Enhanced Perchloroethylene Reduction in Column Systems Using ...

9 downloads 1299 Views 132KB Size Report
Institute of Mining and Technology, Socorro, New Mexico. 87801, Department of .... ronmental Studies, University of West Florida, 11000 University. Parkway ...
Environ. Sci. Technol. 2002, 36, 3597-3603

Enhanced Perchloroethylene Reduction in Column Systems Using Surfactant-Modified Zeolite/ Zero-Valent Iron Pellets P E N G F E I Z H A N G , * ,† X I A N T A O , ‡ ZHAOHUI LI,§ AND ROBERT S. BOWMAN† Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, New Mexico 87801, Department of Geology, China University of Geosciences, Wuhan, Hubei 430074, People’s Republic of China, and Geology Department and Chemistry Department, University of WisconsinsParkside, Kenosha, Wisconsin 53141

Surfactant- (hexadecyltrimethylammonium, HDTMA) modified zeolite (SMZ)/zero-valent iron (ZVI) pellets having high hydraulic conductivity (9.7 cm s-1), high surface area (28.2 m2 g-1), and excellent mechanical strength were developed. Laboratory column experiments were conducted to evaluate the performance of the pellets for perchloroethylene (PCE) sorption/reduction under dynamic flowthrough conditions. PCE reduction rates with the surfactantmodified pellets (SMZ/ZVI) were three times higher than the reduction rates with the unmodified pellets (zeolite/ZVI). We speculate that enhanced sorption of PCE directly onto iron surface by iron-bound HDTMA and/or an increased local PCE concentration in the vicinity of iron surface due to sorption of PCE by SMZ contributed to the enhanced PCE reduction by the SMZ/ZVI pellets. Trichloroethylene and cis-dichloroethylene production during PCE reduction increased with the surfactant-modified pellets, indicating that the surfactant modification may have favored hydrogenolysis over β-elimination. PCE reduction rate constants increased as the travel velocity increased from 0.5 to 1.9 m d-1, suggesting that the reduction of PCE in the column systems was mass transfer limited.

Introduction Zero-valent iron (ZVI) is capable of reductively degrading a wide range of organic contaminants including chlorinated aliphatics (1, 2), chlorinated aromatics (3, 4), nitro aromatics (5, 6), and azo dyes (7, 8). The reduction rates for nitro aromatics, azo dyes, and some chlorinated aliphatics (e.g., carbon tetrachloride, hexachloroethane, 1,1,2,2- and 1,1,1,2tetrachloroethane) are relatively fast, with half-lives (t1/2) based on pseudo-first-order kinetics of less than 1 h (5-9). In contrast, the reduction rates for chlorinated aromatics and most of the chlorinated aliphatics are slower, with halflives on the order of days or more (1, 2, 4, 9). To enhance the reduction rates for chlorinated aliphatics/ aromatics and other organic contaminants, researchers have * Corresponding author present address: Department of Environmental Studies, University of West Florida, 11000 University Parkway, Pensacola, Fl 32514; phone: (850)474-3469; fax: (850)8576036; e-mail: [email protected]. † New Mexico Institute of Mining and Technology. ‡ China University of Geosciences. § University of WisconsinsParkside. 10.1021/es015816u CCC: $22.00 Published on Web 07/09/2002

