EVALUATION OF THREE ORGANOCLAYS FOR AN ADSORPTIVE BARRIER TO MANAGE DNAPL AND DISSOLVED-PHASE POLYCYCLIC AROMATIC HYDROCARBONS (PAHS) IN GROUND WATER

Transcription

1 EVALUATION OF THREE ORGANOCLAYS FOR AN ADSORPTIVE BARRIER TO MANAGE DNAPL AND DISSOLVED-PHASE POLYCYCLIC AROMATIC HYDROCARBONS (PAHS) IN GROUND WATER FINAL REPORT by Craig H. Benson, Seung-Hak Lee, and A. Hakan Ören Geo Engineering Report No Geo Engineering Program University of Wisconsin-Madison Madison, WI USA 30 September 2008

2 EXECUTIVE SUMMARY This study consisted of evaluation of an existing adsorptive barrier (AB), evaluation of three commercially available organoclays that might be used for a full-scale AB, and modeling for preliminary design of a full-scale AB intended to block the flow of DNAPL and to remove dissolved polycyclic aromatic hydrocarbons (PAHs) from ground water. Core samples from the existing AB were inspected for the presence of DNAPL and analyzed for total PAH concentrations. Batch tests were conducted on candidate organoclays using solutions prepared with a single PAH and multiple PAHs to determine adsorption isotherms and to evaluate competition for adsorption sites. Column tests were conducted to evaluate the primary mode of transport (DNAPL flow or aqueous-phase transport) and to evaluate adsorption of PAHs from the aqueous phase onto organoclay under flow-through conditions. Organoclays and sand-organoclay mixtures were evaluated in the column tests. Modeling of flow and transport was conducted using hydraulic and transport properties of the organoclays and aquifer materials to illustrate how the AB may perform in the field. The following are findings of the study: 1. The distribution of DNAPL in the existing AB was heterogeneous and inconsistent with the expectations for an AB containing 25% organoclay. This may be due to heterogeneity of the AB material induced during installation or the presence of DNAPL within the periphery of the AB during construction. 2. Organoclays and organoclay-sand mixtures having at least 25% organoclay that are solvated with DNAPL have conductivities less than 10-8 cm/s, whereas water-saturated organoclays have hydraulic conductivities on the order of cm/s. Thus, the primary mechanism for transport of polycyclic aromatic hydrocarbons (PAHs) in organoclay ABs is advection in the aqueous phase. DNAPL migration through ABs constructed with organoclays is expected to be negligible. 3. Water migration is negligible in organoclays solvated with DNAPL. However, organoclays solvated with DNAPL will release PAHs when contacted with water, even under quiescent conditions. PAHs released from DNAPL-solvated organoclays most likely will be adsorbed by the organoclay through which the water is flowing (see subsequent inferences in Finding 6). 4. Adsorption isotherms for PAHs and organoclays tend to be linear at low concentrations, but non-linear when considered over a broader range. The non-linearity increases as the adsorbed concentration increases, and can be described by the Freundlich isotherm model. Greater adsorption of a given PAH occurs when the aqueous phase contains multiple PAHs. 5. Adsorption of PAHs was greatest for PM-199, followed by EC-199, and ET-1. PM-199 exhibited greater adsorption compared to EC-199 despite having slightly lower organic carbon content. 6. PAHs were detected sporadically in effluent from the column test conducted with 100% ET-1 organoclay at 190 pore volumes of flow. However, consistent breakthrough of PAHs from this column has not occurred at 200 pore volumes of flow. No PAHs have been detected in the effluent from the other columns with organoclay or organoclay-sand mixtures at the time this report was prepared (168 pore volumes for the organoclay-sand i

3 mixtures, 200 pore volumes for the 100% organoclay). 7. DNAPL spreading calculations indicate that active monitoring and maintenance of DNAPL upgradient of the AB will be necessary to ensure that the AB does not become blocked and the DNAPL will not migrate around the ends of the AB over time. 8. For a 0.6-m-thick AB, breakthrough of naphthalene will occur in 5-8 yr depending on the target effluent concentration being considered ( mg/l). For a 0.9-m-thick AB, breakthrough is anticipated in 8-12 yr depending on the target effluent concentration. If solvation of the organoclay is eliminated by actively managing DNAPL upstream of the AB, breakthrough at 3.9 mg/l will occur in 11 yr (0.6-m-thick AB) or 17 yr (0.9-m-thick AB). Recommendations from the study are as follows: 1. Efforts should be made to ensure that the adsorptive material within the full-scale AB is as uniform as practical. Greater uniformity can be achieved using 100% organoclay and by employing open trench or biopolymer slurry trench construction methods. 2. DNAPL spreading calculations indicate that active monitoring and maintenance of DNAPL upgradient of the AB will be necessary to ensure that a significant portion of the AB does not become blocked and the DNAPL. Accordingly, an active monitoring and management scheme is recommended to minimize spreading of the DNAPL in the field. Managing spreading of the DNAPL will result in longer breakthrough times and greater lifespan of the AB. 3. PM-199 is recommended for the full-scale AB. PM-199 has very low conductivity to DNAPL and the greatest affinity for PAHs of the organoclays that were evaluated. 4. The thickness of the AB may be increased in critical areas to increase the breakthrough time. ii

4 ACKNOWLEDGEMENT Financial support for this study was provided by CH2M Hill, Inc. Cores from the existing adsorptive barrier, bulk samples of DNAPL, and a bulk sample of ground water were provided by Thomas Hutchinson of CH2M Hill. Kostas Dovantzis and Thomas Hutchinson of CH2M Hill provided constructive feedback as the study progressed. The findings and inferences in this report are solely those of the authors. Endorsement by CH2M Hill is not implied and should not be assumed. iii

5 LIST OF ACRONYMS AND SYMBOLS DI = deionized (water) DNAPL = dense non-aqueous phase liquid EC-199 = organoclay manufactured by Biomin, Inc. ET-1 = organoclay manufactured by Aqua Technologies, Inc. f oc = organic carbon fraction K oc = organic carbon partition coefficient K ow = octanol-water partition coefficient PAH = polycyclic aromatic hydrocarbon PM-199 = organoclay manufactured by CETCO PTFE = polytetrafluoroethylene PVF = pore volumes of flow USCS = Unified Soil Classification System XRD = x-ray diffraction XRF = X-ray fluorescence = van Genuchten s a parameter for the capillary pressure curve n = van Genuchten s n parameter for the capillary pressure curve r = residual water content in capillary pressure curve s = saturated water content in capillary pressure curve iv

6 TABLE OF CONTENTS EXECUTIVE SUMMARY LIST OF ACRONYMS AND SYMBOLS ACKNOWLEDGMENT i iv ii 1. INTRODUCTION 1 2. MATERIALS Organoclays Aquifer Sands Glass Beads DNAPL PAHs and PAH Solutions 4 3. EXPERIMENTAL METHODS DNAPL Column Tests DNAPL Dissolution Tests DNAPL Solubility Tests Dissolved PAH Batch Tests Dissolved PAH Column Tests Analytical Methods 8 4. EVALUATION OF EXISTING ADSORPTIVE BARRIER 8 5. RESULTS OF LABORATORY EXPERIMENTS DNAPL Transport Primary Migration Pathway DNAPL Conductivity Permanence of DNAPL Dissolved-Phase Transport Adsorption Behavior Flow-Through Transport Behavior PRACTICAL IMPLICATIONS DNAPL Behavior at Upgradient Face Transport of Dissolved PAH CONCLUSIONS AND RECOMMENDATIONS REFERENCES 26 TABLES 28 FIGURES 40 APPENDICES 75 v

7 1. INTRODUCTION Historical activities associated with a former tie-treating facility at the resulted in a dense non-aqueous phase liquid (DNAPL) being discharged to the subsurface. This DNAPL, which is comprised primarily of polycyclic aromatic hydrocarbons (PAHs), is migrating from the former tietreating facility to, and has impacted ground water. An aerial photograph of the site showing the former tie-treating facility and is presented in Fig. 1. A cross-section illustrating migration of the DNAPL toward the lake is shown in Fig. 2a. To reduce the rate at which the DNAPL migrates into, an adsorptive barrier (AB) consisting of 25% organoclay and 75% sand was installed along the shoreline in Fall 2005 as an interim measure (Fig. 1, Fig. 2b). Monitoring since construction has shown that the existing AB has been successful in reducing discharges of DNAPL to. Consequently, a full-scale AB is being evaluated for the site. The study described in this report was conducted to support this evaluation. The study consisted of (i) evaluation of the existing AB, (ii) evaluation of three commercially available organoclays that might be used for the full-scale AB, and (iii) preliminary modeling to assess the effectiveness of a full-scale application of the AB. Evaluation of the existing AB included visual examination of cores collected from within the AB and chemical analysis on sections of the cores to determine the total PAH concentrations associated with the AB solid. Evaluation of the commercially available organoclays included batch and column tests to evaluate transport characteristics of the organoclay as well as physical tests to characterize the geotechnical and hydrological properties of the organoclays. The preliminary modeling consisted of two-dimensional flow and transport simulations to evaluate ground water velocities within the AB and breakthrough curves for various locations on the effluent face along the length of the AB. 1

