Conceptualizing Controlling Factors for PFAS Salting Out in Groundwater Discharge Zones Along Sandy Beaches.

Understanding fate and transport processes for per- and poly-fluoroalkyl substances (PFAS) is critical for managing impacted sites. "PFAS Salting Out" in groundwater, defined herein, is an understudied process where PFAS in fresh groundwater mixes with saline groundwater near marine shorelines, which increases sorption of PFAS to aquifer solids. While sorption reduces PFAS mass discharge to marine surface water, the fraction that sorbs to beach sediments may be mobilized under future salinity changes. The objective of this study was to conceptually explore the potential for PFAS Salting Out in sandy beach environments and to perform a preliminary broad-scale characterization of sandy shoreline areas in the continental U.S. While no site-specific PFAS data were collected, our conceptual approach involved developing a multivariate regression model that assessed how tidal amplitude and freshwater submarine groundwater discharge affect the mixing of fresh and saline groundwater in sandy coastal aquifers. We then applied this model to 143 U.S. shoreline areas with sandy beaches (21% of total beaches in the USA), indirectly mapping potential salinity increases in shallow freshwater PFAS plumes as low (<10 ppt), medium (10-20 ppt), or high (>20 ppt) along groundwater flow paths before reaching the ocean. Higher potential salinity increases were observed in West Coast bays and the North Atlantic coastline, due to the combination of moderate to large tides and large fresh groundwater discharge rates, while lower increases occurred along the Gulf of Mexico and the southern Florida Atlantic coast. The salinity increases were used to estimate potential perfluorooctane sulfonic acid (PFOS) sorption in groundwater due to salting out processes. Low-category shorelines may see a 1- to 2.5-fold increase in sorption of PFOS, medium-category a 2.0- to 6.4-fold increase, and high-category a 3.8- to 25-fold increase in PFOS sorption. The analysis presented provides a first critical step in developing a large-scale approach to classify the PFAS Salting Out potential along shorelines and the limitations of the approach adopted highlights important areas for further research.

Natural source zone depletion (NSZD) insights from over 15 years of research and measurements: A multi-site study.

Site-average Natural Source Zone Depletion (NSZD) rates measured from 40 petroleum light non-aqueous phase liquid (LNAPL) source zone sites were compiled from researchers, project reports, and scientific papers. At each site, the following data were compiled: i) general site location; ii) LNAPL fuel type; iii) measurement method, number of locations, and number of measurements per location; and iv) calculated site-average NSZD rate in liters per hectare per year (L/ha/yr) per site and the associated measurement method (i.e., Gradient Method, Carbon Traps, Dynamic Closed Chamber (DCC), or Thermal Monitoring). The resulting dataset showed site-average NSZD rates that ranged from 650 to 152,000 L/ha/yr (70 to 16,250 gallons per acre per year (gal/acre/yr)), with a median value of 9,540 L/ha/yr (1,020 gal/acre/yr). The median site-average NSZD rate by type of fuel spill did not show a statistically significant difference between fuel types. When comparing the different NSZD measurement methods applied to the same sites, the site-average NSZD rates differed by up to 4.8 times (i.e., ratio of faster rate to slower rate), with a median difference of 2.1 times. No clear bias was observed between NSZD rate measurement methods. At four sites with calculations of NSZD rates by season, NSZD rates were typically higher during summer and fall compared to winter and spring. For these sites, Q10 values (a measure of the increase in NSZD rate associated with a 10 °C increase in temperature) ranged from 0.8 to 15.1, with a median of 2.2. The implications of this study suggest that increasing mean annual soil temperature at a site using engineered methods could potentially increase the biodegradation rate (e.g., an increase of 10 °C could double the NSZD rate). Finally, for five sites with site-average NSZD rates for multiple years, average NSZD rates varied by 1.1 to 4.9 times across years. Overall, the evaluation of NSZD rates measured at 40 LNAPL sites suggests that measurable NSZD occurs across a broad range of LNAPL sites. Although NSZD rates vary across sites, fuel type is not the primary factor explaining observed differences in rates.

Author response to technical commentary: Direct aerobic NSZD of a basalt vadose zone LNAPL source in Hawaii, McHugh et al., JCH #235 (2020).

