Radisson Hotel Universal Orlando
Orlando, Florida
June 12, 2001

INTRODUCTION

On June 12, 2001, members of the Remediation Technologies Development Forum's (RTDF's) Permeable Reactive Barriers (PRB) Action Team met in Orlando, Florida, as part of the 2001 International Containment and Remediation Technology Conference and Exhibition. Information on PRBs was presented during three sessions; talks focused on design and emplacement, hydraulic and geochemical measurements, contaminant removal, and material development. At the end of the day, a panel discussion was held so that audience members could discuss PRB-related issues in greater detail. This report summarizes the formal presentations and highlights key points made during the panel discussion.

BEST PRACTICE FOR PRB DESIGN AND EMPLACEMENT

The Use of Ultrasound To Restore the Dehalogenation Activity of Iron in Permeable Reactive Barriers
Cherie Geiger, University of Central Florida

Cherie Geiger began her presentation by describing ultrasound theory: a transducer converts electrical or mechanical energy into high-frequency sound, i.e., ultrasound. To apply ultrasound in a PRB, one uses a submersible transducer that can be lowered into a well. Ultrasound produces two effects: acoustic cavitation and acoustic streaming. In acoustic cavitation, gas or vapor cavities are formed. These cavities collapse, creating small areas in which temperature and pressure are very high; this has a cleaning effect. In acoustic streaming, the energy from the streaming cleaves the surface of the targeted particles.

Geiger said that her research group conducted bag experiments to examine the effects of ultrasound on rate constants. She found these effects to be dramatic (e.g., the rate constant, normally 0.0025 reciprocal hours, more than doubled, rising to 0.0062 reciprocal hours). Effects increase, she said, as the duration of ultrasound application increases, but only for about an hour. After an hour, there does not appear to be much effect.

In another laboratory experiment, Geiger said, column studies were used to evaluate the effect of ultrasound on the half-life of trichloroethylene (TCE). The experiment ran for a year. Very fine particulates flowed out of the column, and the color of the column changed as it was exposed to ultrasound.

To obtain physical evidence of the effect of ultrasound, Geiger said, she used scanning electron microscopy (SEM) micrographs of oxidized iron filings. Before ultrasound treatment, the iron filings had circular formations of oxidized iron on their surfaces, and the TCE disappeared quickly. Byproducts did not appear quickly, she said, noting that there was a great deal of sorption, but not degradation. (Electron transfer cannot occur through the several layers of debris on the iron's surface.) After 2 hours of ultrasound treatment, however, the iron's surface was clear of most of the debris and was cracked and pitted, which allowed for areas of increased electron transfer.

After conducting laboratory experiments, Geiger and her team decided to apply the technology in the field. To do so, they needed to find an applicable transducer probe that could be submersed into a well in front of a wall. The probe also had to be omnidirectional. The first transducer they used was 20kHz, 1000 watts. As a first step, Geiger simulated field conditions to determine the radius of influence (from the transducer) of the ultrasound's cleaning effect on iron. Geiger used a microphone to accomplish this. After conducting the simulations, Geiger implemented the technology in the field. Having dropped the transducer into a 4-inch well, she saw an effect from 36 to 54 centimeters away. To analyze the results of the field application and determine the effects on the iron, she took core samples before and after the ultrasound application within 54 centimeters of the well.

In the field application, Geiger compared the effects of a 40kHz, 1000 watt transducer to those of a 25kHz, 3000 watt transducer at 30 minutes and 90 minutes of exposure. At different depths, she saw a 21% to 41% improvement in the half-life of contaminants with the 40kHz transducer; with the 25kHz transducer, the improvement was up to 67%.

Geiger also described ultrasound application at a second field site, where one of the gates of a PRB was not working. At this site, her team had to sample at an angle, because there were caps over the wall. Geiger showed the different ultrasound results with a 25kHz transducer (which was 4 feet long and fit into a 4-inch well) and a 40kHz transducer (which was 2 feet long and fit in a 2-inch well) after 90 minutes of treatment. Kinetic studies of core samples showed that the 25kHz transducer improved half-life by 64% to 73% and the 40kHz transducer improved half-life by 40%.

From these laboratory and field applications, Geiger and her team concluded that:

• Using iron (in PRBs) to degrade TCE and other organics in ground water produces various species of inorganic deposits. (This has been verified by other researchers.)
• Applying ultrasound to the iron can remove these deposits and increase the reactivity of the iron.
• There is no evidence of changes in water chemistry due to ultrasound treatment of these inorganic deposits, although there is some initial increase in turbidity.
• The ultrasound process is easy to mobilize, costs little, and requires little equipment and personnel.
  

Audience members asked questions. Someone asked whether ultrasound lowers the permeability of barriers. Geiger did not expect it would: it would only be possible if extremely small fines or silts were produced, but, in her laboratory work, she has not seen evidence that this happens. Instead, the surface area of the barrier is increased without permeability being lowered.

Another attendee asked a question about the migration of particles, the flow rate of the columns used in the laboratory experiments, and the methodology used to measure permeability. Geiger replied that her team used 3-foot columns, and that they had a flow rate of 6 feet per day. To measure permeability, her team looked at pressure change. When ultrasound was applied, particulate flowed out of the column with the effluent, the column color returned to normal, and the pressure drop across the column was reduced.

In response to a question on whether her team checked for subsidence, Geiger responded that there was no evidence of subsidence even with a 40-foot-deep PRB. An audience member asked whether Geiger's team analyzed what was vibrated off from the columns in the laboratory experiments, and Geiger replied that they did not. Answering a question on whether particles were moving elsewhere out of the system due to the ultrasound application, Geiger said that she did not see that it would cause a problem, considering the way the particles moved in the column experiments.

The Innovative Use of High-Pressure Jetting of Thin Diaphragm Walls To Construct Hydraulic
Control Barriers
Richard Landis, DuPont Specialty Chemicals

Richard Landis presented information about a project that was conducted at the National Environmental Test Site at Dover Air Force Base (AFB). The project's objectives were to demonstrate that high-pressure jetting can create thin diaphragm walls in the ground that act as hydraulic control barriers, and then to verify barrier continuity and develop preliminary cost information. Landis said that the site was very well characterized, noting that soils were poorly sorted sand, hydraulic conductivity was 1 × 10-7, and the subsurface included a confining unit and a saturated zone.

Landis said that jetting technology provides a number of advantages. For example, it works especially well in sites where conventional techniques are problematic, such as sites with flowing sand conditions, sites near or around foundations (including underpinning foundations), and sites near underground utilities. The system also makes it possible to jet a discrete zone below the surface. Instead of having to dig all the way down to that discrete zone, one can drill down to the zone and apply the treatment, pull out, move over, and drill again. Also, the technology can be used in deeper applications (100 meters or so).

Landis described the high-pressure jetting system and how it creates a barrier. The jetting system is basically a high-tech drill rig. At the desired depth, one activates the system's series of nozzles to begin the jetting process. Then the drill is slowly extracted. This creates a grouting zone. To create the barrier, Landis said, air and grout are applied at a rate of 160 centimeters per minute. To create panel structures, the drill is not rotated as it is extracted. To create columns, the drill is rotated.

Landis showed the test emplacement panels his team used for the Dover AFB demonstration effort. Before going to the field, he said, the hydraulic conductivities of the barriers were measured in the laboratory. This step was taken so that researchers could evaluate emplacement methodology and barrier permeability.

