SUMMARY OF THE REMEDIATION TECHNOLOGIES DEVELOPMENT FORUM
PERMEABLE REACTIVE BARRIERS ACTION TEAM MEETING

Holiday Inn Select
Niagara Falls, New York
October 15–16, 2003

WELCOME AND INTRODUCTIONS
Bob Puls, U.S. Environmental Protection Agency (EPA)
John Vidumsky, DuPont

Bob Puls and John Vidumsky, co-chairs of the Remediation Technologies Development Forum’s (RTDF’s) Permeable Reactive Barriers (PRB) Action Team, welcomed attendees to the meeting. (The final attendee list is included as Attachment A. (PDF, 50KB)) Vidumsky provided an overview of the agenda, noting that the presentations delivered during the meeting would cover four broad topic areas: (1) hydraulic impacts on PRB performance, (2) long-term PRB performance, (3) applications of PRB technology to source zone management, and (4) novel PRB applications. Vidumsky also noted that a poster session would be held during the evening of October 15, 2003. Before launching into the program, Vidumsky thanked those who played a role in planning the meeting. He also thanked EnviroMetal Technologies, Inc., and GeoSyntec for providing refreshments.

SESSION I: HYDRAULIC IMPACTS ON PRB PERFORMANCE
Session Chair: John Vidumsky, DuPont

Summary of Field Performance of PRB Systems
John Vogan, EnviroMetal Technologies, Inc. (ETI)

John Vogan summarized the status of PRB technology. He said that PRBs are already being used across the world to remediate organic compounds, and that 10 to 15 new PRBs are installed each year. While most PRBs are located in the United States, the technology is also being deployed in Australia, Canada, Europe, and Japan. Vogan said that PRB design and installation practices have changed somewhat over time. For example, over the last 5 to 6 years, designers have moved away from funnel-and-gate designs and toward continuous walls. There has also been a shift in construction methods: while cofferdam and continuous trenching enjoyed popularity between 1995 and 1998, injection-type methods (e.g., hydrofracturing, biopolymer trenching, and jetting) have gained popularity in more recent years. The reason for this shift might be two-fold, Vogan said: injection-type methods are (1) more cost effective than other techniques and (2) capable of emplacing reactive materials at greater depths.

Vogan spent the remainder of his presentation exploring the following question: “How many PRBs are working effectively?” To answer this question, he said, one must first answer the questions below.

• Are PRBs meeting regulatory criteria? Vogan said that ETI evaluated 52 full-scale PRB applications and concluded that 43 of them are meeting regulatory objectives, even though at least 6 of the 43 exhibit deviations from predicted flow fields. At the nine PRB sites that are not meeting regulatory objectives, pump-and-treat systems are being used to address deficiencies at three sites and site retrofits are being considered at the remaining six. For the most part, Vogan said, hydraulic issues are to blame for the problems that have arisen at these nine sites. Vogan said that ETI’s performance review is still in the preliminary stage and that a more rigorous analysis is needed before ETI can release concrete conclusions about the percentage of PRBs that meet regulatory objectives. He said that certain issues, such as the persistence of volatile organic compounds (VOCs), cloud the interpretation of performance results.
  
  
• What are the geochemical performance issues? Much research has been conducted to examine mineral precipitates’ impact on PRB performance. While precipitate buildup has been observed in the field, Vogan said, there is no indication that these precipitates reduce PRB reactivity to the point of necessitating rejuvenation or iron replacement. Given these findings, it seems reasonable to conclude that most PRBs will function for at least 10 to 15 years before they require refurbishing. Addressing concerns that have been expressed about biofouling, Vogan noted that ETI is only aware of 1 PRB out of about 90 (the Denver Federal Center) that has exhibited a significant problem.
  
  
• What are the hydraulic performance issues? Vogan said that hydraulic issues are typically to blame when a PRB exhibits inadequate performance. These problems can be caused by sewer lines that influence flow, construction artifacts that impede flow, bypass around the system, and insufficient residence time. In Europe, he said, concern about hydraulic issues is spurring a return to more heavily engineered PRBs that can achieve greater hydraulic control. Active PRBs of this nature differ from the more passive systems (e.g., continuous walls) that are more common in North America.

In some cases, Vogan said, inadequate PRB performance is attributed to construction problems. Fortunately, over time, construction methods for PRB installation have improved, and this has reduced the frequency of problems.

After Vogan completed his presentation, some audience members returned to his point about active PRBs gaining popularity in Europe. Puls acknowledged that some active PRBs have indeed been installed in Europe to impose hydraulic control, but that in many cases such measures were taken to address problems that arose with funnel-and-gate systems—a design that is fairly popular in Europe. Puls said that the Europeans he has talked to have starting leaning away from funnel-and-gate designs in favor of continuous wall designs. If this trend bears out, Puls said, there would be a shift toward more passive (rather than active) PRB systems in Europe. Grant Hocking, saying that it is important to promote passive systems over active systems, advised placing more emphasis on quality assurance/quality control (QA/QC) checkpoints throughout the construction process. Doing so, he said, would help ensure that passive systems work properly. If more robust QA/QC procedures were followed, he said, it would be possible to fix and address problems at the time of construction.

Puls asked for input on what can be done to address hydraulic performance issues. While better site characterization data could help reduce the potential for unpredicted hydraulic patterns, he said, subsurface heterogeneities will always cause a certain degree of uncertainty. Vogan said that Puls’ point was important, and one that regulators, site owners, and the public need to digest. He did note, however, that there are tools that can help address uncertainty. For example, during the design process, probabilistic modeling tools can be used to account for uncertainty. Also, efforts could be made to develop low-cost construction methods that patch small problem-ridden sections of a PRB.

Evaluation of the Permeable Reactive Wall at the Northeast Corner Operable Unit, Lake City Army Ammunition Plant, Missouri
John Moylan, Consulting Geologist

John Moylan provided information about a permeable reactive wall (PRW) at the Lake City Army Ammunition Plant’s Northeast Corner Operable Unit. This PRW, installed in August 1999, was constructed using the biopolymer slurry method. The PRW is near a creek, is about 400 feet long, ranges in depth from 25 feet to 64 feet, and was constructed in a silty clay alluvial terrace deposit that has thin discontinuous sand layers. The hydraulic conductivity of the subsurface changes with depth, Moylan said; the most conductive materials are in the deepest part of the formation.

The PRW has encountered problems. To be specific, post-construction water levels are as much as 14 feet higher than pre-construction levels. Moylan said that he was asked to identify the cause of the high water levels and to determine how they affect the PRW’s performance. Toward this end, Moylan has evaluated pre-design investigation data, design data, construction reports, ground-water elevation data, chemical data, and pump test results. In summary, Moylan’s investigations revealed the following:

• The PRW design included an appropriate level of conservatism and adequate safety factors, but the orientation of the PRW relative to the ground-water contours was not considered.
  
  
• Construction factors are responsible, at least in part, for the PRW’s subpar performance. The trench collapsed during initial installation, reactive materials were not emplaced through a tremie pipe with water, a significant amount of fine soil was present in the polymer slurry, and an I-beam endstop and a clay sliver were left in the trench. Because of these problems, Moylan postulated, the reactive materials exhibited significant segregation, the permeability of the PRW was less than it should have been, and a low-permeability skin developed along the trench walls.
  
  
• The pump tests revealed that the reactive materials in the PRW are most permeable in the northeast end of the PRW and are least permeable in the deep southwest portion of the PRW, where the trench originally collapsed and where slurry viscosity was the highest.
  
  
• Ground-water level elevation data proved to be the most important source of information when it came to evaluating the effectiveness of the PRB.
  
  
• The chemical data showed that the monitoring transects are not located along a flowline. As a result, water that passes through upgradient wells does not pass through the in-wall wells or the downgradient wells.

Moylan said that he has concluded the following about the site: (1) the flow pattern that has developed since the PRW was constructed is complex; (2) a significant but unknown portion of the plume is bypassing the PRW; (3) plume bypass is occurring because the PRW is not aligned parallel to equipotential contours, the hydraulic conductivity of the PRW’s reactive material is low and varies laterally, and there is a low conductivity skin at the trench walls; (4) the chlorinated compounds that actually pass through the PRW do appear to be degraded below maximum contaminant levels (MCLs); (5) the wells in the performance monitoring transects are not located along flow lines; and (6) the limited pre-design investigation coupled with the delayed implementation of the performance monitoring program prevented a timely recognition of performance problems. Expanding on the latter, Moylan said that it is unforgivable to postpone performance monitoring programs. To avoid holdups, Moylan advised coming to consensus on a monitoring program before initiating PRB construction.

Rich Steimle asked whether the site managers plan to address the plume bypass problem. Moylan said that he is not sure what is planned for the site, but did note that a new contractor has been hired and has been charged with remediating the site to the satisfaction of the regulatory community. Hocking engaged Moylan in a discussion about the factors that led to the PRW’s performance problems. While Moylan said that he thought complex hydraulic flow patterns played a significant role, Hocking suggested that many of the problems stemmed from the fact that the PRW was less permeable than anticipated.

Influence of Sewers on Ground-Water Flow in the Vicinity of a PRB in an Urban Setting
William M. Goodman, Sear-Brown

William Goodman, whose presentation is included as Attachment B (PDF, 774KB), talked about the effects manmade conduits can have on PRB performance. He opened by introducing the term urban epi karst, a phrase coined by academics at the University of Texas at Austin. Examples of urban epi karst include utility trenches, fractures in asphalt, storm drains, manholes, septic tanks, sanitary sewer systems, storm drainage systems, leaky tunnels, and leaky basements, all of which can alter subsurface conductivity and change localized ground-water flow patterns. Goodman said that urban epi karst can have the most pronounced impacts in low-permeability settings.

Goodman discussed a remediation project performed at a former drycleaning facility in Geneva, New York. The site, contaminated with tetrachloroethylene (PCE) and its degradation products, is characterized by low permeability (i.e., a hydraulic conductivity of 10^-4 to 10^-5). A two-pronged approach was used to remediate the site. First, about 1,200 tons of contaminated soil were excavated to remove the most grossly impacted soil. Next, a PRB was installed to address the site’s contaminated ground water. Hydraulic shoring techniques were used to install the PRB, Goodman said; a sand/iron mixture was emplaced into a trench, covered with geotextile, and capped with a bentonite/cement mixture. The PRB, installed about 30 months ago, is about 200 feet long and 16 feet deep. The PRB was installed near a sanitary sewer system and a manhole, Goodman said, and these urban epi karst elements impacted ground-water flow in unanticipated ways. In this particular case, researchers believe, the sewer/manhole system improved the PRB’s performance by creating a localized ground-water depression that encouraged ground water to flow through the PRB in a more longitudinal fashion. This altered flow pattern lengthens residence times, allowing the contaminants in ground water to be subjected to the PRB’s reactive media for longer time periods. Goodman said that monitoring results indicate that the PRB is indeed treating ground water. This is evidenced by the fact that VOCs are rarely detected within or downgradient of the wall. The site owners and regulators have been very happy with the results.

Goodman concluded by saying that the effects of urban epi karst can be understood and characterized through site investigation. In some cases, they might enhance the efficacy of a PRB. In other cases, they could be detrimental. He advised PRB designers to learn about the urban epi karst elements that are present at a given site and to make an effort to either exploit their benefits or avoid the pitfalls they create.