 2002 American Chemical Society

modified ZVI in various ways. Coating iron with Pd or other metals such as Pt, Ni, and Cu (bimetallic couples) dramatically increased reduction rates for chlorinated ethylenes and chlorinated methanes (10-12), N-nitrosodimethylamine (13), and more recalcitrant compounds such as polychlorinated biphenyls (PCBs) and chlorinated phenols (14). Reducing the particle size of ZVI or Pd-coated ZVI from millimeters to nanometers resulted in 1-2 orders of magnitude increase in the surface-area-normalized reduction rate constants for TCE and PCBs (15). Coupling ultrasound with ZVI enhanced the reduction rate of carbon tetrachloride by a factor of 40 (16). Modification of ZVI with surfactants had mixed effects on the degradation rates for chlorinated hydrocarbons, depending on the type of surfactant used. In the absence of surfactant micelles in solution, modification of ZVI with cationic surfactants increased PCE reduction rates by factors of 3-19 (17, 18), while modification with a nonionic surfactant resulted in only a minor enhancement in PCE reduction (19). The enhanced reduction of PCE with surfactant-modified ZVI was attributed to the increased surface PCE concentration due to sorption by the bound surfactant (18). Modification of ZVI with anionic surfactants, however, did not enhance the reduction of PCE (17, 19). When surfactant micelles were present in solution, Sayles et al. (4) found that the reduction of DDT (1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane) by ZVI increased by almost a factor of 2 because of increased solubility and enhanced mass transfer of the hydrophobic compound in the presence of Triton X-114 (a nonionic surfactant). In contrast, the presence of hydroxypropyl-βcyclodextrin in solution lowered the PCE degradation rate because of the partitioning of PCE molecules into the hydrophobic interiors of the cyclodextrin, limiting PCE availability (20). Combining a sorbent such as surfactant-modified zeolite (SMZ) with ZVI in a pelletized form showed promising results in contaminant removal. SMZ is capable of sorbing nonpolar organic contaminants such as chlorinated ethylenes (21, 22), ionizable organics such as phenol and aniline (23), and inorganic anions such as nitrate and chromate (24, 25). Batch studies using granular SMZ/ZVI pellets (2-5 mm diameter) showed that the reduction rates for PCE and chromate increased by factors of 3-9 as compared to raw ZVI or zeolite/ ZVI pellets (26). It was speculated that the sorption of PCE or chromate by SMZ increased the local concentration in the vicinity of the ZVI surface, thereby enhancing the reduction (26). The performance of the SMZ/ZVI pellets for degrading contaminants in dynamic flow-through systems, however, still remains uncertain because of the very different solid to solution ratios in the two systems, possible mass transfer limitations in flow-through systems, and other factors (27). For applications such as in situ permeable reactive walls, it is necessary to conduct experiments under flow-through conditions to evaluate the synergistic effects of SMZ and ZVI. Most of the published laboratory studies with ZVI were either batch or resident-concentration column experiments (1, 28) that did not account for the dynamics (advection and dispersion) of a flow-through system (27, 29). Casey et al. (29) were the first to report effluent TCE breakthrough curves (BTCs) and simulate them with transport models. Casey et al. (29) utilized pulse solute injections, whereas in permeable reactive walls contaminants are constantly flowing through the walls (i.e., constant injection). To our knowledge, relatively few studies have reported BTCs from column experiments with reactive materials using a constant injection of chlorinated hydrocarbons. VOL. 36, NO. 16, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3597

Subsurface permeable barriers require high hydraulic conductivities to effectively capture contaminant plumes. ZVI and ZVI/sand barriers have reported hydraulic conductivities in the range of 10-1-10-2 cm s-1 (30-32). A concern with the long-term performance of ZVI-based barriers is plugging and concomitant hydraulic conductivity reductions after long-term use, as seen in some pilot and in-situ ZVI barriers (33-37). A highly reactive, high-permeability barrier material resistant to plugging offers clear advantages. The purpose of this study was to develop a reactive SMZ/ ZVI pellet with high mechanical strength, high surface area, and high hydraulic conductivity and to test the reactive material for PCE sorption/reduction in flow-through systems (columns packed with the pellets) with a constant PCE injection. We obtained and simulated PCE BTCs, determined PCE reduction rate constants, investigated the effects of possible mass transfer limitations, and determined the effects of surfactant modification on PCE reduction pathways.