Our Paper uses two independent methods (carbon dioxide flux and heat flux) to measure the rates of natural source zone depletion (NSZD) at a petroleum release site in Hawaii. The two methods yielded estimates of the NSZD rate that agreed within a factor of 2. We disagree with the technical commentors (Beckett et al., 2022). Specifically, the available data indicate that the observed NSZD is not occurring through a two-stage process of methane generation at the water table followed by methane oxidation in the vadose zone; rather direct aerobic oxidation of LNAPL in the vadose zone is the simplest and most likely explanation for the observed heat generation. In addition, the agreement between the two independent NSZD rate measurement methods and the temporal consistency in the measured heat flux across two field events provides confidence that the NSZD rates presented in the paper are not greatly over or under-estimated.

Modeling a well-characterized perfluorooctane sulfonate (PFOS) source and plume using the REMChlor-MD model to account for matrix diffusion.

Two of the most important retention processes for per- and polyfluoroalkyl substances (PFAS) in groundwater likely are sorption and matrix diffusion. The objective of this study was to model concentration and mass discharge of one PFAS, perfluorooctane sulfonate (PFOS), with matrix diffusion processes incorporated using data from a highly chemically- and geologically-characterized site. When matrix diffusion is incorporated into the REMChlor-MD model for PFOS at this research site, it easily reproduces the field data for three key metrics (concentration, mass discharge, and total mass). However, the no-matrix diffusion model produced a much poorer match. Additionally, after about 40 years of groundwater transport, field data and the REMChlor-MD model both showed the majority (80%) of the measured PFOS mass that exited the source zones was located in downgradient low permeability zones due to matrix diffusion. As such, most of the PFOS mass is not available to immediately migrate downgradient via advection in the more permeable sands at this site, which has important implications for monitored natural attenuation (MNA). Plume expansion over the next 50 years is forecasted to be limited, from a 350-m plume length in 2017 to 550 m in 2070, as matrix diffusion will attenuate groundwater plumes by slowing their expansion. This phenomenon is important for constituents that do not degrade, such as PFOS, compared to those susceptible to degradation. Overall, this work shows that matrix diffusion is a relevant process in environmental PFAS persistence and slows the rate of plume expansion over time.

Impact of matrix diffusion on the migration of groundwater plumes for Perfluoroalkyl acids (PFAAs) and other non-degradable compounds.

Groundwater fate and transport modeling results demonstrate that matrix diffusion plays a role in attenuating the expansion of groundwater plumes of "non-degrading" or highly recalcitrant compounds. This is especially significant for systems where preferred destructive attenuation processes, such as biological and abiotic degradation, are weak or ineffective for plume control. Under these conditions, models of nondestructive physical attenuation processes, traditionally dispersion or sorption, do not demonstrate sufficient plume control unless matrix diffusion is considered. Matrix diffusion has been shown to be a notable emergent impact of geological heterogeneity, typically associated with back diffusion and extending remediation timeframes through concentration tailing of the trailing edge of a plume. However, less attention has been placed on evaluating how matrix diffusion can serve as an attenuation mechanism for the leading edge of a plume of non-degrading compounds like perfluoroalkyl acids (PFAAs), including perfluorooctane sulfonate (PFOS). In this study, the REMChlor-MD model was parametrically applied to a generic unconsolidated and heterogeneous geologic site with a constant PFOS source and no degradation of PFOS in the downgradient edge of the plume. Low levels of mechanical dispersion and retardation were used in the model for three different geologic heterogeneity cases ranging from no matrix diffusion (e.g., sand only) to considerable matrix diffusion using low permeability ("low-k") layers/lenses and/or aquitards. Our analysis shows that, in theory, many non-degrading plumes may expand for significant time periods before dispersion alone would eventually stabilize the plume; however, matrix diffusion can significantly slow the rate and degree of this migration. For one 100-year travel time scenario, consideration of matrix diffusion results in a simulated PFOS plume length that is over 80% shorter than the plume length simulated without matrix diffusion. Although many non-degrading plumes may continue to slowly expand over time, matrix diffusion resulted in lower concentrations and smaller plume footprints. Modeling multiple hydrogeologic settings showed that the effect of matrix diffusion is more significant in transmissive zones containing multiple low-k lenses/layers than transmissive zones underlain and overlain by low-k aquitards. This study found that at sites with significant matrix diffusion, groundwater plumes will be shorter, will expand more slowly, and may be amenable to a physical, retention-based, Monitored Natural Attenuation (MNA) paradigm. In this case, a small "Plume Assimilative Capacity Zone" in front of the existing plume could be reserved for slow, de minimus, future expansion of a non-degrading plume. If potential receptors are protected in this scenario, then this approach is similar to allowances for expanding plumes under some existing environmental regulatory programs. Accounting for matrix diffusion may support new strategic approaches and alternative paradigms for remediation even for sites and conditions with "non-degrading" constituents such as PFAAs, metals/metalloids, and radionuclides.

