MITIGATION OF CHLORINATED SOLVENT GROUNDWATER PLUMES WITH COLLOIDAL ACTIVATED CARBON

Chlorinated volatile organic solvents (CVOCs) can form a contamination plume in groundwater that needs to be managed to protect human health and the environment. There are many injectable products available to treat contamination plumes in groundwater using a number of specific technologies, individually and in combination. This study evaluates the effectiveness of injectable colloidal activated carbon (PlumeStop® Liquid Activated CarbonTM) at two sites in Rhode Island with chlorinated solvent groundwater contamination plumes. The injectate was introduced via low-pressure injection methods to avoid mobilizing additional contaminants. The initial results show a reduction in contaminant levels downgradient of the injection points for both Sites. The long-term effectiveness of the injectate remains to be seen. GZA will continue to monitor the groundwater quality downgradient of the injections to evaluate breakthrough of the Permeable Reactive Barrier (PRB) formed by the PlumeStop injections. iii ACKNOWLEDGMENTS This thesis is dedicated to Professor Arijit Bose with the Chemical Engineering Department at the University of Rhode Island. I would also like to acknowledge Edward Summerly P.G. and Principal, Karen Kinsella Ph.D. and Senior Technical Consultant, and Richard Carlone P.E. and Senior Project Manager with GZA GeoEnvironmental Inc.


LIST OF TABLES
bedrock below that depth. The bedrock in the area consists of an igneous granite gneiss with no primary porosity or permeability, therefore groundwater movement and associated contaminant migration is limited to bedrock fractures. Figure 4 presents the profiles of ML-11 and the downgradient monitoring well ML-12. Monitoring well 5 ML-12 is a multi-level well with well screened intervals 12A (elevation 317.6 feet to 328.6 feet), 12B (elevation 279.6 to 300.6 feet), MLC (elevation 235.6 to 245.6 feet), MLD (elevation 209.6 to 222.6 feet), and 12E (elevation 148.6 to 158.6 feet).

Figure 4: Profiles of Monitoring Wells ML-11 and ML-12
GZA conducted packer testing of ML-11 in 2003 to evaluate the hydraulic conductivity of bedrock in the vicinity of the borehole. The packer testing consisted of lowering two inflatable packers into the open borehole of the well to isolate 5-foot intervals of the exposed bedrock. The groundwater in the 5-foot interval is then evacuated, and the time to fill the void with groundwater is measured to estimate the hydraulic conductivity of the interval. Bedrock packer permeability testing was performed in approximately 5-foot intervals starting below the steel casing and 6 running the length of the borehole. These data were used to estimate the bedrock hydraulic conductivity. As a part of previous investigations, GZA also conducted a suite of geophysical testing on ML-11 in 2003, including digital acoustic televiewer (ATV) logging. Based on the ATV logging and bedrock packer permeability testing described above, GZA calculated the average hydraulic aperture (i.e., effective fracture opening size) for ML-11. The hydraulic aperture for each ML-11 packer test zone ranged between 15 to 110 microns, this aperture size is large enough to accommodate the 1-2-micron sized PlumeStop particles.
Site II: Former Plating Facility Background Groundwater monitoring results suggested that reductive dechlorination was still occurring onsite. However, the data also suggest that supplemental treatment along the downgradient edge of property would be beneficial for continuing the remediation process and minimizing offsite plume migration. After reviewing alternatives, PlumeStop was selected as the preferred treatment alternative. The PlumeStop injections serve as a PRB to mitigate offsite migration of contaminants.
For both sites the PlumeStop activated carbon injectate was paired with Hydrogen Release Compound (HRC). The HRC stimulates bacterial growth to jumpstart the bioremediation process. Therefore, the overall treatment from the injection of PlumeStop and HRC should reduce contaminant concentrations by adsorption and biodegradation.