Materials and Methods Pellet Preparation. A natural zeolite with a particle size of less than 0.4 mm (-40 mesh) and BET surface area of 15.1 m2 g-1 (determined using a Flowsorb 2300 BET surface area analyzer, Micromeritics, Norcross, GA) was obtained from the St. Cloud Mine, Winston, NM. The zeolite consists of roughly 74% clinoptilolite, 10% feldspar, 10% quartz and cristobalite, 5% smectite, and 1% illite (38). ZVI was obtained from Peerless Metal Powders and Abrasives (Detroit, MI), had a particle size of less than 0.2 mm (-70 mesh), and had a BET surface area of 1.76 m2 g-1. HDTMA-Cl solution (25% HDTMA-Cl by weight in water) was obtained from Aldrich Chemical Co. (Milwaukee, WI). The pellet-production process involved coating zeolite/ ZVI onto a glass foam substrate (Cell-Pore) from Cercona of America, Inc. (Dayton, OH). The patented Cell-Pore technology involves the gelation of soluble silicates and soluble aluminates (39). This high-porosity, high-strength glass substrate had an elemental composition of approximately 15% Na2O, 10% CaO, 5-8% Al2O3, 65% SiO2, and a small amount of MgO, TiO2, Fe2O3, and MnO2. Slabs 22.9 cm square and 2.5 cm thick were slotted (2.5-cm slots, 1.25 cm deep, running perpendicular to one another on opposite sides of the slab) for easy breaking into 2.5-cm cubes following coating with zeolite and ZVI. To coat the glass foam substrate, a homogeneous slurry (by weight) of zeolite powder (14.5%), iron powder (43.5%), and water (42.0%) was prepared in a large mixing bath. Surfactant can be included in the coating slurry along with zeolite and iron powders (40); for the studies performed here, surfactant was added after coating in order to test pellet reactivity with and without surfactant. The foam slabs were dipped into the slurry for about 5 s, removed from the bath, and dried at room temperature for 4 days. Preliminary tests indicated that the dipping process allowed the slurry to flow into the glass foam substrate without significant clogging of the pore structure. When dry, the zeolite/iron mixture adhered tightly to the silica foam substrate. The dried slabs were broken into 2.5-cm cubes. The final cubes consisted of approximately 16.5% zeolite, 49.5% iron, and 34.0% silica foam substrate by weight (determined by the mass fraction (34%) of the silica foam substrate and the mass ratio (1:3) of the zeolite to the ZVI in the slurry). For the laboratory column experiments, zeolite/ZVI cubes were crushed to smaller size pellets (0.5-1.0 cm in one dimension) using a hammer and a screwdriver. During the crushing process, about 50% of the zeolite/ZVI coating was lost, making the final composition of the crushed pellets approximately 12.5% zeolite, 37.5% iron, and 50% silica foam substrate by weight. Column Setup. CHROMAFLEX glass chromatograph columns (Kontes, Vineland, NJ), 30 cm long by 4.8 cm 3598

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 16, 2002

diameter, were used in the transport experiments. Column end caps and most of the fittings and tubing were made of Teflon. Glass and Teflon were chosen because of their limited sorption for chlorinated hydrocarbons. Each end cap was fitted with a three-way polycarbonate valve for influent and effluent sampling. The fluid delivery system consisted of a PHD 2000 programmable multichannel syringe pump (Harvard Apparatus, Holliston, MA) fitted with 10-mL glass gastight syringes (Hamilton, Reno, NV) and fed from a 60-L collapsible Teflon bag (Lab Safety Supply, Janesville, WI). All experiments were performed at 22.0 ( 1.0 °C. Four columns were packed with crushed pellets, purged with CO2 for 1.5 h to promote water saturation, and then fed with degassed synthetic groundwater (0.52 mM NaHCO3, 0.08 mM MgSO4, 0.09 mM CaCl2, 0.01 mM KNO3, 0.12 mM CaCO3, and 0.02 mM Mg(NO3)2) at a linear velocity of 0.5 m d-1. This synthetic groundwater was used as the aqueous matrix for all subsequent experiments. All injections were conducted in an upward flow mode. The weight of each column was determined approximately every pore volume (PV). When the weight difference between two consecutive measurements was less than 0.2%, a column was considered fully saturated, and the porosity of the packed pellets was determined by the difference between the saturated weight and the dry weight of the column. The bulk density of the packed pellets was calculated based on the dry mass of the packed pellets and the volume of the column. Surfactant Modification. After the columns were fully water-saturated, 10.0 mM HDTMA-Cl solution was fed into two columns at a linear velocity of 0.5 m d-1 to modify the pellets, and the effluent HDTMA concentrations were monitored. The sorption of HDTMA by the zeolite/ZVI pellets reached equilibrium after 10 PV of injection, as indicated by the constant effluent HDTMA concentration. Surfactant-free synthetic groundwater water was then injected at the same flow rate to rinse nonsorbed surfactant from the system. After 10 PV of rinsing, HDTMA concentration reached a constant value below the critical micelle concentration. The final HDTMA loadings for these two surfactant-modified columns, SM1 and SM2, were 13.3 and 15.3 mmol kg-1, respectively. During the surfactant loading of columns SM1 and SM2, the other two columns (unmodified, UM1 and UM2) received surfactant-free water at the same flow rate. Transport Experiments. PCE (12.0 µM) sorption/reduction in the SM and UM columns was determined at three different flow rates (0.49 ( 0.02, 0.97 ( 0.05, or 1.89 ( 0.09 m d-1). For each experiment, the effluent concentrations of PCE and its daughter products (TCE, 1,1-DCE, trans-DCE, cis-DCE, and vinyl chloride (VC)) were monitored. A step injection of about 10 PV was used for each PCE transport experiment. Between each separate experiment, the columns were flushed with the contaminant-free water until the effluent concentrations of the chlorinated ethylenes reached nondetectable levels. Six-milliliter effluent samples were collected every 0.150.20 PV by attaching 10-mL Hamilton gastight syringes to the effluent sampling points and switching the three-way valves. The samples were immediately transferred to 10-mL glass headspace vials and capped with Teflon-lined butyl septa. Separate effluent samples for pH determination were collected in the same manner every 0.5 PV. Influent samples were collected every pore volume to determine input concentrations. Flow rates were determined from the cumulative volume of effluent over time. Separate transport experiments with the conservative tracer tritiated water were conducted at the low and high flow rates to determine the longitudinal dispersivity (RL) values of the packed pellets. A 1-PV pulse of tritiated water (11 800 counts min-1 mL-1) was injected, and effluent samples

