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

Sheraton Oceanfront Hotel
Virginia Beach, Virginia
September 18­19, 1997

WELCOME AND INTRODUCTIONS
Dr. Robert Puls, U.S. Environmental Protection Agency, National Risk Management Research Laboratory

Dr. Puls, the Co-chair for the Remediation Technologies Development Forum (RTDF) Permeable Reactive Barriers (PRBs) Action Team, opened the meeting by welcoming the participants. He announced that the next Action Team meeting will be held in March 1998. The next meeting will focus on PRBs that use alternative media (other than zero-valent iron [ZVI]) to treat contaminant classes other than chlorinated solvents. The location for this meeting has yet to be determined, but the meeting will include a field trip to an alternative media/contaminant PRB site.

Dr. Puls introduced the keynote speaker, Dr. Bob Gillham. He described Dr. Gillham as the pioneer of PRB technology.

KEYNOTE ADDRESS: CURRENT STATE OF THE SCIENCE AND TECHNOLOGY OF PERMEABLE REACTIVE BARRIERS
Dr. Bob Gillham, University of Waterloo

Because many of the other speakers at the meeting were addressing the current state of PRB technology, Dr. Gillham said he would instead speak about what might be on the horizon for PRBs. Dr. Gillham began with a brief history of PRB development:

Dr. Gillham said that while the scientific community now generally accepts that the desired chlorinated hydrocarbon (cVOC) degradation reactions do occur in situ, several areas of PRB technology require further research:

• Mechanisms. The electrochemistry of the reactions is not well understood. Research is also needed to extend the technology to remediate a wider range of contaminants.
• Limitations. How precipitation onto the reactive surfaces may limit degradation reactions is a major question. Limitations of hydraulic behavior of the two major PRB design types---funnel and gate systems and continuous reactive walls---are not completely understood.
• Installation Methods. With the exception of the PRB at Elizabeth City, which was installed with a deep trencher, all PRBs to date have been installed through excavation and replacement. This method is relatively expensive, and the cost becomes prohibitive at greater depths. Alternative installation methods under investigation include high pressure jetting, deep soil mixing, and hydrofracturing (horizontal and vertical).
• Enhancements. There may be ways to improve the reactivity of ZVI. Possible methods include using bi-metals (iron plated with palladium, nickel, copper, or other metals) and sorbent materials plated to iron.
• Scope of Application. The range of treatable contaminants could be expanded. Addition of reactive materials to clay barriers is also under investigation. So far, PRBs have been used to remediate only dissolved plumes; there is now also research on PRBs as a method for removing contaminant source zones (e.g., dense-nonaqueous phase liquids [DNAPL]). Research is also focused on the use of PRBs for sequential treatment of multiple contaminants. Remediation efforts to date have been directed at single contaminant classes or simultaneous treatment of multiple contaminants. However, iron reactive media cannot treat many contaminant classes, such as aromatic hydrocarbons. Sequential treatment systems, with multiple reactive zones for different contaminants, are an avenue for further study.

Objectives of Ongoing Research at the University of Waterloo
Dr. Gillham described some of the current research efforts he is involved with at the University of Waterloo. Researchers there are studying the electrochemical mechanisms of reductive dehalogenation reactions by ZVI, including the influence of cVOCs on the corrosion behavior of iron and the effect of cVOCs on the structure of the iron/solution interface. The study methods involve supplying current, through an iron cathode, to a reaction solution and measuring the changes on the reactive iron surfaces through in situ raman spectroscopy. Two chlorinated compounds were added to the reaction solution: carbon tetrachloride (CT), which degrades quickly, and carbon dimethyl chloride (DCM), which degrades slowly, if at all. CT accelerated oxidation of the iron electrode, while DCM did not. A variation of this experiment indicated that allowing the iron electrode to equilibrate in the reaction solution, before supplying current, enhanced the oxidation reaction. This suggests that the iron oxide/oxyhydroxide materials precipitated on the reaction surface may have a catalytic effect on the dehalogenation reaction.

Dr. Gillham and colleagues are also investigating iron-nickel bimetal as a reactive medium. While the half life of trichloroethylene (TCE) in iron reaction columns is typically 30 to 45 minutes, it is reduced to approximately 3 minutes in iron-nickel columns. Some trials have indicated a loss of reactivity in this bi-metal with greater numbers of pore volumes. The rate of reactivity loss is apparently proportional to the mass of TCE treated. The rate of reactivity loss is also related to the concentration of dissolved inorganics in the water, with the presence of calcium carbonate in the water slowing the rate of reactivity loss. Thus, while precipitation of calcium carbonate is believed to reduce iron reactivity, the presence of dissolved calcium carbonate may benefit the bi-metal as reactive medium.

Researchers at Waterloo are assessing the utility of PRBs for source zone containment. One method of source remediation has been to enclose the source zone with an impermeable barrier (e.g., sheet piling, clay, or high density polyethylene [HDPE]). Dr. Gillham's research shows that cVOCs can, in fact, rapidly diffuse through HDPE or clay and form a plume of contaminated ground water with concentrations almost as high as those inside the barrier. Dr. Gillham's group is investigating whether adding metals to clay barriers can prevent diffusion of cVOCs. In laboratory tests, the addition of large amounts of nickel-plated iron essentially prevented diffusion of TCE over an 80-day period. A bentonite barrier without metal additives did not deter diffusion. A larger-scale experiment is now underway to test further these metal additives.