Hydraulic Pulse Testing of PRBs
Grant Hocking, GeoSierra LLC

Grant Hocking, whose presentation is included as Attachment C, (PDF, 866KB) talked about QA/QC procedures that can be used to determine whether PRBs are meeting design criteria. He opened by noting that PRB construction methods can cause smearing, filter cake clogging, and skin effect issues, and that problems of this nature can cause a PRB to have a permeability that differs from that which was called for in the original design. In such cases, ground-water flow patterns could be impacted and PRB performance may be seriously compromised. Given the seriousness of these ramifications, Hocking said, it is essential to conduct construction verification testing to determine whether a PRB has indeed been constructed as originally planned.

Hocking provided information about three QA/QC methodologies: (1) using ground-water monitoring wells to examine water levels, (2) performing slug tests within the PRB wall, and (3) performing hydraulic pulse interference tests. He said that the first two methods listed are not sensitive enough to quantify PRB permeability or to identify clogging issues; thus, he advised against relying on either of them as a tool for QA/QC. He said that the hydraulic pulse interference test is a far better alternative. Developed by the oil industry more than 30 years ago, this test is ideal for quantifying the hydraulic impact of PRBs and identifying partial clogging issues. The test can also quantify PRB skin effects, and it is very sensitive to hydrogeological conditions between source and receiver wells. It is simple and straightforward to use.

Hocking’s mention of a PRB skin effect attracted attention from the audience. John Moylan asked whether the hydraulic pulse interference test can be used to calculate the skin’s thickness. Hocking said that the test does not provide such information, but added that this should not be viewed as a drawback. He said that skin thickness is an irrelevant measurement—hydraulic resistance, a value that can be measured with the hydraulic pulse interference test, is the more important parameter to quantify. One attendee asked Hocking what can be done to address skin effects. Hocking said that he does not have an answer yet. The ideal solution, he said, is to use construction methods that do not yield such effects. For PRBs that already exhibit skin effects, however, no clear retrofitting solution has been identified.

Modeling of Downgradient Reverse Diffusion Effects
Tom Sale, Colorado State University

Tom Sale talked about work that is being performed to understand why downstream plume concentrations, in at least some cases, remain above MCLs even after a source zone has been eliminated. He said that recent research suggests that reverse diffusion processes may play a role in this phenomenon, and he supported his statement by presenting information gleaned through mass discharge modeling exercises, laboratory studies performed under the auspices of the Air Force Center for Environmental Excellence Source Zone Initiative, analytical modeling studies, and field results collected from the PRB¹ at the F.E. Warren Spill Site 7. Detailed information about these projects is presented in Attachment D (PDF, 1.2MB). In summary, the data suggest that:

• Significant amounts of contaminant mass diffuse into stagnant zones. Some reverse diffusion is observed once a source zone is shut off. Inward and outward diffusion can occur simultaneously.
  
  
• Even if a PRB has a zero-flux boundary (in terms of contaminant mass), contaminants might persist downstream as a result of reverse diffusion.

Sale said that these findings could help researchers address tough questions that they have been struggling with for years, such as: (1) Why don’t observed plume concentrations match predicted concentrations? (2) Why do source areas persist for so long? (3) Why can’t researchers find dense non-aqueous phase liquid (DNAPL) at some sites even when contaminants persist? (4) Why do some sites exhibit very limited water quality improvements even though they have been subjected to aggressive treatment? and (5) Why is rebound observed at some sites? Sale concluded by noting that the information that is emerging about reverse diffusion effects might force researchers to change some of their basic understanding about plumes. For example, the following paradigm shifts might be necessary:

Puls asked Sale whether regulators and site owners are satisfied with the performance exhibited by the PRB at the F.E. Warren site. Although the downgradient contaminant concentrations remain above MCLs, Sale said, the wall has met its primary objective—protecting a nearby surface-water body. Other attendees commented on the importance of distributing Sale’s findings to the regulatory community so that false expectations about remedial objectives are not perpetuated.

Evaluation of Ground-Water Hydraulics at a Zero-Valent Iron (ZVI) PRB at the Somersworth Superfund Site
Suzanne O’Hara, GeoSyntec Consultants

Suzanne O’Hara, whose presentation is included as Attachment E, (PDF, 768KB) provided information about a PRB that was installed at the Somersworth Sanitary Landfill Superfund site in 2000. She opened by providing background information on the site and describing the remedial approach being used. The subsurface, she said, consists of an overburden layer (40 feet of clay and gravel) and a weathered bedrock layer. Contaminated ground water flows through both layers and discharges to an adjacent wetland. (Trichloroethene [TCE] and its degradation products are the primary contaminants of concern.) In an effort to remediate the site, a 915-foot-long ZVI PRB was installed to treat the ground water in the overburden layer. In addition, an extraction well was installed upgradient of the PRB to remove contaminated ground water from the bedrock layer. The extracted water is passed through an infiltration gallery and reinjected on top of the landfill. The water percolates through the landfill’s permeable cap and moves into the overburden, where it is treated as it passes through the PRB.

O’Hara focused her talk on hydraulic issues. She described the PRB as a “hanging wall,” noting that it is not keyed into an impermeable layer and that this raises concern about the potential for ground water to flow around or under the PRB. She also said that concerns about the potential for flow diversion are compounded by the fact that the hydraulic conductivity of the PRB is similar to that of the surrounding subsurface formation. (This means that flow will likely be diverted if any portion of the PRB is less permeable than called for in the design specifications.) Given these concerns, O’Hara said, hydraulic testing is performed annually to determine whether there has been any unexpected change in flow patterns or any change in the PRB’s porosity. She said that pump tests are used to collect data, noting that attempts to use slug tests and pneumatic tests proved to be problematic. In addition, the wells that immediately surround the PRB are monitored regularly to analyze ground-water elevation and chemistry. The data reveal the following:

• During certain times of the year, a small amount of mounding occurs on the upgradient side of the PRB and some water goes around the PRB. Sampling results indicate, however, that the VOC flux around and beneath the PRB is insignificant.
  
• No significant permeability changes have been detected along the northern and central sections of the PRB, but some changes have been detected in the southern portion of the wall. Site investigators are unsure of the cause: while true permeability reductions are a definite possibility, O’Hara said, the permeability changes observed between 2001 and 2002 might have resulted from bioslurry breakdown effects. Site investigators plan to continue monitoring the situation.

Before closing, O’Hara noted that hydraulic gradients at the site are influenced by the seasons. While winter gradient reversals obscure the geochemical profile within the PRB, she said, they do not appear to have a negative impact on the PRB’s performance.

Biopolymer Liquid Shoring: General Characteristics and Use
Lloyd Marsden, Rantec Corporation

Lloyd Marsden provided information about biodegradable liquid shoring, a technique used to construct PRBs. He said that foundation contractors have used bentonite liquid shoring to assist in excavation efforts since the 1940s. Biopolymer-based liquid shoring was first used in the United States in 1986. Marsden provided detailed information (see Attachment F (PDF, 1.41MB)) about the characteristics and properties of guar gum, noting that: (1) it is derived from a legume, (2) it is a complex carbohydrate consisting of galactose and mannose sugars, and (3) it has an abundance of hydroxyl radicals attached. The chemical composition of guar gum, Marsden said, explains the unique fluid characteristics that make the material valuable for maintaining excavation stability.

Marsden described how guar gum is used to construct PRBs: a trench is dug, guar gum is pumped into the trench to keep it open, reactive materials are poured into the trench, and the guar gum breaks down or is displaced in the process. The guar gum prevents ground water from entering the trench, limits the movement of solids, and forms a wall cake. Expanding on the latter, Marsden explained that guar gum forms weak hydrogen bonds with silicas and clays, and that this prevents water damage penetration, swelling, and dispersion. When it comes to constructing a PRB, Marsden said, it is essential to break down the guar gum once the reactive materials have been emplaced within the trench. This is no problem, he said: guar gum can be broken down by interacting with indigenous molds, fungi, and bacteria, or by introducing enzymes, bleach, or other oxidizing agents.

Marsden identified the following as topics to address before implementing a biopolymer liquid shoring project:

• Identifying an appropriate injection/circulation system. Guar gum should be pumped into the subsurface in a manner that maximizes its contact throughout the media zone and the trench walls.
  
  
• Preparing the biopolymer. Guar gum is sold as a powder. It must be wetted, mixed, and given sufficient time to hydrate before it can be used. Thus, it is important to have a storage tank available.
  
  
• Determining whether extra measures are needed to stabilize the biopolymer. Guar gum breaks down quickly. Several measures can be taken to stall breakdown, including altering the pH to inhibit microbial activity and adding a biostat to reduce microbial growth.
  
  
• Identifying a plan for biopolymer breakdown. Although microbes and enzymes can be relied upon to break down guar gum, in some special cases, more mechanical methods might also be required to facilitate breakdown.
  
  
• Deciding on quality control procedures. Marsden advised monitoring the slurry’s pH and viscosity throughout the construction project to ensure that the slurry is performing according to plan.

Before closing, Marsden provided a list of some recent PRB installations that have made use of the biopolymer slurry construction method. The list is included in Attachment F. (PDF, 1.41MB)

SESSION II: LONG-TERM PERFORMANCE OF PRBS
Session Chair: Bob Puls, EPA

Field Monitoring of the Performance of a PRB at the Vapokon Site, Denmark
Irene M.C. Lo, Hong Kong University of Science and Technology

Irene Lo provided information about a funnel-and-gate PRB that has been installed at the Vapokon Site to address a variety of contaminants, including PCE, TCE, 1,1,1-trichloroethane (1,1,1-TCA), benzene, toluene, ethylene, and xylene. The wall, which uses iron as a reactive medium, was installed in 1999. Monitoring data have been collected on five occasions since that time. The data are being used to (1) evaluate the PRB’s ability to treat chlorinated organics; (2) identify changes in ground-water geochemistry and identify precipitates; (3) investigate any contributions from microbial degradation or any iron adsorption in the PRB; and (4) investigate variations in porosity, hydraulic gradient, and water table distribution across the PRB. Lo presented a detailed analysis (see Attachment G (PDF, 2.8MB)) of the data collected to date. In summary, the data reveal the following:

• The PRB exhibits a 98 percent removal rate for PCE, TCE, and 1,1,1-TCA, and after operating for 3 years, it shows no significant signs of deteriorating in its ability to degrade these contaminants. All three compounds have been detected at high concentrations in downgradient wells. Lo suspects that these contaminants might have desorbed from the downgradient aquifer.
  
  
• There has been a significant deterioration in the PRB’s ability to degrade daughter products, such as cis-dichloroethene (DCE) and 1,1-dichloroethane (1,1-DCA). For example, while removal rates for cis-DCE were between 80 to 94 percent during the first half of the monitoring period, the rates fell to 8.3 to 49 percent during the second half of the monitoring period. Researchers believe that the reduction in removal rates is attributed to mineral precipitate formation. Based on the data collected, calcite and pyrite are considered the main culprits. Calculations suggest that mineral precipitates caused a loss of 2.32 cubic meters of void volume inside the PRB and a loss in porosity of 3.71 percent.
  