were collected using Retriever II fraction collectors (ISCO Inc., Lincoln, NE). Methods of Analysis. PCE, TCE, 1,1-DCE, trans-DCE, cisDCE, and VC were analyzed using a Hewlett-Packard 5890 gas chromatograph equipped with a 0.53 mm by 30 m GSQPLOT capillary column (J&W Scientific, Folsom, CA) and a flame ionization detector (FID). Helium was used as the carrier gas with a flow rate of 15 mL min-1. A Hewlett-Packard 7694 headspace autosampler was used for sample injection. The samples in the headspace vials were equilibrated at 60 °C for 30 min with vigorous shaking. The gaseous sample was injected splitless at 210 °C and was detected by the FID at 240 °C. The oven temperature program profile was 50 °C for 2 min, ramp 30 °C min-1 to 180 °C, and hold at 180 °C for 8 min. Calibration curves were produced using standards obtained from Supelco (Bellefonte, PA). The detection limits of this method for PCE, TCE, three DCE isomers, and VC were 0.016, 0.01, 0.005, and 0.006 µM, respectively. The linear response ranges for PCE, TCE, three DCE isomers, and VC were up to 15, 15, 10, and 8 µM, respectively. HDTMA was analyzed via high-performance liquid chromatography using the method of Li and Bowman (41). Tritium was analyzed by liquid scintillation counting (Beckman LS 6500, Palo Alto, CA) after mixing 1 mL of the effluent sample with 5 mL of Ecolite+ scintillation cocktail (ICN Biomedical, Costa Mesa, CA). Solution pH was measured using a Beckman Φ 45 pH meter with a Beckman combination pH electrode. The pH electrode was calibrated daily. Hydraulic conductivity of the pellets was measured using a constant head method (42). Sixteen well-shaped cubes were used to estimate the internal porosity (intra-pellet porosity). The internal porosity of the pellets was determined as the difference between the total cube volume (calculated from the dimensions) and the volume of solids (determined from the volume of water displaced by the cubes after saturation). Transport Model. The numerical model HYDRUS-1D version 2.0 (43) was used to simulate PCE breakthrough curves. The following partial differential equation was used to describe the one-dimensional transport of a solute undergoing reversible sorption and first-order decay in porous media (43):

Fb ∂ 2C ∂C Fb ∂S ∂C + ) D 2 - v - µ LC - µ S S ∂t θ ∂t ∂x θ ∂x

(1)

where C is the solute concentration in aqueous phase, S is the solute concentration on solid phase, t is time, Fb is the sediment bulk density, θ is the porosity, D is the hydrodynamic dispersion coefficient, v is the travel velocity, x is distance, µL is the aqueous-phase first-order decay constant, and µS is the solid-phase first-order decay constant. Previous studies indicated that the sorption of PCE and other nonpolar organics by SMZ followed a linear sorption isotherm (21, 22):

S ) KdC

(2)

where Kd is the distribution coefficient. Coupling eq 1 with the linear sorption isotherm yields

(

1+

)

Fb ∂C ∂ 2C ∂C Kd )D 2-v - µTC θ ∂t ∂x ∂x

(3)

where µT is the overall first-order reaction rate constant given by

(

µT ) µL +

F b Kd µ θ S

)

(4)