Dr. Gillham's group is also testing the ability of granular iron additives to remove cVOC source zones. A conventional method of remediation is to run fluids through the source zone and treat these fluids. One problem with this approach is that heterogeneity of the soil can create preferential flow paths and prevent complete treatment of the soil. In addition, flushing cannot treat cVOCs tightly bound to fine-grained materials such as silts and clays; cVOCs may then diffuse from these materials into the surrounding aquifer after treatment has stopped. Dr. Gillham's group is looking at deep soil mixing as an alternative to flushing. Mixing soil with iron and bentonite (as a lubricant) homogenizes the soil. In addition, the bentonite reduces soil hydraulic conductivity, effectively isolating the source zone from the surrounding ground water as the dehalogenation reaction with iron proceeds. (There has been no indication that increased alkalinity, caused by the bentonite, affects reactivity.) Short-term experiments indicate that TCE is reduced at a constant rate in this process. Future longer-term trials will analyze the performance of this method as more and more of the TCE is broken down and the concentration of resulting chloride increases. Dr. Gillham said that deep soil mixing can be performed easily to a depth of 90 feet, after which the cost increases significantly.

Dr. Gillham is also involved in a study of deep remediation through vertical soil fracturing and injection of ZVI in a guar polymer at the Massachusetts Military Reservation. This method, developed by Golder Associates, does not actually fracture the soil, but rather forces the guar mixture through the soil. A line of injection wells will overlap the injected material to form a continuous reactive barrier. These walls will be installed at a depth of 80 to 120 feet.

In conclusion, Dr. Gillham said that research over the last 8 years has only scratched the surface of the potential of PRB technology. The future should bring numerous advances in PRB applications.

LONG-TERM PERFORMANCE INVESTIGATIONS

Analysis of Granular Iron Cores From a Field Pilot Wall
Dr. Timothy Sivavec, General Electric

Dr. Sivavec has been studying the long-term performance of PRBs by analyzing precipitation effects in system entrance zones and reactive zones of PRB field installations. Dr. Sivavec began by describing recent findings on the effects of dissolved oxygen scavenging in the system entrance zone. In aerobic aquifers, dissolved oxygen in the reaction zone can cause precipitation of oxides and hydroxides that may limit reactivity of the iron at the upgradient edge. Reducing oxygen in the system entrance zone can thus limit precipitation in the reaction zone. A system designed with a dilute mixture of iron and sand or pea gravel in the system entrance zone can "scrub" dissolved oxygen and maintain anaerobic conditions in the reactive zone. This dilute iron mixture will not, however, appreciably reduce cVOC concentrations in the system entrance zone. This type of design will be implemented in a pilot project at Dover Air Force Base in Delaware, which is the first PRB field site to be studied that has an aerobic aquifer.

In measuring long-term PRB performance, Dr. Sivavec focuses on precipitates of calcium carbonate, iron oxides/hydroxides, and ferrous carbonate on the ZVI. The speciation of these precipitates at a given field site depends on the ground-water chemistry (e.g., hardness, alkalinity). The effects of precipitation can be measured in decreased porosity of iron in the reactive zone. This effect is important because porosity plays a role in reactive zone hydraulics and reactivity.

Dr. Sivavec studied these precipitation effects by pumping large volumes of site ground water through a series of ZVI columns; this method of accelerated aging subjects the reactive medium to 10,000 pore volumes, or the equivalent of 10 years of field operation. Dr. Sivavec pointed out that this laboratory method differs from site conditions, however, because residence time in the reactive zone in the columns is short due the high-volume flow, while residence times are longer under low-flow site conditions. Precipitates on the iron surface can be measured and analyzed by surface analytical techniques, including: cross-sectional scanning electron microscopy (SEM); X-ray photoelectron spectroscopy (XPS), and wavelength dispersive spectroscopy (WDS).

Porosity losses of approximately 10 percent were observed in these studies. However, Dr. Sivavec has not seen the decrease in reactivity expected to be caused by precipitation. Dr. Sivavec suspects the precipitates pack loosely on the iron surface without impeding electron transfer. It is still unknown how much decrease in porosity is required to reduce iron reactivity. Sivavec said porosity loss may be offset by designing PRBs for higher reactive zone porosity. General Electric has achieved reactive zone porosities of up to 70 percent.

Dr. Sivavec has compared these laboratory results with precipitation effects in the field by studying cores taken from PRBs at three sites: Sherburne, New York; Lowry, Colorado; and Elizabeth City. Because iron materials differ between sites, this study compared core material to virgin material from the same sites, rather than comparing core materials from different sites. Cores were taken at an angle through the entire barrier to obtain a profile of the reactive zone along the direction of ground-water flow. The sampled PRBs had been operating for 1 to 2 years.

Dr. Sivavec quantified precipitation in the cores by measuring water alkalinity and hardness as a function of residence time, measuring porosity, and studying surface effects with SEM, XPS, and WDS. Dr. Sivavec observed 30-micron-thick rinds of calcium carbonate on iron from the first few upgradient inches of the Sherburne and Lowry cores. Calcium precipitation in the Elizabeth City core was only 10 microns thick, possibly because this wall is newer than the other PRBs. Much thinner calcium layers, similar to those seen on virgin material, were observed in iron from further downgradient in the cores. Dr. Sivavec's group also correlated iron/oxygen ratios in the iron material with precipitation and depth; there was much less iron relative to oxygen observed in the first few inches of upgradient material. Thus, both laboratory and field data showed that precipitation is greatest early in the treatment zone, and that equilibrium arises further into the treatment zone, after which only minimal precipitation occurs. Observed porosity losses in these cores were minimal.

Dr. Sivavec stated that it is still unclear how many cores are needed to characterize a given PRB. More research is also needed to correlate ground-water chemistry with precipitation effects.

A participant asked Dr. Sivavec if core-handling procedures might affect the surface of the iron material. Dr. Sivavec said it is important to avoid rusting after removing the core---his laboratory dries the cores with acetone and stores them in an inert atmosphere to preserve their in situ conditions.

Results of Recent Iron Zone Coring Activities
Mr. John Vogan, EnviroMetal Technologies, Inc.

Mr. Vogan described additional results of coring activities at the Sherburne and Lowry pilot projects. These funnel-and-gate systems consist of 15-foot-wide sheet piling on either side of a 10-foot PRB gate with an iron zone thickness of 3 feet parallel to the direction of ground-water flow.