Mass-Based, Field-Scale Demonstration of PFAS Retention within AFFF-Associated Source Areas.

Transport of poly- and perfluoroalkyl substances (PFAS) at aqueous film-forming foam (AFFF)-impacted sites is limited by various processes that can retain PFAS mass within the source area. This study used concentration data obtained via a high-resolution sampling and analytical protocol to estimate the PFAS mass distribution in source and downgradient areas of a former firefighter training area. The total PFAS mass present at the site was approximately 222 kg, with 106 kg as perfluoroalkyl acids (PFAAs) and 116 kg as polyfluorinated precursors. Zwitterionic and cationic PFAS represented 83% of the total precursor mass and were found primarily in the source and up/side-gradient areas (75%), likely due to preferential hydrophobic partitioning, electrostatic interactions, and diffusion into lower-permeability soils. Based on the release history and the high percentage of total PFAS mass represented by precursors (primarily electrochemical fluorination-derived compounds), the estimated conversion rate of precursors to PFAAs was less than 2% annually. Eighty-two percent of the total PFAS mass was encountered in lower-permeability soils, which limited the potential for advection and transformation. This contributed to a 99% decrease in the mass discharge rate at the far-downgradient plume (0.048 kg/yr compared to the near-source area (3.6 kg/yr)). The results provide field-scale evidence of the importance of these PFAS retention processes at sites where AFFF has been released.

Direct aerobic NSZD of a basalt vadose zone LNAPL source in Hawaii.

In recent years, a number of methods have been used to measure the biodegradation of petroleum light non-aqueous phase liquids (LNAPL) at petroleum release sites, a process known as natural source zone depletion (NSZD). Most commonly, NSZD rates have been measured at sites with unconsolidated geology and relatively shallow groundwater (<50 ft. bgs, <15 m bgs). For this study, we have used two methods (1. carbon dioxide flux measured using carbon traps and 2. heat flux based on subsurface temperature gradients) to measure NSZD rates at a petroleum release site in Hawaii with basalt geology and deep groundwater (>300 ft. bgs, >100 m bgs). Both methods documented the occurrence of NSZD at the facility and the two methods yield estimates of the NSZD rate that agreed within a factor of 2 (4600 to 7400 gal/yr; 17,000 to 28,000 L/yr for the flux method and 8600 to 13,000 gal/yr; 33,000 to 49,000 L/yr for the temperature method). Soil gas samples collected directly above the water table and at shallower depths within the vadose zone indicated aerobic conditions throughout the vadose zone (oxygen >13%) and no detectable methane. These results indicate that NSZD occurs at this site through the direct aerobic biodegradation of LNAPL rather than the two-step process of anaerobic methanogenesis followed by methane oxidation at a shallow depth interval documented at other sites.

Monitoring, assessment, and prediction of microbial shifts in coupled catalysis and biodegradation of 1,4-dioxane and co-contaminants.

Microbial community dynamics were characterized following combined catalysis and biodegradation treatment trains for mixtures of 1,4-dioxane and chlorinated volatile organic compounds (CVOCs) in laboratory microcosms. Although a few specific bacterial taxa are capable of removing 1,4-dioxane and individual CVOCs, many microorganisms are inhibited when these contaminants are present in mixtures. Chemical catalysis by tungstated zirconia (WOx/ZrO2) and hydrogen peroxide (H2O2) as a non-selective treatment was designed to achieve nearly 20% 1,4-dioxane and over 60% trichloroethene and 50% dichloroethene removals. Post-catalysis, bioaugmentation with 1,4-dioxane metabolizing bacterial strain,Pseudonocardia dioxanivorans CB1190, removed the remaining 1,4-dioxane. The evolution of the microbial community under different conditions was time-dependent but relatively independent of the concentrations of contaminants. The compositions of microbiomes tended to be similar regardless of complex contaminant mixtures during the biodegradation phase, indicating a r-K strategy transition attributed to the shock experienced during catalysis and the subsequent incubation. The originally dominant genera Pseudomonas and Ralstonia were sensitive to catalytic oxidation, and were overwhelmed by Sphingomonas, Rhodococcus, and other catalyst-tolerant microbes, but microbes capable of biodegradation of organics thrived during the incubation. Methane metabolism, chloroalkane-, and chloroalkene degradation pathways appeared to be responsible for CVOC degradation, based on the identifications of haloacetate dehalogenases, 2-haloacid dehalogenases, and cytochrome P450 family. Network analysis highlighted the potential interspecies competition or commensalism, and dynamics of microbiomes during the biodegradation phase that were in line with shifting predominant genera, confirming the deterministic processes guiding the microbial assembly. Collectively, this study demonstrated that catalysis followed by bioaugmentation is an effective treatment for 1,4-dioxane in the presence of high CVOC concentrations, and it enhanced our understanding of microbial ecological impacts resulting from abiotic-biological treatment trains. These results will be valuable for predicting treatment synergies that lead to cost savings and improve remedial outcomes in short-term active remediation as well as long-term changes to the environmental microbial communities.