8 This report is divided into 8 sections. Section 2 describes the materials used in the study. Experimental methods that were employed are described in Section 3. Section 4 describes the existing AB. Transport properties of the three organoclays are described in Section 5 and the practical implications of preliminary modeling results are described in Section 6. Conclusions and recommendations are presented in Section 7. References are provided in Section MATERIALS 2.1 Organoclays Three organoclays were used in this study: ET-1 (Aqua Technologies of Wyoming, Inc., Casper, WY, USA), PM-199 (CETCO, Arlington Heights, IL, USA), and EC-199 (Biomin, Inc., Ferndale, MI, USA). Particle size distribution curves (ASTM D 422) for the organoclays are shown in Fig. 3a. Mineralogical composition of the organoclays (determined by X-ray diffraction) is summarized in Table 1. The organic carbon content (f oc ) and specific gravity of solids (ASTM D 854) are summarized in Table 2. Organic carbon content was measured by combustion at 925 C using a LECO CNS carbon analyzer following the procedure in WSPL (2005). All three of the organoclays consist of sand-size granules and classify as poorly graded sands (SP) in the Unified Soil Classification System (USCS). PM-199 contains more uniform and smaller granules than ET-1 and EC-199. Each of the organoclays consists primarily of montmorillonite, but also contains appreciable amounts of quartz, cristobalite (ET-1), and feldspar (PM-199) (Table 1). EC-199 has the highest montmorillonite content (83%), whereas PM-199 (62%) and ET-1 (64%) contain less (and comparable) amounts of montmorillonite. The organic carbon content of the organoclays ranges from 15.5% (ET-1) to 26.9% (EC-199). All three organoclays have lower specific gravity of solids ( ) than is typical of montmorillonite (2.65, Jo et al. 2004), which reflects the organic carbon amended to the mineral 2

9 solid (Table 2). Hydraulic conductivity to water for each organoclay was determined using ASTM D The hydraulic conductivities are high ( cm/s) compared to those characteristic of natural montmorillonites, which typically have hydraulic conductivities to water on the order of to 10-8 cm/s (Mesri and Olson 1971). The organoclays have high hydraulic conductivity because they do not hydrate in water. As a result, the hydraulic conductivity is characteristic of the size and distribution of the sand-size granules in the organoclay rather than clay-size particles typically associated with a natural clay. For example, EC-199 has the largest median granule size, the least amount of fine granules, and is the most permeable of the organoclays. Capillary pressure curves for each organoclay corresponding to an air (non-wetting fluid) and water (wetting fluid) system were measured using the hanging column procedure described in ASTM D These curves ultimately were not used in the analysis because the properties of the organoclay changed markedly when solvated with DNAPL, precluding a conventional multiphase flow and transport analysis (see Section 6). The capillary pressure curves are provided in Appendix A. 2.2 Aquifer Sands Aquifer sands used in the study were obtained from cores collected from the field site (see discussion in Section 4). Analysis was conducted on samples obtained from portions of cores T2-B2, T2-B3, and T2-B4 that were beneath the existing AB (Fig. 2b). Particle size distribution, specific gravity of solids, and hydraulic conductivity of the sands were determined using the same methods employed for the organoclays. All three sand samples had a specific gravity of The hydraulic conductivities ranged between (T2-B3) and cm/s (T2-B2). Particle size distribution curves for the sands are shown in Fig. 3b. These distributions are essentially identical, with sand fractions ranging from 60% to 70% and gravel fractions from 28% to 38%. 3

10 2.3 Glass Beads Beads composed of soda-lime glass that are virtually chemically inert were used as control materials. The glass beads had a specific gravity of 2.50, uniform gradation with particle sizes between 0.4 and 0.8 mm, and roundness between 0.65 and 0.90 (as defined by ASTM D 1155). 2.4 DNAPL DNAPL samples from the field site were shipped to the University of Wisconsin- Madison in sealed metal cans and then transferred into a 20-L carboy. The DNAPL density is 1.08 kg/l and the dynamic viscosity is 0.85 Pa-s at room temperature (21 C). PAH concentrations in the DNAPL, determined using USEPA Method 8270C, are presented in Table 3. Naphthalene is the dominant PAH present in highest concentration in the DNAPL. However, phenanthrene, fluoranthene, 2-methylnaphthalene, acenaphthene, fluorene, and pyrene are also abundant. 2.5 PAHs and PAH Solutions Naphthalene (99% purity), acenaphthene (99% purity), fluoranthene (99% purity), and phenanthrene (98% purity) obtained from Aldrich Chemical, Inc. (Milwaukee, WI, USA) were used to make aqueous PAH solutions for the batch and column tests. PAH solutions were prepared by dissolving naphthalene, acenaphthene, fluoranthene, and/or phenanthrene in DI water. 3. EXPERIMENTAL METHODS 3.1 DNAPL Column Tests Column tests were conducted in cylindrical stainless-steel columns that had a diameter 4

11 of 60 mm and a height of 52 or 153 mm. The tests employed the same design as those in Cope and Benson (2008). The longer columns were used for tests conducted with organoclaysand mixtures, whereas the shorter columns were used for tests conducted solely with organoclay. Stainless steel plates containing a PTFE O-ring were placed on either end of the column. PTFE tubing (4.8 mm inside diameter) was used for the influent and effluent lines, except at the pump head, where flexible Viton peristaltic tubing (0.89 mm ID) was used. A glass microfiber filter was used to distribute flow uniformly along both ends of the specimen. Organoclays or organoclay-sand mixtures were placed into the columns in three layers, with each layer tamped lightly with a rubber rod to simulate the modest densification that would occur in the field if the material was dumped or poured into a trench. Short-term preliminary column tests were initially conducted to identify the primary mechanism for PAH transport through the organoclays (i.e., DNAPL flow or dissolved phase transport). Duplicate tests were prepared using PM-199 and ET-1. Deionized (DI) water was initially pumped through the organoclays using a peristaltic pump (Minipuls 2, Gilson, France). After the flow rate was steady, the influent liquid was switched to DNAPL from the site. The tests were conducted in an upflow mode at a constant flow rate (8 ml/hr). Pressure gages were used to monitor the influent pressure, and the effluent was collected in a graduated cylinder at atmospheric pressure. Additional longer-term DNAPL column tests were conducted to characterize the conductivity of the organoclays to the DNAPL. These tests were conducted in similar columns as the preliminary tests. However, due to the very low conductivity of the organoclays to DNAPL, the longer-term tests employed a constant head apparatus rather than a peristaltic pump (Fig. 4). A hydraulic gradient of 5 was used for the longer columns, and 15 was used for the shorter columns. As with the preliminary tests, organoclays in the longer-term DNAPL column tests were saturated with DI water prior to introducing the DNAPL. 5

12 3.2 DNAPL Dissolution Tests Dissolution tests were conducted on DNAPL-solvated organoclays from the longer-term column tests after the flow-through tests were terminated. The purpose of these dissolution tests was to evaluate whether the PAHs solvating the organoclay would be released into ground water contacting the AB. Samples of organoclay (27 g) were placed in a beaker filled with 200 ml of DI water. The beaker was sealed and stored in a quiescent state similar to that existing in situ (no shaking or movement was permitted). Samples of the aqueous phase were collected periodically with a syringe, mixed with 0.5 ml of HPLC-grade acetonitrile (Fisher Scientific, Chicago, IL, USA) to enhance desorption from the HPLC column, and analyzed for PAH concentrations using USEPA Method DNAPL Solubility Tests DNAPL solubility tests were conducted to evaluate maximum concentrations of DNAPL that might be observed at the influent face of an AB. DNAPL-water mixtures (1:1 and 1:7) were prepared using DI water in 40-mL glass vials without head space and sealed with PTFE-lined screw caps. The vials were placed in a tumbler at 28 rpm. After 24 or 48 hr the vials were removed from the tumbler and set on the bench for 1 hr to separate the aqueous and DNAPL phases. A 0.5-mL aqueous sample was then collected and analyzed for PAHs using USEPA Method PAH concentrations from the solubility tests are summarized in Table Dissolved PAH Batch Tests Adsorption of PAHs to the three organoclays under equilibrium conditions was evaluated with batch tests conducted with a single PAH (naphthalene, acenaphthene, phenanthrene, or fluoranthene) and with a mixture of PAHs. The latter tests were conducted to evaluate potential competition for sorption sites. Organoclay ( g) was placed in a 40- ml amber glass vial that was subsequently filled with an aqueous PAH solution. The vials 6