The Sherburne PRB, installed in May 1995, treats cVOCs in concentrations in the hundreds of parts per billion (ppb). Three angled cores were collected from the iron zone with a driven hammer core sampler. cVOC data from the upgradient (influent), iron (treatment), and downgradient (effluent) zones have indicated consistent performance removing contaminants, although there is an unexplained rise in cVOC levels between the iron zone and the downgradient zone. The results of precipitation analysis (by SEM and raman spectroscopy) agreed with data from Sherburne cores presented by Dr. Sivavec: calcium carbonate precipitate content was greatest (12 percent) at the upgradient edge of the core, and decreased rapidly to 1 percent only 6 inches into the core. Porosity of core material was 0.4 to 0.6, which is higher than the expected value of 0.4, indicating field conditions may cause higher porosity than laboratory conditions.

Mr. Vogan's group analyzed microbial activity in the cores by performing both aerobic and anaerobic microbial cultures, and by measuring lipid biomass. Results indicated relatively low microbial activity, and there appeared to be no microbial fouling of the iron zone.

The Lowry installation has operated for 18 months. Angled cores collected by Geoprobe™ did not yield a clear upgradient interface, so Mr. Vogan focused analysis on vertical cores. A 5 to 10 percent loss of porosity was observed in the vertical core 6 inches in from the upgradient edge. As at Sherburne, carbonate content decreased with increased distance downgradient in the reaction zone. Microbial activity was also low at Lowry.

Dominant mineral precipitates detected at both sites were calcite, aragonite, iron oxides, ferrous hydroxides, and magnetite. The carbonates were predominant in the first 6 inches of the reactive zones. Green rusts were present in cores from both sites, and may have been produced by sulfate-reducing bacteria. Although sulfates are present in influent ground water at both sites, and these sulfate levels decrease across the reactive zones, no pure-phase sulfide precipitates have been detected in the iron zones. The fate of these sulfates is still unknown.

Alternative Materials in PRBs
Mr. Shawn Benner, University of Waterloo

At the University of Waterloo there has been considerable research under the direction of Dr. David Blowes using PRBs with non-ZVI reactive media to treat non-cVOC contaminants. Dr. Blowes and colleagues have looked at treating inorganic contaminants including: acid mine drainage, trace metals, nutrient contaminants (nitrate and phosphate), selenium, and arsenic. In addition to looking for media that are reactive, long-lasting, and hydraulically permeable, the researchers have searched for very low cost materials. After developing a theoretical concept for a material, the group runs materials through laboratory batch tests. The most promising materials are then run through column tests, some lasting as long as three years. Suitable media are then put through pilot-scale field tests, modeling studies, and full-scale implementation.

Acid Mine Drainage
Acid drainage of mine wastes is the largest environmental problem facing the North American mining industry and affects an estimated 20,000 miles of streams. Acidic drainage is caused by oxidation of residual sulfides in mining waste products. The iron and sulfates produced in this reaction are released to the ground water, which discharges either at the toe of the waste pile or at a distant source. While some acidity occurs in the ground water, some ferrous iron will not oxidize and form acid until it reaches surface water. The goal of PRB treatment is to intercept the iron and sulfates in the ground water and reduce them back into iron sulfides through biological remediation induced by organic carbon. To find the best reactive medium, Dr. Blowes' group tested a number of material mixtures including leaf compost, wood chips, and sawdust. After batch, column, and pilot tests, the researchers implemented a full-scale PRB with a compost/gravel mixture. This PRB was positioned to treat acid mine drainage discharging to a highly acidified lake (pH=3). The barrier removes iron and sulfate from the influent primarily through sulfate reduction. This PRB has decreased influent iron concentrations of 300-1,300 mg/L to 0.4-100 mg/L in the treatment zone, below regulatory guidelines. Data show the PRB is effectively reducing the acid-generating potential of the effluent ground water.

Metals
The sulfate reduction mechanism can also reduce trace metals such as lead, copper, zinc, and cadmium. A full-scale installation in Vancouver, using a compost/gravel medium, has achieved greater than 99 percent reduction of copper and zinc concentrations.

Nutrient Contaminants
Research in this area is focused on treating effluent from on-site treatment systems such as septic systems. These systems treat biological contaminants, but do not remove nitrates and phosphates, which can cause eutrophication in surface water. Phosphate can be removed by adsorption onto a metal oxide followed by precipitation of a mineral phase. The same reaction removes nitrate, presumably converting it to nitrogen gas (although the nitrogen product of this reaction has not been identified, data show it is not ammonia). Similar mechanisms may be used to remove arsenic and selenium. At Waterloo, column tests of an oxidized iron byproduct mixed with agricultural limestone have achieved more than 80 to 90 percent reduction of phosphates. Precipitation does not appear to reduce reactivity or porosity of the medium, even after almost 1,500 pore volumes. A field installation of this medium at a septic system leach field has achieved similar results. While this application is in an uncontrolled setting, the treatment process can also be used in controlled systems (e.g., to remove nutrients from effluent of a bioreactor). An added benefit of this type of application is that this PRB medium also effectively reduces coliform bacteria levels.

A Pilot Test of a Surfactant-Modified Zeolite Permeable Barrier for Chromate and Perchlorethylene Removal
Dr. Zhaohui Li, New Mexico Institute of Mining and Technology

Dr. Li is currently involved in a Phase II pilot-scale PRB study using a surfactant-modified zeolite (SMZ) reactive medium to remove perchlorethylene (PCE) and chromate from ground water. Zeolites are micro-porous crystalline solids with large surface areas and high internal and external cation exchange capabilities. Natural or synthetic zeolites can be tailored for permeability of a range of particle sizes. In this study, Dr. Li has treated natural zeolite with HDTMA, an amphiphilic surfactant which, in aqueous solution, associates in micelles due to a hydrophobic effect. When added to zeolites, HDTMA sorbs to the negatively charged zeolite surface and forms a bilayer of HDTMA molecules in a "pseudo-organic" phase similar to a lipid bilayer. The positively charged outer surface of the HDTMA bilayer can bind oxyanions (e.g., chromate, arsenate, nitrate) while the hydrophobic, hydrocarbon interior of the bilayer binds organic molecules such as PCE. Studies have shown that surfactant modification does not alter zeolite permeability. With increased numbers of pore volumes, SMZ exhibits a chromate sorption plateau, after which reactivity is limited. Dr. Li's group has developed a regeneration process of flushing the SMZ with sodium dithionite solution to restore the medium's reactivity to chromate.