Landis said that he used a cofferdam design that was somewhat circular. Twelve intersecting panels had to be created for it. He said that each wall was 9.3 feet, the cofferdam had a diameter from 3 to 5 feet, the jetted zone was below grade between 2 and 43 feet, and the grout protocol included water, cement, and bentonite. A guidance tool was used to ensure that the cofferdam barrier would not have gaps. The tool was used to adjust the nozzles' orientation, inclination, and heading. To make the guidance tool work, Landis said, an electric connection was established in the slurry using a wet connect.

Landis showed the construction results of the pilot test. The guidance tool provided information on where the cofferdam walls actually wound up and determined where the design was compromised. Landis then went back in and patched the design in several places, added monitoring wells and flood wells, then covered the cofferdam with two layers of geotextiles and geomembranes.

Landis said that he also conducted a variety of flood tests on the cofferdam barrier (up to 15 gallons per minute) and ran these until steady-state conditions were met. Based on Darcy's law, Landis said, he predicted that the cofferdam barrier would have an average hydraulic conductivity of 1 × 10-7. The actual results, however, indicated that conductivity was 2.52 × 10-6. This indicates that there was a leak somewhere. In an effort to find it, Landis ran hydraulic pulse tests. This helped pinpoint the location of the leak—the spot where the pulse signal was strongest.

Landis said that the results from the pilot test have lead him to conclude that jetting is cost-effective, especially for situations in which conventional techniques are problematic. He has also determined that to construct adequate barriers, one must have automated and onboard data acquisition (this type of data acquisition is becoming available in the United States; currently, it is more available in Europe), and that deep emplacements might require additional guidance tool adaptation and development.

Wall-and-Curtain for Subsurface Treatment of Contaminated Ground Water
David Lee, Atomic Energy of Canada

David Lee noted that his team has worked at a site along the Chalk River in Ontario: a former storage area for nuclear waste, now used to produce isotopes for medical use. The site has a contaminated plume, he said, noting that it was formed in the 1950s when process upsets occurred at a fuel-processing plant. Following the accident, studies were performed to predict how long it would take for the plume to reach a nearby wetland. The predictions suggested that the wetlands would be impacted in the 1990s; data collected from the area confirm that this actually occurred. Lee presented data on different contaminant transport rates. For example, he provided information on strontium-90's rate, noting that this contaminant is of great concern in low-calcium wetlands because it tends to accumulate and replace the calcium that plants and invertebrates conserve.

Lee said that his team constructed a wall-and-curtain barrier to address the ground-water plume. This PRB, he said, had two special elements. First, the PRB allowed for direct monitoring of permeability. Second, the width and depth of the capture zone was adjustable. Lee showed the PRB's design: it included a 100-foot-long wall with steel joints, a series of 10 vertical wells, steel sheet piling, a flexible hose that adjusted flow rate, and a curtain of zeolite material. The latter, which created an ion exchange reaction, had a hydraulic conductivity 10 times that of the aquifer (1 × 10-2 cm/sec versus 1 × 10-3 cm/sec).

Before constructing the PRB, Lee's team created a sandbox model of the barrier to determine if it would remediate contaminants. To ensure that particulates were moving in the aquifer and that reactions would be what they expected, Lee's team conducted in situ column testing using a string of columns loaded with strontium-90. Using the results of the column tests, Lee did modeling to determine a Kd, which was around 2,000. With these data, Lee showed that even if the barrier ran at 2 meters per day they would see a centimeter per month of improvement through the material. (The test did not provide information about preferential flow or packing.)

Lee said that his team used the column test results to determine the technical specifications for the PRB. Once this was accomplished, the PRB was constructed. During the course of construction, Lee said, bedrock was encountered and this prevented the team from installing the PRB as deep as they had hoped.
The PRB has been operational since December 1998. Flow measurements are collected monthly. In addition, the water table is monitored every few months, and Lee frequently measures discharge versus capture width. Lee indicated that his measurements show that the PRB is effectively reducing gross beta—the effluent gross beta was down to below 0.6 Bq/L of strontium-90, while the inflow was as high as 85 Bq/L (although this will increase in the future). Lee's team is now measuring suspected leakage at the wall, and is planning to do grouting to prevent leakage.

Performance of a Deep Iron Permeable Reactive Barrier for Ground-Water Remediation of Volatile Organic Compounds
Grant Hocking, Golder Sierra LLC

Grant Hocking said that his team constructed a PRB at a Superfund site in Iowa that had a contaminant plume extending 1,500 feet. The plume's predominant contaminant was TCE; concentrations were as high as 10,000 ppb. Hocking said that the PRB installed at the site was 240 feet long and 75 feet deep. The reactive material utilized was iron, he said, noting that reductive dehalogenation reactions that occur between TCE and iron render the contaminant nontoxic.

When designing the PRB, Hocking said, his team performed column reactivity tests on the site's ground water to determine degradation rates for various chlorinated solvents and to find out what daughter products would be generated. The tests were also used to address precipitation and clogging issues. One test involved a first-order reduction by zero-valent iron. Hocking showed his measurement of the degradation of TCE and the generation of cis-dichloroethylene (cis-DCE).

Hocking also described how the PRB was constructed. First, his team installed a specialized casing in the ground. Then they created a fracture at a particular azimuth orientation. Vertically oriented hydraulic fractures were used to create the PRB. (Hocking provided details about the fracturing process, noting that it requires specialized equipment and fluids. An audience member asked what formation properties are conducive to vertical hydraulic fracturing. Hocking said the process can be completed in silts and sands, but not in overconsolidated materials such as clays.) Hocking showed the mixing and pumping equipment his team used, as well as the injection control and data acquisition systems that determined where the material was going in the subsurface. The fluid that carried iron filings into the ground had to have a very high viscosity, he said, or the iron filings would have dropped out of it. A clean-breaking fluid was also used to achieve optimal barrier permeability. Construction costs were about $700,000.

Hocking said that the PRB's performance is being assessed using downgradient monitoring wells. Currently, the PRB is monitored every quarter. Results indicate that TCE concentrations have been lowered a great deal. The ground-water flow within the 3-inch wall has also slowed to half its normal speed, or 30 feet per day.

Successful Remediation of Solvent-Contaminated Ground Water Using a Funnel-and-Gate System
Constructed by Slurry Trench Methods
S.R. Day, Geo-Solutions Inc.

S.R. Day described a site in Ballard, Washington (just north of Seattle), in an industrial area near a railroad. Windows were once manufactured at the site, he said, noting that various chlorinated solvents were used in the process. A ground-water plume, contaminated with tetrachloroethylene (PCE) and DCE, has formed under the site.

Day said that remediating this site has been challenging because: (1) the subsurface has a complex mixture of soils, (2) the ground water pH is high, and (3) the ground water has low flows and a high gradient. Previous attempts to clean the site with pump-and-treat technologies have failed; thus Day decided to assess alternative technologies. After considering a number of technologies, Day chose to install a zero-valent iron funnel-and-gate PRB. (Each gate is about 40 to 50 feet long each, and the funnel is 330-feet long.) Before implementing the technology, Day conducted a dehalogenation bench test to determine how much iron was needed and how thick the walls should be.

Day said that his team encountered several construction challenges; these arose mainly because of space restrictions and complex subsurface soil structures. Day's team needed to work in a very limited space that was 70 feet at its widest along a railroad track that was still in use. Also, Day needed a place to put contaminated soils. A mobile test laboratory on site was used to determine if some of the excavated contaminated soils needed to be removed from the site. To deal with the soil conditions, Day used the slurry trench method during construction activities. This method allows one to excavate deep, narrow vertical trenches in soils that cannot otherwise stand unsupported without shoring and/or dewatering. During excavation, the vertical walls were supported by keeping the trench filled with bentonite slurry.
Day used a biodegradable slurry and self-hardening cement-bentonite (CB) cutoff walls for the funnel section.