13 were sealed with screw caps containing PTFE-lined septa, and placed in a tumbler for 48 hr at 28 rpm. After 48 hr, a sample of the supernatant was collected with a syringe, mixed with 0.5 ml of HPLC-grade acetonitrile, and analyzed for PAHs using USEPA Method Control tests conducted without sorbent indicated that losses were less than 0.1% of the initial concentration. Thus, no correction for losses was made. Kinetic sorption tests were conducted initially to determine the time required to reach equilibrium. These tests were conducted using similar methods as the equilibrium tests, except multiple replicate vials were prepared and dissolved-phase concentrations typical of field conditions were used for the initial condition. Vials were removed from the tumbler periodically and analyzed for naphthalene, acenaphthene, phenanthrene, and fluoranthene using USEPA Method Concentrations from these tests are shown in Fig. 5 as a function of time. Equilibrium is established in approximately 10 hr. Thus, the 48-hr reaction time used in the equilibrium tests was sufficient to achieve equilibrium. Naphthalene was adsorbed at greater solid-phase concentrations than the other PAHs used in the kinetic batch tests because the initial dissolved-phase concentration of naphthalene was higher than the concentrations for the other PAHs Dissolved PAH Column Tests Column tests were conducted with organoclays, organoclay-sand mixtures, and glass beads (control) to evaluate transport of dissolved PAHs under flow-through conditions (Fig. 6). These tests were conducted with a peristaltic pump in essentially the same manner as the preliminary DNAPL column tests. Influent solution (DI water followed by a PAH solution) was introduced into the column in an upflow mode at a constant rate (6 ml/hr). The influent solution was contained in a flexible 3-L PTFE bag and effluent solutions were collected in sealed flasks. Samples of the column influent and effluent were collected periodically, mixed with 0.5 ml of acetonitrile, and analyzed for PAH concentrations using USEPA Method

14 3.6 Analytical Methods HPLC analysis for PAH concentrations was conducted with a SPD-M10Avp detector, LC-10ATvp pump, and CTO-10Avp oven (Shimadzu, Kyoto, Japan). The HPLC was equipped with a Pinnacle TM II PAH 4 μm column (150 x 3.2 mm, Restek, Bellefonte, PA, USA) and a Pinnacle TM II PAH guard cartridge (10 x 2.1 mm, Restek, Bellefonte, PA, USA). A mixture of acetonitrile and HPLC water was used as the mobile phase at a flow rate of 1.2 ml/min. The oven temperature was set at 40 C and the run time was 25 min, during which naphthalene was detected at 6.57 min, acenaphthene at 9.10 min, phenanthrene at min, and fluoranthene at min. The method detection limit (MDL) was estimated to be 4.9 μg/l for naphthalene, 2.4 μg/l for acenaphthene, 0.3 μg/l for phenanthrene, and 1.6 μg/l for fluoranthene. 4. EVALUATION OF EXISTING ADSORPTIVE BARRIER Eleven core samples were collected from the AB and adjacent aquifer by. Locations where the cores were collected are shown in Fig. 7. Eight cores were collected from the AB and three cores were collected from the aquifer on the upgradient side of the AB. All cores were collected with a direct push technique to a depth of approximately 9 m. The cores were delivered to UW in clear acrylic tubes 45 or 55-mm in diameter and approximately 1 m long. Each core was extruded from the acrylic tube by hand and examined visually for the presence of DNAPL. Contrast between the dark color of the DNAPL and the light color of the sand or AB was used to identify zones where DNAPL was present. Presence of DNAPL in each core was classified as high, medium, or low based on the amount of DNAPL that was visually observed. The distribution of DNAPL in the existing AB based on the visual inspection is shown in Fig. 8. The greatest presence of DNAPL was anticipated near the upgradient face of the AB, and was expected to diminish rapidly with increasing distance from the entrance face as DNAPL 8

15 was adsorbed by the organoclay. Greater presence of DNAPL was observed in cores obtained closer to the entrance face (B2 series cores in Fig. 8). However, the DNAPL did not consistently diminish with distance from the entrance face. For example, complete penetration of DNAPL occurred along transect T2, and the vertical distribution of DNAPL in core B2 from transect T2 was inconsistent with the vertical distribution of DNAPL in the aquifer at the entrance face (core B1 in transect T2). Similarly, greater DNAPL presence was observed in core B4 (near the exit face) than in core B3 (mid point in the AB) in transect T3. DNAPL was also observed in the aquifer material beneath the AB, which is consistent with the presence of a lower zone of DNAPL upgradient of the AB (Fig. 2b). Nine samples of core from the AB corresponding to locations with high, medium, and low presence of DNAPL were analyzed for total PAH concentrations using USEPA Method One sample of core from the aquifer sand having high DNAPL presence (T1-B3, m bgs) was also analyzed. Total concentrations of PAHs in the core samples are summarized in Table 5. On average, good agreement exists between the PAH concentrations and visually observed presence of DNAPL (i.e., higher PAH concentration is present in samples with greater visual presence of DNAPL). For example, the average naphthalene concentration is 2970 mg/kg in the samples with high presence, 1960 mg/kg in the samples with moderate presence, and 530 mg/kg in the samples with low presence. However, on an individual core basis, the PAHs concentrations are not always consistent with the amount of DNAPL that was visually present. For example, the naphthalene concentration in T3-B2 (3.0~3.3 m bgs) with low DNAPL presence is 1400 mg/kg, whereas the naphthalene concentration in T3-B2 (2.6~2.9 m) is 530 mg/kg, even though this section of T3-B2 appeared to contain a greater amount of DNAPL (moderate DNAPL presence). The reason for these periodic inconsistencies cannot be determined with certainty. However, they probably reflect differences in the rate of transport of PAHs in the DNAPL and aqueous phases. That is, elevated naphthalene concentrations could be present even if DNAPL was not visibly evident, if advection in the aqueous phase was the 9

16 primary mechanism controlling naphthalene transport through the existing AB. A uniform AB containing 25% organoclay should have had very low conductivity to DNAPL (see subsequent discussion in Section 5), which would have forced most of the DNAPL to be located near the entrance face of the AB and would have resulted in a relatively sharp DNAPL front in the AB. The irregular pattern of DNAPL in the AB and the broad distribution of PAH concentrations suggests that the organoclay content in the AB was heterogeneous or that the DNAPL was present within the periphery of the AB during installation of the organoclay. Attempts were made to determine the distribution of organoclay content by conducting analyses on core samples using X-ray fluorescence (XRF). Ratios of Al-to-Si were computed from the XRF data in an attempt to identify regions with high organoclay content (high Al/Si ratio) and low organoclay content (low Al/Si ratio). These attempts, however, were unsuccessful as no consistency was present in the Al/Si ratios. Regardless, the evaluation of the existing AB suggests that effectiveness of the full-scale AB will depend significantly on the degree of homogeneity that can be achieved. 5. RESULTS OF LABORATORY EXPERIMENTS 5.1 DNAPL Transport Primary Migration Pathway. The preliminary column tests with DNAPL (Section 3) were conducted with 100% organoclay to evaluate whether the primary pathway for migration of PAHs would be in the DNAPL or aqueous phases. These tests were conducted with PM-199 and ET-1 and employed a peristaltic pump to provide a constant flow rate. Pressure rapidly increased at the influent end during these column tests. The influent pressure for the column with PM-199 increased to 324 kpa within 6 hr, which burst the tubing at the pump head and forced cessation of the test. For ET-1, DNAPL was observed at the effluent end of the column in approximately 21 hr, which corresponded to 0.9 pore volumes of flow (PVF). Flow from the effluent end diminished, and the pressure at the influent port 10