Phase I study results showed SMZ quickly removes a range of organic and inorganic contaminants, is biologically and chemically stable, has regenerable reactivity, and has minimal impact on microorganisms in the subsurface.

One objective of the Phase II study, to be completed by May 1998, is to demonstrate the feasibility of producing large quantities of SMZ; estimated cost of this SMZ is $0.23/cubic foot (most of the cost is attributable to the HDTMA). The Phase II study will involve three separate SMZ reaction zones. One will be saturated with chromate and regenerated with sodium dithionite flushing, while another will be saturated with PCE and regenerated through air sparging. This study will also test SMZ/ZVI pellets, which should combine the reactive properties of iron with the sorptive properties of SMZ. SMZ/ZVI thus has potential as a combined immobilization/degradation barrier.

Analytical and Numerical Modeling of Iron-Based PRBs
Dr. Alan Rabideau, State University of New York at Buffalo

For 4 years, Dr. Rabideau has worked with DuPont to develop design-oriented models for contaminant transport, low-permeability materials, and PRBs. Dr. Rabideau focuses on factors affecting long-term performance of PRBs. Dr. Rabideau presented an advective-dispersive-reactive equation (ADRE) to describe the relationship of these three types of mechanisms. In the simplest ADRE, the user treats the model as steady-state only, neglects dispersion, and provides idealized boundary conditions. To better describe field conditions, a more complicated model would include dispersion, while the most complex equation will also include realistic boundary conditions. Comparison of solutions from these three forms of the ADRE, however, indicates that the simplest batch model predicts the more complicated solutions with error of only a few percent, and should be adequate for even complicated design purposes. Rabideau's group is preparing for publication a generalized analytical model that includes sorption kinetics useful for establishing pre-steady-state data.

The ultimate goal of Dr. Rabideau's research is to develop a numerical model for field application and determination of long-term performance. This model should be able to consider variations in velocity or source concentrations with time (e.g., seasonal or regional ground-water flow variations). Dr. Rabideau's group is in the early stages of such a model, which is based on a split-operator approach using a DuPont one-dimensional model, Trans 1D. Their model should be able to link results of laboratory batch tests directly to transport models for one-dimensional columns or walls. This model will include a number of long-term performance conditions, including: plugging, precipitation (porosity reduction, hydraulic conductivity, and loss of reactive surface), reaction rate changes, and changes in regional ground-water flow. Development of this model involves several conceptual issues, including: multiple precipitates, precipitate mass/volume relationships, other mechanisms of porosity reduction, permeability reduction, and buffering by reactive media. Dr. Rabideau's model will eventually be expanded for three-dimensional field applications using RT3D, an algorithm developed by Pacific Northwest Laboratories to relate chemistry to hydraulics. The Ground-water Modeling System (GMS) is a computer application which allows the user to work with spatial data to design field-scale ground water projects. Once the PRB model is developed, it could be added as a module to the GMS to allow PRB design with that system. Dr. Rabideau's group is in an early stage of developing such a module. In closing, Dr. Rabideau stated that for these models to accurately predict long-term performance, more detailed field data about PRB performance is needed.

A participant asked Dr. Rabideau how many pore volumes are needed to reach a steady state (some researchers have reported 20 to 40 pore volumes are required). Dr. Rabideau replied that when sorption, kinetics, and other factors are neglected, decay equations predict a steady state after as few as five pore volumes. When those factors are considered, however, it may indeed take 20 to 40 pore volumes to reach a steady state.

The Interstate Technology Regulatory Cooperation Work Group
Mr. Matthew Turner, New Jersey Department of Environmental Protection

Mr. Turner described the Interstate Technology Regulatory Cooperation (ITRC) as a national coalition of state regulatory agencies working with federal agencies and stakeholders to promote the deployment of innovative hazardous waste management and remediation technologies. ITRC includes 25 states, the Western Governors Association, the Southern States Energy Board, the Department of Energy (DOE), the Department of Defense (DOD), and EPA, as well as public stakeholders and industry representatives. ITRC tries to reduce state barriers, increase receptivity to new technologies, and facilitate transfer of data from technology demonstrations and full-scale applications among states to reduce duplication of effort. ITRC also develops protocols for innovative technology deployment for use by regulators and technology vendors. Like the RTDF, ITRC is divided into work groups addressing different technologies. There are currently eight ITRC Work Groups: Permeable Barrier Walls, In Situ Bioremediation, Accelerated Site Characterization, Low-Temperature Thermal Desorption, Metals in Soil, Technology Verification Program, Plasma Technologies, and Policy Issues.

The ITRC process involves two phases: product development, followed by state consensus/implementation. There are three types of ITRC products: technical/regulatory guidance documents, informational case studies, and consultation services. After developing a product, ITRC uses a tiered approach to try to gain state consensus. ITRC tries to obtain a letter from each state indicating the level of consensus and implementation the state plans for the document.

Eight states (New Jersey, Colorado, Florida, Massachusetts, Washington, New York, Nevada, and California) are represented on the PRB Work Group, which was formed in September 1996. The RTDF and public stakeholders also participate in the Work Group. The Work Group recently produced a document, Regulatory Guidance for Permeable Barrier Walls Designed to Remediate Chlorinated Solvents, and assisted the Air Force Center for Environmental Excellence (AFCEE) with development of its PRB design guidance document. The Work Group's regulatory guidance document includes sections on site characterization, bench scale testing, modeling, permitting, monitoring, maintenance and closure, health and safety, stakeholder concerns, and variances. Mr. Turner said permitting and monitoring are perhaps the most important sections of the document.