  
• There has been a 55 percent increase in the hydraulic gradient across the barrier, and nonuniform ground-water distribution has been observed.
  
  
• Over the last 3 years, there has been no significant change in chloride concentration, a 2.5 to 3.3 unit increase in pH, a 65 to 69 percent reduction of electrical conductivity, and nearly a 300 millivolt decrease in Eh.
  
  
• The PRB is removing dichloromethane and 1,2-dichloroethane (1,2-DCA) even though iron does not reduce these contaminants. Researchers are not sure why this is occurring, but two possible explanations have been suggested: microbial degradation processes and adsorption to the iron.

Impact of Mineral Fouling on the Long-Term Performance of PRBs
Craig Benson, University of Wisconsin-Madison

Craig Benson, whose presentation is included as Attachment H, (PDF, 1.95MB) noted that mineral precipitates form within ZVI-containing PRBs over time and that these minerals have the potential to reduce a PRB’s porosity, reactivity, and hydraulic conductivity—changes that can alter local hydraulic characteristics and reduce the treatment capacity and life-span of a PRB. To gain a better understanding of the potential impacts of mineral fouling, Benson said, the University of Wisconsin–Madison performed a modeling study to (1) estimate the degree of fouling that occurs in PRBs that are installed in heterogeneous aquifers, (2) evaluate how flow heterogeneity impacts fouling, and (3) examine the impact fouling has on PRB reactivity, ground-water flow, and residence time.

Benson provided an overview of the study, noting that numerical models were used to simulate flow, transport, and geochemical reactions in a heterogeneous aquifer. MODFLOW was used to simulate flow, and a modified version of the RT3D model was used to simulate geochemical reactions and predict mineral fouling in the PRB. (See Attachment H (PDF, 1.95MB) for additional details.) Field data from the PRBs at Moffett Field and the Elizabeth City site were used to validate the model, Benson said, noting that there was general agreement between actual field measurements and the values predicted by the model. After the validation exercise was completed, simulations were performed to examine mineral precipitation and porosity reduction under a variety of scenarios. Calcium carbonate, ferrous hydroxide, magnesium carbonate, and ferrous carbonate were identified (through sensitivity analysis) as the four key minerals that warranted inclusion in the assessments.

While the modeling results do show that mineral fouling has an adverse effect on PRB performance, Benson said, the results also indicate that performance problems are unlikely to be major until the PRB is at least 25 years old. He said that the following conclusions can be drawn from the modeling exercise:

• Mineral fouling causes porosity reductions. After 10 years of operation, Benson said, porosity reductions in the PRB are not likely to be dramatic. After 30 years, however, about 50 percent of the pore space in the front half of the PRB could be filled with mineral precipitates. At the 50-year mark, some portions of the wall will be completely filled with mineral precipitates.
  
• Porosity reductions are spatially variable due to flow heterogeneities. Porosity reductions were not uniform throughout the wall, Benson said, noting that the greatest porosity losses were recorded in areas with the highest ground-water flow rates. In some ways, Benson said, the fact that clogging occurs the most rapidly in high-flow areas helps heal the problem of preferential pathways.
  
  
• Mineral fouling impacts hydraulic conductivity. While some subtle changes in hydraulic conductivity were observed at the 10 year mark, Benson said, the changes were not substantial enough to impact site hydraulics. After 30 years, while some portions of the PRB were found to be less permeable than the aquifer, the majority of the PRB was still more permeable than the surrounding formation. In contrast, at the 50-year mark, large segments of the PRB were impervious and local hydraulics were impacted.
  
  
• Mineral fouling impacts hydraulic performance. Benson said that mineral fouling, at least after the 10-year mark, did cause reductions in residence time. In addition, after the 30-year mark, some flow bypass was observed. Also, the modeling exercise suggested that mineral precipitate buildup leads to hydraulic gradient buildup, although the changes recorded were subtle.

The model was also used to explore the utility of strategies that remedy mineral fouling buildup. For example, modelers examined the effect of (1) setting up sacrificial upgradient iron/gravel zones and (2) using an auger to stir up PRB materials and knock off precipitates. The results suggest that these practices only have a modest effect in preventing porosity reductions and hydraulic changes.

Implications from Long-Term Monitoring of Two ZVI Reactive Barriers in Germany
Markus Ebert, Christian-Albrechts-Universität zu Kiel

Markus Ebert, whose presentation is included as Attachment I, (PDF, 881KB) provided information about some of the experiences Germany has had with PRBs. (Volker Birke, one of Ebert’s colleagues, also planned to provide information about the status of PRBs in Germany during the meeting. Because he was unable to attend, Birke sent a PowerPoint presentation, which is included as Attachment J (part 1, (PDF, 1.3MB) part 2 (PDF, 1.42MB)) to accompany Ebert’s presentation.) Ebert said that seven PRBs have been installed in the field in Germany since 1998 and four more field applications are in the planning stages. Most of these PRBs use ZVI or granulated activated carbon and are designed as funnel-and-gate systems. While some of the PRBs are research and development (R&D) projects, others have been installed to address commercial sites. At some sites, the PRBs are not performing as well as originally hoped. For example, the following have been observed at some sites: (1) hydraulic problems, (2) barrier bypass, (3) failure to remediate TCE byproducts, and (4) the appearance of downgradient contaminants a short time after barrier construction. Ebert said that the cause behind the problems is not always clear, a fact that is due, at least in some cases, to substandard monitoring efforts. As a result, those who oppose PRB technology are starting to gain a stronger voice in Germany.

Ebert focused the remainder of his presentation on the Rheine site and the Tübingen site, both of which are being evaluated through RUBIN—one of Germany’s R&D networks. Both sites have been subjected to laboratory testing, “standard” ground-water monitoring, multi-level flow sampling, passive sampling, coring, stable carbon isotope analyses, and pump and tracer tests. The results are presented in detail in Attachment I. (PDF, 881KB) In summary, Ebert said, the following information has been collected:

• The Rheine site. The PRB constructed at this site has been in place since June 1998. It is a continuous wall and consists of two sections. One of the sections is filled with iron sponge, and the other is filled with a mixture of gravel and 30 percent grey cast iron. While the iron-sponge section is performing well, Ebert said, the same cannot be said for the section that contains the iron/gravel mixture. Ebert noted the following about each section:
  
  
• The iron-sponge section. Excellent PCE degradation rates have been observed in this section of the PRB, Ebert said, noting that the rates match those predicted in laboratory column tests. There is no sign of decreased reactivity in this part of the PRB and the life-span is expected to exceed 5 years. Ebert also said that a microbial community was established within 1 year of the PRB’s installation, and that the microbes, as well as the ZVI, have impacted inorganic ground-water constituents.
  
  
• The iron/gravel section. Although PCE reduction rates were initially high in this section of the PRB, the wall’s reactivity decreased significantly within 6 months. In addition, downgradient contaminant concentrations have increased. A forced gradient tracer test was performed to examine the wall, Ebert said, noting that the results suggest the following: (1) the wall’s permeability is still high, (2) there is no significant difference in reactivity between the front and the back part of the PRB, (3) PCE concentration increases are due to reactivity loss rather than flow problems, (4) nitrate reduction is less than in column experiments that used fresh iron, and (5) it is impossible to make a distinction between a general loss in reactivity and strong passivation along single flow paths. Given these findings, Ebert said, researchers believe that the problems exhibited by this section of the PRB are due to nonuniform distribution of the reactive media and heterogeneities in contaminant loading—problems which are likely attributable to construction and design failures.
  
  
• The Tübingen site. The funnel-and-gate PRB constructed at this site has been in place since autumn 1998. Within the first year of construction, Ebert said, increased contaminant concentrations were detected in downgradient monitoring wells. Forced gradient tracer tests were performed at Gate 1 to determine why this is the case. Ebert said that the results suggest that the poor performance could be attributed to the fact that a preferential flow path has been established along a portion of the PRB that has high permeability and low iron content. Construction and design failures could be to blame.

Ebert made two parting comments before concluding his presentation: PRB performance problems usually reveal themselves shortly after construction, and forced gradient tracer tests help identify problems.

Long-Term Performance of PRBs: Lessons Learned
Bob Puls, EPA

Bob Puls, whose presentation is included as Attachment K, (PDF, 790KB) provided information about a tri-agency initiative that EPA, the Department of Defense, and the Department of Energy undertook to evaluate the long-term performance of PRBs. The evaluation was performed on three continuous reactive barriers and five funnel-and-gate systems. Of those PRBs examined, Puls said, most were installed to treat chlorinated compounds, but some were installed to treat metals, such as hexavalent chromium (Cr[VI]), and radionuclides. The evaluation, which was conducted over a 3-year period, is now finished. Its results will be published in the near future as a Federal Remediation Technologies Roundtable (FRTR) document. Puls provided a brief summary of the results:

• Hydraulic performance evaluations. During the PRB evaluation, Puls said, investigators analyzed water levels and flow meters and performed slug tests and tracer studies. In the process, the principal investigators formulated some opinions about the utility of different investigation methodologies. Speaking for them, Puls said that water level measurements provide the best indicator of bulk flow and are effective at identifying major flow problems. (Water level measurements are particularly useful when collected from a dense network of monitoring wells with uniform screened intervals.) Tracer tests, although expensive to conduct, also provide useful information about flow conditions. Puls also listed key points that became clear after investigators examined the hydraulic performance of a number of PRBs. For example: (1) plume concentrations vary spatially and seasonally, a point that should be taken into account when estimating a PRB’s residence time requirements; (2) one should consider a full range of hydraulic scenarios (rather than just an average) when designing a PRB; (3) estimating hydraulic conductivity to within half an order or one order of magnitude is difficult (if not impossible), so ground-water velocity estimates may vary by at least a half-order or one-order of magnitude; (4) seasonal effects impact ground-water flow directions; (5) investigators have experienced difficulties finding clean fronts on the downgradient side of PRBs because, at some sites at least, flow is bypassing the PRB, contaminants trapped in finer sediments are diffusing into the bulk flow, and/or the number of pore volumes flowing through the PRB is relatively low; and (6) one can incorporate variability and uncertainty into the design of a PRB by modeling multiple flow scenarios and using appropriate safety factors. Puls also noted that the following became clear after analyzing the PRB at the Elizabeth City site: it is important to know whether plumes are stable, where the sources are located, and what kind of concentrations are reasonable to anticipate over time.
  
  
• Geochemical performance evaluations. Spatial and temporal trends in ground-water chemistry were analyzed during the course of the PRB evaluation, and coring and mineralogical analyses were also performed. The following was revealed through these analyses: (1) all of the PRBs (with the exception of Denver Federal Center Gate 2) exhibited consistent contaminant degradation, (2) minerals and biomass accumulated in the PRBs in a spatially heterogeneous manner, (3) most mineral and biomass buildup was observed in the first 10 centimeters of the wall, (4) mineral and biomass buildup correlated to ground-water chemistry and ground-water flow rates, and (5) porosity loss rates ranged from 1 to 4 percent per year of the original available volume. Puls said that the latter finding is consistent with modeling results, as well as field results collected in other nations. The findings have led Puls to believe that most PRBs will have a life span that exceeds 10 years, and that some might even last for 30. He also noted that the following was concluded about porosity loss in PRBs: (1) the dominant carbonate minerals in PRBs—aragonite/calcite, iron hydroxy carbonate, and carbonate green rust—account for most of the pore space loss in PRB systems; (2) mackinawite is the dominant sulfur mineral, but it accounts for little pore space loss; and (3) if oxidation is excessive, iron metal transformation can lead to significant loss of pore space. Puls also said that investigators learned the following during the geochemical performance evaluations: (1) vertically resolved hydro/geochemical data should be collected during site characterization; (2) high sulfate, carbonate, and nitrate concentrations can effect PRB performance and longevity; and (3) in fine-textured formations, extra care should be taken during installation to ensure good hydraulic connection between the PRB and the aquifer sediments.
  