17 increased to 63 kpa, at which time the test with ET-1 was ceased to prevent bursting of the tubing at the pump head. After terminating the preliminary column tests, the organoclay was extruded and inspected. For PM-199, the DNAPL migrated less than 10 mm distance from the influent port. Most of the organoclay appeared unchanged by permeation, having retained the original granular texture and color (Figs. 9a and b). However, at the very bottom of the column (influent end), a thin layer (1 mm thick) was present where the organoclay had been transformed from its original coarse-grained texture to a dark plastic gel-like material (Fig. 9c) that had similar consistency as Na-montmorillonite hydrated with water. Above this gel-like layer, another layer approximately 9-mm thick was present that transitioned in color from brown to yellow. Constituents in the DNAPL apparently sorbed onto the organoclay surfaces in this layer (Fig. 9c), although adsorbed PAH concentrations in this layer were not measured. Distinct granules of organoclay were present in this 9-mm-thick zone. The specimen of ET-1 removed from the column had different appearance (Fig. 10). The entire column appeared saturated with DNAPL. The core was stained dark brown (Fig. 10b), including the influent and effluent faces (Fig. 10a). In contrast to PM-199, granules of ET-1 remained visible after penetration by DNAPL (Fig. 10c). The very low flow rate observed in these preliminary column tests, even with high pressure at the influent end, suggests that DNAPL transport in the organoclay will be minimal. Moreover, the hydraulic conductivity of water saturated organoclay is very high (Table 2). Thus, the primary pathway for PAH transport should be advection of PAHs dissolved in the aqueous phase (ground water). This is particularly true under natural conditions, where the hydraulic gradient driving flow would be orders of magnitude lower than hydraulic gradient applied in the laboratory, resulting in a very low rate of migration of DNAPL in the organoclay. 11

18 5.1.2 DNAPL Conductivity. Longer-term column tests (Section 3) were conducted with DNAPL as the permeant liquid to confirm that the very low DNAPL conductivity observed in the preliminary tests would persist when the organoclays were permeated over longer periods of time. Tests were conducted with 100% PM-199, ET-199, and ET-1 as well as with mixtures of sand and PM-199 having organoclay contents of 10, 25, and 50%. A test was also conducted with sand alone. As mentioned in Section 3, these tests were conducted using a constant head applied using a burette filled with DNAPL to more accurately represent a field condition. Test conditions for these experiments are summarized in Table 6. DNAPL hydraulic conductivities corresponding to near equilibrium conditions are summarized in Table 7. Graphs showing DNAPL conductivity vs. time for the tests conducted with 100% organoclay are shown in Fig. 11. Similar behavior was obtained for each organoclay. The DNAPL conductivity diminishes over time, with a rapid decrease occurring initially that is followed by a gradual decrease that trends towards an equilibrium condition. In all cases, the hydraulic conductivity was decreasing slightly when permeation with DNAPL ceased. Thus, a true equilibrium condition (i.e., conductivity steady over time) was not achieved in any of the column tests, and the DNAPL conductivities reported in Table 7 are a conservative upper bound on the long-term equilibrium DNAPL conductivity. The DNAPL conductivities are correlated to the organic carbon content of the organoclays (Fig. 12), which reflects the ability of the DNAPL to solvate the mineral surface. Very low DNAPL conductivities were obtained for EC-199 (3.7x10-10 cm/s, 26.9% organic carbon) and PM-199 (7.6x10-10 cm/s, 25.0% organic carbon). The DNAPL conductivity for ET-1 (3.4x10-9 cm/s, 15.5% organic carbon) was approximately one order of magnitude higher than those of PM-199 and EC-199 (Table 7). Graphs of DNAPL conductivity vs. time for the sand and organoclay-sand mixtures are shown in Fig. 13. The temporal trends for the sand and organoclay-sand mixtures are similar to those obtained from the tests conducted with 100% organoclay, with a rapid decrease in hydraulic conductivity followed by a more gradual decrease (organoclay-sand mixtures) or 12

19 constant hydraulic conductivity (sand). The initial rapid decrease was observed in all of the longer-term column tests, including the sand. Because the sand is essentially inert compared to the organoclays, the initial drop in conductivity most likely reflects hydraulic stabilization to the imposed influent boundary condition rather than solvation mechanisms. However, the subsequent gradual decrease in DNAPL conductivity observed in all tests conducted with organoclay most likely is related to solvation of the organoclay. DNAPL conductivity vs. organoclay content is shown in Fig. 14. Addition of organoclay to the sand causes a remarkable reduction in the DNAPL conductivity in a manner analogous to the effect that addition of Na-montmorillonite has on the hydraulic conductivity of sand (e.g., Abichou et al. 2002). The most rapid reduction in DNAPL conductivity is obtained as the organoclay content is increased from 0 to 25%, and DNAPL conductivities less than 10-8 cm/s are obtained for organoclay contents greater than or equal to 25%. That is, from a practical perspective, DNAPL flow is essentially arrested when the organoclay content is at least 25%. In fact, when the organoclay content was at least 50%, no DNAPL emanated from the effluent end of any of the column tests during the testing period (only pore water was expelled, and only within the first 10 d). The organoclays and organoclay-sand mixtures were inspected after the column tests were terminated. Photographs of specimens from column tests on 100% organoclay are shown in Fig. 15 (PM-199), Fig. 16 (EC-199), and Fig. 17 (ET-1). Penetration of DNAPL was evident in all of the columns, and DNAPL was present as spots or as a layer at the effluent end of the columns prepared with PM-199 (Fig. 15a) and EC-199 (Fig. 16a). No DNAPL was present at the effluent end of the column conducted with ET-1, but this column was terminated 29 d earlier than the other columns. Distinct granules of organoclay remained in all of the column tests conducted with 100% organoclay (Fig. 15d for PM-199, Fig. 16c for EC-199, Fig. 17b for ET-1). Particles in the columns of PM-199 and EC-199 generally were surrounded by DNAPL (black liquid in Figs

20 and 16) and had a plastic consistency that could be smeared (Figs. 15c and 16b). In contrast, ET-1 retained a granular and non-plastic friable structure after permeation (Fig. 17b). The difference in texture suggests that the DNAPL solvates the surface of the PM-199 and EC-199 organoclays to form a plastic material, but interacts with the ET-1 organoclay in another manner. This difference in behavior may be due to differences in the compounds used to organically modify the clays. Photographs of the granules of PM-199 (Fig. 15d) and EC-1 (Fig. 16c) also suggest that the organoclay granules are not fully solvated by the DNAPL. Most of the solvation appears to occur on the exterior surfaces of the granules, whereas the interior of the granules appears to be unsolvated. A very low conductivity rim may be forming around the exterior of the organoclay granules, which restricts the rate at which DNAPL can migrate into the interior of the granule along micropores and solvate interior surfaces. Consequently, organoclays having smaller granule size may be more readily solvated and be less permeable to DNAPL than those tested in this study. However, a detailed examination of the micropores and solvation processes within the granules was beyond the scope of this study. For the organoclay-sand mixtures with at least 25% organoclay, a distinct zone of DNAPL penetration approximately 50-mm thick (Fig. 18b) was observed that was very similar to the penetration observed in the columns with 100% organoclay. However, in contrast to the columns with 100% organoclay, complete solvation of the organoclay granules (including the interior surfaces) apparently occurred in the organoclay-sand mixtures. A dark gel-like structure was evident in the pore space that was devoid of remnant granules (Fig. 18d). The mixture also was plastic and could be smeared (Fig. 18c) Permanence of DNAPL. Two sets of tests were conducted to evaluate whether the DNAPL contained in the organoclay would be retained in the presence of water (e.g., when the DNAPL source was no longer present and only ground water was contacting the AB). One set of consisted of permeating the DNAPL solvated columns of 100% PM-199 and EC-199 with tap 14

21 water after the DNAPL permeation phase was complete. The other tests consisted of quiescent dissolution experiments conducted with DNAPL-solvated organoclays. The dissolution tests were conducted with 100% PM-199 and EC-199, and the 50% PM-199-sand mixture. Hydraulic conductivities obtained from the water permeation tests are shown in Fig. 11 and are summarized in Table 7. Water permeation was conducted for 29 d after 46 d of DNAPL permeation. Very low hydraulic conductivities to water were obtained for both clays throughout the duration of the test (9.6x10-10 cm/s for PM-199, 1.1x10-9 cm/s for EC-199). These hydraulic conductivities are slightly higher than the DNAPL conductivities because of the difference in viscosity and density of water relative to the DNAPL. Nevertheless, the hydraulic conductivities are very low, which indicates that water will not migrate through the DNAPLsolvated organoclay in appreciable amounts. Moreover, no DNAPL or water was discharged in the effluent from the columns during the 29-d permeation period. Thus, the DNAPL should be retained within the organoclay even after the source is removed and ground water is contacting the AB. Aqueous-phase concentrations of naphthalene, acenaphthene, and phenanthrene recorded during the dissolution tests are shown in Fig. 19 along with the maximum concentrations of each compound observed at the field site. The concentration increases rapidly, and then levels off near the maximum concentration observed in the field after approximately 100 hr. When compared with the 100% organoclay columns, higher concentrations were obtained from the tests conducted with the 50% organoclay-sand mixture, which may reflect less avid binding of the PAHs when less organoclay is present. Slightly higher concentrations were also obtained for the tests conducted with PM-199 compared to the tests with EC-199, which reflects the lower organic carbon content of PM-199 relative to EC-199. The findings from the dissolution tests indicate that DNAPL-solvated organoclays in an AB can serve as a long-term source for dissolution of PAHs in contacting ground water. 15