The permitting section identifies major permitting considerations that may be applicable to PRBs in a given state, including: underground injection control permits, National Pollution Discharge Elimination System (NPDES) permits, and air permits. While jet grouting may require an underground injection permit, the Work Group found most of these types of permits were not applicable to PRBs in most states. Mr. Turner stressed, however, that it is important to perform a thorough review of permitting issues required for site-specific and local permitting consideration.

The monitoring section provides guidelines for analytical parameters and frequency of analytical sampling and water level monitoring at PRB installations. This section also presents a rationale for various monitoring well locations at both continuous permeable walls and funnel-and-gate installations.

The Permeable Barrier Walls Work Group's goals for the coming year include:

• Obtaining regulatory consensus on both the ITRC regulatory and AFCEE design guidance documents.
• Developing a document regarding the use of PRBs to treat metals and radionuclides. Mr. Turner said he would like the RTDF Action Team to contribute to and review this document.
• Reviewing any documents forthcoming from the RTDF PRB Action Team.

Information on ITRC and copies of the regulatory guidance document are available through the World Wide Web @http://www.westgov.org/itrc.

[Dr. Puls announced that the RTDF PRB Action Team is developing a separate, broader guidance document that will address alternative media/contaminants, performance and compliance monitoring objectives, and other issues surrounding PRB technology. All members of the Action Team will be welcome to review this document, and the document will also undergo the EPA peer review process. The RTDF will endorse the final product. Dr. Puls said this document will compliment the ITRC regulatory guidance.]

PRB Applications
Mr. James Paulson, DOE

Mr. Paulson presented an overview of DOE and non-DOE PRB work using the following approaches: chemical zones, reactive walls, funnel-and-gate systems, gravel trenches, and reactive columns and wells. Mr. Paulson said DOE projects differ from commercial sector sites in that they deal with radionuclides as well as metals and VOCs.

• Hanford 100D Area, Washington. This installation is an injected chemical barrier of sodium dithionite for reduction of chromium. The treatment zone, installed through five injection wells, is 150 feet long, 50 feet wide, and 80 to 100 feet below ground surface.
  
• Rocky Flats Mound Site Plume, Colorado. This EPA SITE Program demonstration is a funnel-and-gate system, 8 to 15 feet deep, and consists of 270 feet of wall with two reactor cells for treating cVOCs, metals, and trace uranium and plutonium. This site is complicated by low-permeability soil and a hilly topography that creates a hillside seep area.
• Oak Ridge National Laboratories Y-12 Plant South Plume, Tennessee. This installation is a funnel-and-gate system treating uranium, technetium-99, PCE, nitrate, and total dissolved solids (TDS).
• Oak Ridge National Laboratories Y-12 Plant Channel Plume, Tennessee. At this site, a plume of uranium, technetium, nitrate, and TDS is treated in a ground-water capture system consisting of 20 feet of gravel cutoff trench and a treatment zone. This capture system creates a preferential zone of ground-water flow.
• Savannah River D Area Coal Pile Basin, South Carolina. Ground water at this site has low pH, trace metals, TCE, and sulfate. This project is a retrofit of an earlier funnel-and-gate system. Gravel backfill along the existing funnel has caused a high-permeability zone of preferential ground-water flow which pulls in area ground water. The retrofit will consist of a reactive canister in a 2-foot-diameter well that will serve as a run-off basin for the ground water. DOE is currently performing treatability studies to determine the best reactive medium.

Mr. Paulson said that through these sites the DOE has learned the importance of detailed site characterization for PRB projects, especially for less proven applications of the technology such as metal and radionuclide remediation. Overall, PRBs are proving to be encouraging alternatives to conventional pump-and-treat systems. In the future, Mr. Paulson would like to see methods for deeper installation of PRBs.

In Situ Reactive Barriers for Simultaneous Treatment of Radionuclides and Chlorinated Organic Contaminants
Dr. Baohua Gu, Oak Ridge National Laboratory

Dr. Gu spoke about ongoing PRB work at the Oak Ridge National Laboratory Y-12 Plant Bear Creek Valley site. From 1951 to 1983, the S-3 pond at this site received nitric acid waste containing uranium, technetium, PCE, and heavy metals. Over 2.5 million gallons of waste per year were discharged to the site until it was closed and capped in 1984. The goal of the PRB project is to evaluate ZVI and other media for simultaneous treatment of the various contaminants present in the ground water.

The researchers have focused laboratory studies on reaction kinetics and mechanisms of interaction between the various contaminants and ZVI and sorbent materials. Dr. Gu's group has performed a number of spectroscopic studies (e.g., fluorescence, X-ray diffraction, energy dispersive spectroscopy [EDX], SEM) to analyze behavior of these reactions. Dr. Gu found that ZVI efficiently removes uranium but has relatively slow reaction kinetics, while iron oxide shows fast kinetics but lower total capacity. With long enough reaction time, ZVI can remove almost all present uranium. Fluorescence spectroscopy and EDX showed that uranium is removed primarily by precipitation onto the ZVI surface, and a small percentage precipitates as a suspended solid.

Dr. Gu's group observed that nitrate removal was faster in a mixture of ZVI and peat material than it was in ZVI alone; ammonium, nitrogen gas, and nitrous oxide were the major products of this nitrate removal. Study of cVOC degradation with ZVI showed that cVOC reduction was effective but slow, and produced chlorinated breakdown products, especially with high initial concentrations of primary contaminants.

In summary, Dr. Gu stated that uranium, chromium, and other redox-sensitive metals can be effectively reduced by ZVI, as can nitrate. Degradation of cVOCs by ZVI is relatively slow but can be achieved with longer residence times; ZVI may generate toxic byproducts, particularly with high influent concentrations. Installation of a pilot-scale ground-water collection/iron-barrier trench is now underway to test these findings in the field.