  
• Microbiological evaluations. Puls said that some microbial investigations were performed at the two EPA sites that were included in the tri-agency study. The results indicated that: (1) microbial biomass increases upgradient, downgradient, and beneath ZVI PRBs; (2) biomass numbers are low within the wall; and (3) microbial communities are dominated by anaerobic, sulfate-reducing, and metal-reducing bacteria.
  
  
• Geochemical modeling. Geochemical modeling exercises were performed to examine mineral precipitation. Puls said that the model’s results generally agreed with those measured in the field.

OPEN DISCUSSION

At the end of the day, the speakers who participated in Sessions I and II were invited to the front of the room to form a panel and participate in a question-and-answer session. During this session, panel members provided insight on the following topics:

• Using PRBs to address contaminated source zones. Hocking said that interest has grown in using PRBs to treat source zone areas. Audience members confirmed Hocking’s statement; one person said he knew of a site owner who is considering installing one PRB in the source zone, one at the leading edge of a plume, and one in the middle of the plume.
  
  
• Setting goals for source zone treatment. While PRBs can remove large quantities of DNAPL from source zones, Vogan said, it is unclear whether they can reduce contaminants to levels that will be acceptable to the regulatory community. When deciding whether to choose a PRB as a remedial system, he said, regulators might not be interested in hearing what percentage of mass removal it can achieve. Instead, they will probably want to know whether the system can reduce contaminant concentrations below MCLs. Vidumsky said that DuPont’s managers also want to know the bottom line, noting that managers will not be happy to hear that contaminants above MCLs are expected to persist even after a $1 million source control system has been in place for 10 years. Puls said that the regulatory community knows the difficulties associated with DNAPL source removal and is aware that residuals can persist for long periods. This realization has led some regulators to stop focusing quite as much on MCLs as a final goal and to start taking a closer look at the actual risks associated with residual contamination.
  
  
• Setting realistic expectations. Vogan, Sale, and Hocking stressed the importance of being honest about projected outcomes. For example, it is important to let people know that contaminants may persist for long periods after PRB installation and to explain why this is the case. Also, panel members said, it is important to ensure that regulators understand that PRBs will not last forever. This point is particularly important to make, Moylan said, in cases where a PRB is being used to treat a source area: contaminants in source areas are likely to persist long after a PRB has reached the end of its life span.
  
  
• PRB longevity. Based on available studies, Benson and Vogan said, it appears that PRBs have reasonable life spans. While a porosity loss of 10 percent might result after 10 years of operation, Vogan said, such a change is unlikely to exert substantial effects on PRB performance. Significant performance problems are not likely to surface until the PRB experiences major clogging (which might not occur until it has operated for at least 20 years).
  
  
• Incorporating safety factors into design. Benson revisited a point that Puls made during his presentation: one can incorporate variability and uncertainty into the design of a PRB by modeling multiple flow scenarios and using appropriate safety factors. Hocking said that he does not think the scientific community has acquired enough experience to set safety factors for PRBs yet. As an alternative, he suggested using a probabilistic approach when trying to design a system. He said that he uses the 85th percentile as a goal. This means that he aims to design PRBs so that they have an 85 percent probability of meeting the objectives that the regulatory community has designated for them.

SESSION III: APPLICATION OF PRB TECHNOLOGY TO SOURCE ZONE MANAGEMENT
Session Chair: Bob Gillham, University of Waterloo

DNAPL Source Zone Characteristics in Sandy Aquifers
Bob Gillham, University of Waterloo

Standing in for Beth Parker (of the University of Waterloo), Bob Gillham talked about DNAPL source zone characteristics and the problems one encounters when trying to clean up source zones. He opened by presenting some graphical representations of DNAPL source zones. When DNAPL is spilled, he said, it migrates down to the unsaturated zone, eventually reaching the underlying aquitard and forming pools in the aquitard’s surface. As it moves downward, a trail of residual DNAPL is left behind. While some of the DNAPL might be mobile—and therefore easier to pull back out of the subsurface—other portions are likely to be immobile, trapped in low-permeability areas, and difficult to recover. Expanding on the latter point, Gillham said that investigators who have flushed fluids (e.g., oxidants or cosolvents) into the subsurface have experienced limited success in remediating source zone areas. This is because it is difficult for the fluids to penetrate the low-permeability areas that harbor residual DNAPL. Gillham stressed that geological heterogeneity yields complicated DNAPL distribution patterns, making it difficult to remediate DNAPL source zones.

To drive home his point, Gillham presented a case study about a site in Connecticut where a TCE spill led to the formation of a DNAPL source area in the aquifer. He said that a contaminated ground-water plume emanates from the site’s source area and discharges to a river. Core sampling shows that the geometry of the DNAPL source zone is extremely variable: while there is residual DNAPL in some places, most of the DNAPL occurs as discontinuous pools at the bottom of the aquifer. When the geometry of a source area is this complex, Gillham said, it is important to invest substantial effort in source zone characterization activities to avoid making inaccurate estimates of the subsurface DNAPL volume. This point is clearly demonstrated at the Connecticut site. For example, when researchers used conventional long-screen well characterization techniques to characterize the site, they concluded that 165 to 330 55-gallon drums’ worth of DNAPL were present in the source area. When more sophisticated depth-discrete core sampling was performed, however, the researchers concluded that there were only 9 to 30 drums’ worth of DNAPL in the aquifer. In addition to sampling the aquifer, Gillham said, researchers at the Connecticut site collected cores from the underlying aquitard and detected a significant amount of TCE (about 85 gallons) in the aquitard. If a flushing technology were used to remediate DNAPL at this site, Gillham said, the technology would have little impact on the TCE in the aquitard. As a result, even after removing a large mass of DNAPL from the aquifer, TCE would probably continue to bleed (through reverse diffusion) from the aquitard and cause the contaminated ground-water plume to persist. For this reason, Gillham said, it would be difficult to achieve MCLs at this site unless something were done about the TCE stored in the fine-grained lenses of the aquitard.

Gillham said that he actually does not know of any DNAPL source zones that have been cleaned up to MCLs. In an effort to confront this reality, people in the remediation field set their goals on reducing source zones rather than cleaning them up. In recent years, however, some people have expressed more optimism about the idea of cleaning up source zones to MCLs. Such optimism would be warranted, Gillham said, if remediators identified remedial approaches that are better suited to address source zone contamination, or if methods (such as soil mixing) are identified to resolve the complexities that arise from geological heterogeneities. Gillham said that the rest of the talks presented in this session would touch on these very ideas.

Laboratory Tests and Field Investigations of DNAPL Source Zone Remediation Using Granular Iron
Sharon L.S. Wadley, University of Waterloo

Sharon Wadley, whose presentation is included as Attachment L, (PDF, 990KB) talked about using iron-bentonite containment walls to remediate DNAPL source zones. She opened by describing how the technology works. First, iron-bentonite slurries are mixed into the ground using soil-mixing augers. During the mixing exercise, not only is reactive slurry material emplaced into a source zone, but the subsurface is homogenized in the process. Wadley said that an iron-bentonite wall serves two purposes: it contains contaminants and treats them over time. The bentonite initially serves as a lubricant and viscosifer and facilitates the injection of iron. Once in the subsurface, however, the bentonite’s function shifts: its main purpose is to reduce the hydraulic conductivity of the source zone. As for the iron, Wadley said, it is included in the slurry because of its ability to dechlorinate contaminants.

To examine the efficacy of iron-bentonite walls as a source zone treatment, the following studies were conducted:

• Preliminary batch experiments. Laboratory experiments were performed to examine: (1) whether granular iron and bentonite promote the degradation of TCE when mixed in contaminated sandy material, and (2) whether bentonite inhibits TCE degradation. The results were encouraging, Wadley said. TCE degraded, exhibiting a half-life of about 11 hours; chloride concentrations increased; and the concentration of contaminant degradation products dropped.
  
  
• A field demonstration. After obtaining positive results in the laboratory, researchers decided to take the iron-bentonite technology to the field in order to evaluate its ability to remove DNAPL, as well as the feasibility of producing a uniform iron-bentonite-soil-DNAPL mixture in the field. The field demonstration was performed in a 9-meter by 9-meter PCE-contaminated sheet-piled cell at the CFB Borden site². Wadley said that several iron-bentonite slurry walls were installed, each of which contained a different percentage of iron. (Iron content ranged from 0 to 10 percent dry weight.) Augers with tapered mixing blades were used to install the walls. Providing additional detail about the installation process, Wadley noted that (1) the iron-bentonite slurry was premixed before being injected, (2) the iron-bentonite slurry was delivered through drill rods, (3) the mixing blades were raised and lowered several times to homogenize the region, and (4) the tapered blades prevented materials from surfacing. After the walls had been in place for some time, Wadley said, soil cores were collected, stored in methanol and acetonitrile, and analyzed for PCE, TCE, DCE, and vinyl chloride. Chloride concentrations were also analyzed. The analyses showed that PCE concentrations decreased and the iron-bentonite slurry walls were well-mixed but not completely homogenized. The chloride data yielded confusing results, however, which made it difficult to determine whether the PCE reductions were actually attributable to the iron-bentonite walls. Therefore, the decision was made to return to the laboratory and perform additional investigations.
  
  
• Cell experiments. More laboratory experiments were performed to examine the effect that iron-bentonite mixtures have on DNAPL. Toward this end, three cells were established, each of which had CFB Borden sands added to it, bentonite, and 0.5 liters of PCE. In addition, iron (ranging from 5 to 10 percent dry weight) was added to two of the three cells, and the cells were allowed to sit for 1,200 days. Samples were collected from the cells and analyzed on a periodic basis. PCE removal rates of 70 percent and 96 percent were observed in the cells that contained 5 percent iron and 10 percent iron, respectively. In addition, chloride concentrations responded as expected.

Wadley concluded her presentation by encouraging future research on iron-bentonite walls and moving forth with a large-scale field study.

Use of ZVI and Clay for Source Zone Remediation
Michael Liberati, DuPont

Michael Liberati, whose presentation is included as Attachment M, (part 1 (PDF, 1.15MB) part 2 (PDF, 2MB) part 3 (PDF, 1.46MB)) talked about ZVI Source Treatment Technology, a process that DuPont patented and recently donated to Colorado State University. This technology, also referred to as “Saturation Bombing,” was developed to destroy contaminants in source zone areas, with the ultimate goal being to improve downgradient ground-water plume conditions. The technology involves using a large-diameter auger to emplace a column of ZVI-clay mixture into the subsurface. Subsurface materials are mixed continuously throughout the emplacement process to help remove site heterogeneities. Once in the subsurface, Liberati said, the ZVI destroys contaminants and the clay creates a plug. As a result of the latter, ground water flows around the column rather than through it.