REVIEW OF LITERATURE
There are many alternatives for implementing PRBs for in-situ treatment of groundwater contamination plumes. Traditional PRBs require trench excavation and fill utilizing the proposed treatment compound, such as granular ZVI, activated carbon, or other reactive materials (Tosco, et al. 2014). Due to limitations (e.g., high cost, spatial constraints, etc.) for installation of PRB via excavation, injectable treatment materials have been developed. Nanoscale ZVI particles are commonly used for remediation because they can be injected into the soil matrix, and due to their large surface area Nano-ZVI has been shown to be more effective at degrading contaminants than granular ZVI (Zhang, et al. 1998). Although Nanoscale ZVI particles are small enough to pass through the soil matrix, the mobility of these particles is limited due to agglomeration and adhesion to the soil matrix (O-Carroll, et al. 2013). Nano-ZVI particles can be modified with various coatings and metal catalysts or suspended in emulsions to improve their mobility and efficacy (Comba, et al. 2011). Similar to Nano-ZVI, granular activated carbon is also an effective material for PRBs (Ruiz, et al. 2014  The PlumeStop is intended to form a bio-remediation barrier in the soil matrix.
As contaminated groundwater migrates through the PRB formed by the PlumeStop particles, contaminants adsorb to the carbon surface and should degrade more effectively due to the longer residence time and direct contact with bacteria.
According to the manufacturer, as the contaminants degrade, adsorption sites again become available to adsorb additional contaminant (Birnstingl, et al. 2014) allowing for long term mitigation of contaminant plumes. In order for the long-term treatment to be effective, biodegradation of the contaminants of concern must be possible, and the rate of biodegradation must be greater than or equal to the rate of adsorption under saturation conditions. The sorption capacity of PlumeStop depends on the surface area of the activated carbon particles. Information provided by Regenesis does not provide the surface area to mass ratio for PlumeStop. However, published data for surface areas of activated carbon products can range from 34 m 2 /g (Jaroniec et al. 1989) to 2696 m 2 /g (Lih, E. Y. et al. 2016).
PlumeStop is an aqueous-phase amendment (i.e., active only in the saturated portion of an aquifer) that is typically injected via direct push injection methods but can also be emplaced using monitoring wells and open boreholes.
This study evaluates the effectiveness of PlumeStop injections at two sites. Both sites contain chlorinated solvents. The main contaminants of concern at Central Landfill are 1,2-dichlorobenzene and chlorobenzene. Both 1,2-dichlorobenzene and chlorobenzene have been found to biodegrade from bacterial growth/metabolism (Reineke et al. 1984) (Haigler et al. 1988) (Seignez et al. 2001) (Ziagova et al. 2007).

12
The main contaminants of concern at the Former Plating Facility are PCE, TCE, DCE, and VC. Under anaerobic conditions PCE undergoes reductive dichlorination/biodegradation to form TCE, which is dechlorinated to DCE, and ultimately VC (Ni, Z. et al. 2014) (Bradley 2000). Given that the contaminants of concern for both sites can undergo bioremediation, long-term treatment should be possible. During the injections the hydraulic head was limited to approximately 10 feet (4.3 pounds per square inch) above the static water level in the well. Dense Non-Aqueous Phase Liquid (DNAPL) is known to be present within the bedrock aquifer in the Hot Spot area, therefore, it was important to avoid injecting any of the additives under significant pressure because of the potential to mobilize contaminant source material to areas that are currently un-impacted. Based on GZA's evaluation of hydraulic conductivities and fracture spacing in ML-11, it was determined that injection within a specific well bore zone using zone isolation packers was not necessary and would increase the risk of applying a large hydraulic head to the borehole fractures.
The ML-11 extraction system (HSHCS) was removed from service in April 2016, and the static water level was allowed to equilibrate (2 months) prior to the injection.
14 In preparation for the in-situ treatment program, the injection well (ML-11) was developed to remove accumulated sediment to increase the injection rate. The well development process essentially cleans the borehole to remove sediment and improve the hydraulic conductivity of the borehole. The injectate was mixed in 50-gallon batches, and pumped from the base of the landfill, in the vicinity of the Hot Spot treatment system up to ML-11. Flow rates were adjusted during the injection event to maintain less than 10 feet of hydraulic head above the static water level. Each 50-gallon batch required approximately 1.5 to 2 hours to inject. During the first day of injections, GZA attempted to continuously circulate water within the borehole using a small submersible pump 1 . However, the submersible pump failed shortly after the first injection began. GZA repeated this attempt to circulate water in the borehole using submersible pumps, the second and third day of injections, however these pumps also failed. During the fourth and fifth days of injections, the water in the borehole was not circulated.  As shown in Table 2, after the first injection, the concentrations of 1,2dichlorobenzene in ML-11 decreased from 14,000 µg/L to less than 2,000 µg/L during the August and September sampling rounds. The concentration of 1,2-dichlorobenzene then rebounded to 13,000 µg/L during the October sampling round. This rebound suggests that the PlumeStop migrated downgradient of ML-11 as intended. The concentration of chlorobenzene decreased from 6,400 µg/L in April to 1,500 µg/L in August and 2,100 µg/L in September. The chlorobenzene concentration in October rebounded to 10,000 µg/L. The rebound observed in October is considerably higher than the concentration of chlorobenzene prior to the injection, which may be the result of mobilizing contaminants due to the injection.
Due to the different formulation of PlumeStop used for the second injection, the impacts of the second injection are expected to be more localized around ML-11.