Potential Use of Jetting to Emplace PRBs
Mr. Richard Landis, DuPont

High pressure jetting is an alternative PRB emplacement technique that may achieve greater depths than conventional excavation and replacement. There are three types of high pressure jet grouting systems:

• Single-fluid system. This involves direct injection of grout into the soil.
• Double-fluid system. This system surrounds the grout stream in a sheath of high pressure air to achieve greater penetration into the soil. This is the most cost-effective system and was used in this study.
• Triple-fluid system. This system achieves almost complete excavation of soil and replaces it with grout.

All of these systems begin by drilling a hole to the desired jetting depth, lowering the jet nozzle to the bottom, and jetting the grout while gradually pulling the drill string up out of the hole. Aiming the nozzle in one direction produces a grout diaphragm, while spinning the jet as it rises out of the hole forms a column. Overlapping these structures creates longer, continuous barriers. For these pilot tests, 90 gallons of grout per minute were pumped at 5,800 pounds per square inch (psi).

In a field study at a DuPont site, the researchers tried two grout formulas: a guar gum/granular cast iron slurry and a kaolinite clay/iron slurry. Landis's group installed both columns and diaphragms at this site. Columns exhibited a permeability of 1.66 x 10^-3 cm/sec and emplaced density of 116 lbs/cubic foot; diaphragms had a permeability of 1.38 x 10^-4 cm/sec and an emplaced density of 143 lbs/cubic foot. The reactivity of these barriers is under investigation, as is iron distribution in the emplaced barriers versus the virgin grout.

A field study at Dover Air Force Base involved injection of bentonite cement to form deeper and larger barriers suitable for a funnel-and-gate system. Mr. Landis tested both low- and high-cement content slurry formulas. The high-cement slurry formed approximately 5-foot-long walls with permeability of 2.4 x 10^-7 to 3.3 x 10^-9 cm/sec, and an (unconfirmed) strength of 2,170 psi. (Mr. Landis said the sampling process of cutting a cement sample with a diamond saw may damage the sample by causing micro-fractures. The true strength of this wall thus may not be reflected in the sample.) The low-cement slurry formed a longer diaphragm with a permeability of 2.5 x 10^-7 to 8.4 x 10^-8 cm/sec and an (unconfirmed) strength of 57 psi. During the course of installation, the researchers found that a jet nozzle angled 5 degrees downward achieved more penetration and formed a much longer, more complete diaphragm than did a fully horizontal jet nozzle. Mr. Landis suspects the slight downward angle of the jet creates a vortex in the cutting zone, which clears out the in situ soil more effectively.

A participant asked if this jetting is feasible in other lithologies. Mr. Landis said this technique is not as effective in more cohesive or rocky soils. In a cohesive soil like a clay, jet grouting will form much thinner diaphragm walls.

SITE UPDATES
Moderated by Dr. Dale Schultz, DuPont

Field Evaluations of a PRB Containing ZVI at the Denver Federal Center
Dr. Peter McMahon, U.S. Geological Survey

At the Denver Federal Center, Colorado, a PRB containing ZVI has been operational for 11 months to treat ground water contaminated with cVOCs and prevent it from migrating off site. The PRB is a 1,200-foot-long funnel-and-gate system consisting of four 40-foot-wide permeable ZVI treatment cells and impermeable sheet piling keyed into siltstone bedrock to a depth of approximately 25 feet below ground surface.

Dr. McMahon said that analysis of cVOCs in influent and effluent ground water indicates that the treatment zones are reducing cVOCs to below drinking water standards. The only cVOC exiting the gates at concentrations above reporting levels (5 ppb) is 1,1-dichloroethane (maximum concentration 8 ppb), a breakdown product of trichloroethane.

Annual porosity loss has been 0.5 percent and is likely caused by calcite and siderite precipitation. Dr. McMahon believes precipitation in the reaction zone is limited partly because ZVI mixed into the upgradient pea gravel may precipitate the available calcium and inorganic carbon before it reaches the reaction zone.

There is an increased hydraulic head in the surficial aquifer upgradient of the barrier, due in part to leakage from a nearby upgradient reservoir. The 10-foot head differential across the barrier has increased the potential for contaminated ground water to flow under, over, or around the barrier. Concentrations of cVOCs have, in fact, increased in ground water flowing around the southern end of the barrier. However, this ground water discharges to a stream which effectively dilutes the cVOC concentrations. In addition, leakage from the reservoir is diluting cVOC concentrations in the upgradient ground water. Dr. McMahon said there is evidence that cVOCs may also be moving underneath the barrier at one location (gate 2) where the sheet piling is keyed into weathered bedrock rather than competent, confining bedrock. Dr. McMahon expects the reservoir leakage eventually to dilute the cVOCs moving underneath the barrier at this location. Water is flowing over gate 2 with cVOC concentrations much lower than in the upgradient water. The perched water above this gate has been caused by recent heavy rain.

Dr. McMahon said that within the next year his group will attempt to reduce the head build-up around gate 2 by augering into soil and clay against the gate that may be reducing its permeability.

Performance Monitoring of PRB at USCG Support Center, Elizabeth City, North Carolina
Dr. Robert Puls, EPA NRMRL

The Elizabeth City PRB is a full-scale demonstration being performed by members of EPA NRMRL and the University of Waterloo. Installed in June 1996, the PRB is a continuous ZVI wall placed between an old plating shop and the Pasquotank River. The PRB was emplaced to treat chromium- and TCE-contaminated ground water flowing towards the river. Influent contaminant concentrations include: 12 mg/L TCE, 1 mg/l cis-dichloroethylene (DCE), 100 ug/L vinyl chloride, 8 mg/L hexavalent chromium. The source of the chromium is a hole in the floor of the plating shop, while the source of TCE has not been identified.

The demonstration project used a new method of PRB installation. The wall was installed with a deep trencher that simultaneously excavated soil with a cutting head and replaced it with iron from a hopper. The wall was keyed into a layer of relatively low-permeability sandy clay at 22 to 26 feet below ground surface.