Day showed pictures of the CB slurry wall installation. (He mentioned that, when they installed the trench, his team broke a storm sewer—they had been told it was steel-encased, but it was actually encased in vitrified clay. They were asked to abandon it in place instead of fixing it.) Once the team had installed the slurry wall, they filled the trench up with slurry.

Day also showed the wall profile and pictures of iron and sand mixing to produce the gates of the PRB. Day said that his team used a conveyer to transport the iron and a tremie pipe to place the mix into the excavation pit. The slurry needed to be broken, Day said; this was accomplished by adding enzyme breakers and pumping and recirculating the slurry. After completing the trench in November 1999, Day conducted monitoring to determine how well the funnels were directing ground water to the gates. He also analyzed changes in chlorinated hydrocarbon concentrations. The results indicate that the contaminants, including PCE, are much lower than the initial concentration of 24,000 ppb. Day noted that the PRB was relatively cheap for the owner (about $300,000), construction took only a month, and the barrier achieved 99% removal efficiency.

An audience member asked how the high pH is affecting the iron in the PRB. Day said he has not seen any changes yet in iron and that the cause of the high pH has never been determined. He expects the iron to change in the future. In response to another question, Day said it was not a problem to get separation of iron and sand at a 1:1 ratio. Day's team chose this ratio because of cost considerations.

Use of Blast Fracturing and In Situ Treatment Agents for Passive Treatment of a Chlorinated Solvent Plume in Bedrock
V.B. Dick, Haley & Aldrich, Inc.

V.B. Dick opened his presentation by noting that it is challenging to remediate contaminants in bedrock and other difficult geologic formations. At the Princeton Plasma Physics Lab (PPPL), he said, a combination of three remediation technologies are being used to address a ground-water plume in a heterogeneous bedrock geology. Dick said that the plume is contaminated with chlorinated solvents, but no dense nonaqueous-phase liquid (DNAPL) is present. Dick showed isopleths for PCE for the plume—the highest concentration was at 280 ppb—and described the subsurface conditions. He said that the Remedial Action Selection Report for the site determined monitored natural attenuation to be a suitable management approach because there was not a high concentration of contaminants and the migration of the contaminants was controlled. PPPL, however, wanted to reduce its baseline costs for long-term monitoring by shrinking the ground-water plume. As a result, PPPL has suggested an alternative approach: combining three well-demonstrated but previously uncombined technologies. The technologies are bedrock blast fracturing, hydrogen release compound, and zero-valent iron.

Dick explained what is involved with blast fracturing, noting that it involves explosives. Test blasting is always conducted first, he said; blasting can be performed safely around sensitive utilities. The technology is used to overcome highly variable subsurface permeability. Once applied, a single zone of very high permeability is created and water can then be collected and distributed through that zone. This blast fracturing approach, Dick said, is similar to funnel-and-gate flow diversion, but uses high-permeability material in the formation. Dick said that the project involves blasting two parallel trenches, and adding zero-valent iron to one and hydrogen release compound to the other. He did note, however, that there are still some outstanding issues slowing implementation of the remediation method. The technology is almost cleared, in terms of regulation, but some technological issues (involving implementation and design criteria) remain.

An audience member asked about the size of the remediation zone. The system, said Dick, is 30 feet deep overall, with 20 to 25 feet in bedrock. Only the bedrock section is blasted. In response to a question about reduction in permeability due to the presence of iron, Dick acknowledged that this was a distinct possibility. From his experience with other projects using blasted bedrock fracturing, however, Dick believed such reduction would be rather low (10% or less). One attendee asked whether sufficient retention time will be obtained. Dick said that retention time depends on the amount of iron added. To address this issue, Dick is using two to three times the thickness of iron necessary for the source concentrations at the site.

PRB FIELD CASE STUDIES—HYDROLOGIC AND GEOCHEMICAL MEASUREMENTS, CONTAMINANT REMOVAL

Performance Monitoring of Permeable Reactive Barriers: Hydrologic and Geochemical Assessment Methods
G.R. Moline, Oak Ridge National Laboratory

G.R. Moline said that he conducted a study of existing PRB field demonstrations and column studies. He said that his study had many objectives: evaluating the impact of heterogeneity in both the aquifer and permeable barrier on PRB performance, assessing the relationship between hydrological and geochemical processes within PRBs, evaluating the relative effectiveness of different monitoring methods, and evaluating data needs for hydrological and geochemical models.

Moline discussed performance assessment briefly, listing the following as useful criteria: (1) the barrier's ability to capture the contaminant plume and prevent bypass or underflow, (2) the adequacy of resident times, and (3) the change in performance over time. Data needs for evaluating success in meeting these criteria, he said, include ground-water flux, PRB residence times, contaminant flux, reaction rates, and dissolved gas concentrations. These data must be obtained over space and time. An important question to answer is how often data need to be obtained and how much data is needed.

Moline's study investigated PRBs that are already in place. For example, Moline looked at a funnel-and-gate system in Monticello, Utah, that is treating uranium. The permeable barrier in this system, consisting of iron and gravel, is keyed into the bedrock. Intense hydrological and geochemical monitoring has been conducted spatially and temporally at the site for some time. Colloidal borescope tests, Moline said, show that heterogeneities in water flow and hydraulic gradients are not remarkable; however, bromide and iodide tracer injection tests showed that contaminant effects were occurring. Moline said that he used data from the tracer tests to calculate transport velocities and determined that there was a high amount of heterogeneity in these velocities. Given the results of the tracer tests, Moline said, colloidal borescope tests might not be applicable to funnel-and-gate PRB systems. (The convergence of flow lines in these systems probably affects the accuracy of colloidal borescope tests.)

Alkalinity and pH contour maps, Moline said, were used as geochemical indicators at the Monticello site. Previous studies have shown that when ground water is equilibrated with zero-valent iron, the pH of the water rises. Using the pH contour map, Moline saw that, researchers showed that there was a general increase in pH as water flowed through the barrier. pH did not increase everywhere: Moline suggested this might be because of underflow that was limiting residence time in the barrier. The alkalinity contour map was consistent with the pH contour map.

Moline provided a brief summary of the results that have been obtained at the Monticello site. Uranium concentrations drop to nondetectable levels as the contaminated water flows through the barrier. Calcium shows the same pattern, but there is some breakthrough that may be due to underflow or bypass. Geochemical tests reveal significant temporal variability in influx of uranium into the barrier.

Moline concluded that heterogeneity within sediments above and below the Monticello PRB and within the PRB itself can impact flux distributions and velocities; heterogeneity can develop within the PRB, causing the development of preferential flowpaths, faster velocities, and longer residence times; hydraulic heads alone are not sufficient to determine the magnitude and direction of flow through the barrier; and geochemical parameters appear to be linked to hydrological parameters and can be used as indicators of hydrologic performance.

In response to a question on whether an increase in gradient and velocity through the gate structure might have caused the heterogeneity in the Monticello PRB, Moline said this was a significant factor. Moline said the PRB was designed to handle 2 feet per day of ground-water flow, but the flow has been higher.

Another audience member asked whether the zone where iron density is lower than in other areas is allowing shorter residence time and causing a breakthrough. Moline said this is possible. He suspects that there was some smearing at the interface of the barrier and aquifer material that might be causing this.