24
Rebound is not observed in ML-11 after the second injection, as expected because the formulation of the second injection was designed to fall out of suspension faster and would therefore provide treatment closer to the injection well.

ML-12 Results
As depicted in Figures 11 through 20 below, the 12-month average concentration of chlorobenzene and 1,2-dichlorobenzene in downgradient well ML-12 also appear to be trending downward since the first injection. However, the downward trend observed in ML-12 is less pronounced than the trend for ML-11. Note, this well is approximately 230 feet downgradient of ML-11, and there is a delay of approximately 3-9 months (depending on groundwater table fluctuations) before any changes in ML-11 appear in ML-12.
The well screen for ML-12A is installed from elevation 317.6 to 328.6 (i.e., 136 feet to 125 feet bgs). During the operation of the HSHCS, ML-12A experienced less drawdown then the lower well screen sections, which indicates that ML-12A is not as hydraulically connected to ML-11 as the other well screens in ML-12. Therefore, the reduction in contaminant concentrations in ML-12A is not observed. Prior to the injection, the concentrations of 1,2-dichlorobenzene in ML-12A ranged from approximately 800 µg/L to 3,900 µg/L and the concentrations of chlorobenzene ranged from approximately 500 µg/L to 3,600 µg/L. Since the third injection the concentration of 1,2-dichlorobenzene ranged from 730 µg/L to 2,500 µg/L, and the concentration of chlorobenzene ranged from 500 µg/L to 1,800 µg/L. The shape of the 12-month moving averages depicted in Figures 11 and 12    This well screen is considered to be hydraulically connected to the borehole in ML-11.
The concentration of 1,2-dichlorobenzene prior to the injection ranged from approximately 40 µg/L to 160 µg/L. The concentration of chlorobenzene prior to the injection ranged from approximately 500 µg/L to 1,000 µg/L.

Central Landfill Statistical Analysis
A t-test analysis was performed to compare pre and post injection data for 1,2dichlorobenzene and chlorobenzene in ML-11 and ML-12. For ML-11 the pre and post injection data was separated based on the date of the first injection. For ML-12 pre and post injection data was separated based on a 6-month delay (assumed 6 months due to the 3 to 9-month range of the expected delay) from the date of the first injection.
Eight (8) samples have been collected from ML-11 since the first injection, so the t-test compared the eight (8) sample results pre and post the injection. Accounting for a 6-month delay there have been ten (10) post injection sampling events and nine (9) pre-injection sampling events. The p-values from the t-test are presented in Table 1. Based on t-test using 90% confidence (p = 0.1), there is a statistically significant difference between the pre and post injection results for 1,2-dichlorobenzene in ML-11, ML-12B, ML-12C, and ML-12D, but not ML-12A or ML-12E. There is a statistically significant difference between the pre and post injection results for chlorobenzene in ML-11, ML-12B, and ML-12D but not for ML-12A, ML-12C, or ML-12E.

GZ-3 Results
During the first sampling event after the injection, activated carbon particles were observed in GZ-3. The appearance of PlumeStop in GZ-3 after the injection may be an indication that the PlumeStop has migrated past GZ-3. The concentration of contaminants in GZ-3 decreased initially, but appears to show signs of rebound. This rebound may be occurring because the injectate has migrated downgradient of GZ-3, therefore the results are reflective of the source area without treatment. Alternatively, the increase may be because the PlumeStop in the vicinity of GZ-3 is saturated with contaminants. Note, that GZ-3 has historically shown contaminant concentrations reflective of the source area at the Site. As shown in Figure 21, the TCE is the predominant contaminant in GZ-3. The results for PCE, TCE, DCE, and VC in GZ-3 are still in exceedance of action levels.