Dr. Puls said that performance monitoring entails studying physical, chemical, and mineralogical parameters to ensure that the wall meets design criteria. Dr. Puls noted that changes in reactivity, hydraulics, and site conditions are of particular concern. The array of monitoring wells at the site includes 10 compliance monitoring wells and 15 multi-level samplers, located upgradient, downgradient, and within the barrier. The researchers have collected three rounds of performance monitoring samples to date, with another round to be collected during the Action Team's field trip to the site on September 19. Sampling techniques are in many cases non-traditional and are intended to capture small water volumes to analyze discrete intervals within and around the wall. Samples are analyzed for primary contaminants and breakdown products. Samples indicate that the majority of TCE breakdown occurs in the first foot of the barrier and reduces TCE to below drinking water limits. TCE is detected, however, in a well downgradient of the barrier, indicating that, while the barrier is successfully degrading TCE, some TCE is flowing underneath the barrier. The ZVI is removing all chromate in the first few inches of the wall. While sulfate in the influent water is greatly reduced across the wall, it is unclear how the sulfate is reduced because no sulfides have been observed in the cores. The sulfate may be precipitated in a green rust observed in the cores, but the researchers have not yet analyzed this material.

Dr. Puls's group is conducting hydrologic (through head measurement, a tracer test, and in situ flow measurement) and geochemical assessment of the PRB. There is little head difference across the wall. A bromide tracer test indicated travel times of between 5 and 9 inches/day,---close to design criteria. The wall significantly reduces Eh (to > -400 mV), increases pH (to > 10), and decreases the specific conductance. These values begin to rebound as water flows from the wall.

Researchers have collected angled and vertical cores from the wall for reactive surface analyses and microbiological characterization. Cores are wrapped in plastic and frozen in liquid nitrogen for transport to the laboratory, where they are stored in a freezer before being distributed to research sites or processed in liquid nitrogen.

Dr. Puls's group has tried several methods to verify proper emplacement of the wall. Because the iron wall has much higher electrical conductivity than does the native soil, patches of the wall exhibiting lower conductivity through a conductivity probe may indicate locations where the trencher failed to deposit iron or to pack it tightly. Vertical cores from these areas of lower conductivity have shown gaps in the wall where soil had not been completely replaced with iron. Conductivity data indicates the wall is from 45 to 55 cm thick, compared with the 60 cm design, and that wall depth also varies. Researchers from the University of Waterloo will use ground penetrating radar and cross-borehole tomography to gain a clearer picture of the barrier and identify any gaps or variations in depth, wall thickness, or iron packing that may have been caused by difficulties with this new method of barrier installation.

Dr. Puls pointed out similarities and differences between the pilot-scale and full-scale systems at this site. The pilot project was a fence system using a mixture of iron and native soil. The full-scale system is a continuous wall of pure iron. While both systems eliminate chromium, more ferrous iron was detected in the fence system than in the full-scale wall. Eh is much lower and pH is much higher in the pure iron wall, while the iron/soil mixture apparently buffered Eh and pH levels. Sulfides were present in the pilot barrier but have not yet been detected in the full-scale barrier.

Moffett Field Permeable Barrier Ground-Water Monitoring and Tracer Testing
Mr. Charles Reeter, Naval Facilities Engineering Service Center
Dr. Arun Gavaskar, Battelle Memorial Institute

Mr. Reeter presented an overview of the site and described recent monitoring data. The Naval Facilities Engineering Services Center, Navy Engineering Field Activity West, Battelle Memorial Institute, and PRC Environmental Management initiated this pilot-scale demonstration project in April 1996. The project was later sponsored by the DOD Environmental Security Technology Certification Program to collect extensive performance data and assess cost effectiveness of PRB technology. Researchers developed several evaluation criteria:

• cVOCs reduction
• Elevated inorganic ions (e.g., iron, chlorides)
• Production of gaseous analytes (e.g., ethanes, ethenes, propane)
• Hydraulic capture efficiency and flow through the treatment cell (determined through water levels measurement, velocity gauging, slug/tracer testing, precipitate analysis, continuous monitoring, and ground-water modeling)

The pilot project is a funnel-and-gate system with two 20-foot-long funnels and a ZVI gate 10 feet long and 6 feet thick. Performance is monitored through an array of single-level, multi-level, and clustered monitoring wells, with a total of 10 wells in the iron zone and 23 wells outside the iron zone. Influent concentrations of TCE (1,400 ug/L), PCE (50 ug/L), and DCE (260 ug/L) have all been reduced below detection limits within the first 2 to 3 feet of the reactive zone. While vinyl chloride is not present in the influent, a low concentration of vinyl chloride was detected in one sampling event in the front of the reaction zone. Vinyl chloride has not been detected in wells further downgradient in the reaction zone. Alkalinity and calcium levels decrease as water flows through the cell. Decreasing sulfate levels may be a result of microbial activity in the reaction zone. Eh decreases and pH increases through the reaction zone.

Early in the project, the researchers were concerned about a possible ground-water mounding effect upgradient of the barrier that could cause backflow. The ground-water levels eventually stabilized and there is currently no evidence of backflow conditions. Concern about possible backflow, however, prompted the researchers to perform bromide tracer tests on the barrier. Mr. Reeter introduced Dr. Gavaskar to describe the tracer test project.

To investigate ground-water mounding, Dr. Gavaskar's group placed continuous water level monitors in three wells near the barrier. Over a 3-week period, data indicated that ground-water level was strongly correlated to rainfall, and resulted in some temporary instances of mounding. The site is hydrogeologically heterogeneous, with channels of sand and gravel surrounded by inter-channel deposits. Most of the ground-water flow occurs in discrete channels of high permeability sand. Dr. Gavaskar said the barrier appears to be well designed to intercept this section of high permeability. Slug tests results were as expected, with higher hydraulic conductivity seen in sand channels surrounded by layers of deposits with lower conductivity.