Moline did residence time calculations and determined that the design flow rate was lower than what the tracer tests showed—2 feet per day compared to as much as 8 to 13 feet per day. Either way, the residence time was still sufficient to immobilize uranium. Underflow could be occurring because the PRB is keyed to the bedrock and might be fractured.

Carbon and Sulphur Accumulation and Iron Mineral Transformations in Permeable Reactive
Barriers Containing Zero-Valent Iron: Impact on Long-Term Performance
Rick Wilkin, U.S. Environmental Protection Agency (EPA)

Rick Wilkin said that he has investigated long-term performance issues—specifically, the way mineral and microbial biomass buildup can impact the performance of PRB systems. This buildup can cause infilling of porosity and other problems. Wilkin's project represents the EPA contribution to an initiative, involving three federal agencies, that is facilitating the exchange of technical information. Field sites include a U.S. Coast Guard site in Elizabeth City, New Jersey, and a site at the Denver Federal Center in Lakewood, Colorado. At both sites, PRBs using peerless iron were installed in 1996, but the sites have different PRB designs (continuous wall versus funnel-and-gate) and site conditions (hydrologic and geochemical). At the Elizabeth City site, after 5 years, there has been consistent degradation and removal of contaminants with one exception: one cell has not been as successful, and has also had a buildup of precipitation and microbial biomass.

At both sites, Wilkin said, he wanted to identify the types of mineral phases forming, quantify how much mineral and biomass accumulation has occurred, evaluate the spatial and temporal trends of accumulation, and determine the impacts of precipitates and biomass on the performance of the systems. Wilkin also wanted to gather data that could help form predictive models for this phenomenon and identify parameters that can be measured to serve as early warning indicators.

To determine the type of mineral precipitates that might form, Wilkin examined the thermodynamics of the systems. He saw many of the minerals he expected. Different patterns emerged at each site. An x-ray diffraction study showed, however, that at both sites, green rust (a type of carbonate) and magnetite formed, but not calcium carbonates.

Wilkin said that he also examined carbon and sulfur systematics. For carbon systematics, iron corrosion drives the production of hydroxide, which increases pH and leads to precipitation of calcium carbonate and iron carbonate. For sulfur systematics, iron corrosion drives the production of hydrogen, which is used by bacteria to break down sulfur, leading to higher concentrations of iron sulfate. For both carbon and sulfur, Wilkin saw increased concentrations of mineral precipitates at the upgradient interface and decreases in the PRB itself.

Microbial biomass accumulation showed the same pattern as mineral precipitates. To examine the effects of microbial biomass accumulation on long-term PRB performance, Wilkin said, he used a model to estimate porosity loss.

Wilkin concluded that zero-valent iron is a long-term sink for carbon, sulfur, calcium, magnesium, and silicon; mineral accumulation varies spatially and most precipitation occurs at the upgradient interface; porosity loss at the interface depends on total dissolved solids, ground-water composition, and flow rate; iron is largely conserved in PRB systems even though it is involved in many reactions; zero-valent iron is an iron(II) source to downgradient regions with increasing flow rate; and mineral and biomass accumulation correlate positively with total dissolved solids and negatively with flow rates.

Geochemical Investigation of Three Permeable Reactive Barriers To Assess Impact of Precipitation on Performance and Longevity
B. Sass, Battelle Memorial Institute

B. Sass opened his presentation by expressing a concern that is commonly heard: PRBs might not last as long as necessary, given how long contaminants are expected to last at particular sites. To address this concern, Sass said, Battelle has been researching how to better understand the geochemical factors that contribute to the longevity of PRBs.

Sass said that Battelle's performance evaluation of PRBs involve both field and laboratory investigations. In the field, Battelle looked at contaminant concentrations in ground-water inflow and effluent at three PRB sites, conducted geochemical modeling of the ground water, took iron cores from the PRBs to look at precipitates, and performed hydraulic modeling. In the laboratory, using column tests, Battelle simulated long-term field conditions to estimate PRB performance and looked at degradation rates of TCE as the iron aged. At the conclusion of the study, Sass offered a detailed analysis of corrosion compounds.

Sass presented an evaluation of two field sites. The first site—at Moffett Federal Airfield in California—is a pilot-scale funnel-and-gate design that uses 75 tons of Peerless iron. It was constructed in April 1996. Sass looked at cores from this site collected 15 months after it was constructed and also cores from 5 years later. The second site Sass evaluated is the former Lowry Air Force Base in Denver, Colorado. This site, constructed in 1995, also has a funnel PRB design. It uses 45 tons of Master Builder's iron. Cores were collected from this site in 1997 and 1999.

Sass said that the preliminary geochemical findings of the ground-water analysis indicate a very significant loss of inorganic compounds within the iron reactive cell of the PRB. Sass performed some geochemical modeling to obtain insight into the reactions that might be occurring. An analysis of core samples indicated that the mass of precipitate buildup was much lower than expected.

Using ground-water samples from the Moffett Federal Airfield site, Sass evaluated the inflow concentrations of inorganic species and examined the effluent precipitates. The concentrations of many inorganics (calcium, magnesium, bicarbonate, nitrate, and sulfate) were lowered significantly; however, concentrations of chlorides did not drop much. Sass made predictions based on geochemical modeling in which iron was incrementally reacted with site-characteristic ground-water.

Next, Sass performed core sampling of the PRBs in the field to see if his predictions could be confirmed. In addition, in the laboratory, Battelle conducted some accelerated column testing of ground water from each field site to assess the long-term performance of PRBs. Sass showed the results of the sampling and tests for each field site. He saw similar results from Moffett Federal Airfield and Lowry Air Force Base, except that the Master Builder's iron (used at Lowry) seemed to be aging faster than the peerless iron (used at Moffett).

Sass also looked at the silt depositing on the bottom of some of the ground-water monitoring wells. Using a scanning electron microscopic, he saw that the silt contained calcium and ettringite, which modeling did not predict. Using an energy dispersive spectrometry microprobe, Sass developed maps of the silt. These showed that a significant amount of calcium carbonate is present.

Removal of TCE and Chromate in a Permeable Reactive Barrier Using Zero-Valent Iron
Peter Kjeldsen, Technical University of Denmark

Peter Kjeldsen opened his presentation by noting that TCE and chromate, both frequently used at electroplating sites, are mobile contaminants commonly found in aquifers. They both can be removed by PRBs with zero-valent iron. One concern, however, is that when chromate is reduced it precipitates, potentially leading to passivation of the zero-valent iron material in the PRB.

Kjeldsen said that he performed a study to determine the chromate-removal capacity of zero-valent iron using different iron types, chromate concentrations (20 ppm and 300 ppm), ground-water composition, pH values, iron content, and flow rate. Kjeldsen also studied the effects of chromate removal on the simultaneous TCE degradation rate.

Kjeldsen said that he carried out column experiments; 30 columns were used, some of which were controls. He looked at TCE degradation rates before and after the addition of chromate. Kjeldsen showed the results of the experiments: they show that pH might be a good indicator of the remaining capacity of iron for removing chromate. Kjeldsen also saw that there was a slightly lower chromate-removal capacity when chromate concentrations were high (300 ppm) rather than low (20 ppm). Kjeldsen also conducted laboratory experiments that investigated the removal of magnesium and calcium by PRBs, as well as the relationship between pH and relative chromate concentration. From the results of his study, Kjeldsen predicted the lifetime of PRBs affected by chromate to be between 15 and 20 years.