22 However, PAHs released from the organoclay most likely will be adsorbed by unsolvated organoclays in other portions of the AB where ground water is flowing. 5.2 Dissolved-Phase Transport Batch and column tests were conducted to assess adsorption of PAHs dissolved in the aqueous phase to organoclays, and to evaluate transport under flow-through conditions as would occur in the field. Naphthalene (logk ow =3.30), acenaphthene (logk ow =3.92), phenanthrene (logk ow =4.46), and fluoranthene (log K ow =5.16) were selected for testing because these compounds have relatively high concentrations in the field relative to the generic clean-up criteria (Fig. 20), their octanol-water partition coefficients (K ow ) span a relatively broad range for PAHs, and analysis for these compounds was available and reliable. Concentrations of these compounds and other PAHs present in ground water at the field site are summarized in Fig. 20 along with the relevant clean up criteria in, Part 201. Target concentrations for batch and column testing were selected so that they would represent concentrations near the upper bound observed in the field. The following target concentrations were selected: naphthalene 10 mg/l (batch) and 20 mg/l (column), acenaphthene 1 mg/l (batch) and 3.5 mg/l (column), fluoranthene 1 mg/l (batch), and phenanthrene 1 mg/l (batch and column). Slightly higher target concentrations were used in the column tests to offset losses ( 26% for naphthalene, 29% for acenaphthene, and 60% for phenanthrene) in the column system. Only a limited number of tests were conducted with fluoranthene because its extremely avid adsorption to organoclay rendered concentrations in the batch equilibrium tests near the detection limit. Consequently, data for fluoranthene are not presented in this section. 16

23 5.2.1 Adsorption Behavior. Adsorption isotherms for PM-199 for a single PAH and multiple PAHs are shown in Fig. 21. Tests with a single PAH were only conducted with PM-199. Each isotherm was fit with the Freundlich isotherm model, which can be described as n q e = K F C e where q e is the concentration adsorbed on the solid (mg/kg), C e is the equilibrium concentration in the aqueous phase (mg/l), K F is the Freundlich distribution coefficient (L/kg), and n is the sorption exponent. A summary of the Freundlich parameters is in Table 8. A linear isotherm (n = 1) was also fit to the data corresponding to lower concentrations where the isotherm was approximately linear. Partition coefficients for the linear isotherm (K d ) are in Table 9. As shown in Fig. 21, greater adsorption occurs in the presence of multiple compounds than with a single compound in all cases, making competition for sorption sites between compounds moot. In addition, for multiple compounds, the Freundlich n parameter is always greater than 1, indicating that propensity for adsorption increases as the amount of PAH adsorbed increases. These findings for multiple compounds are particularly important, as a mixture of PAHs exists in the field. Greater adsorption was also obtained with the more hydrophobic compounds (higher K ow ). For example, the highest K F was obtained with phenanthrene (logk ow = 4.46) and the lowest with naphthalene (logk ow = 3.30) (Table 8). The effect of hydrophobicity is illustrated in Fig. 22, which the shows a strong positive relationship between logk oc and logk ow. The K oc in Fig. 22 were computed as K d f oc using data from the tests conducted with PM-199 using a single PAH. A comparison of isotherms for all three organoclays is shown in Fig. 23 for tests conducted with multiple PAHs. PM-199 consistently has the highest adsorption, and ET-1 the lowest. For example, the Freundlich distribution coefficient for naphthalene (3888 L/kg) is 2 times higher than the Freundlich distribution coefficient for EC-199, and 10 times higher than for 17

24 ET-1. PM-199 and EC-199 were expected to have greater adsorption than ET-1 because of the lower organic carbon content of ET-1. However, greater adsorption with PM-199 relative to EC-199 was not anticipated given that EC-199 has slightly higher organic carbon content (Table 2). This difference is likely due to greater specific surface area for PM-199, which is comprised of smaller granules than EC-199 (Fig. 2). The higher surface area of PM-199 probably offsets the modest difference in organic carbon content that exists between PM-199 and EC Flow-Through Transport Behavior. Column tests are being conducted with each of the organoclays and with organoclay-sand mixtures having 25% and 50% PM-199. A non-reactive control test with glass beads is also being conducted. Effluent concentrations from the columns are shown in Fig. 24 (100% organoclay) and Fig. 25 (organoclay-sand mixtures). At the time this report was prepared, more than 200 PVF had passed through the columns with 100% organoclay and more than 168 PVF had passed through the columns with the organoclay-sand mixtures. PAH concentrations in the effluent have remained below detection limits in all but one of the columns. The exception is the column containing 100% ET-1 organoclay (Fig. 24), for which PAHs have been detected sporadically since 190 PVF (complete breakthrough has not occurred in the ET-1 column). These findings are consistent with the isotherm parameters for the batch tests conducted with multiple PAHs. Based on the partition coefficients obtained from the batch tests, complete naphthalene breakthrough (effluent influent concentration) is not anticipated until approximately 360 PVF for ET-1, 1880 PVF for EC-199, and 3890 PVF for PM-199. For the organoclay-sand mixtures, complete naphthalene breakthrough is not anticipated for 970 PVF (25% PM-199) or 1940 PVF (50% PM-199). 6. PRACTICAL IMPLICATIONS Modeling was conducted using the adsorption parameters in Section 5 to assess the 18

25 expected performance of a full-scale AB. The AB was located along the shoreline from SB to SB as shown in Fig. 26, and to be keyed into the underlying clay layer (Fig. 2). The AB was assumed to be filled with 100% PM-199 organoclay. PM-199 was selected because it exhibited the greatest affinity for PAHs, and had very low hydraulic conductivity to DNAPL. Thus, PM-199 will block the flow of DNAPL and be the most effective in removing dissolved PAHs from ground water passing through the AB. Similar results would be obtained for Abs containing well-blended mixtures of sand and organoclay, except that breakthrough would occur earlier in an AB containing a mixture ( breakthrough time for 100% organoclay x percentage of organoclay in sand-organoclay mixture). 6.1 DNAPL Behavior at Upgradient Face Column testing showed that DNAPL flows into organoclay at a very slow rate. Thus, the full-scale AB is expected to act as a barrier to DNAPL flow, which will cause spreading of DNAPL along the upgradient face. Simulations of this behavior were attempted with the multiphase flow and transport simulator MOFAT (RSI 1991). However, these attempts were not successful because solvation of the organoclay with DNAPL alters the hydraulic properties of the organoclay in a manner that is fundamentally different than the assumptions employed in conventional multiphase flow theory. Thus, spreading along the upgradient face was estimated using an analytical mass-balance method, where the volume of DNAPL spreading within the AB was equated to the volume of DNAPL flowing to the AB in the aquifer. The two primary DNAPL layers were considered in the analysis (elevations 575 and 565 ft in Fig. 27). For each layer, the DNAPL contacting the surface of the AB was assumed to be an ellipse, the DNAPL mobility was assumed to be 1.4 m/yr (upper bound of the range in the field; ), and the DNAPL was assumed to spread isotropically within the AB. DNAPL penetration into the AB was assumed to occur for one year, resulting in 0.3 m of penetration at the rate observed in the 19