Response and recovery of microbial communities subjected to oxidative and biological treatments of 1,4-dioxane and co-contaminants.

Microbial community dynamics were characterized following combined oxidation and biodegradation treatment trains for mixtures of 1,4-dioxane and chlorinated volatile organic compounds (CVOCs) in laboratory microcosms. Bioremediation is generally inhibited by co-contaminate CVOCs; with only a few specific bacterial taxa reported to metabolize or cometabolize 1,4-dioxane being unaffected. Chemical oxidation by hydrogen peroxide (H2O2) as a non-selective treatment demonstrated 50-80% 1,4-dioxane removal regardless of the initial CVOC concentrations. Post-oxidation bioaugmentation with 1,4-dioxane metabolizer Pseudonocardia dioxanivorans CB1190 removed the remaining 1,4-dioxane. The intrinsic microbial population, biodiversity, richness, and biomarker gene abundances decreased immediately after the brief oxidation phase, but recovery of cultivable microbiomes and a more diverse community were observed during the subsequent 9-week biodegradation phase. Results generated from the Illumina Miseq sequencing and bioinformatics analyses established that generally oxidative stress tolerant genus Ralstonia was abundant after the oxidation step, and Cupriavidus, Pseudolabrys, Afipia, and Sphingomonas were identified as dominant genera after aerobic incubation. Multidimensional analysis elucidated the separation of microbial populations as a function of time under all conditions, suggesting that temporal succession is a determining factor that is independent of 1,4-dioxane and CVOCs mixtures. Network analysis highlighted the potential interspecies competition or commensalism, and dynamics of microbiomes during the biodegradation phase, in line with the shifts of predominant genera and various developing directions during different steps of the treatment train. Collectively, this study demonstrated that chemical oxidation followed by bioaugmentation is effective for treating 1,4-dioxane, even in the presence of high levels of CVOC mixtures and residual peroxide, a disinfectant, and enhanced our understanding of microbial ecological impacts of the treatment train. These results will be valuable for predicting treatment synergies that lead to cost savings and improved remedial outcomes in short-term active remediation as well as long-term changes to the environmental microbial communities.

Implications of matrix diffusion on 1,4-dioxane persistence at contaminated groundwater sites.

Management of groundwater sites impacted by 1,4-dioxane can be challenging due to its migration potential and perceived recalcitrance. This study examined the extent to which 1,4-dioxane's persistence was subject to diffusion of mass into and out of lower-permeability zones relative to co-released chlorinated solvents. Two different release scenarios were evaluated within a two-layer aquifer system using an analytical modeling approach. The first scenario simulated a 1,4-dioxane and 1,1,1-TCA source zone where spent solvent was released. The period when 1,4-dioxane was actively loading the low-permeability layer within the source zone was estimated to be <3years due to its high effective solubility. While this was approximately an order-of-magnitude shorter than the loading period for 1,1,1-TCA, the mass of 1,4-dioxane stored within the low-permeability zone at the end of the simulation period (26kg) was larger than that predicted for 1,1,1-TCA (17kg). Even 80years after release, the aqueous 1,4-dioxane concentration was still several orders-of-magnitude higher than potentially-applicable criteria. Within the downgradient plume, diffusion contributed to higher concentrations and enhanced penetration of 1,4-dioxane into the low-permeability zones relative to 1,1,1-TCA. In the second scenario, elevated 1,4-dioxane concentrations were predicted at a site impacted by migration of a weak source from an upgradient site. Plume cutoff was beneficial because it could be implemented in time to prevent further loading of the low-permeability zone at the downgradient site. Overall, this study documented that 1,4-dioxane within transmissive portions of the source zone is quickly depleted due to characteristics that favor both diffusion-based storage and groundwater transport, leaving little mass to treat using conventional means. Furthermore, the results highlight the differences between 1,4-dioxane and chlorinated solvent source zones, suggesting that back diffusion of 1,4-dioxane mass may be serving as the dominant long-term "secondary source" at many contaminated sites that must be managed using alternative approaches.