Kjeldsen concluded that zero-valent iron has limited chromate-removal capacity: 1 mg to 3 mg of chromate per gram of iron, slightly more at lower chromate concentrations. Mixing iron in sand neither increases nor decreases chromate-removal capacity, decreasing the pH of inflow to 4 significantly increases removal capacity, and higher hardness gives lower chromate-removal capacities. Kjeldsen also determined that pH is a good indicator of remaining chromate-removal capacity and that limited removal capacity might affect the longevity of PRBs at high chromate loadings (not concentrations).

In response to a question, Kjeldsen said that, in his study, the TCE degradation rate changed after chromate was loaded. This indicates that full passivation of the PRB by the chromate also affects the PRB's ability to degrade TCE.

Performance Monitoring of a Permeable Reactive Barrier at the Somersworth, New Hampshire,
Landfill Superfund Site
Timothy Sivavec, General Electric

Timothy Sivavec described a performance monitoring study that was conducted on two pilot-scale PRB installations at the Somersworth Landfill Superfund site in New Hampshire. He said that the remediation approach being used at the site is complex, noting that it involves PRB technologies, natural attenuation, ground-water extraction, and a permeable cover. The PRBs are being used to treat the landfill's overburden ground water. The first pilot-scale PRB, which became operational in 1996, used a funnel-and-gate caisson. The second pilot-scale installation was installed in late 1999, and used a biopolymer slurry in the construction phase. In 2000, the bioslurry polymer was used again; this time, to install a full-scale 915-foot continuous wall.

Sivavec said that the pilot-scale PRBs' performance was monitored using an innovative sampling methodology that included passive and semi-passive sampling techniques. These techniques are preferable to others, he said, because they minimize hydraulic perturbations.

Sivavec said that ground-water monitoring wells were installed and sampled to monitor the first pilot-scale installation. The analytical results indicated a significant reduction of VOCs—50%—between the upgradient aquifer and the reactive wall. Sivavec said that this was likely due to biodegradation via sequential anaerobic and aerobic processes, and a reduction in alkalinity. The results also revealed some precipitation around the barrier perimeter, a phenomenon that could affect porosity and the PRB's iron reactivity. Sivavec said that he quantified mineral precipitation using: (1) a simple aqueous inorganic profile analysis, and (2) a more precise reactive zone coring and iron analysis. The latter employed quantitative methods such as auger spectroscopy and x-ray photoelectron spectroscopy, or XPS. As for biofouling, Sivavec said, no significant biomass formed, but some sulfate-reducing bacteria were identified. There was no indication that microbial biomass affected the PRB's performance. According to molar volume calculations, said Sivavec, the calcium and alkalinity profile indicated a porosity loss of less than 3% over a 1-year period. XPS showed a range of 0.5% to 2.5% loss of porosity over 18 months. At first, Sivavec believed that porosity losses in the iron zones were due to hydrogen evolution and the fact that ferrous hydroxide production was higher than expected. Later, he realized that the losses were due to mineral precipitation. In low-alkalinity and low-hardness waters, Sivavec found ferrous hydroxide and ferrous carbonate. In waters with higher alkalinity and hardness, there was more calcium carbonate.

To control mineral precipitation in the iron zones, Sivavec proposed controlling the rise in pH by adding acidic minerals. Sivavec discovered a ferrous sulfide mineral that is highly reactive and reducing. However, even when his team was able to control bulk pH in the aqueous phase, they were not controlling precipitation.

For the second PRB pilot-test, Sivavec said, 6 months of data were collected using in situ ground-water quality probes to determine the effects of biopolymer on hydraulics and reactivity. Diffusion samplers were used for VOC analysis of ground water, which is a passive technique especially useful for PRB sites.

Sivavec said that the performance monitoring studies provided useful information. For example, the studies led him to make the following conclusions: (1) mineral precipitation had less of an effect than expected based on other sites and accelerated aging column studies, (2) biofouling is not significant, and (3) ground-water parameter probes are useful. He said that the extent of performance monitoring that is needed for a site should decrease as PRBs are advanced and become a more widely accepted technology.

In Situ Treatment of Acid Mine Drainage in Ground Water Using Permeable Reactive Materials
David Smyth, University of Waterloo

David Smyth said that organic carbon can be an effective material to use in PRBs; in fact it is the best material to use when treating acid mine waste and high-sulfate waters in situ. Adding organic carbon to these types of waste, he said, generates hydrogen sulfide and causes iron sulfide to precipitate. In the process, about 10 to 20 mg of sulfate can be removed per meter per day. The process is slow, however; thus, in order to be effective, the wastes must spend a substantial amount of time in the PRB. (The residence time of contaminated ground water in the PRB is critical to the level of sulfate reduction.)

Smyth said that an organic carbon thermoreactive barrier is being used to treat ground water at the Nickel Rim mine, a site with sulfate concentrations in excess of 4,000 mg/L, iron concentrations up to 1,200 mg/L, slightly increased acidity, relatively low alkalinity, and ground-water flow rates of 16 meters per year. Smyth said that a PRB has been installed at the site, in the ground adjacent to a tailings pile. The organic carbon material used in the PRB consists of wood chips, municipal compost, and limestone. The barrier, installed in June 1995, is 15 meters long and 8 meters thick (with 4 meters of reactive material). Sampling results, Smyth said, indicate that the PRB is reducing sulfate considerably: several thousand milligrams per liter of sulfate have been reduced each year in the water within the barrier. Despite these successes, Smyth said, a number of challenges are also being encountered with the PRB. For example, the barrier has heterogeneities within it, including a high-flow zone in the central part of the barrier. Also, the location chosen for the PRB does not optimize residence time. In addition, Smyth said, there is some evidence that the PRB's reactivity is decreasing with time, although it does appear that the PRB has many years of reactive potential remaining.

Smyth described two other field sites at which he has used organic carbon. At one site, a storage tank leaked copper, cadmium, zinc, nickel, and sulfate into the ground water. Concentrations of the latter, he said, ranged from 300 mg/L to 1,000 mg/L. After installing a trial PRB (5 meters long, 2.5 meters thick, and 6 meters deep), copper was reduced from 20 mg/L to 10 mg/L, with similar results for zinc. At the other site, Smyth said, tailings were present and the site was contaminated with natrojarosite, which can release very high levels of sulfate and iron. Sulfate concentrations at the site ranged up to 3,000 mg/L and iron concentrations reached up to 10,000 mg/L. At this site, an organic carbon mixture was placed directly into the tailing materials. Significant reductions in sulfates and iron concentrations were observed.

MATERIAL DEVELOPMENT OF PRBs

The In Situ Treatment of DNAPL With Zero-Valent Iron Emulsions
Christian Clausen, University of Central Florida

Christian Clausen began by describing the problems with current methods for treating DNAPL contamination. Steam injection requires a great deal of energy and has a limited distance of effectiveness. Oxidation technologies, such as KMnO4, enter the subsurface as a hydrophilic phase, and since DNAPL is hydrophobic, makes it difficult to effectively treat large quantities of DNAPL. A need exists for a hydrophobic remediation technology that can incorporate itself within the DNAPL phase. This would allow for the remediation of the entire contaminant source.

Iron has been successfully used for the past decade to remediate dissolved phase chlorinated aliphatic compounds; however, by itself it is hydrophilic. In order to create a method that would allow the iron to interact with DNAPL phase contaminant, it was theorized that an emulsion with an oil exterior could carry the iron to the contamination. Laboratory work has involved synthesizing a variety of emulsions using corn oil, surfactant, water, and microsize or nanosize iron. Since the delivery method for this technology requires some kind of pumping, experiments also have focused on the stability of the emulsions when pumped through soil columns.