The technology is being used as a remedial strategy at DuPont’s Martinsville plant, a former nylon manufacturing facility located on the Virginia-North Carolina border. For many years, Liberati said, laboratory wastes, such as nitric acid, formic acid, phenol, and carbon tetrachloride, were dumped into two open-bottomed waste pits at this site. Not surprisingly, the contaminants migrated and now impact the site’s soil, ground water, and surface water. To provide some perspective on the magnitude of the problem, Liberati noted that: (1) concentrations as high as 30,000 parts per million (ppm) have been detected in the soil, (2) about 8,000 cubic yards of soil are impacted, (3) about 20 tons of carbon tetrachloride are present in the vadose zone, (4) contaminants have been detected in nearby surface water at increasing concentrations, and (5) a carbon tetrachloride plume that has formed under the site is unstable. Although the site’s surface water and ground water are not used for drinking water purposes, Liberati said, concern has been expressed about these natural resources. Thus, it became clear that a management plan had to be developed for the site. When choosing a management strategy, DuPont considered several approaches, including (1) performing long-term ground water and surface water monitoring, (2) treating the plume and exerting downgradient controls, and (3) containing the contaminant source zone. After filling out a decision matrix, DuPont chose source control. This strategy was selected because the source area is relatively small and well-defined, and because source control has the potential to control migration and plume growth in a cost-effective manner. After much consideration, DuPont chose the ZVI Source Treatment Technology over other source control technologies, such as excavation, capping, and soil vapor extraction. (In response to a question from the audience, Liberati said that excavation was ruled out because it would have cost $7 million.)

DuPont performed laboratory studies and a pilot study before implementing the ZVI Source Treatment Technology full-scale. The researchers who performed the studies concluded that ZVI-clay mixtures destroy carbon tetrachloride almost immediately and eventually destroy the contaminant’s daughter products as well. (One unexpected result was recorded in the laboratory, however: low levels of PCE and hexachlorobutadiene formed.) Based on the laboratory and pilot findings, DuPont decided to move forward with full-scale implementation in late 2002.

The installation, which required 8 to 10 crew members and took 10 weeks, was performed using a link-belt crane, a Casagrande Mixing Unit with an 8-foot-diameter auger, an excavator, a batch plant, and a fork lift. Seventy-six columns of ZVI-clay were installed (down to a depth of 35 feet) and three separate treatment zones were established. (The treatment zones differed in the ratio of iron used, with the greatest quantities being used in the most contaminated zones.) The contractor mixed the clay into the ground first and then mixed the iron down into the column. After each column was in place, samples were collected at 10 feet, 20 feet, and 30 feet below ground surface (bgs) to ensure that the proper ratio of ZVI to clay had been maintained throughout the column. In many instances, the contractor found that the iron concentrations were below that which was called for in the design specifications. To fix this problem, more iron was mixed into the column. As a result, the contractor used about 1.5 times the amount of iron that was originally anticipated. By the project’s end, Liberati said, about 225 tons of iron, 340 tons of clay, and 250,000 gallons of water were used to install the 76 columns. This was done at a cost of about $650,000. Liberati said that DuPont learned the following lessons during the installation phase: (1) it is important to be prepared for subsurface obstacles and to remove them in advance if possible, (2) it may have been better to incorporate the ZVI and clay together rather than emplacing them separately, (3) it is difficult to account for all of the iron, (4) water management is an important consideration, (5) additional soil volume might be generated, and (6) it is important to anticipate the release of air emissions or odors if contamination is located close to the ground surface.

Some post-remedial samples were collected in late September 2003. The results, which are very preliminary at this point, show some impressive findings. For example, Liberati said, the maximum concentrations detected for carbon tetrachloride are now only about 0.7 ppm. TCE concentrations have also decreased dramatically: while the maximum pre-remedial concentrations were recorded at 16 ppm, the maximum post-remedial concentrations detected were 0.9 ppm. Post-remedial sampling efforts do show that some daughter products (e.g., methylene chloride and TCE) have increased in concentration since the columns have been in place.

Liberati concluded his presentation by noting that surface water and downgradient ground water will be monitored at this site over the long term and that additional source zone sampling will be performed. In addition, the ZVI Source Treatment Technology will be implemented at more sites in the future.

Emulsified ZVI Treatment of Chlorinated Solvent DNAPL Zones
Suzanne O’Hara, GeoSyntec Consultants

Suzanne O’Hara, whose presentation is included as Attachment N, (PDF, 700KB) presented information about emulsified ZVI (EZVI), a technology developed to enhance the degradation of DNAPLs in source zones. EZVI consists of nanoscale (or microscale) iron particles that are suspended in water and encapsulated within a liquid oil membrane droplet. The droplet, O’Hara said, consists of food-grade surfactant and biodegradable vegetable oil. The technology—developed by the University of Central Florida (UCF) and NASA—promotes DNAPL degradation by enhancing contact between the DNAPL and ZVI.

O’Hara said that ZVI must be in the presence of water to promote active reductive dehalogenation. As a result, when ZVI is injected into the subsurface as a powder, it can only treat dissolved-phase contaminants at the edges of a DNAPL. In contrast, when the ZVI is encapsulated in a hydrophobic emulsion droplet, the DNAPL and the ZVI are able to interact because the DNAPL is miscible with the droplet: chlorinated VOCs in the DNAPL diffuse through the droplet’s oil membrane, contact the aqueous ZVI, and undergo reductive dechlorination. In addition to EZVI’s abiotic remedial activities, O’Hara said, the droplet’s vegetable oil and surfactant serve as long-term electron donors and promote anaerobic degradation. Also, O’Hara said, although EZVI was developed specifically to treat DNAPLs, the technology will also treat contaminants in dissolved-phase plumes if these contaminants come into contact with the emulsion droplets.

GeoSyntec, UCF, and NASA completed a field-scale demonstration of the technology at NASA’s Launch Complex 34 (LC34) site, O’Hara said, noting that the results were independently evaluated by Battelle. Hydraulic controls were used to ensure that contaminants in the 15-foot by 10-foot pilot area were contained and to maintain consistent ground-water velocity in the treatment zone. During the demonstration project, about 750 gallons of EZVI were injected into the subsurface and the effects were examined by (1) comparing pre-treatment soil cores with post-treatment soil cores and (2) evaluating ground water data. Pressure pulse technology (PPT) was used to inject the EZVI into the subsurface, O’Hara said. PPT was chosen over other injection technologies because the research team believed it would produce a more even distribution of emulsion by (1) causing instantaneous dilation of the pore throats in the porous media and (2) increasing fluid flow and minimizing the “fingering” effect that occurs when a fluid is injected into a saturated medium. Although PPT did succeed in distributing EZVI through a portion of the test zone, O’Hara said, the emulsion tended to migrate to shallow areas and it did not spread as widely or as homogeneously as originally hoped. Nevertheless, the pilot project did meet its primary objective, which was to remove at least 50 percent of the total TCE present in the test area. In fact, according to soil core sampling results, the average TCE reduction rate across the entire test area was 58 percent. More significant reductions (i.e., greater than 80 percent) occurred in areas where EZVI was actually confirmed as being present. The ground-water data told a similar story: significant TCE reductions (60 to 100 percent) were observed at target depths and mass flux was reduced by 56 percent.

O’Hara concluded by noting that GeoSyntec and NASA plan to perform more research on EZVI technology. For example, research will be performed to determine what percentage of EZVI-induced contaminant degradation is attributable to abiotic versus biotic degradation pathways. In addition, in the fall of 2003, a project was initiated to investigate four injection methods, the goal being to find ways to improve EZVI delivery. O’Hara said that the PPT approach will be modified and included in the evaluation, along with hydraulic fracturing, pneumatic fracturing, and direct injection. While it is possible that all of these methods will work, researchers might find that one method works better than another at a specific site.

At the conclusion of her presentation, O’Hara took questions from audience members. In response to their queries, she made the following points: (1) the iron in EZVI, which is shipped from Japan, costs about $12 per pound; (2) no offgassing from the emulsion has been observed; and (3) some consideration is being given to freezing the ground before attempting the pneumatic hydraulic fracturing injection approach.

Source Zone Treatment Using Injection of ZVI Into a Fractured Rock Aquifer
John Liskowitz, ARS Technologies, Inc.

John Liskowitz, whose presentation is included as Attachment O, (PDF, 825KB) provided information about a source zone remediation technology called Ferox^SM. This technology, which involves injecting a liquid atomized ZVI powder into the subsurface, can be used to remediate deep contaminants, as well as contaminants located under structures and utilities. Liskowitz described how the technology works. First, dry microscale ZVI powder is fed into a colloidal mixer, blended with water, and kept in suspension through rapid circulation. The resulting slurry is then fed into a pump and blended with nitrogen gas before being injected into a contaminated source area. The nitrogen gas atomizes the iron-water mixture, which facilitates distribution in the subsurface. Once in the subsurface, the ZVI remediates contaminants. The ZVI used in the Ferox^SM process, a sponge powder that is certified (by FDA) as being 95 percent pure, does have some trace carbon and other impurities. The latter, Liskowitz said, might actually enhance the reactivity of the iron. The iron costs about $1.10 to $1.70 per pound.

Liskowitz said that the Ferox^SM process has been piloted at Hunter Point, a Brownfields site located in San Francisco, California. This site, established in 1869 as the first dry dock on the Pacific Coast, was operated by the Navy between 1940 and 1976 and was used to build ships and maintain, service, and test submarines. Six land parcels (referred to as Parcels A through F) have been identified as requiring cleanup. The Navy is eager to address the contamination as quickly as possible because much interest has been expressed in turning the land over to the City of San Francisco for redevelopment and revitalization. For this reason, the Navy agreed to test the Ferox^SM process to determine whether it can reduce the TCE/PCE source area underlying Parcel C in a cost-effective manner. In addition to measuring mass reduction, Liskowitz said, the following were identified as key goals for the pilot project: (1) determine whether Ferox^SM mobilizes DNAPL or displaces dissolved-phase plumes, and (2) measure the treatment radius achieved. Liskowitz said that the geology at Parcel C is complicated, noting that a 10-foot layer of artificial fill overlies fractured bedrock. He also noted that contaminant concentrations in the subsurface are high. For example, before the pilot project was performed, TCE concentrations in ground water exceeded 88,000 parts per billion (ppb).

More than 16,000 pounds of ZVI were introduced into the source zone area during the pilot project, Liskowitz said. The iron-water mixture was injected into four boreholes to treat soil and ground-water contamination in areas extending from the ground-water table (encountered about 8 feet bgs) to about 30 feet bgs. During the installation process, Liskowitz said, the research team proved that Ferox^SM can be safely applied within and adjacent to structures. One difficulty was encountered, however: in the shallow areas, the iron-water mixture tended to rise. One can overcome this problem, Liskowitz said, by incorporating pulse injection. Monitoring, performed about 3 weeks after injection, revealed some dramatic findings. Most notably, target chlorinated VOCs had been reduced by 95 to 99 percent, and the treatment achieved a radius of influence of about 15 feet. The results also showed little or no evidence of contaminant displacement. (Contaminants did not appear in significant quantities in the four wells that had been established outside of the test area or in the two wells screened below the test area.) Although significant contaminant mass reductions were achieved, Liskowitz said, MCLs have not been achieved at this site. Some consideration is being given to injecting more iron-water mixture in an effort to achieve further reductions.