Time vs. Money: A Quantitative Evaluation of Monitoring Frequency vs. Monitoring Duration.

The National Research Council has estimated that over 126,000 contaminated groundwater sites are unlikely to achieve low ug/L clean-up goals in the foreseeable future. At these sites, cost-effective, long-term monitoring schemes are needed in order to understand the long-term changes in contaminant concentrations. Current monitoring optimization schemes rely on site-specific evaluations to optimize groundwater monitoring frequency. However, when using linear regression to estimate the long-term zero-order or first-order contaminant attenuation rate, the effect of monitoring frequency and monitoring duration on the accuracy and confidence for the estimated attenuation rate is not site-specific. For a fixed number of monitoring events, doubling the time between monitoring events (e.g., changing from quarterly monitoring to semi-annual monitoring) will double the accuracy of estimated attenuation rate. For a fixed monitoring frequency (e.g., semi-annual monitoring), increasing the number of monitoring events by 60% will double the accuracy of the estimated attenuation rate. Combining these two factors, doubling the time between monitoring events (e.g., quarterly monitoring to semi-annual monitoring) while decreasing the total number of monitoring events by 38% will result in no change in the accuracy of the estimated attenuation rate. However, the time required to collect this dataset will increase by 25%. Understanding that the trade-off between monitoring frequency and monitoring duration is not site-specific should simplify the process of optimizing groundwater monitoring frequency at contaminated groundwater sites.

Evidence of 1,4-dioxane attenuation at groundwater sites contaminated with chlorinated solvents and 1,4-dioxane.

There is a critical need to develop appropriate management strategies for 1,4-dioxane (dioxane) due to its widespread occurrence and perceived recalcitrance at groundwater sites where chlorinated solvents are present. A comprehensive evaluation of California state (GeoTracker) and Air Force monitoring records was used to provide significant evidence of dioxane attenuation at field sites. Temporal changes in the site-wide maximum concentrations were used to estimate source attenuation rates at the GeoTracker sites (median length of monitoring period = 6.8 years). While attenuation could not be established at all sites, statistically significant positive attenuation rates were confirmed at 22 sites. At sites where dioxane and chlorinated solvents were present, the median value of all statistically significant dioxane source attenuation rates (equivalent half-life = 31 months; n = 34) was lower than 1,1,1-trichloroethane (TCA) but similar to 1,1-dichloroethene (1,1-DCE) and trichloroethene (TCE). Dioxane attenuation rates were positively correlated with rates for 1,1-DCE and TCE but not TCA. At this set of sites, there was little evidence that chlorinated solvent remedial efforts (e.g., chemical oxidation, enhanced bioremediation) impacted dioxane attenuation. Attenuation rates based on well-specific records from the Air Force data set confirmed significant dioxane attenuation (131 out of 441 wells) at a similar frequency and extent (median equivalent half-life = 48 months) as observed at the California sites. Linear discriminant analysis established a positive correlation between dioxane attenuation and increasing concentrations of dissolved oxygen, while the same analysis found a negative correlation with metals and CVOC concentrations. The magnitude and prevalence of dioxane attenuation documented here suggest that natural attenuation may be used to manage some but not necessarily all dioxane-impacted sites.

The new potential for understanding groundwater contaminant transport.

The groundwater remediation field has been changing constantly since it first emerged in the 1970s. The remediation field has evolved from a dissolved-phase centric conceptual model to a DNAPL-dominated one, which is now being questioned due to a renewed appreciation of matrix diffusion effects on remediation. Detailed observations about contaminant transport have emerged from the remediation field, and challenge the validity of one of the mainstays of the groundwater solute transport modeling world: the concept of mechanical dispersion (Payne et al. 2008). We review and discuss how a new conceptual model of contaminant transport based on diffusion (the usurper) may topple the well-established position of mechanical dispersion (the status quo) that is commonly used in almost every groundwater contaminant transport model, and evaluate the status of existing models and modeling studies that were conducted using advection-dispersion models.