The ideal emulsion must have a hydrophobic membrane, contain reactive iron and water, and be more dense than water. Over one hundred combinations of emulsion components were tested to determine the most active, stable formula. Anionic, cationic, and nonionic surfactants were evaluated. Initial studies focused on Rhodacal N, which appeared to provide the best overall performance, but later studies focused on food grade surfactants.

The emulsions were found to fall into a Winsor Type II category, which indicates an oil exterior membrane and a water interior. Micrographs of the emulsions (both microscale iron and nanoscale iron) verified the structure of the micelles. Visual studies were performed where emulsion (or unemulsified iron) was added to dyed TCE and water. Unemulsified nanoscale iron simply floated on top of the TCE; however, the emulsion enveloped and combined with the TCE. There was no visibly separate TCE phase in the vial containing water, the emulsion, and TCE.

Studies were also performed to determine if the emulsion would challenge TCE in a sand matrix. Jars were filled with sand, water, and dyed TCE. When the emulsion was added, the TCE migrated into the emulsion. This led to the eventual near disappearance of the TCE.

Clausen presented kinetic studies that were performed on a variety of the most stable emulsions. For dissolved phase TCE, the first order rate constant for the disappearance of TCE was less than one order of magnitude less than the rate constant for unemulsified nanoscale or microscale iron. In kinetic studies with DNAPL phase TCE, headspace analysis was used to measure ethylene production as a measure of TCE destruction. The production of ethylene was two to six times greater for emulsified iron than for unemulsified iron. Chlorinated byproducts were not detected in these studies (by simple headspace analysis), indicating that the emulsion micelles are holding on to the byproducts interiorly until the nonchlorinated byproducts form and diffuse out of the micelles. When the vials were exposed to ultrasound for 3 minutes, cis-DCE and vinyl chloride were detected in concentration of less than 400 µg/L and less than 80 µg/L, respectively.

Pumping experiments showed that both nanoscale and microscale iron emulsions can be pumped with water through a two foot long by two inch diameter column packed with native soil (obtained from Cape Canaveral Air Station) with 32-36% porosity.

From these studies, Clausen concluded the following: the emulsion system can challenge a DNAPL pool as shown by visual addition experiments and closed jar experiments. Degradation rates for DNAPL TCE are much higher with emulsified iron than with unemulsified iron. The emulsions were observed to enter the DNAPL pool, resulting in a dehalogenation reaction within the emulsion-DNAPL phase. The emulsion micelles retain their structural integrity even after pumping through a soil matrix.

Future work includes a field-scale test at a DNAPL site located at launch complex 34, Cape Canaveral Air Station, Florida, later this year.

Permeable Reactive Barrier for Metals Treatment at the Newport, Delaware, Superfund Site
John Wilkens, DuPont Central Research and Development

John Wilkens said that a pilot demonstration PRB was installed at a site in Newport, Delaware, the former home of a pigment plant and landfill. Lithopone (ZnS-BaSO4) white pigment was produced at the plant and wastes and spent ores were deposited in the landfill from 1902 to 1953. Contaminants of concern that have been detected at the site, Wilkens said, include barium, zinc, copper, cadmium, lead, cobalt, nickel, and manganese. Only barium, zinc, and manganese, however, have been recorded at concentrations that exceed ground-water standards. Wilkens said that site characterization had shown a situation too complex to delineate areas as rich in one constituent or another.

Wilkens described the steps that he undertook when developing an appropriate PRB for the site. First, he carried out laboratory batch tests to investigate how well different materials treat barium and zinc. After analyzing the results, Wilkens decided to use Peerless zero-valent iron to treat zinc, and gypsum to treat barium. Then, he performed a series of column tests and found that the chosen materials would reduce barium and zinc below cleanup goals, but that they would not do the same for manganese. To find a way to treat the manganese contamination, Wilkens determined the conditions where manganese is stable by developing a Pourbaix diagram. With the information from the diagram, Wilkens tried to precipitate the manganese with a number of substances and found that magnesium carbonate worked well.

Wilkens then projected the life of the field wall to be used in the PRB. He investigated wall plugging potential and found there would not be any permeability loss over the entire life of the PRB. The results of Wilkens' tests and calculations helped him identify an optimal reactive mix for the site's PRB. This was created and then installed as a field demonstration project. The in situ reactive demonstration well reduced barium, zinc, and manganese below standards. A full-scale PRB will be implemented in 2002.

Wilkens determined the life of the PRB as a function of the cap that is placed on top of the landfill. In this case, Wilkens wanted a cap with a maximum permeability of 1 × 10-7 cm/sec, so he chose to use a geomembrane in combination with a geosynthetic clay liner. Wilkens predicts that the PRB will have a 600-year lifetime based on demonstrated field performance. (A separate calculation based on reaction and solubility losses predicts a life measured in millennia.) He estimated the cost of the project to be $5 million, noting that conventional remediation technologies would have cost $17 million. Wilkens concluded that this PRB achieves the desired performance standards and is essentially a permanent remedy.

An Examination of Zero-Valent Iron Sources Used in Permeable Reactive Barriers
Richard Landis, DuPont Specialty Chemicals

Landis said that the commercial iron (or "end products") used in PRBs comes from a mix of feedstocks and sources, including automotive parts. Thus, each end product is produced by different processing methods, and may have different characteristics. Landis said that he wants to know whether these differences affect the success of PRBs or have other unforseen effects.

In the field, he said, the average degradation rates of influent compounds passing through a PRB have been found to be different depending on the iron source used. To investigate this phenomenon further, Landis conducted a study focused on the relationships between the physical and chemical properties of iron feedstocks, end products, and degradation rates. Landis obtained samples of different end products and feedstocks and then determined the properties of each. Surface area, density, and porosity were measured, and scanning electron microscopy (SEM) and other microscopy methods were used to determine the physical properties. SEM showed a higher abundance of oxide coatings and other differences in morphology between the two types of iron. The end products were also analyzed through column tests to determine the difference in degradation rates of contaminants (10 mg/L TCE and 1 mg/L vinyl chloride).

Landis also looked at the forms of carbon in the iron and noticed two major forms (plate-like carbon and nodular carbon). Although the end products and the individual feedstocks had both forms, one iron feedstock had more plate-like carbon than nodular carbon.

A significant finding of the study was the apparent lack of correlation of degradation rates to the Brunauer, Emmett, and Teller surface areas. The column data showed there were slightly higher cis-DCE and vinyl chloride degradation rates in one end product than another.

Landis said the gross elemental composition of feedstocks from both commercial sources appeared similar; however, historical differences in the reaction rates between the different feedstocks was confirmed in recent samples.

Future work includes investigating the correlation between reactivity, carbon structure, and trace metal content; taking alternative measurements of the reactive surface area; characterizing the influence of milling and other processing procedures on reactivity; and determining the water corrosion rate.

Factors Affecting Long-Term Performance of a Granular Iron Permeable Reactive Barrier
Robert Gillham, University of Waterloo

Gillham opened his presentation by saying that some people are concerned that PRBs will not last long enough to be cost effective. He said that he believes this concern is unfounded, noting that there are many reasons to believe that PRBs (at least iron PRBs) are cost effective. PRBs have low operating and maintenance costs and—in the case of granular iron—exhibit long-term, maintenance-free performance. For example, using a study performed by Reardon in 1985, Gillham calculated that iron would persist in a PRB for 82 years (assuming that the PRB was 100% granular iron and had 50% porosity, and that the iron was consumed from corrosion by water). This would be a best-case scenario: in the field, PRB lifetime would be shorter.

