.05 g/L Triton this competition may be neglected. Due to the long X-100; (m) 0.15 g/L Triton X-100; ( ) 1.5 g/L Triton X-100; ( ) 15.0 g L Triton X-100. Curves fit using Langmuir half-life of PCE in this system (53 h), it can be equation (outliers were determined by residual analysis assumed that partitioning equilibrium conditions and omitted from fitting), samples were taken after 2, 24, 48, were reached and the rate of the surface reaction 72, 96, 168, and 216 h. was the rate controlling step. Under these conditions the data can be modeled using a Langmuir equation. pseudo-first order rate for PCE loss is 39% greater d½PCE kÂKaÂ½PCEAqu than in water alone. At higher TX concentrations the ¼ V ¼ ð1Þ dt 1 þ K Â½PCE rates diminish perhaps due to partitioning of PCE a Aqu from the iron surface into mobile micelles. where k is the rate of reaction per surface area and Were Triton X-100 to be used in soil washing Ka is the adsorption equilibrium constant of PCE applications for a PCE contaminated aquifer, the per gram iron. Values of k ¼ 1:07 nM/s-m2, and 1458 Gregory A. Loraine

Table 4. Pseudo ﬁrst-order rate constants for TCE ([TCE]0= 22–49 mg/L) and PCE ([PCE]0=35–40 mg/L) reduction in alcohol/ 0 2 water solutions, [Fe ]0=5.00 Æ 0.005 g, SA=1.25 m /g, in 14 mL DI water or alcohol solution, well mixed at 208C

TCE PCE

Sample Rate ( Â 107/s m) Rate ( Â 107/s m)

DI 16.7 Æ 0.16 4.35 Æ 0.42 MeOH (57%) 1.42 Æ 0.16 0.46 Æ 0.02 IPA (57%) 0.97 Æ 0.16 0.46 Æ 0.16 EtOH (57%) 1.82 Æ 0.11 0.47 Æ 0.03 EtOH (28%) 15.3 Æ 0.01 NDa EtOH (100%) 0.005 Æ 0.016 ND

a NC – not determined.

Fig. 5. Langmuir–Hinshelwood plot of PCE reduction by Fe0 in Triton X-100 (1.5 g/L). The surfactant allows increased loading of PCE in the system and more PCE is of PCE reduced in DI and 1.5 g/L TX (mass balance degraded per gram of iron than is possible in water at this for DI=77%, TX=79%). The yield of TCE is higher solid/liquid ratio. However, even in the presence of in TX than in DI, and the production of ethene, surfactant, solubility limits the maximum loading before ethane, and DCE is approximately equal. Fractiona- the active sites are saturated. tion of TCE and PCE to the surface may have aﬀected the relative aqueous concentrations. In addition, TCE was reacting as it was formed and was in competition for active surface sites with PCE and other products. Arnold and Roberts (1999) used a modiﬁcation of the Langmuir–Hinshelwood–Hou- gan–Watson (LHHW) equation to model the formation of reactive products from parallel reactions on zero-valent iron. This same method was used to estimate the true yield of TCE. d½TCE ¼ YÂk Â½PCE dt P Aqu K Â½TCE aT P Aqu kTÂ½TCEAqu ð2Þ 1 þ KaiCi

where Y is the fraction of PCE going to TCE, kP and kT are the measured pseudo-first order reaction rates of PCE and TCE, respectively, KaT is the adsorption coefficient of TCE from the Langmuir isotherm (Table 1), and SKaiCi is the sum of the concentra- Fig. 6. Product yields of PCE reduction by zero-valent iron tions times adsorption coefficient of the parent and in DI (a) and [TX]=1.5 g/L (b). (^) Ethene; (&) Ethane; all the products. Since the major products were TCE, (m) TCE, (*) cis-DCE; ( ) 1,1-DCE. ethene, and ethane only PCE and TCE were considered, this might have resulted in an over- estimation of Y. However, this method was accurate Ka ¼ 0:24 m/M were determined. This equation was enough to demonstrate the significant difference in used to describe the system empirically and cannot be TCE yield between DI and TX. taken as proof of the assumptions made above. Solubility constraints of PCE in 1.5 g/L TX pre- d½TCE ¼ YÂkPÂ½PCEAqu kTÂ½TCEAqu vented determining the reduction rates at higher dt initial PCE concentrations. KTÂ½TCE Product analysis of PCE reduction in both Triton Â Aqu X-100 and in DI water showed that the major 1 þ KPÂ½PCEAqu þ KTÂ½TCEAqu products were TCE, ethene, and ethane. Minor products were methane, cis-DCE, 1,1-DCE, and ð3Þ acetylene. Trace amounts of trans-DCE were de- Table 5 lists the values of Y fitted by Eq. (3) to data tected. Vinyl chloride and chloroacetylene concen- from five matrices, DI, [TX]=0.15 g/L, [TX]=1.5 g/ trations were below detection limits. The only L, [SDS]=1.7 g/L, [SDS]=4.6 g/L (product concen- significant difference between DI and the surfactant trations from the alcohol experiments were at or solutions was the amount of TCE that built up in the below detection limits and were not modeled). The system. Figure 6 shows the yield of products per mole presence of SDS did not affect the formation of TCE Effects of PCE and TCE in alcohols and surfactants 1459