26 laboratory ( 25 mm/month). The calculated spreading at the upgradient face is illustrated in Fig. 28. The spreading calculation shows that the DNAPL-solvated zone at the upgradient face (heavy black line) widens over time, and that a significant fraction of the AB (open rectangle) will be blocked eventually if DNAPL is allowed to accumulate. Moreover, DNAPL may flow around the northern end of the AB after 5 yr if spreading is unabated. However, these computations are conservative, because all of the DNAPL flowing towards the AB is assumed to enter the AB and then spread laterally. In the field, the DNAPL zone upgradient of the AB may widen as the AB blocks flow, which will reduce spreading along the upgradient face and reduce the area of the AB that becomes essentially impervious to water flow. Nevertheless, the potential exists for spreading to occur relatively rapidly, which would block flow through the AB and significantly alter the hydrology and transport pathways. 6.2 Transport of Dissolved PAHs Flow and transport of dissolved PAHs through the AB was simulated with the variably saturated flow and transport program HYDRUS (v1.02; im nek et al. 2007). HYDRUS solves the variably saturated water flow equation and advection-dispersion-reaction equation for dissolved contaminants using the finite-element method. HYDRUS was selected because it accommodates both non-linear and linear isotherm models, and permits flow and transport problems to be solved in multiple dimensions using a single software package having a graphical user interface. For this analysis, saturated conditions were assumed for all simulations, which is consistent with the field scenario. Simulations were conducted for the two-dimensional domain shown in Fig. 29. The AB is surrounded by 10 m of aquifer material in all directions, and a portion of the upgradient face is blocked by DNAPL-solvated organoclay (the breadth of this blockage depends on the amount of 20

27 DNAPL spreading that is permitted, as described subsequently). Constant head conditions were applied at the left and right side boundaries to induce the average hydraulic gradient (i) observed in the field (i=0.01, ). No flow boundaries were applied on the north and south surfaces of the domain, which were located sufficiently far from the AB to prevent alterations in flow patterns within the AB. Hydraulic parameters used as input are summarized in Table 10. Parameters for naphthalene were used for the transport modeling. Of the PAHs considered, adsorption of naphthalene was the least favorable. Thus, naphthalene is anticipated to breakthrough the earliest. The dry density of the AB was assumed to be 0.84 Mg/m 3 and the porosity was assumed to be 0.47, which correspond to conditions in the column test conducted with PM-199. Multicomponent isotherms obtained from the batch tests with multiple PAHs were used to simulate the field condition. Diffusion was ignored due to the high ground water velocities. The longitudinal dispersivity was assumed to be one-tenth of the barrier thickness, and the transverse dispersivity was assumed to be one-tenth of the longitudinal dispersivity (Fetter 1999). Concentration at the upgradient face was assumed to be uniform and invariant with time. Three different concentrations were used at the upgradient face: 12 mg/l, which is slightly higher than the maximum concentration obtained from the dissolution test on DNAPL-solvated PM-199; this concentration is intended to represent conditions where water is contacting the DNAPL-solvated organoclay near the edges of the DNAPL-solvated zone, 7 mg/l, which is a typical naphthalene concentration existing in the field ( 2007), and 0.1 mg/l, which represents the low concentrations observed in the vicinity of the southern portion of the AB. Ground water velocities predicted in the middle of the AB are shown in Fig. 30 as a function of distance from the north end of the AB for three cases of flow blockage due to DNAPL solvation of the organoclay: (a) immediately after construction (i.e., DNAPL 21

28 directly upstream of the AB inducing immediate solvation of the organoclay), (b) after 1 yr of DNAPL spreading, and (c) after 2 yr of DNAPL spreading. Velocities are higher at the edges of the DNAPL-solvated zone (marked NAPL in Fig. 30) due to the focusing of flow that occurs as ground water migrates around the DNAPL-solvated organoclay, which is nearly impervious to water flow. These peak velocities increase over time as the zone of DNAPL solvation becomes wider, which enhances focusing. In fact, if spreading is allowed to persist unabated, the velocity will increase as much as 20% over five years (to 0.9 m/d at the northern edge of the zone solvated with DNAPL, not shown in Fig. 30). Three velocities within the profiles were selected as characteristic of flow through the AB (noted as squares in Fig. 30): 0.75 m/d, which represents a peak velocity where flow focuses at the northern edge of the DNAPL-solvated zone (0.78 m/d for 2 yr of DNAPL spreading), 0.60 m/d, which represents peak velocities in the northern and southern ends of the AB away from the focused flow, and 0.55 m/d, which represents typical velocities in the northern and southern ends of the AB, Effluent concentrations from a 0.6-m-thick AB are shown in Fig. 31 for adsorption following the Freundlich isotherm model. Similar predictions are shown in Fig. 32 for a 0.6-mthick AB with adsorption following a linear model, in Fig. 33 for a 0.9-m-thick AB with adsorption following a Freundlich model, and in Fig. 34 for a 0.9-m-thick AB with adsorption following a linear model. Initial and more rapid breakthrough occurs earlier along with a more rapid rise in effluent concentrations when the ground water velocity is higher (Fig. 31c) or the concentration at the upgradient face is higher (e.g., Fig. 31a). The most rapid breakthrough and the highest effluent concentrations are anticipated near the northern edge of the DNAPL-solvated zone (Fig. 22

29 31c, C o = 12 mg/l) due to the high velocity in this region (Fig. 30) and the elevated concentrations anticipated as ground water contacts the DNAPL-solvated organoclay. In contrast, much longer breakthrough times are anticipated in the southern regions of the AB, where velocities and concentrations are lower (Fig. 31a, C o = 0.1 mg/l). Intermediate breakthrough times and concentrations are anticipated in the northern portion of the AB in regions away from the DNAPL-solvated zone (Fig. 31b, C o = 7 or 12 mg/l). Comparison of Figs. 31 and 32 indicates that initial breakthrough occurs later and a more rapid rise in concentration occurs when a linear isotherm is used. The field condition is anticipated to fall between the conditions predicted with the linear and Freundlich isotherm models. Later initial breakthrough times associated with the linear model are anticipated, as linear adsorption behavior should occur at the low concentrations associated with initial breakthrough. The less rapid rise in concentrations associated with the Freundlich model should also be anticipated, because the Freundlich isotherm accounts for the non-linearity of the isotherm at higher concentrations. Predicted breakthrough times corresponding to target effluent concentrations of 1.3, 2.6, and 3.9 mg/l ( 10, 20, and 30% of the aqueous solubility for naphthalene) are summarized in Tables This range of target concentrations is lower than observed naphthalene concentrations downstream of the proposed AB and thus would not induce a downstream concentration gradient. Earliest breakthrough at 1.3 mg/l is anticipated to occur in approximately 5 yr at the northern end of the DNAPL-solvated organoclay (0.6-m-thick AB) and at approximately 8 yr for a 0.9-m-thick AB. These breakthrough times increase as the target effluent concentrations increase (Tables 11-12); at 3.9 mg/l, the earliest breakthrough time is approximately 8 yr (0.6-m-thick AB) or 12 yr (0.9-m-thick AB) (Table 13). Comparison of the initial breakthrough times shows that the initial breakthrough time can be increased approximately linearly by increasing the thickness of the AB. Thus, the AB could be made 23

30 thicker in the most critical regions (e.g., edges of the DNAPL-solvated organoclay) to increase the overall lifetime. The importance of managing DNAPL solvation at the influent face of the AB is illustrated by comparing the breakthrough times for 3.9 mg/l that are reported in Tables 14 and 15. Breakthrough times in Table 15 correspond to an AB where DNAPL is actively captured before reaching the face of the AB (e.g., in an upstream trench), which precludes solvation of the organoclay and formation of an impermeable zone on the upstream face of the barrier. As a result, focusing of flow is greatly eliminated as illustrated by the velocity profile shown in Fig. 15 (some focusing exists within the AB due to the higher hydraulic conductivity of the organoclay relative to the aquifer materials). The lower flow velocities result in considerably longer breakthrough times ( 11 yr for a 0.6-m-thick AB; 17 yr for a 0.9-m-thick AB). 7. CONCLUSIONS AND RECOMMENDATIONS This study was conducted to support the full-scale design of an adsorptive barrier (AB) to retain DNAPL and dissolved PAHs at the. The study consisted of an evaluation of an existing AB at the site, evaluation of three commercially available organoclays (PM-199, EC-199, and ET-1) that might be used for a full-scale AB, and modeling for preliminary design. Evaluation of the candidate organoclays included column tests to quantify the primary transport mode and to confirm transport conditions under flow-through conditions, and batch tests to quantify adsorption of dissolved PAHs. The following conclusions follow from the findings of these activities: 1. The distribution of DNAPL in the existing AB was heterogeneous and inconsistent with the expectations for an AB containing 25% organoclay, which suggests that the organoclaysand mixture used in the existing AB was heterogeneous or that DNAPL was within the periphery of the AB during installation. 2. DNAPL migration occurs at a very low rate in organoclay and organoclay-sand mixtures having at least 25% organoclay. Thus, the primary mechanism for PAH transport in an 24