Gillham spoke about the use of nanoscale iron for dehalogenation. He estimated that nanoscale iron's rate of consumption would be 1,500 times that of 100% granular iron; however, he noted that the former would only persist in the subsurface for 20 days. Longevity of the PRB would likely be more of an issue with highly active iron.

Gillham said that two factors that play a large role in determining a wall's longevity are iron surface activity and precipitate generation. With regard to the activity of iron surfaces, initial studies and investigations have used iron of high purity while field applications have used commercial-grade iron or processed scrap metal. (Pure iron is more reactive than commercial-grade iron.) Commercial-grade Connelly iron has a thick oxide film and coating. The inner coating, Fe3O4, is conducting, while the outer coating, Fe2O3, is passivating.

Gillham said that he has started performing tests to study commercial-grade iron, noting that he has set up column tests specialized for Raman spectroscopy to measure open circuit potential. Four different influent solutions (i.e., millipore water, 1.5 mg/L TCE, 100 mg/L nitrate, and TCE plus nitrate) were used in the study, and the effects that these influents have on outer iron coatings were evaluated. Raman spectroscopy indicated no difference between the water and the 1.5 mg/L TCE; however, the addition of nitrate caused passivation in the outer coating. In the presence of nitrate, the degradation of TCE occurred briefly but did not persist due to this passivation. Gillham said the column tests showed rapid (in a matter of minutes) autoreduction of Fe2O3 on the surface of commercial iron in the presence of water and TCE solution. There was persistent degradation of TCE over a 3-day period after the oxides were removed, indicating that Fe2O3 is unstable in the TCE solution. He also noted a minor amount of nitrate reduction and minor TCE reduction in the presence of nitrate (indicating Fe2O3 passivation).

As a follow up to this study, Gillham examined factors that could make Fe2O3 persist on iron. He looked at the effects of gas and precipitate formation on reactivity and permeability. To do this, he simulated the front 10 cm of a PRB, where most precipitates are apt to accumulate. In this simulation (another column test), Gillham used four treatments: distilled water, water with 300 ppm CaCO3, water with 10 ppm TCE, and water with CaCO3 and TCE. While running this experiment, Gillham evaluated the changes in column weight. There was an initial loss in weight, due to hydrogen gas generation within the column. Next, precipitation of CaCO3 caused column weight to increase. Gillham also looked at the effects on hydraulic conductivity. For the columns with CaCO3, it was difficult to tell whether hydraulic conductivity was any less than in those columns without it, even when the CaCO3 content took up 10% to 15% of the pore volume. On the other hand, hydrogen gas caused a large decrease in hydraulic conductivity initially. Gillham found that CaCO3 precipitated in iron columns supplied with carbonate solutions. The mass of the CaCO3 accumulated to a certain value and then leveled off. Gillham predicted that CaCO3 would reduce the reactivity of the iron and would accumulate to 0.1 g/cm3 of iron. He said the precipitate front would advance through the PRB at a rate of 1.3 cm/yr.

Gillham concluded, from his laboratory studies, that iron persists for long periods of time in 100% pure iron PRBs, but persistence might be an important issue if the iron is highly dispersed in the PRB material or is highly reactive. The autoreduction of oxides is an essential process that determines the performance of commercial iron, but other oxidants such as nitrate might interfere with the autoreduction process. Calcium carbonate precipitates form as a progressing front. Although precipitates do not cause a major decline in hydraulic conductivity, they can reduce iron reactivity where they form.

RTDF PANEL DISCUSSION
Led by Robert Puls, EPA
Panel members: Jacqueline Quinn (National Aeronautics and Space Administration), Richard Landis (DuPont), Liyuan Liang (Oak Ridge National Laboratory), Robert Gillham (University of Waterloo), and Stephan Jefferis (University of Surrey)

Summaries Provided By Panel Members

Robert Puls asked the five panel members to provide some introductory comments. The following comments were offered:

• Liyuan Lang said that numerous PRB field case studies are ongoing, and that more and more data are being released about them. She noted that many of the day's presentations had focused on field emplacement. She also said that she has noticed that there are more geochemical models being used to predict the types of minerals that might precipitate—in some cases quantitatively. Although coring is currently necessary to monitor the effectiveness of PRBs, Liang said, it might not be necessary in 5 years; at that point, only a simple sampling of indicators may be necessary. She mentioned that many complex geochemical models are being used that describe the PRB process very well; however, there is a concern that these geochemical models might be too complex for people in the field to use. Liang also mentioned that heterogeneity is still a major issue that must be dealt with in the field.
• Jacqueline Quinn made two general observations: (1) most of the day's discussion focused on PRB applications in the dissolved phase, and (2) PRBs are gaining regulatory acceptance. She also spoke about research that NASA is conducting, noting that an emulsified nanoscale demonstration project has recently been initiated at a NASA cleanup site. Over the next year, Quinn said, she hopes to share data on the use of nanoscale and microscale iron placed within emulsions. Another data set that NASA hopes to share in the next year will be from a project involving long-term pumping of a reactive barrier. Quinn said that she and EPA's Rich Steimle discussed initiating this long-term pumping effort to increase degradation and minimize precipitate fouling.
• Richard Landis noted that there is much interest in PRBs and that, considering the quality of the day's presentations, the future looks bright for this technology. Landis has been perplexed, however, because the rate of PRB installation appears to be leveling off. He is wondering why this might be. After considering a number of issues, he has determined that cost might be the central issue: the initial capital costs for PRBs are more than for pump-and-treat, although the payback is much more quick. PRBs' biggest capital costs at this point are for emplacement and materials. One way to save money is to use innovative emplacement methods and alternative materials. Landis believed that it is important to drive down the costs for PRB installation since, from an industry standpoint, cost is a central issue. He noted that many of the day's presentations had focused on improving the economics of PRB installation. DuPont, ETI, and other organizations are looking at improving the reactivity of the iron used in PRBs. Landis was encouraged by NASA's efforts to increase the reactivity of iron through the use of ultrasound. Increasing iron reactivity by just 20% would mean significant cost savings.
• Stephan Jefferis provided a European perspective on PRBs. He pointed out that PRBs can be strategically placed to maximize natural attenuation processes. For some contaminants, natural attenuation can usefully occur upstream of the barrier as well as downstream. When considering deposition of precipitates within a PRB, he said, it is necessary to consider chemical engineering issues, particularly mass balance. This is especially important for PRBs with high flow rates. Jefferis noted that a zero-valent iron PRB has been installed in Belfast. At this site, a significant plume existed downstream of the barrier so the remediation team pumped some of the downstream plume back to the upstream side of the reactor so that it could be treated by the PRB. After a few years of operation, no significant build up of mineral deposits has occurred, and reactivity of the iron has increased. It is clear, he said, that the scientific community's understanding of the in situ behavior of iron is still developing. While the Belfast reactor has shown an increase in iron reactivity, loss of reactive capacity may be particularly important for some species, such as chromium or uranium. Jefferis was not sure why increased reactivity had not been reported in any U.S. investigations. In Europe, he said, several reactor designs (including the one in Belfast) are designed with the understanding that reactive media may need to be replaced. As a result, funnel-and-gate configurations are more popular in Europe than the continuous walls which are preferred in the United States. Jefferis said that European designs have also addressed issues pertaining to soil and flow heterogeneities. In several reactors, flow has switched from horizontal to vertical. If vertical flow is used and the inflow is arranged to be at the top of the reactor, he said, then exhausted or clogged reactive media may be accessed and replaced.
• Robert Gillham said that the scientific community has put PRBs under a lot of scrutiny—more than any other innovative remediation technology. As a result, the organic chemistry and degradation processes occurring with PRBs are well understood. Current research questions are centering on the inorganic chemistry of PRBs, and much has been learned about this issue. There are 5 to 6 years of performance data at some commercial field sites, and results indicate that performance has been satisfactory and according to design. PRBs are now being constructed at greater depths, using, for example, guar gum slurry wall methods that can place PRBs deeper than 100 feet. Furthermore, contractors' innovations have improved installation methods. Those who implement PRB technology are in a stronger position to say whether the technology should or should not be used for any particular set of field conditions. Gillham asked why PRBs are not being used more frequently, given that no fatal flaw has been found as a result of scientific scrutiny and that there are many sites requiring cleanup in the United States. Gillham believes that the regulatory environment has posed an obstacle because it encourages use of the least expensive remedial option, which is to do nothing (not to be confused with monitored natural attenuation). Gillham believes that the market for PRBs will continue to be sluggish until the regulatory environment begins to insist upon cleanup rather than doing nothing.