The researchers conducted a first tracer test in March 1997 and began a second test in August. The first test involved injecting bromide into the upgradient pea gravel and attempting to track it through the reactive cell. Before the field test, the researchers performed a column test on the bromide tracer and confirmed that the tracer was not retarded by the column. The researchers injected 3,000 mg/L of tracer into a well in the upgradient pea gravel and observed the site for 1 month. They traced the bromide with 16 continuous probes and one mobile probe, and confirmed these results with laboratory analysis of discrete samples. The bromide dispersed rapidly through the pea gravel and appeared in the reactive zone after 6 days. Flow of the bromide was not straight, however, and exhibited a more complex behavior than predicted through modeling. No backflow of bromide into upgradient monitoring wells was observed. While measured flow velocities differed among several methods, the range of results was close to the expected value of 3 to 4 feet/day.

Dr. Gavaskar said that at this point the group is relatively confident of its understanding of chemistry in the reactive zone, and for the remainder of the project will concentrate on questions of hydraulic performance. The researchers began a second tracer test in August, this time injecting bromide into the upgradient aquifer to trace it to the upgradient pea gravel zone. This effort is still underway. The final round of quarterly sampling at this demonstration project will be completed in October 1997, to be followed by coring for analysis of surface effects. The researchers will complete the Final Performance Evaluation and Technology Demonstration Report in March 1998.

Installed PRB Applications
Mr. John Vogan, EnviroMetal Technologies, Inc.

Mr. Vogan presented the status of current and planned PRB systems.

Pilot-Scale Systems

• Borden, Ontario, Canada. Monitoring was discontinued after 5 years of consistent performance, and results will be presented in an upcoming issue of Groundwater.
• Somersworth Landfill, New Hampshire. After a setback with removal of a caisson, the project encountered further problems when contaminant concentrations in the surrounding ground water dropped to low levels, making assessment of the pilot project difficult.
• Naval Air Station, Alameda, Texas. The University of Waterloo is directing this installation consisting of a sequence gate to treat cVOCs and VOCs. The system was installed in December 1996. Influent cVOC concentration have been higher than anticipated, and the effects of these concentrations on degradation in the iron zone remain to be seen.
• Westinghouse Savannah River Company, South Carolina. The Geosyphon system has been operational since July 1997.

Full-Scale Systems
Mr. Vogan said there is a recent fact sheet available on the RTDF Web site which describes many of the following installations:

• Sunnyvale, California. The installation has consistently exhibited non-detectable concentrations of cVOCs in the iron zone. As Dr. Gillham mentioned, the system has paid for itself in savings over pump-and-treat costs.
• Mountainview, California. This system is placed in a cVOC source zone. Data indicate cVOC removal of 85 to 99 percent, but some of this may be due to ground-water pumping adjacent to the site.
• Belfast, Ireland. This in situ reactor was installed in December 1995. It is performing well in terms of TCE degradation, but dry conditions and adjacent pumpage may be causing flow reversal through the system.
• Coffeyville, Kansas. This single-gate, funnel-and-gate system was installed in late 1995. It consists of a 500-foot slurry wall with a 25-foot-wide treatment zone. It has been reducing TCE from 100 ug/L to below detection limits. Hydraulic studies of this site are upcoming.

Near-Term Installations
Mr. Vogan says he has learned to take PRB construction dates with a grain of salt, but several systems are planned for the next 6 to 12 months:

• An in situ collection/treatment system, mentioned by Mr. Paulson, is planned at Rocky Flats, Colorado.
• A small, single-gate, funnel-and-gate system is planned at a private site in Colorado.
• A continuous PRB will be installed by hydrofracturing in New Jersey.
• Continuous PRBs are planned in New York, South Carolina, and Montana.
• Ten or more other installations could be constructed by the summer of 1998.

Mr. Vogan said that, since 1993, the average cost of a PRB project has dropped from $1 million to $500,000, as a result of decreased construction costs and iron material costs.

DISCUSSION

Mr. Donald Marcus (EMCON) informed the participants that testing of the reactive sand fracturing technology, under development at EMCON for several years, is progressing towards a field study. EMCON will begin installation of the reactive propents at a field site contaminated with TCE and chromium. The propents will be injected by hydrofracture into three treatment zones. Mr. Marcus hopes to make a presentation at the next Action Team meeting.

FIELD TRIP TO USCG SUPPORT CENTER STUDY SITE

About 50 people attended presentations and a performance sampling demonstration at the Elizabeth City field site on Friday, September 19. The day began with a series of brief presentations by the following people:

• Captain Gregory Magee, Commanding Officer, USCG Support Center
• Commander Michael Herring, USCG Support Center Environmental Compliance Division
• Mr. Jim Vardy, Civil Engineering Unit, Cleveland, Ohio
• Mr. Bill Schmithorst, Parsons Engineering Science, Cary, North Carolina
• Ms. Surabhi Shah, North Carolina Department of Environmental Health and Natural Resources Division of Waste Management
• Dr. Robert Puls, EPA NRMRL

These presentations were followed by a site visit to the continuous iron wall installation. Dr. Puls divided the group into four subgroups to present how the research team of NRMRL and the University of Waterloo conducts performance sampling. Exhibits or demonstrations included: sampling from conventional and multi-level monitoring wells; field chemical analyses performed on site in a mobile laboratory; and use of a Geoprobe for angled coring and electrical conductivity measurements.

RTDF Permeable Reactive Barriers
Action Team Meeting

Sheraton Oceanfront Hotel
Virginia Beach, VA
September 18-19, 1997

Final Speaker List

RTDF Permeable Reactive Barriers
Action Team Meeting

Sheraton Oceanfront Hotel
Virginia Beach, VA
September 18-19, 1997

Final Attendee List

RTDF logistical and technical support provided by:

Colin Devonshire
Environmental Scientist
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02713-3134
781-674-7274
Fax: 781-674-2851

Susan Brager Murphy
Conference Manager
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02173-3134
781-674-2906
E-mail: sbmurphy@erg.com

Laurie Stamatatos
Conference Coordinator
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02173-3134
781-674-7320
Fax: 7810674-2906
E-mail: lstamata@erg.com