34
The results for PCE in GZ-3 shown in Figure

35
As shown in Figure 23, DCE increases immediately after the injection. TCE is the predominant contaminant in GZ-3. TCE degrades into DCE as a result of reductive dechlorination. The increase in DCE therefore, may the result of degradation of TCE.

Figure 23:Downgradient Well GZ-3 DCE Results
The results for VC shown in Figure 24 appear to be decreasing. As indicated above this decrease may be indicating that biodegradation is not occurring because the PlumeStop is downgradient of GZ-3.

GZ-4 Results
VC results for GZ-4 are not presented because all of the results for VC to date have been below the method detection limit (i.e. non-detect). As shown in Figure

38
The DCE results in GZ-4 also appear to have decreased since the injection.

GZ-5 Results
TCE is also the predominant contaminant in GZ-5. As shown in Figure

40
As shown in Figure 30, the results for DCE in GZ-5 also appear to have decreased since the injection. DCE degrades to VC. Figure 31 shows a sharp increase in VC concentration after the injection, which may be the result of DCE degradation.
However, as previously described it is unclear whether degradation products remain adsorbed to the PlumeStop surface.

MW-2D Results
As shown in Figure 32, DCE is the predominant contaminant found in MW-2D.
VC results for MW-2D are not presented because all of the results for VC to date have been below the method detection limit (i.e. non-detect). The concentration of DCE and TVOCs initially increases after the injections. There is not enough TCE observed in MW-2D to account for the increase in DCE as a result of degradation. The increase may be the result of mobilizing a pocket of contamination as a result of the injections.
After the initial spike in DCE and TVOC concentrations, the concentration of both DCE and TVOCs appear to decrease.

GZ-6 Results
Downgradient well GZ-6 is the furthest from the injection points. We expect to see a delay of approximately 12 to 14 months for groundwater to travel from the PRB injection area to GZ-6. As shown in Figure 35, TCE is the predominant contaminant observed in GZ-6. Based on the results thusfar, it appears as though the concentrations of PCE, TCE, DCE, and VC in GZ-6 are beginning to decrease. As shown in Figure   36, the results for PCE have been decreasing since 2011, which may be a result of the ZVI treatment in addition to the PlumeStop treatment. Figure 37, depicts a sharp decrease in DCE concentrations during the latest sampling round. As shown in Figure   38, the VC was beginning to increase and then decreased, due to the delay it appears that this increase and decrease is related to the ZVI treatment and not the PlumeStop treatment. Long-term monitoring will be required to determine if the PlumeStop injections effectively reduce offsite migration of contaminants.

Former Plating Facility Statistical Analysis
A t-test was performed for the sampling results from GZ-3, GZ-4, GZ-5 and MW-2D. A t-test was not performed for GZ-6. As described previously there is a delay of 12-14 months between the time of the injection, and when the impacts of the injection are expected to appear in GZ-6. As a result of that delay only three (3) samples have been collected that reflect the impact of the treatment in GZ-6. With only three (3) data points the results of a t-test would not be valid.
For GZ-3, GZ-4, GZ-5, and MW-2D six (6) samples have been collected since the injection. The t-test was performed with the six (6) Table X. The results for VC were non-detect for GZ-4 and MW-2D, therefore, a t-test was not performed for these results. Based on t-test using 90% confidence (p = 0.1), there is a statistically significant difference between the pre and post injection results for TVOC, PCE, TCE, and VC in GZ-3. There is a statistically significant difference between the pre and post injection results for TVOC, PCE, and TCE in GZ-4. There is a statistically signficant difference between the pre and post injection results for TVOC, PCE, and DCE in GZ-5. There was not a statistically significant difference between the pre and post injection results There appears to be some interference from the PlumeStop that affects the submersible pumps. As described above, the injection into upgradient well MW56 failed to infiltrate into the bedrock. This well was not developed prior to the injection which may have contributed to the injectate not infiltrating. Additionally, MW56 only extends into the shallow bedrock which may have contributed to the low infiltration rate.
The groundwater at Central Landfill contains many more contaminants of concern than those discussed in this paper. Some of these contaminants may interfere with the performance of PlumeStop.
The injection at the former plating facility also appears to have successfully The long-term effectiveness of the injectate remains to be seen. GZA will continue to monitor the groundwater quality downgradient of the injections to evaluate breakthrough of the PRB.