Open Discussion

Puls opened the floor and invited audience members to provide comments on the day's discussion. Attendees obliged, focusing their discussions on two major topics: factors that place PRBs at a higher risk of "failure," and obstacles to PRB implementation.

One audience member said that, in his experience, a number of PRB projects have had problems, ranging from minor to significant, and these have mostly been caused by hydraulic issues. The problems have involved poor site characterization, pre-design investigation, and construction methodology. He said that practitioners and regulators hold many misconceptions about PRB technology because of these problems; a major reason PRBs have not been accepted as widely as they should be is that these issues have not been analyzed and discussed enough. In response to this comment, Puls said that the RTDF Web site was recently updated to include case studies on 17 additional PRB sites and that existing case studies on the site were updated as well. This was welcome news to the audience member. Quinn also responded to the audience member. She said that from the perspective of project management at NASA, it generally takes a long time to pass through regulatory hoops and it has taken awhile to develop appropriate site characterization tools. At one site, she said, NASA started investigating PRB use in 1992, but the remedy is only now, in 2001, being implemented. Part of the delay has occurred because NASA had to develop tools to characterize the DNAPL at the site. Landis commented that site characterization is costly and believes more money needs to be spent on it. He believed that the lack of adequate site characterization is one of the reasons for the failure of some PRBs. The audience member who had just commented agreed, saying that the major problem with PRBs has been the everyday issues involving hydraulics and other parameters, such as ground-water flow directions, hydraulic conductivities, soil properties, and soil reactions.

Another audience member said that the hydraulics issue is very important. He said that it is hard to show how much residence time PRBs are actually getting, even though barriers are probably working. He also said that it is very hard to design PRBs to ensure a particular residence time. Gillham agreed that hydraulics is the major issue in the debate over the effectiveness of PRBs, although he was not sure what to do about it. Gillham said that there is an effort to develop more tools to directly measure flow velocity, but said that there is still a need for instrumentation to measure or confirm hydraulics, such as flow direction (sometimes errors in determining flow direction can be as much as 45 degrees). Hydraulics need to be considered more in the pre-design stage. One audience member said that pump-and-treat faces the same hydraulic problems as PRBs. Gillham responded that when pump-and-treat has been used for cleanup (rather than control), the success record has been worse than for PRBs.

Another audience member said that, when a PRB is to be designed, regulators and the client need to establish a thorough definition of their goals. Once the PRB is constructed, the evaluator's view of success can be "ill defined" if a definition was not reached initially. The audience member asked if there were going to be any modifications or updates to the Interstate Technology and Regulatory Cooperation's performance monitoring criteria and standards. A regulator replied that there are no such plans now, though he believed it might be something to consider in the future.

Steimle responded to a comment that Gillham made during the first half of the panel session pertaining to the suggestion that the regulatory community might be responsible for the sluggish PRB market. Steimle said the government cannot dictate what remediation technology to use. EPA currently considers pump-and-treat to be the fallback technology. Steimle mentioned a potentially major issue: sometimes, regulators suggest PRBs but consultants do not agree to use them, perhaps because a PRB would reduce the amount of work available for them. Pump-and-treat systems provide more work for consultants, since these are long-term projects that require a lot of engineering to install and maintain. Steimle said potentially responsible parties usually rely on their consultants for advice on what technology to use. So there appears to be some favoritism on behalf of pump-and-treat. Gillham responded by saying that a Nuclear Regulatory Commission report indicates that existing incentives do not encourage cleanup. Current incentives favor delay, for example, by encouraging the testing of another technology to wait and see if it might work better.

Landis said that from an owner standpoint, it would be a very hard sell within his company to put in another pump-and-treat system ever again. Although DuPont only has one PRB in so far, three others will be in place shortly. Puls noted though that not all site owners are as environmentally responsible and forward-looking as DuPont.

Jefferis suggested developing recommendations for PRB implementation that address the average, straightforward site only—the "middle ground." In Europe there are national specifications for these average sites. Although academics and consultants would lose work, PRB technology would be used more often.

Quinn told Steimle that NASA wanted to shut down a pump-and-treat system they had, but it took them 3 years to do so: it took a lot of work to convince EPA regulators that shutting the system down was appropriate. Steimle replied that regulators feel most comfortable with pump-and-treat systems and that regulators do not prefer technologies that appear more risky and about which less is known. Quinn said that NASA and other federal agencies would never just use PRBs before all of the important scientific questions were answered for a site. Even when these questions have been answered, she said, federal agencies have been reticent to implement PRBs at full scale. Quinn expressed hope that in the next 5 years, a flurry of PRBs will be implemented.

Another audience member said that his company, DuPont, shut down a pump-and-treat system and it took them less than a year. This is because the regulatory environment was favorable. To (successfully) argue for the use of a PRB rather than pump-and-treat, his company showed that it would be better from an environmental and financial stand point. The project was successful because the regulator involved with the site was very informed, holding a clear understanding of PRB technologies and its limitations. In addition, there was a high level of trust between DuPont and the regulator. Puls asked whether the public had been amenable to the idea of shutting down the pump-and-treat system. The DuPont audience member said he had done a presentation for the public that looked at the energy consumption used in a pump-and-treat system. He demonstrated that the energy used by that system could be used to power 250 homes, a fact particularly relevant to the public given that the system was being operated in California.

One audience member asked whether part of the difficulty with PRBs is their lack of presumptive status. Gillham responded by saying that one Canadian agency is close to considering PRBs as a presumptive solution. Jefferis said the United Kingdom has a policy that all government contracts must consider sustainability, that is, the environmental, economic, and social impacts of a technology and its effects on future generations. In Britain, Jefferis believes, pump-and-treat technology will soon cease to be considered sustainable, while PRB technology will be. Sustainability is a strong political lever, he said; he encouraged its use in the United States. Puls said that there is an increasing movement in the U.S. to look at sustainability as a decision criterion for choosing remediation technologies.

Puls concluded the session by saying that although PRB technology might not have appeared to be moving forward in the last few years, it has taken time for regulators, site owners, and consultants to get enough information about the technology to feel comfortable about using it. RTDF and other groups have gone a long way in trying to educate all of these groups. Puls believes that the future is bright for PRBs.

RTDF Permeable Reactive Barriers
Action Team Meeting

Radisson Hotel Universal Orlando
Orlando, Florida
June 12, 2001

Final Attendee List