31 organoclay AB will be advection of PAHs dissolved in the aqueous phase. DNAPL conductivities on the order of 10-9 cm/s (ET-1) and cm/s (PM-199, EC-199) were obtained for the organoclays. Organoclay-sand mixtures containing at least 25% organoclay had DNAPL conductivities less than 10-8 cm/s. In contrast, hydraulic conductivity of water-saturated organoclay is on the order of 0.1 cm/s. DNAPL migration in solvated organoclay should be negligible. 3. Water migration is negligible in organoclays solvated with DNAPL. Hydraulic conductivities to water on the order of 10-9 cm/s were obtained for the DNAPL-solvated organoclays. 4. Organoclays solvated with DNAPL will release PAHs when contacted with water, even under quiescent conditions. For the dissolution experiments conducted in this study, concentrations in the aqueous phase under equilibrium conditions were similar to the maximum PAH concentrations observed in the field. PAHs released from DNAPLsolvated organoclays likely will be adsorbed by the organoclay through which the water is flowing. 5. Adsorption isotherms for PAHs and organoclays tend to be linear at low concentrations, but non-linear when considered over a broader range. The non-linearity increases as the adsorbed concentration increases, and can be described by the Freundlich isotherm model. Greater adsorption also occurs when the aqueous phase contains multiple PAHs. These findings suggest that competition for sorption sites is not an issue when organoclays are used to adsorb PAHs from the aqueous phase. 6. Adsorption of PAHs was greatest for PM-199, followed by EC-199 and ET-1. PM-199 exhibited greater adsorption compared to EC-199 despite having slightly lower organic carbon content. Greater surface area associated with the smaller granules comprising PM-199 is believed to be responsible for the greater adsorption behavior of PM PAHs were detected in effluent from only one of the column tests (100% ET-1 organoclay) after 190 PVF. However, these detections have been sporadic and complete breakthrough has not occurred. No PAHs have been detected in effluent from any of the other organoclays or organoclay-sand mixtures after 200 PVF and 151 d (organoclays) or 168 PVF and 141 d (mixtures). These observations are consistent with the partitioning observed in the batch adsorption tests. 8. Ground water velocities will vary along the alignment of the AB due to the different hydraulic conductivities in regions of the AB that are solvated or not solvated with DNAPL. The highest velocities are anticipated at the northern edge of the AB where the flow rate increases due to focusing that occurs as water migrates around the DNAPL-solvated organoclay. 9. For a 0.6-m-thick AB, breakthrough of naphthalene will occur at the northern edge of the 25

32 DNAPL-solvated organoclay in 5-8 yr depending on the target effluent concentration being considered ( mg/l). For a 0.9-m-thick AB, breakthrough at the northern edge of the DNAPL-solvated organoclay is anticipated in 8-12 yr depending on the target effluent concentration. If solvation of the organoclay is eliminated by actively managing DNAPL upstream of the AB, breakthrough at 3.9 mg/l will occur in 11 yr (0.6-m-thick AB) or 17 yr (0.9-m-thick AB). Based on the findings of this study, the following recommendations are made: 1. Construction methods should be selected that will enhance the likelihood of achieving a uniform AB. Methods to increase uniformity include construction with an open trench or a biopolymer slurry trench and use of 100% organoclay. 2. DNAPL spreading calculations indicate that active monitoring and maintenance of DNAPL upgradient of the AB will be necessary to ensure that a significant portion of the AB does not become blocked by DNAPL solvation. Accordingly, an active monitoring and management scheme is recommended to retain spreading of the DNAPL in the field. Managing spreading of the DNAPL (or eliminating DNAPL solvation) will also result in longer breakthrough times. 3. PM-199 is recommended for the full-scale AB. PM-199 has very low conductivity to DNAPL and the greatest affinity for PAHs of the organoclays that were evaluated. 4. The thickness of the AB may be increased in critical areas to increase the breakthrough time. 8. REFERENCES Abichou, T., Benson, C., and Edil, T. (2002), Micro-Structure and Hydraulic Conductivity of Simulated Sand-Bentonite Mixtures, Clays and Clay Minerals, 50(5), Cope, D. and Benson, C. (2008), Grey-Iron Foundry Slags as Reactive Media for Removing Trichloroethylene from Groundwater, Environmental Science and Technology, in review. Fetter, C. (1999), Contaminant Hydrogeology, Upper Saddle River, NJ, Prentice Hall, 1999 Jo, H., Benson, C., and Edil, T. (2004), Hydraulic Conductivity and Cation Exchange in Non- Prehydrated and Prehydrated Bentonite Permeated with Weak Inorganic Salt Solutions, Clays and Clay Minerals, 52(6), Mesri, G. and Olson, R. (1971), Mechanisms Controlling the Permeability of Clays, Clays and Clay Minerals, 19,

33 RSI (1991), MOFAT: A Two-Dimensional Finite Element Program for Multiphase Flow and Multicomponent Transport, Resources & Systems International, Inc., Blacksburg VA, 1991 im nek, J.; ejna, M.; van Genuchten, M. (2007), The HYDRUS Software Package for Simulating the Two- and Three-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably-Saturated Media, Version USEPA (2008), Estimation Programs Interface (EPI) Suite for Microsoft Windows, v 3.20, United States Environmental Protection Agency, Washington, DC, USA. WSPL (2005), Carbon (Total, Organic, and Inorganic), Standard Method, Wisconsin Soil and Plant Analysis Laboratory, University of Wisconsin, Madison, WI. 27

34 TABLES 28

35 Table 1. Mineralogical composition of the organoclays as determined by XRD. Relative Abundance (%) Mineral Constituents PM-199 ET-1 EC-199 Quartz Cristobalite Tridymite Plagioclase Feldspar K-Feldspar 2 <1 2 Calcite 1 <1 - Dolomite Halite 5-5 Gypsum - - <1 Hornblende Illite 1 - <1 Smectite Organoclay Table 2. Geotechnical properties of organoclays. Specific Gravity of Solids Organic Carbon Content (%) Hydraulic Conductivity (cm/s) PM ET EC

36 Table 3. PAH concentrations in DNAPL [PAHs account for 50% (by wt.) of total liquid analyzed]. Compound Concentration (mg/kg) 2-Methylnaphthalene Acenaphthene Acenaphthylene 420 Anthracene 7200 Benzo(a)anthracene 5900 Benzo(a)pyrene 2800 Benzo(b)fluoranthene 4300 Benzo(g,h,i)perylene 1000 Benzo(k)fluoranthene 1500 Chrysene 4100 Dibenzo(a,h)anthracene <49 Fluoranthene Fluorene Indeno(1,2,3-cd)pyrene 910 Naphthalene Phenanthrene Pyrene

37 Table 4. PAH concentrations from DNAPL solubility tests along with maximum concentrations observed in ground water at the field site, effective solubilities computed using Raoult s Law, and solubilities reported by USEPA s EPI Suite (v.3.20). Compound Acenaphthene Naphthalene DNAPL- Water Ratio Mixing Time (hr) Conc. (mg/l) Approximate Maximum Concentration in Ground Water (mg/l) 1 Effective Solubility (mg/l) EPI Aqueous Solubility at 25 C (mg/l) Phenanthrene Note: 1 Maximum concentration obtained from data shown in Fig

38 Table 5. Total PAH concentrations (mg/kg) for core samples from existing AB and adjacent aquifer material. Compound DNAPL Presence High High High High Medium Medium Medium Low Low Low T2-B4 ( ) T1-B3 ( ) T3-B2 ( ) T1-B3 ( ) Location of Sample (m bgs) T2-B4 ( ) T3-B2 ( ) T1-B3 ( ) T3-B2 ( ) T2-B4 ( ) 2-Methylnaphthalene Acenaphthene < < 14 < Acenaphthylene <32 < 16 < 16 < 34 < 18 < 18 < 15 < 16 < < 19 Anthracene Benzo(a)anthracene Benzo(a)pyrene < < Benzo(b)fluoranthene < T1-B2 ( ) Benzo(g,h,i)perylene < < 3.4 < 6.5 < 3.8 < < 3.3 < < 4.1 Benzo(k)fluoranthene < < 2.2 < Chrysene Dibenzo(a,h)anthracene < 5.2 < 3.4 < 3.4 < 5.4 < 3.8 < 3.8 < 3.2 < 3.3 < < 4.1 Fluoranthene Fluorene < Indeno(1,2,3-cd)pyrene < < 3.3 < 2.5 < < 2.2 < Naphthalene Phenanthrene Pyrene Note: All samples from existing AB cores except T1-B3, which is aquifer sand beneath AB.