Progress in remediation of groundwater at petroleum sites in California.

Quantifying the overall progress in remediation of contaminated groundwater has been a significant challenge. We utilized the GeoTracker database to evaluate the progress in groundwater remediation from 2001 to 2011 at over 12,000 sites in California with contaminated groundwater. This paper presents an analysis of analytical results from over 2.1 million groundwater samples representing at least $100 million in laboratory analytical costs. Overall, the evaluation of monitoring data shows a large decrease in groundwater concentrations of gasoline constituents. For benzene, half of the sites showed a decrease in concentration of 85% or more. For methyl tert-butyl ether (MTBE), this decrease was 96% and for TBE, 87%. At remediation sites in California, the median source attenuation rate was 0.18/year for benzene and 0.36/year for MTBE, corresponding to half-lives of 3.9 and 1.9 years, respectively. Attenuation rates were positive (i.e., decreasing concentration) for benzene at 76% of sites and for MTBE at 85% of sites. An evaluation of sites with active remediation technologies suggests differences in technology effectiveness. The median attenuation rates for benzene and MTBE are higher at sites with soil vapor extraction or air sparging compared with sites without these technologies. In contrast, there was little difference in attenuation rates at sites with or without soil excavation, dual phase extraction, or in situ enhanced biodegradation. The evaluation of remediation technologies, however, did not evaluate whether specific systems were well designed or implemented and did not control for potential differences in other site factors, such as soil type.

Membrane interface probe protocol for contaminants in low-permeability zones.

Accurate characterization of contaminant mass in zones of low hydraulic conductivity (low k) is essential for site management because this difficult-to-treat mass can be a long-term secondary source. This study developed a protocol for the membrane interface probe (MIP) as a low-cost, rapid data-acquisition tool for qualitatively evaluating the location and relative distribution of mass in low-k zones. MIP operating parameters were varied systematically at high and low concentration locations at a contaminated site to evaluate the impact of the parameters on data quality relative to a detailed adjacent profile of soil concentrations. Evaluation of the relative location of maximum concentrations and the shape of the MIP vs. soil profiles led to a standard operating procedure (SOP) for the MIP to delineate contamination in low-k zones. This includes recommendations for: (1) preferred detector (ECD for low concentration zones, PID or ECD for higher concentration zones); (2) combining downlogged and uplogged data to reduce carryover; and (3) higher carrier gas flow rate in high concentration zones. Linear regression indicated scatter in all MIP-to-soil comparisons, including R(2) values using the SOP of 0.32 in the low concentration boring and 0.49 in the high concentration boring. In contrast, a control dataset with soil-to-soil correlations from borings 1-m apart exhibited an R(2) of ≥ 0.88, highlighting the uncertainty in predicting soil concentrations using MIP data. This study demonstrates that the MIP provides lower-precision contaminant distribution and heterogeneity data compared to more intensive high-resolution characterization methods. This is consistent with its use as a complementary screening tool.

Perfluorooctanoic acid degradation in the presence of Fe(III) under natural sunlight.

Due to the high bond dissociation energy (BDE) of CF bonds (116 kcal/mol), perfluorooctanoic acid (PFOA) is a highly recalcitrant pollutant. Herein, we demonstrate a novel method to decompose PFOA in the presence of sunlight and ferric iron (Fe(III)). Under such conditions, 97.8 ± 1.7% of 50 µM PFOA decomposed within 28 days into shorter-chain intermediates and fluoride (F(-)), with an overall defluorination extent of 12.7 ± 0.5%. No PFOA was removed under visible light, indicating that UV radiation is required for PFOA decomposition. Spectroscopic analysis indicates that the decomposition reaction is likely initiated by electron-transfer from PFOA to Fe(III), forming Fe(II) and an unstable organic carboxyl radical. An alternative mechanism for the formation of this organic radical involves hydroxyl radicals, detected by electron paramagnetic resonance (EPR). The observation that PFOA can be degraded by Fe(III) under solar irradiation provides mechanistic insight into a possibly overlooked natural attenuation process. Because Fe(III) is abundant in natural waters and sunlight is essentially free, this work represents a potentially important step toward the development of simple and inexpensive remediation strategies for PFOA-contaminated water.