In Situ Cr(VI) Source and Plume Treatment Using a Ferrous Iron Based Reductant
Ralph Ludwig, EPA

Ralph Ludwig, whose presentation is included as Attachment P, (PDF, 1.13MB) talked about remediation efforts at the Macalloy Corporation Superfund site, a former ferrochrome production facility located in Charleston, South Carolina. During its years of operation, a variety of wastes, including slag, conditioning tower sludge, and electrostatic precipitator dust, were disposed of at the site. As a result, large quantities of this chromite ore processing residue (COPR) persist at the site and a large dissolved-phase Cr(VI) plume is migrating toward a nearby tidal area. Ludwig provided details about the site’s source zone area, noting that it includes approximately 20 acre-foot of saturated zone contaminated sediments and ground water. The sediments are characterized as follows: (1) average total chromium concentrations are about 3,500 ppm, (2) maximum Cr(VI) concentrations are about 550 ppm, (3) hydrosulfite (dithionite) reducible iron is present at low levels; and (4) magnesium hydroxide content is high, which gives the sediments a strong pH buffering capacity. As for the ground water in the source area, Ludwig said, it is characterized by Cr(VI) concentrations ranging from 3.0 ppm to 57 ppm, pH values ranging from 8.5 to 11.5, dissolved oxygen concentrations less than 1.0 ppm, and relatively high dissolved solids.

Batch laboratory experiments were performed in an effort to identify a reductant that could be used to treat the Cr(VI) at this site. Ludwig said they found that sodium dithionite was ineffective, but that ferrous salts, such as ferrous sulfate and ferrous chloride, were highly effective. While the latter finding was accepted as good news, they were immediately confronted with a problem: delivering ferrous salts to the subsurface is normally impractical, particularly under conditions of high pH, because iron tends to precipitate rapidly out of solution and cause clogging. To address this problem, Ludwig said, they combined the ferrous salts with sodium dithionite (which acts to stabilize ferrous iron and thereby serve as a carrier fluid) to examine whether this combination would effectively treat Cr(VI). The results were positive and indicated that the “ferrous sulfate-sodium dithionite” combination was better than the “ferrous chloride-sodium dithionite,” Ludwig said: the former, in addition to being an effective treatment solution, was better at reducing Eh and pH values. Ludwig provided some background information about the chemistry involved in the treatment reaction. When ferrous iron reacts with Cr(VI), the chromium is converted into a stable trivalent form.

Once a reductant had been selected for the site, pilot projects were initiated to treat:

• The source zone. To determine whether the chosen reductant blend is capable of treating Cr(VI) in the COPR source area, researchers injected about 4,800 gallons of a 0.2 molar (M) ferrous sulfate/0.2M sodium dithionite blend into the source zone in 2002. The materials, injected over about 5 hours, were introduced to the subsurface through 2-inch PVC wells (screened 7.5 to 15 feet bgs) at an average flow rate of 15 gallons per minute and an average injection pressure of 10 psi. Ground-water and solid-phase samples were collected before and after the injection. Although the results are still being analyzed, preliminary findings suggest that the ferrous iron disseminated out to a radius of more than 8 feet, clogging was not a problem, and dissolved-phase and solid-phase Cr(VI) were effectively treated.
  
  
• The dissolved-phase Cr(VI) plume. Ludwig said they were also interested in seeing whether the reductant blend could help address the site’s dissolved-phase Cr(VI) plume. Specifically, they wanted to know: (1) Can ferrous iron (in the presence of sodium dithionite) treat Cr(VI) in a dissolved plume? (2) Will the hydraulic conductivity of the formation be adversely impacted by the injection of ferrous iron? and (3) How long will the ferrous-iron-enriched zone remain reactive? In an effort to answer these questions, in 2002, about 2,600 gallons of 0.2M ferrous sulfate/0.2M sodium dithionite blend were injected into the pathway of the dissolved-phase Cr(VI) plume. The materials were introduced to the subsurface through two 2-inch injection wells (screened at 10 to 15 feet bgs and spaced 8 feet apart) at an average flow rate of 15 gallons per minute and an average injection pressure of 10 psi. Ground-water and solid-phase samples were collected before and after the injection. The results suggest that: (1) an effective ground-water treatment zone was established, (2) no well or aquifer formation clogging occurred, and (3) some metals were mobilized but remained confined to the boundaries of the ferrous iron treatment zone.

SESSION IV: NOVEL PRB APPLICATIONS
Session Chair: Tom Krug, GeoSyntec Consultants

Arsenic Removal From Ground Water Using a PRB of Basic Oxygen Furnace (BOF) Slag at DuPont’s East Chicago Site
John A. Wilkens, DuPont

John Wilkens, whose presentation is included as Attachment Q, (PDF, 636KB) provided information about a PRB installed at DuPont’s East Chicago site. The PRB is unique in that it is the first full-scale PRB that has ever used basic oxygen furnace (BOF) slag as a reactive medium to remove arsenic from ground water. Wilkens provided background information, noting that hundreds of different chemicals have been manufactured at the East Chicago site over the years and that an arsenic plume has formed underneath the land parcel. Arsenic concentrations, which have been detected as high as 1 to 2 ppm, increase with depth and have been recorded at depths exceeding 25 feet.

In an effort to identify an economical way to contain the ground-water plume, DuPont started researching the possibility of using a PRB at the site. As a first step, DuPont performed laboratory tests on a variety of commercially available iron byproducts to determine which ones effectively treat arsenic. Many of the products proved to be ineffective, but promising results were obtained with iron filings and BOF slag, both of which reduced arsenic concentrations to below 10 ppb. Although the laboratory tests provided useful information, DuPont felt it was necessary to test the byproducts under actual field conditions before making a final decision about which to use as a reactive medium. Thus, DuPont installed several reactive test wells and used them to examine the efficacy of ZVI, BOF slag, and iron oxide in the field. The BOF slag proved to be the best performer, so DuPont chose it.

Attachment Q (PDF, 636KB) presents detailed information about the PRB design specifications and the installation methods that were used to emplace the system. In summary, the PRB, which consists of 100 percent BOF slag, is designed to last for 20 years and to reduce arsenic concentrations in the 1 to 3 ppm range to levels below 10 ppb. The PRB, which took 3 months to install, is 2,000 feet long and 35 feet deep. It consists of two parallel walls, Wilkens said, noting that 15 feet lie between the centers of each wall. Installation proved to be a challenge because the site was characterized by a high water table and flowing sands. These factors, along with the fact that BOF slag has a high pH, made it necessary to rule out bioslurry injection as an installation method. As an alternative, trenching was chosen as the installation method and a phased construction approach was used to overcome the 25-foot depth limitation associated with this installation methodology. Deep trenching and extensive dewatering (1,000,000 gallons of water were removed daily) made it possible to install the PRB with a trenching machine. During the installation process, about 26,500 cubic yards (43,000 tons) of BOF slag were emplaced in the subsurface. Wilkens indicated that he was impressed with the contractors who installed the PRB, noting that they completed their work a week ahead of schedule, used innovative approaches to address challenges, and practiced solid health and safety procedures.

Wilkens said that a post-installation monitoring program has been initiated for the East Chicago site. DuPont and the University of Waterloo are working together to collect data and analyze the findings. The first set of results will be presented to the public in spring 2004.

Waste Green Sands as Reactive Media for PRBs
Craig Benson, University of Wisconsin-Madison

Craig Benson, whose presentation is included as Attachment R, (PDF, 256KB) discussed the possibility of using waste green sands as a reactive medium for ground-water treatment. This material is a byproduct of the gray iron industry, an industry that produces radiators, manholes, and a variety of other products. Benson said that foundries use green sand to cast products into their desired shapes. The process works as follows: green sand is pressed into a mold, a molten metal is poured into the mold, the metal is allowed to harden, and the green sand is stripped off the finished product and reused. Green sand is made up of sand, as well as small amounts of binding material (e.g., bentonite), water, and an organic additive (e.g., coal dust). About 2 to 12 percent of waste green sand consists of iron particles that adhere to the sand during the metal molding process. Additional sand and other materials are constantly added to the green sand mixture to make sure it has the correct composition. Excess sand is generated in the process. When it becomes too cumbersome, the foundries send the excess sand to landfills. For this reason, Benson said, some people call green sand a waste material even though it is not truly a waste product.

The iron and the organic additive in green sand give it the potential to serve as a reactive medium for ground-water treatment. While iron can react with contaminants, organic additives provide sorptive capacity. Benson said that there are several compelling reasons to consider using green sand as a remedial agent. For starters, green sand is economical: foundries are happy to give away green sand for free and are sometimes even willing to transport and deliver the material at no charge. Also, green sand is easy to handle. In addition, Benson said, identifying ways to reuse an industrial byproduct supports the sustainable development paradigm, a concept that warrants more attention in North America.

Given these benefits, researchers decided to perform laboratory experiments on green sand to assess its potential as a treatment agent. Their objectives were to (1) assess the hydraulic conductivity, reactivity, and sorptive capacity of green sands; (2) evaluate the material’s long-term reactivity; (3) evaluate potential field scenarios; and (4) determine whether metals and polycyclic aromatic hydrocarbons are likely to leach out of green sand. Toward this end, batch tests and column studies were performed to examine the properties of a conventional ZVI (Peerless iron) and 12 green sands that were donated by foundries across the Midwest region. As part of this effort, researchers examined how effective these materials were at treating chlorinated herbicides (e.g., alachlor, acetyl alachlor, and metolachlor) as well as TCE and some of its degradation products. The results suggest that (1) the hydraulic conductivity of green sands varies across foundry sources, (2) green sands have high sorptive capacity for TCE and chlorinated herbicides, (3) the normalized rate constant for TCE degradation in the presence of iron particles that were extracted from green sand was comparable to the rate observed with Peerless iron, (4) the long-term reactivity of green sand appears to be comparable to that of Peerless iron, (5) rate constants and partition coefficients obtained from the batch tests were comparable to those observed with the column tests, (6) analytical modeling shows that reactive barriers containing green sand have the potential to treat contaminated ground water in the field, and (7) metals did leach out of the green sand but the leaching was less pronounced than that observed with natural soils or conventional ZVI.

Impacts of a ZVI PRB on Downgradient Biodegradation Processes
John E. Vidumsky, DuPont

John Vidumsky, whose presentation is included as Attachment S, (PDF, 148KB) opened by noting that residual contamination is frequently observed downgradient of ZVI PRBs, and that these residuals result from steady-state processes as well as transient processes (e.g., desorption of contaminants from downgradient aquifer solids and reverse diffusion of contaminants from stagnant/low-permeability areas). Expanding on the steady-state processes, Vidumsky said that contaminants can be found in treated PRB effluent for the following reasons: (1) some contaminants (e.g., 1,2-DCA) pass right through the wall because ZVI is ineffective against them, (2) some stable daughter products—which become contaminants in their own right—form when parent compounds pass through the ZVI wall, and (3) some parent compounds are not completely treated as they pass through the wall. The bottom line, Vidumsky said, is that contaminant concentrations can persist downgradient of PRBs for a long time. Given this reality, it is important to identify approaches that address residual contamination. Vidumsky said that biodegradation (both through natural means and through engineered enhancements) has the potential to do so. He spent the remainder of his presentation describing how biological processes remediate chlorinated compounds and explaining how ZVI barriers actually help create conducive conditions for biodegradation.