39 Table 6. Dry densities and pore volumes of organoclays and organoclay-sand mixtures used in DNAPL conductivity tests. All mixtures prepared with PM-199 organoclay. Percentage shown in materials column is for organoclay fraction of organoclay-sand mixture. Material Column Height Dry Density Pore Volume (mm) (Mg/m 3 ) (ml) PM ET EC % PM % PM % PM % PM Table 7. Summary of conductivities for organoclays and organoclay-sand mixtures. All mixtures prepared with PM-199 organoclay. Percentage shown in materials column is for organoclay fraction of organoclay-sand mixture. Material Permeant Test Hydraulic Liquid Duration (d) Conductivity (cm/s) PM-199 DNAPL/Water X10-10 (for DNAPL) 9.6X10-10 (for water) ET-1 DNAPL X10-9 EC-199 DNAPL/Water X10-10 (for DNAPL) 1.1X10-9 (for water) 0% PM-199 DNAPL 1 4.1X % PM-199 DNAPL 1 2.6X % PM-199 DNAPL X % PM-199 DNAPL X

40 Table 8. Parameters of Freundlich isotherm model fitted to batch test data. K F (L/kg) n R 2 Organoclay PAHs Single Multiple Single Multiple Single Multiple PM-199 ET-1 EC-199 Naphthalene Acenaphthene , Phenanthrene 57,480 3,259, Naphthalene Acenaphthene Phenanthrene - 13, Naphthalene Acenaphthene - 10, Phenanthrene - 263, Table 9. Partition coefficient for linear isotherm (K) and normalized distribution coefficient for the organic carbon fraction (K oc ) for PM-199 organoclay. Compound K d (L/kg) R 2 K oc (L/kg) Naphthalene ,780 Acenaphthene 11, Phenanthrene 72, Table 10. Hydraulic parameters used in HYDRUS for flow simulations. Material Hydraulic Properties Saturated Hydraulic θ r θ s α (m -1 ) n Conductivity Water Saturated AB Aquifer DNAPL-Solvated AB x

41 Table 11. Barrier thickness (m) Elapsed time when effluent concentrations exceed 1.3 mg/l at the downgradient face of AB with partial blockage by NAPL-solvated organoclay. Ground Water Velocity (m/d) Initial Concentration at Upgradient Face (mg/l) Initial breakthrough time (yr) - Freundlich model Initial breakthrough time (yr) - linear model

42 Table 12. Barrier thickness (m) Elapsed time when effluent concentrations exceed 2.6 mg/l at the downgradient face of AB with partial blockage by NAPL-solvated organoclay. Ground Water Velocity (m/d) Initial Concentration at Upgradient Face (mg/l) Initial breakthrough time (yr) - Freundlich model Initial breakthrough time (yr) - linear model

43 Table 13. Elapsed time when effluent concentrations exceed 3.9 mg/l at the downgradient face of AB with partial blockage by NAPL-solvated organoclay. Barrier thickness (m) Ground Water Velocity (m/d) Initial Concentration at Upgradient Face (mg/l) Initial breakthrough time (yr) - Freundlich model Initial breakthrough time (yr) - linear model

44 Table 14. Barrier thickness (m) Elapsed time when effluent concentrations exceed 2.6 mg/l at the downgradient face of AB without blockage by NAPL-solvated organoclay. Ground Water Velocity (m/d) Initial Concentration at Upgradient Face (mg/l) Initial breakthrough time (yr) - Freundlich model Initial breakthrough time (yr) - linear model

45 Table 15. Elapsed time when effluent concentrations exceed 3.9 mg/l at the downgradient face of AB without blockage by NAPL-solvated organoclay. Barrier thickness (m) Ground Water Velocity (m/d) Initial Concentration at Upgradient Face (mg/l) Initial breakthrough time (yr) - Freundlich model Initial breakthrough time (yr) - linear model

46 FIGURES 40

47 Existing AB Fig. 1. Aerial view of site showing former tie-treating facility and location of existing AB (adapted from ). 41

48 (a) (b) Fig. 2. Cross sections B-B (a) and D-D (b) showing DNAPL zone and existing AB. Refer to Fig. 1 for spatial referencing. 42

49 PM-199 ET-1 EC-199 Percent Finer (%) Particle Diameter (mm) 0.1 (a) T2-B2 T2-B3 T2-B4 Percent Finer (%) Particle Size (mm) 0.1 (b) 0.01 Fig. 3. Particle size distributions of organoclays (a) and aquifer materials (b). 43

50 Fig. 4. Apparatus for DNAPL conductivity tests conducted with PM-199, ET-1, and EC

51 Amount Sorbed (mg/kg) Naphthalene Acenaphtene Phenanthrene Fluoranthene Amount Sorbed (mg/kg) Amount Sorbed (mg/kg) (a) Elapsed Time (hours) 50 (b) Elapsed Time (hours) 50 (c) Elapsed Time (hours) Fig. 5. Sorption kinetics of PM-199 (a), ET-1 (b), and EC-199 (c) for naphthalene, acenaphthene, phenanthrene, and fluoranthene. 45

52 Fig. 6. Photograph of column tests for dissolved PAHs (naphthalene, acenaphthene, and phenanthrene) using glass beads (control), PM-199, ET-1, EC-199, 25% organoclaysand mixture, and 50% organoclay-sand mixture. 46

53 0.45 m North T3 B1 B2 B3 B4 T2 B1 81 m B2 B3 B4 7.5 m T1 B1 B2 B3 7.5 m G.W. flow South Fig. 7. Location of cores from existing AB. 47

54 0.0 m B1 B2 B3 B4 T3 4.5 m 9.0 m 0.0 m B1 B2 B3 B4 PAB depth (3.3 m) T2 4.5 m T1 9.0 m 0.0 m 4.5 m B1 B2 B3 High Medium Low 9.0 m Fig. 8. DNAPL distribution in core samples from existing AB. 48

55 (a) (b) (c) Fig. 9. Top view (a), side view (b), and cross-sectional view of influent end (c) of PM-199 after preliminary column test. Black zone at bottom in (c) is thin (1 mm) layer of gel-like organoclay solvated with DNAPL. 49

56 (a) (b) (c) Fig. 10. Top view (a), side view (b), and cross-sectional view (c) of ET-1 after preliminary column test. 50

57 NAPL Conductivity (cm/s) NAPL (a) NAPL Conductivity Q out /Q in Days Water Q out /Q in NAPL Conductivity (cm/s) (b) Days Q out /Q in NAPL Conductivity (cm/s) NAPL (c) Days Water Q out /Q in Fig. 11. DNAPL conductivity of PM-199 (a), ET-1 (b), and EC-199 (c), and subsequent hydraulic conductivity to water; open circles are ratio of incremental effluent volume to influent volume; conductivity was calculated based on the influent flow rate. 51

58 DNAPL Conductivity (cm/s) K DNAPL (cm/s) = 7.5x OC Organic Carbon Content (%) Fig. 12. DNAPL conductivity of organoclays as a function of organic carbon content 52

59 10-3 (a) (b) 2.0 NAPL Conductivity (cm/s) NAPL Conductivity Q Q out/ in Pore Volumes of Flow (PVF) Q out /Q in NAPL Conductivity (cm/s) Pore Volumes of Flow (PVF) Q out /Q in 10-3 (c) (d) 2.0 NAPL Conductivity (cm/s) Q out /Q in NAPL Conductivity (cm/s) Q out /Q in Days Days Fig. 13. DNAPL conductivity for organoclay-sand mixtures with organoclay contents of 0 (a), 10 (b), 25 (c), and 50% (d) by weight; open circles are ratio of incremental effluent volume to influent volume; conductivity calculated using influent flow rate. 53

60 DNAPL Conductivity (cm/s) Organoclay Content in the Mixture (%) Fig. 14. DNAPL conductivity as a function of organoclay content in organoclay-sand mixture. 54

61 (a) (b) (c) (d) Fig. 15. Top view (a), side view (b), cross sectional view (by wire saw) (c), and cross sectional view by cracking (d) of PM-199 after DNAPL conductivity test. 55

62 (a) (b) (c) Fig. 16. Top view (a), cross sectional view (by wire saw) (b), and cross sectional view by cracking (c) of EC-199 after DNAPL conductivity test. 56

63 (a) (b) Fig. 17. Top view (a) and cross sectional view (b) of ET-1 after DNAPL conductivity test. 57

64 (a) (b) (c) (d) Fig. 18. Top view (a), side view (b), cross sectional view (by wire saw) (c), and cross sectional view by cracking (d) of 25% mixture of organoclay and sand after DNAPL conductivity test. 58