FIGURE 1. Simulations showing effects of longitudinal dispersivity (rL), overall first-order degradation rate constant (µT), and distribution coefficient (Kd) on breakthrough concentrations of PCE. The units of rL, µT, and Kd are cm, h-1, and L kg-1, respectively. Other transport parameter values are L ) 0.3 m, v ) 0.5 m d-1, θ ) 0.81, and Gb ) 0.67 g cm-3. Parameter values were estimated using eq 3 with leastsquares parameter optimization. For the tritium tracer, Kd, µL, and µS were set to 0, and the longitudinal dispersivity (RL ) D/v while neglecting molecular diffusion) (43) values for each column at both low and high flow rates were determined. For PCE, the dispersivity values from the tritium tests were used, and the values of Kd and µT were estimated. The overall first-order degradation constant µT was evaluated since µL and µS cannot be uniquely determined from inverse modeling of the data. A sensitivity analysis showed that, for a given RL, the steady-state concentration was only affected by µT, e.g., a slight decrease in µT (constant Kd) caused a significant increase in the steady-state concentration (Figure 1, curve 1 versus 2), whereas an increase in Kd (constant µT) did not affect the steady-state concentration (Figure 1, curve 1 versus 3). The increase in Kd, however, did cause a delay in the initial breakthrough (Figure 1, curve 1 versus 3). This analysis indicated that the parameters µT and Kd were independent. In fact, when several optimization trials (with different initial µT and Kd values) were carried out, µT and Kd always converged to the same unique values. Notice that a change in RL (constant µT and Kd) would change the BTC significantly (Figure 1, curve 1 versus 4); therefore, it is essential to determine RL values independently.

Results and Discussion Pellet Properties. The BET surface area of the pellets was 28.2 m2 g-1, indicating that the thin layer of SMZ/ZVI coating maintained the high surface area of the SMZ and ZVI powders. The iron rust (40.3 m2 g-1) on the pellets likely contributed to this high surface area. The columns packed with the crushed pellets had a very high porosity (0.81-0.82) and a very low bulk density (0.66-0.67 g cm-3). The internal porosity (intra-pellet porosity) of the pellets was 0.4, approximately 50% of the total porosity. The hydraulic conductivity of the crushed pellets was 9.7 cm s-1; the original larger pellets (2.5-cm cubes) would have an even higher hydraulic conductivity. The ∼10 cm s-1 hydraulic conductivity of the pellets is 2-3 orders of magnitude greater than that of granular ZVI, granular ZVI mixed with coarse sand, or foamed aluminosilicate-bound ZVI pellets (30-32). The pellets also showed excellent mechanical strength, with no physical deterioration evident after passing more than 100 PV of water (including ∼30 PV that contained PCE) through the columns. No changes in the macroscopic characteristics of the pellets were observed after the experiments (e.g., the pellets remained black, and no bleeding of ferric oxyhydroxide was evident). PCE Reduction. The inlet PCE concentration was constant at 11.0 ( 0.5 µM over the course of the column experiments. VOL. 36, NO. 16, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3599

FIGURE 2. Effluent PCE concentrations during PCE reduction at various flow rates: (a) 0.5, (b) 1.0, and (c) 1.9 m d-1. The smooth curves are the best-fit simulations using HYDRUS 1-D. Effluent PCE concentrations were significantly lower in the surfactant-modified (SM) columns than in the unmodified (UM) columns. This concentration was lower than that in the Teflon bag reservoir (12.0 µM), probably due to sorption of PCE onto the tubing and fittings of the delivery system. For the unmodified columns, PCE breakthrough concentrations at steady-state were about 2.8, 3.8, and 5.7 µM at velocities of 0.5, 1.0, and 1.9 m d-1, respectively (Figure 2). Steady-state PCE breakthrough concentrations for the surfactant-modified columns were significantly lower than those for the unmodified columns (i.e., by factors of 9, 4, and 3 at the three flow rates, respectively; Figure 2). Since the inlet PCE concentration, effluent pH (9.2 ( 0.1), mass of packed pellets (362.0 ( 1.5 g), porosity, and bulk density for the four columns were almost identical, the enhanced PCE reduction by the surfactant-modified pellets was clearly due to the presence of surfactant. The enhanced PCE reduction by the SMZ/ZVI pellets observed in this column study was consistent with the results from an earlier batch study, where enhanced PCE reduction (a 3-fold increase in the pseudo-first-order reduction rate constant) was achieved using a different formulation of SMZ/ ZVI pellets (26). The observed enhancement of PCE reduction by the SMZ/ZVI pellets may be attributed to (i) increased sorption of PCE directly onto iron surfaces by iron-bound HDTMA, accelerating the surface reduction of PCE (26), and/ or (ii) an increased local PCE concentration in the vicinity of iron surface due to the sorption of PCE by SMZ, promoting mass transfer of PCE onto the ZVI surface. Enhanced PCE reduction was observed when iron was modified with cationic or nonionic surfactants. For example, 3600

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 16, 2002

Li (17) observed in a batch study that PCE reduction rate constant increased by a factor of 3 when the ZVI (U.S. Metal,