Table 5. Fraction of PCE going to TCE estimated using Eq. (3). The surfaces (Campbell et al., 1997). It may be that in a fraction of TCE formed in SDS did not differ significantly from DI Triton X-100 hemimicelle the hydrogenolysis path- water. Higher concentrations [TX]bound (Fig. 1(b)) corresponded to higher yields of TCE from PCE reduction way was favored over b-elimination. The TX may have acted as a H donor. Arnold and Roberts (1999) Matrix Est. TCE yield r2 TCE reported that hydrogenolysis is more important in PCE reduction than in TCE reduction. This is one DI 0.04 Æ 0.04 0.79 [TX]=0.15 g/L 0.09 Æ 0.01 0.94 possible explanation for why reduction of PCE is [TX]=1.5 g/L 0.22 Æ 0.02 0.99 enhanced by TX and TCE is not. If the TX matrix [SDS]=1.7 g/L 0.02 Æ 0.01 0.56 favors hydrogenolysis then the formation of vinyl [SDS]=4.6 g/L 0.04 Æ 0.02 0.42 chloride would also be enhanced. Vinyl chloride is of concern due to its greater toxicity. If the hydrogenolysis pathway was favored then more DCE and VC should have been formed. No Vinyl chloride was detected in these experiments (detection limit >4 Â 109 M). Perhaps the reduction rates of DCE and VC in TX are also faster. Verification of this hypothesis has yet to be obtained.

Alcohol matrices Pseudo-first-order rate constants for the reduction of TCE and PCE by metallic iron in DI water, MeOH (57% v/v), IPA (57% v/v), and EtOH (28, 57, and 100% v/v) are shown in Table 4. The dehalogenation rates varied with the type and concentration of alcohol. None of the alcoholic solutions tested improved the removal rate of either TCE or PCE. However, the lowest concentration of ethanol used (28%) did not significantly inhibit TCE dehalogenation. Application of zero-valent iron permeable barriers with ethanol soil washing may be practical Fig. 7. Estimation of yield of TCE using LHHW equation if TCE could be efficiently solubilized at relatively & using data from Fig. 6. DI: ( ) PCEDI;(&) TCEDI, (yield) low EtOH concentrations. Surface sorption of TCE 2 * 0.042 Æ 0.042, (rTCE) 0.79. TX=0.15 g L: ( ) PCE0.15,(*) 2 m and PCE in alcoholic matrices was inhibited even in TCE0.15, (yield) 0.09, (rTCE) 0.94. TX=1.5 g L: ( ) PCE1.5; 2 28% EtOH (data not shown). (4) TCE1.5, (yield) 0.224 Æ 0.017, (rTCE) 0.99.