Relative contribution of DNAPL dissolution and matrix diffusion to the long-term persistence of chlorinated solvent source zones.

The relative contribution of dense non-aqueous phase liquid (DNAPL) dissolution versus matrix diffusion processes to the longevity of chlorinated source zones was investigated. Matrix diffusion is being increasingly recognized as an important non-DNAPL component of source behavior over time, and understanding the persistence of contaminants that have diffused into lower permeability units can impact remedial decision-making. In this study, a hypothetical DNAPL source zone architecture consisting of several different sized pools and fingers originally developed by Anderson et al. (1992) was adapted to include defined low permeability layers. A coupled dissolution-diffusion model was developed to allow diffusion into these layers while in contact with DNAPL, followed by diffusion out of these same layers after complete DNAPL dissolution. This exercise was performed for releases of equivalent masses (675 kg) of three different compounds, including chlorinated solvents with solubilities ranging from low (tetrachloroethene (PCE)), moderate (trichloroethene (TCE)) to high (dichloromethane (DCM)). The results of this simple modeling exercise demonstrate that matrix diffusion can be a critical component of source zone longevity and may represent a longer-term contributor to source longevity (i.e., longer time maintaining concentrations above MCLs) than DNAPL dissolution alone at many sites. For the hypothetical TCE release, the simulation indicated that dissolution of DNAPL would take approximately 38 years, while the back diffusion from low permeability zones could maintain the source for an additional 83 years. This effect was even more dramatic for the higher solubility DCM (97% of longevity due to matrix diffusion), while the lower solubility PCE showed a more equal contribution from DNAPL dissolution vs. matrix diffusion. Several methods were used to describe the resulting source attenuation curves, including a first-order decay model which showed that half-life of mass discharge from the matrix-diffusion dominated phase is in the range of 13 to 29 years for TCE. Because the mass discharge rate shifts significantly over time once DNAPL dissolution is complete, a Power-Law model was shown to be useful, especially at later stages when matrix diffusion dominates. An assessment of mass distribution showed that while relatively small percentages of the initial source mass diffused into the low permeability compartment, this mass was sufficient to sustain concentrations above drinking water standards for decades. These data show that relatively typical conditions (e.g., 50-year-old release, moderate to high solubility contaminant) are consistent with late stage sources, where mass in low permeability matrices serves as the primary source, and fit the conceptual model that mass in low permeability zones is important when evaluating source longevity.

Chlorinated ethene source remediation: lessons learned.

Chlorinated solvents such as trichloroethene (TCE) and tetrachloroethene (PCE) are widespread groundwater contaminants often released as dense nonaqueous phase liquids (DNAPLs). These contaminants are difficult to remediate, particularly their source zones. This review summarizes the progress made in improving DNAPL source zone remediation over the past decade, and is structured to highlight the important practical lessons learned for improving DNAPL source zone remediation. Experience has shown that complete restoration is rare, and alternative metrics such as mass discharge are often useful for assessing the performance of partial restoration efforts. Experience also has shown that different technologies are needed for different times and locations, and that deliberately combining technologies may improve overall remedy performance. Several injection-based technologies are capable of removing a large fraction of the total contaminant mass, and reducing groundwater concentrations and mass discharge by 1 to 2 orders of magnitude. Thermal treatment can remove even more mass, but even these technologies generally leave some contamination in place. Research on better delivery techniques and characterization technologies will likely improve treatment, but managers should anticipate that source treatment will leave some contamination in place that will require future management.

Groundwater remediation: the next 30 years.

Groundwater remediation technologies are designed, installed, and operated based on the conceptual models of contaminant hydrogeology that are accepted at that time. However, conceptual models of remediation can change as new research, new technologies, and new performance data become available. Over the past few years, results from multiple-site remediation performance studies have shown that achieving drinking water standards (i.e., Maximum Contaminant Levels, MCLs) at contaminated groundwater sites is very difficult. Recent groundwater research has shown that the process of matrix diffusion is one key constraint. New developments, such as mass discharge, orders of magnitude (OoMs), and SMART objectives are now being discussed more frequently by the groundwater remediation community. In this paper, the authors provide their perspectives on the existing "reach MCLs" approach that has historically guided groundwater remediation projects, and advocate a new approach built around the concepts of OoMs and mass discharge.