Vidumsky said that reductive dechlorination breaks down chlorinated compounds. The process, mediated by microbes, involves transferring electrons from an electron donor to an electron acceptor, Vidumsky said. (Chlorinated compounds are an example of the latter.) Hydrogen and reduced organic acids, such as acetate, formate, and pyruvate, are important electron donors. In addition, some complex organic substrates (e.g., lactate and benzoate) and even some chlorinated compounds (e.g., methylene chloride and vinyl chloride) can generate electron donors (formate and acetate) when they undergo fermentation. Vidumsky noted that different microbes promote different processes. For example, while some microbes use electron donors to support reductive dechlorination, others use electron donors to promote methanogenesis or sulfate reduction. All of the microbes compete with one another for the electron donors. The biodegradation process that prevails, Vidumsky said, is determined by the makeup of the microbial community, a factor that is in turn determined by environmental conditions (e.g., geochemistry and the mix of contaminants present).

Vidumsky said that ZVI PRBs can alter environmental conditions in the aquifer, cause shifts in microbial community structure, and promote contaminant biodegradation in downstream areas. They do so by:

• Creating geochemical conditions that are conducive to anaerobic reductive dechlorination. Vidumsky noted that ZVI has the potential to eliminate oxygen and nitrates, both of which compete with chlorinated compounds as electron acceptors.
  
  
• Produce hydrogen and other electron donors. Vidumsky said that hydrogen (an electron donor) is generated when iron corrodes. Thus, ZVI provides a direct source of hydrogen. In addition, when ZVI reacts with certain contaminants, such as carbon tetrachloride and chlorofluorocarbon 11, two different reactions take place: a portion of the contaminant is biodegraded and transformed into a daughter product, and the remainder is converted to an unstable radical that later turns into acetate (an electron donor). In some cases, the daughter product ferments and yields even more electron donors in the process. The addition of all these electron donors promotes further reductive dechlorination, sulfate reduction, and methanogenesis.
  
  
• Reduce overall contaminant loading. By reducing contaminant load in downgradient areas, Vidumsky said, PRBs help ensure that the contaminants do not exceed the aquifer’s attenuation capacity.
  
  
• Remove “inhibitory” compounds. PRBs have the potential to eliminate certain compounds, such as chloroform, that inhibit the reductive dechlorination of chlorinated ethenes.
  
  
• Change the mixture of contaminants. The mix of contaminants found upgradient of a PRB differs from the mix found downgradient, Vidumsky said, and this has a profound effect on the type of microbial community on each side of the PRB. He listed the Gibbs Free Energy values associated with a variety of dechlorination reactions. Certain reactions yield more energy than others—for example, while 53 kilojoules per mole are generated when TCE is converted to 1,2-DCE, only 38.3 kilojoules per mole are generated when 1,2-DCE is converted to vinyl chloride. Microbes that use higher-energy electron acceptors outcompete those that use lower-energy electron acceptors. As a result, 1,2-DCE will probably not be converted to vinyl chloride if TCE is available as an electron acceptor. By removing TCE from the contaminant mix, however, PRBs tend to favor microbes that break down 1,2-DCE.
  
  
• Convert parent contaminants to more biodegradable daughter products. Vidumsky noted that some fully halogenated parent compounds, like PCE and carbon tetrachloride, can only be biodegraded through the reductive dechlorination pathway. If the PRB succeeds in partially dechlorinating these parent compounds, the resulting daughter products can be treated through a variety of biological pathways, including reductive dechlorination, anaerobic oxidation, cometabolic processes, aerobic oxidation, and fermentation.

Vidumsky closed by encouraging attendees to start thinking of PRBs and biodegradation as part of an integrated remediation system design. If a site’s baseline attenuation capacity is not sufficient, he said, consideration should be given to applying biostimulation or bioaugmentation technologies in areas downgradient of the PRB.

One attendee asked whether increasing biodegradation rates could lead to biofouling, a phenomenon that can reduce subsurface permeability and thus threaten the efficacy of a PRB. Vidumsky said that problems of this nature have not been observed in the field.

Delivery Approaches for Groundwater Amendments
Tom Krug, GeoSyntec Consultants

Tom Krug’s presentation, included as Attachment T, (PDF, 2.3MB) picked up where Vidumsky’s presentation left off—exploring the topic of biostimulation or bioaugmentation. He introduced the concept of biobarriers, biocurtains, and biologically active zones: these systems are similar to PRBs in that they are established as subsurface zones that remediate contaminants as ground water flows through them. Krug focused his talk on ground-water amendments that promote biodegradation, but he did pause to remind audience members that nanoscale iron, which promotes abiotic degradation, is also a useful ground-water amendment. As for amendments that stimulate or augment biological activity, Krug said, remediators have experimented both with injecting microorganisms and electron donors to the subsurface. (Krug presented a list, which is included in Attachment T, (PDF, 2.3MB) of the different types of electron donors that can be used as ground-water amendments.) Although the concept of enhanced biodegradation has come a long way, implementation remains a challenge, Krug said. It can be difficult to distribute ground-water amendments uniformly throughout the subsurface. Delivery and mixing are issues that still warrant attention.

Krug said that several different delivery approaches can be used, including temporary probes, injection wells, injection/extraction wells, and trenches. He focused his talk on the first two methods, and broke the delivery approaches into three categories: passive, semi-passive, and active. With the passive approach, he said, ground-water amendments are added to injection points and allowed to migrate downward, sometimes with the assistance of water flushing. With the semi-passive approach—a system that uses injection and extraction wells—temporary or intermittent recirculation is used to drag amendments across the contaminated plume. The active approach, which does indeed lead to the broadest dispersal patterns, is an aggressive approach that involves extracting ground water, adding amendments to it, and reinjecting ground water back into the subsurface. Krug said that there are pros and cons with each approach, and that one method might work better than another at a specific site. Krug said that all three approaches have been used in the field, and he presented a case study for each:

• Case study #1—passive injection. A passive injection system has been used to introduce lactate, soybean oil, food-grade emulsions, and Dehalococcoides into the subsurface of an industrial site in California. These amendments were added to promote the biodegradation of TCE, a contaminant detected in the site’s ground water at depths ranging from 15 to 35 feet. Many injection wells were used. They were spaced 15 feet apart from each other. Water was flushed down each injection well in an effort to promote distribution of the amendments. (About 10,000 gallons were flushed down each well.)
  
  
• Case study #2—semi-passive injection. The semi-passive injection system has been used to introduce methanol and sodium lactate into the subsurface of an industrial site in Massachusetts. These amendments were added to promote the biodegradation of TCA, a contaminant detected in the site’s ground water at depths ranging from 25 to 35 feet. The delivery system consisted of an injection well flanked by extraction wells on either side. The extraction wells were about 20 feet away from the injection well. Intermittent pumping (8 hours per week at 4 gallons per minute) was performed to promote distribution of the amendments. This delivery system helped drag the amendments through the contaminated ground-water plume. Krug said that methanol was detected in the extraction wells, which proves that this amendment traveled at least 20 feet. Lactate was not detected in extraction wells, but this was probably because it broke down into acetate quickly. Krug said that the amendments do appear to be successful, noting that TCA degradation has been observed.
  
  
• Case study #3—active recirculation. Krug said that active recirculation has been used to introduce ethanol to the subsurface of a rocket manufacturing site in California. This amendment was added to promote the degradation of perchlorate and TCE, contaminants that have been detected in the site’s sand aquifer down to depths of 100 feet. The delivery system consisted of an injection well with extraction wells situated 200 feet away on either side.

Krug also provided information about some upcoming work, noting that two In Situ biodegradation approaches—an active biobarrier and a semi-passive biobarrier—will be investigated under the Environmental Security Technology Certification Program (ESTCP). The project is being performed to gain a better understanding of the effectiveness of each approach and site conditions that would favor one approach over another in remediating perchlorate-impacted ground water.

Krug concluded with the following points: (1) a variety of approaches can be used to introduce ground-water amendments; (2) the best method to use at a particular site will vary based on depth, plume dimensions, water quality issues, and other site characteristics; (3) passive injection systems are often suitable to use when target areas are shallow; and (4) semi-passive or active methods may be more appropriate at sites that have deeper ground-water contamination and at sites where it is critical to establish tight control over amendment distribution patterns.

Remediation of Perchlorate, NDMA, and Chlorinated Solvents in Ground Water Using Nanoscale ZVI
Neal Durant, GeoSyntec Consultants

Neal Durant, whose presentation is included as Attachment U, (PDF, 443KB) talked about experiments that have been performed to determine whether nanoscale ZVI can be used to remediate areas contaminated with perchlorate, chlorinated solvents, and N-nitrosodimethylamine (NMDA). Durant provided an overview of nanoscale ZVI, describing its properties, treatment potential, and developmental status. He said that nanoscale iron particles, which are typically less than 100 nanometers in size, have a high specific surface area and treat contaminants at a rate 10 to 100 times faster than is observed with conventional ZVI powder. In addition, nanoscale iron is easy to deploy and does not have the same depth restrictions that granular ZVI does.

Durant talked briefly about the factors that have prompted researchers to start focusing attention on nanoscale ZVI as a potential reactive medium. To set the stage, he noted that perchlorate plumes are gaining national attention and are increasingly being viewed as a serious threat to drinking water. He also said that perchlorate plumes often contain NMDA and chlorinated solvents, and that few conventional remediation technologies can treat all of these contaminants simultaneously. While researchers have shown that granular ZVI can abiotically degrade all of them, granular ZVI exhibits a serious limitation: it does not degrade perchlorate at rates that are fast enough to warrant its use as a remedial agent. Durant said that researchers are hopeful that nanoscale ZVI, with its higher rate of reactivity, will overcome this limitation and prove to be an effective in situ remediation technology for addressing perchlorate. To determine whether this is the case, Durant said, researchers have launched a technology demonstration project at an aerospace site in Sacramento, California (hereinafter “the Site”). The project, which will be executed in two phases, consists of laboratory studies and a field demonstration. Researchers are currently in the middle of Phase I—the laboratory portion of the project.

Durant described Phase I, the goals of which are to (1) confirm treatability, (2) determine site-specific nanoscale ZVI loading requirements, (3) estimate site-specific transformation rates, (4) determine treatment capacity under site-specific conditions, and (5) measure retardation and transport of nanoscale ZVI in the aquifer column. As part of this effort, researchers are evaluating two different types of nanoscale ZVI: one produced by Lehigh University and one produced by the Japanese company Toda Kogyo Corporation. (These two products differ in some aspects. For example, the Toda product has a higher oxide content than the Lehigh material, and the particles in each product differ in shape. While the Lehigh particles are roughly spherical, the Toda particles are more angular.) Attachment U (PDF, 443KB) provides detailed information about the Phase I results collected thus far. (The results are still considered preliminary.) In summary, the following has been observed:

• Nanoscale ZVI does appear to be an effective treatment agent for perchlorate, TCE, and NMDA. Perchlorate reduction is the rate-limiting step for this mixture of compounds.
  