RECOMMENDATIONS
Prior to performing future injections with PlumeStop it is recommended to perform bench scale tests of the product using the actual groundwater and soils from the site of interest. The bench scale testing will evaluate the performance of the product in an environment that is easily monitored. This would help define the quantity of PlumeStop required to achieve the project specific goals Understanding the hydraulic conductivity of the site is also important to planning the injections. Tracer tests may also be helpful in evaluating/anticipating the flow paths that the PlumeStop will follow.

50
APPENDICES Appendix A - Tables 3 through 18   Table 3        Orange indicates exceedence of action levels Bold indicates compound detected above method detection limit                                                                                                  The total volume of the first injection was approximately 1,250 gallons.

RIDEM GB EXCEEDANCES ARE IN BOLD AND HIGHLIGHTED GREEN
The total volume of the second injection was approximately 800 gallons. The total volume of the third injection was approximately 2,000 gallons. The groundwater hydraulic conductivity is expected to range from 0.85 to 2.5 feet per day (based on the 3-9 months for changes in ML-11 to appear in ML-12 which is 230 feet downgradient). Based on literature from Regenesis, PlumeStop is expected to migrate at least 2 meters (6.5 feet) from the injection point. Meaning the radius of influence (ROI) would be at least 6.5 feet. Using an interconnected matrix porosity for gneiss of 0.01 (INTERA Environmental Consultants, Inc. 1983), the total flow of groundwater through the radius of influence can be estimated using the following equation. Note, the equation below assumes that the flow is perpendicular to an area that is equivalent to the depth of water in the borehole by twice the ROI.
= ℎ * 2 * * * = (189 − 83 ) * 13 * 0.01 * 0.85 ⁄ = 11.7 3 ⁄ The first injection was performed over five weeks, and the first sample after the injection was collected 77 days after the start of the first injection. The total volume of groundwater that passed through the ROI during that time period is estimated below. 117 = * . = 11.7 3 * 77 = 900.9 3 6,738.7 The estimated dilution due to the addition of the injectate is below ℎ ℎ = 1,250 6,738.7 = 0.185 Therefore, the percent reduction in contaminant concentrations due to dilution alone for the first injection is estimated to be at most 18.5%. Note, the radius of influence may be larger which would lower the percent reduction in contaminant concentrations due to dilution alone. It is not possible to determine the actual ROI of the injections without performing soil borings.
Additionally, the lowest value of hydraulic conductivity was used to determine the flow through the ROI as a conservative estimate. If the 2.5 feet per day hydraulic conductivity was used, then the percent reduction in contaminant concentrations due to dilution would be lower.
The concentration of 1,2-DCBz prior to the injection was 14,000 µg/L and the concentration after the first injection was 1,100 µg/L. The concentration of CBz prior to the first injection was 6,400 µg/L and the concentration after the first injection was 1,500 µg/L. The percent reduction in contaminant concentrations can be calculated as follows Given that the percent reductions for 1,2-DCBz and CBz are more than 18.5%, it is unlikely that the reduction in contaminant concentrations is due to dilution alone, although dilution may be a contributing factor.
Calculations for the second injection Again, the reduction in contaminant concentrations cannot be contributed to dilution alone. These calculations cannot be completed for the third injection because there is no sampling data after the third injection.

Former Plating Facility
The total volume of injectate at the Former Plating Facility was approximately 5,300 gallons. The groundwater hydraulic conductivity is expected to range from 0.12 to 0.14 feet per day (based on the 12-14 months for groundwater to reach GZ-6, 50 feet downgradient). The injections were performed along the corner of the Site. The length of the injections perpendicular to groundwater flow direction is approximately 90 feet, and the injections were completed to 20 feet below grade. The depth to groundwater at the site is typically 9 to feet below grade. The soil at the site consists of silty sand, the porosity of silty sand is 0.25. The estimated dilution due to the addition of the injectate is below ℎ ℎ = 5,300 11,713.7 = 0.452 Therefore, the percent reduction in contaminant concentrations due to dilution alone for the first injection is estimated to be at most 45.2%. Dilution could be influencing the results however, the concentration of some contaminants increased after the injection, therefore it is unlikely that dilution is occurring.