CONCLUSIONS relative to DI. The estimated formation of TCE increased with increasing [TX] ([TX]=0.15 g/L Increasing the concentrations of PCE/TCE near Y ¼ 0:09, [TX]=1.5 g/L Y ¼ 0:22). Data from ex- the iron surface by sorption into surfactant layers periments with DI, [TX]=0.15 g/L, [TX]0=1.5 g/L, was insufficient to increase reduction. The reduction and the fitted lines are shown in Fig. 7. The observed rates observed in SDS remained low in spite of enhancement in aqueous TCE concentration cannot increased [PCE]bound and [TCE]bound. Triton X-100 be attributed to the differences in absorption of T enhanced PCE reduction by as much as 40%, CE/PCE in DI and TX. This implies that TX dependent on the surfactant concentration. The influences the reaction mechanism of PCE. correlation between [TX]bound and reduction rate Hydrogenation reactions catalyzed by heteroge- suggests that Triton X-100 hemimicelles on iron may neous catalysts are greatly affected by the solvent have acted as a reactive phase for PCE reduction. As (Rylander, 1980). The solvent can competitively bind the surface coverage of the surfactant increased the to active sites, change hydrogen availability, di- reduction rate of PCE increased. It was only when electric constant or reactant solubility. Changes TX concentrations reached a level where unbound in reaction rates and product yields in catalytic micelles are formed that the rates declined. The TX hydrogenations are often manipulated by means of affected product yields as well as rates. The observed changing the solvent. The surfactant hemimicelles on yield of TCE from PCE reduction increased as [TX] the iron surface provides a solvent phase very increased, which may be a potential drawback in the different than water. Thus, changes in product yield application of this technique in the field. and reaction rates are not unexpected. The difference in the reduction rates of PCE in Previous work has shown that both TCE and PCE SDS and TX may be due to electrostatic repulsion undergo concerted two electron b-elimination and hindering electron transfer into anionic SDS surface hydrogenolysis via stepwise electron addition on iron aggregates (Horvath and Stevenson, 1999). All of the 1460 Gregory A. Loraine alcohol solutions tested decreased reduction rates, ganic contaminants from within cationic surfactant probably due to inhibited mass transfer to surface enhanced sorbent zones. 1. Experiments. Environ. Sci. active sites. Technol. 31, 1277. Horvath O. and Stevenson K. L. (1999) Micellar effects on photoinduced redox reactions of Fe(bpy)2(CN)2. Acknowledgements}This work was supported by a Na- J. Photochem. Photobio. A. 120, 185. tional Research Council Resident Research Associateship Jawitz J. W., Annable M. D., Rao P. S. C. and Rhue R. D. provided by the Air Force Office of Scientific Research. (1998) Field implementation of a winsor type 1 surfactant/alcohol mixture for in situ solubilization of a complex LNAPL as a single-phase microemulsion. REFERENCES Environ. Sci. Technol. 32, 523. Ko S.-O., Schlautman M. A. and Carraway E. R. (1998) Arnold W. and Roberts A. L. (1999) Pathways and kinetics Partitioning of hydrophobic organic compounds to of chlorinated ethylene and chlorinated acetylene reaction sorbed surfactants. 1. Experimental studies. Environ. Environ. Sci. Technol. with Fe(0) particles. 34, 2850. Sci. Technol. 32, 2769. Burris D. R., Allen-King R. M., Manoranjan V. S., Li Z. (1998) Degradation of perchloroethylene by zero- Campbell T. J., Loraine G. A. and Deng B. (1998) valent iron modified with cationic surfactant. Adv.Envir. Chlorinated ethene reduction by cast iron: sorption and Res. 2, 244. J. En r. En r. mass transfer. v g 1012. Liu M. and Roy D. (1995) Surfactant-induced interactions Burris D. R. and Antworth C. P. (1992) In situ modification and hydraulic conductivity changes in soil. Waste Manag. of an aquifer material by a cationic surfactant to enhance 15, 463. retardation of organic contaminants. J. Cont. Hydrol. 10, Mukerjee P. and Mysels K. J. (1971) Critical Micelle 325. Concentrations of Aqueous Surfactant Systems. National Burris D. R., Campbell T. J. and Manoranjan V. S. (1995) Bureau of Standards, Washington, DC. Sorption of trichloroethylene and tetrachloroethylene in a Possidonio S. and El Seoud O. A. (1999) Effects of charge En iron. Sci. batch reactive metallic iron–water system. v and structure of surfactants on kinetics of water Technol. 29, 2850. reactions: the pH independent hydrolysis of bis(2,3- Campbell T. J., Burris D. R., Roberts A. L. and Wells J. R. dinitrophenyl) carbonate. J. Mol. Liquids 18, 231. (1997) Trichloroethylene and tetrachloroehtylene reduc- Rusling J. F. (1997) Molecular aspects of electron tion in a batch metallic iron/water/vapor system. Environ. transfer at electrodes in micellar solutions. Coll. Surf. Toxicol. Chem. 16(4), 625. 123–124, 81. Chemical and Catalytic Reaction Carberry J. (1979) Rylander P. N. (1980) Solvents in catalytic hydrogenation. En ineerin g g. McGraw-Hill, New York, NY. In Catalysis in Organic Syntheses, ed W. H. Jones, pp. Chen H., Li Y., Chen J., Cheng P. L., He Y. and Li X. 155. Academic Press, New York. (1999) Micellar effect in high olefin hydroformylation Sayles G. D., You G., Wang M. and Kupferle M. J. (1997) catalyzed by water-soluble rhodium complex. J. Mol. Cat. DDT, DDD, and DDE dechlorination by zero-valent A. 149,1. iron. Environ. Sci. Technol. 31, 3448. Deng B. (1998) Chlorinated ethene reduction by cast iron: Smith J. A., Sahoo D., McLellan H. M. and Imbrigiotta J. En iron. En r. sorption and mass transfer. v g 124(10), T. E. (1997) Surfactant-enhanced remediation of a tri- 1012. chloroethene-contaminated aquifer. 1. Transport of triton Gillham R. W. and O’Hannesin S. F. (1994) Enhanced X-100. Environ. Sci. Technol. 31, 3565. degradation of halogenated aliphatics by zero-valent iron. Sposito G. (1980) The Surface Chemistry of Soils. Oxford Ground Water 32(6), 958. Press, New York. Gu B., Liang L., Cameron P., West O. R. and Korte N. Sun S., Inskeep W. P. and Boyd S. A. (1995) Sorption of (1997) Degradation of trichloroethylene (TCE) and nonionic organic compounds in soil–water systems polychlorinated biphenyl (PCB) by Fe and Fe-Pd containing a micelle-forming surfactant. Environ. Sci. bimetals in the presence of a surfactant and a cosolvent. Technol. 29, 903. In International Containment Technology Conference, Verschueren K. (1983) Handbook of Environmental Data on February 9–12, 1997, St. Petersburg, FL. Organic Chemicals, 2nd ed. Van Nostrand Reinhold Co, Hayworth J. S. and Burris D. R. (1997) Nonionic New York, NY. surfactant-enhanced solubilization and recovery of or-