  
• The optimal nanoscale ZVI particle loading requirements for the Site are between 2 and 5 grams per liter.
  
  
• In these trials, the Lehigh nanoscale ZVI degraded chloroethenes and nitrate 2 to 3 times faster than the Toda nanoscale ZVI. Durant said that this is likely due to the fact that the former has fewer oxide impurities than the latter.
  
  
• Perchlorate degradation was preceded by a lag period for both types of nanoscale ZVI, although the lag period was shorter for the Toda material. Durant said that the cause of the lag period was not determined in this study, but he hypothesized that the lag period was due to an adsorption step that precedes dechlorination. He further speculated that the shorter lag period for the Toda material was due to the presence of iron oxides on the nanoparticle surfaces. In support of these hypotheses, Durant referenced the work of Moore et al. (2003), who reported that mixed valence states (i.e., Fe[0] and Fe[III]) enhance the adsorption and abiotic degradation of perchlorate by ZVI. Durant acknowledged that biodegradation (e.g., by autotrophic mechanisms) may have contributed to removal of perchlorate in the trials with the Lehigh material. In the reactors amended with Toda material, however, perchlorate degradation was not affected by the addition of biocide (250 mg/L of mercury chloride), indicating an abiotic reduction mechanism.

Durant also talked briefly about Phase II, the field portion of the technology demonstration. It is scheduled to start in summer 2004; its goal will be to evaluate nanoscale ZVI delivery and performance in both an active recirculation cell and a passive PRB configuration.

Durant concluded by stressing that much more research is needed to determine whether nanoscale ZVI is a feasible and cost-effective remediation tool. For example, research still needs to be performed in the following areas: (1) cost efficacy and the development of a more affordable nanoscale ZVI product, (2) particle transport, (3) material longevity, (4) the relationship between particle shape and transport characteristics, and (5) optimal modes of particle delivery. In addition, Durant said, before nanoscale ZVI technology can be endorsed as a viable option for perchlorate and chloroethene remediation, it should be subjected to rigorous cost and performance comparisons with In Situ bioremediation approaches.

Electrolytic Reactors (e-barriers) for the In Situ Treatment of Contaminated Ground Water
Matthew Ballaban, University of Waterloo
Tom Sale, Colorado State University

Matthew Ballaban and Tom Sale provided information about e-barriers, a technology that is still in its early development stages. Their presentation is included as Attachment V. (PDF, 1.1MB) Ballaban, who kicked off the presentation, explained the theory behind the technology and presented some field and laboratory data. He said that e-barriers work as follows: electrodes are installed in the subsurface, they are powered by an electrical source, they alter in situ oxidation and reduction conditions, chlorinated compounds are dechlorinated in the process, and hydrogen gas and oxygen are generated as well. Starting in 1998, Ballaban said, proof-of-concept experiments have been performed to determine whether the process really works. In the fall of 2001, the technology was actually tested in the field. The main goals of the field experiment were to (1) gain a better understanding of the electrode materials, (2) determine what voltage is required for dechlorination reactions to occur, (3) identify and evaluate techniques that can be used to manage precipitates, (4) examine overall treatment performance, and (5) investigate the upgradient and downgradient impacts of the e-barrier. To examine these issues, a prototype e-barrier was installed 50 meters downgradient of a PCE/TCE source zone at the CFB Borden site and operated at a range of different voltages. The e-barrier consisted of three electrodes that were surrounded by an HDPE Geonet mesh and encased within an HDPE geotextile fabric. (The first and second electrode operated in an anode-cathode sequence. The third electrode, a temporary cathode, was only in operation during a polarity reversal exercise that was used to clean the primary cathode [i.e., the second electrode].) To examine the e-barrier, the researchers collected water samples using a multilevel sampling network; monitored PCE, TCE, and breakdown products upgradient and downgradient of the e-barrier; and calculated removal rates at operating voltages of 0, 5, 7 and 10 volts. Much of the data could be accessed remotely, Ballaban said: the e-barrier was outfitted with a data logger, a cell phone, and an antenna to enable researchers to view data from the comfort of their own desks.

Attachment V (PDF, 1.1MB) provides detailed information about the results that were obtained from the field study. In summary, researchers concluded the following:

• Large oxidation and reduction potentials were recorded near the electrodes, with voltage in excess of 1.7 volts observed at the anode and -2.8 volts recorded at the cathode. The system was operated for $0.93 per day per square meter.
  
  
• Through physical observations, researchers concluded that the first and third electrodes were free of precipitate buildup. According to the results of a bromide tracer test, the second electrode, which is hidden from view, does not have significant mineral precipitate buildup either.
  
  
• E-barriers homogenize contaminants. For example, although the PCE plume entered the e-barrier as a vertically stratified plume, it exited as a vertically mixed plume. To be more specific, upgradient of the e-barrier, PCE concentrations ranged from 100 ppb to about 2,500 ppb, with concentrations increasing with depth. Upon exiting the e-barrier, however, PCE concentrations were about 300 ppb across the entire vertical profile. Ballaban said that researchers believe that the mixing is facilitated by the hydrogen and oxygen gas that is produced at the electrodes.
  
  
• PCE and TCE concentrations were consistently reduced as ground water flowed through the e-barrier. As the ground water moved through the e-barrier, Ballaban said, the system achieved removal rates of 85 percent and 34 percent for PCE and TCE, respectively. Concentrations were not reduced below MCLs, however. Thus, Ballaban recommended performing additional laboratory and field experiments before taking this technology full scale.
  
  
• Data suggest that biodegradation processes occur downgradient of the e-barrier, and that the hydrogen gas that is produced at the electrodes plays a role in enhancing microbial dechlorination. Ballaban noted that PCE concentrations continued to decrease with distance from the e-barrier, but that the opposite was true for TCE. Also, he noted that cis-DCE—a compound that appears when TCE biodegrades—was observed downgradient but not upgradient. All of these findings support the idea that biodegradation processes were at work downgradient of the PRB. With this information in hand, Ballaban said, researchers decided to perform microcosm studies in the laboratory to test the following hypothesis: in the presence of excess hydrogen (which is generated at the e-barrier’s electrodes), indigenous organisms degrade PCE to TCE to cis-DCE. The microcosm study results did support this hypothesis.

Tom Sale briefly discussed other cutting-edge projects that are being performed to assess e-barriers. Under the ESTCP program, for example, a large e-barrier (consisting of 17 panels) has been installed to address a TCE plume at the F.E. Warren site. Results from this study will be released in summer 2004. In addition, Sale said, the Army Corps of Engineers and the Strategic Environmental Research and Development Program are funding studies to determine whether e-barriers can be used to remediate energetics, such as RDX and TNT. The results collected thus far are promising, Sale said, and efforts are underway to identify a site for a field demonstration.

CLOSING REMARKS

Vidumsky and Puls thanked the speakers and the attendees for their participation and said that they thought the meeting had been very informative. They also thanked those who had helped plan the meeting, and expressed appreciation to EnviroMetal Technologies, Inc., and GeoSyntec for providing refreshments. In closing, they encouraged attendees to provide feedback on the meeting and suggestions (in terms of dates, locations, and topics) for future meetings..

Attachments A through V

Attachment A: Final Attendee List (PDF, 50KB)

Attachment B: Influence of Sewers on Ground-Water Flow in the Vicinity of a PRB in an Urban Setting (William M. Goodman) (PDF, 774KB)

Attachment C: Hydraulic Pulse Testing of PRBs (Grant Hocking) (PDF, 866KB)

Attachment D: Modeling of Downgradient Reverse Diffusion Effects (Tom Sale) (PDF, 1.2MB)

Attachment E: Evaluation of Ground-Water Hydraulics at a Zero-Valent Iron (ZVI) PRB at the Somersworth Superfund Site (Suzanne O’Hara) (PDF, 768KB)

Attachment F: Biopolymer Liquid Shoring: General Characteristics and Use (Lloyd Marsden) (PDF, 1.41MB)

Attachment G: Field Monitoring of the Performance of a PRB at the Vapokon Site, Denmark (Irene M.C. Lo) (PDF, 2.8MB)

Attachment H: Impact of Mineral Fouling on the Long-Term Performance of PRBs (Craig Benson) (PDF, 1.95MB)

Attachment I: Implications from Long-Term Monitoring of Two ZVI Reactive Barriers in Germany (Markus Ebert) (PDF, 881KB)

Attachment J: Mainstreams and Lessons Learned at Nine German PRB Sites Over Five Years—An Interim Report (Volker Birke) Part 1 (PDF, 1.3MB) and Part 2 (PDF, 1.42MB)

Attachment K: Long-Term Performance of PRBs: Lessons Learned (Bob Puls) (PDF, 790KB)

Attachment L: Laboratory Tests and Field Investigations of DNAPL Source Zone Remediation Using Granular Iron (Sharon L.S. Wadley) (PDF, 990KB)

Attachment M: Use of ZVI and Clay for Source Zone Remediation (Michael Liberati) Part1 (PDF, 1.15MB), Part 2 (PDF, 2MB), and Part 3(PDF, 1.46MB)

Attachment N: Emulsified ZVI Treatment of Chlorinated Solvent DNAPL Zones (Suzanne O’Hara) (PDF, 700KB)

Attachment O: Source Zone Treatment Using Injection of ZVI Into a Fractured Rock Aquifer (John Liskowitz) (PDF, 825KB)

Attachment P: In Situ Cr(VI) Source and Plume Treatment Using a Ferrous Iron Based Reductant (Ralph Ludwig) (PDF, 1.13MB)

Attachment Q: Arsenic Removal From Ground Water Using a PRB of Basic Oxygen Furnace (BOF) Slag at DuPont’s East Chicago Site (John A. Wilkens) (PDF, 636KB)

Attachment R: Waste Green Sands as Reactive Media for PRBs (Craig Benson) (PDF, 256KB)

Attachment S: Impacts of a ZVI PRB on Downgradient Biodegradation Processes (John E. Vidumsky) (PDF, 148KB)

Attachment T: Delivery Approaches for Groundwater Amendments (Tom Krug) (PDF, 2.3MB)

Attachment U: Remediation of Perchlorate, NDMA, and Chlorinated Solvents in Ground Water Using Nanoscale ZVI (Neal Durant) (PDF, 443KB)

Attachment V: Electrolytic Reactors (e-barriers) for the In Situ Treatment of Contaminated Ground Water (Matthew Ballaban and Tom Sale) (PDF, 1.10MB)

FOOTNOTES

1 PRB field data were included in the analysis because PRBs—which in many cases reduce contaminant concentrations to nondetect values in the areas immediately downgradient of the wall but cannot produce the same results further downstream—serve as a good analog for studying souce zone cleanup.

2 The cell was contaminated in 1991, when 771 liters of PCE were purposely released for research purposes. Since that time, several remediation technologies, including potassium permanganate, have been used to remove PCE from the cell. By 1998, the year when the iron-bentonite walls were installed, about 200 liters of PCE were still in the aquifer and about 150 liters remained in the aquitard.