Concentrations of arsenic (As) in groundwater at many Department of Defense (DoD) facilities exceed the U.S. drinking water standard of 10 μg/L. Known sources of arsenic contamination at DoD facilities include disposal of fly ash and past usage of arsenical biocides. In addition, naturally occurring arsenic present in soils at DoD sites can be mobilized by the inadvertent release or intentional introduction of organic carbon. Arsenic-contaminated groundwater may not, however, pose a threat to human health if the arsenic plume is effectively attenuated before it intercepts a drinking water source. In such cases, monitored natural attenuation (MNA) may be a viable strategy for the management of sites with arsenic-contaminated groundwater.

Acceptance and implementation of MNA as a remedy requires a mechanistic understanding of the processes by which arsenic may be sequestered in the subsurface environment and of the conditions that favor sequestration or that might allow subsequent remobilization of sequestered arsenic. The overall objective of this project was the development of a sound scientific basis for the evaluation of MNA for arsenic, particularly with regard to DoD facilities. The specific objectives of this work were:

• Development of a conceptual model for arsenic attenuation in the subsurface environment that can be coupled with site characterization to determine whether attenuation processes are operative at a given site
• Assessment of methods for demonstrating arsenic sequestration
• Laboratory investigation of strategies for augmenting natural arsenic sequestration
• Application of this paradigm to assess the environmental fate and exposure for arsenic at two DoD sites representing situations where arsenic was either introduced or mobilized as a result of human activities

A one-year project extension was conducted to identify and characterize end-member As- and iron (Fe)- sulfide precipitates and co-precipitated As+Fe sulfide phases as potential sinks for arsenic in sulfate-reduced environments (Final Report Addendum)

The fate and transport of arsenic were examined at two DoD sites. At Tyndall Air Force Base, Florida, arsenic contamination is the result of application of arsenical herbicides. At the Devens Reserve Forces Training Area (Fort Devens), Massachusetts, naturally occurring arsenic has been mobilized by application of enhanced reductive dechlorination (ERD) technology designed to remediate chlorinated solvent contamination. These two sites were compared with respect to plume evolution, mechanism of arsenic mobilization in the source areas, and potential sequestration mechanisms.

At both sites, groundwater was collected from existing monitoring wells and one or more new wells were drilled. Drilling also was performed at both sites to obtain sediment core samples for characterization and laboratory experiments. Groundwater collected at both sites was analyzed for a suite of chemical constituents and was also used in some mobilization experiments. Other mobilization experiments were conducted using synthetic groundwater solutions (based on the composition of background groundwater collected at the sites). Sediments were used for chemical and spectroscopic characterization, for batch and column studies of arsenic mobilization and sequestration, and for batch studies of augmentation of arsenic sequestration.

At Tyndall, elevated levels of arsenic persist in the source area soils, decades after herbicide use was discontinued. Arsenic concentrations in groundwater near the source area exceed 1 mg/L. A plume of arsenic-contaminated groundwater was delineated based on groundwater collected from existing monitoring wells. Comparison with historical monitoring data suggests that the plume of arsenic contamination is stable or even retreating. Contaminated soil in the source area soils was partially excavated in 2003. Laboratory experiments examining arsenic mobilization and sequestration were conducted with source area soils and sediments from the surficial aquifer and the Jackson Bluff Formation (JBF), a presumed confining layer. Substantial mobilization of arsenic was observed under near-ambient conditions in column experiments and in selective extraction under mild conditions that would not be expected to result in dissolution of the solid matrix. Sorption experiments with both surficial aquifer and JBF sediments indicated that significant accumulation of arsenic in the sediments occurred only at very elevated dissolved arsenic concentrations. Thus sorption was excluded as an effective mechanism for arsenic sequestration at this site. The persistence of arsenic in the source area soils was attributed to limited infiltration and leaching (i.e., hydrologic control) rather than to geochemical constraints on arsenic mobility. The basis for the apparent stability of the arsenic plume could not be established based on the available field data. The laboratory results, however, suggest that the assumption that the arsenic contamination is confined to the surficial aquifer by the JBF should be tested by further field investigations.

At Devens, the low concentrations of arsenic in groundwater upgradient of the ERD treatment zone and in the farthest downgradient wells indicated that the naturally occurring arsenic is immobile under ambient conditions (i.e., in the absence of anthropogenic inputs of organic bioavailable carbon). Laboratory experiments in which background sediments were inoculated with a known As- and Fe-reducing bacterium and amended with lactate confirmed that the arsenic and iron in the sediments are subject to microbially mediated mobilization. Comparison of historical monitoring data suggested that the extent of migration of organic carbon, arsenic, iron, and manganese (Mn) at the field site was slower than would be expected based on estimates of groundwater flow velocities. Attenuation of organic carbon could be attributed to microbial mineralization, but the attenuation of inorganic species would require that they be sequestered into the solid phase. Selective extraction of Fe(II) from the sediments did provide some direct evidence for Fe(II) sequestration along the transect of groundwater wells, but the putative sequestration of arsenic, (total) iron, or manganese did not result in any enrichment in these elements detectable against the background concentrations. Examination of the sediments by X-ray absorption spectroscopy (XAS) did not show any evidence of As(III) adsorption under in situ conditions, although sorbed As(III) could be detected spectroscopically in sediments used in batch and column sequestration studies. Sorption of arsenic onto Devens sediments appeared to be enhanced in the presence of Fe(II). The Devens sediments were found to possess some (limited) natural capacity for As(III) and Fe(II) oxidation; oxidative precipitation of Fe(III) oxyhydroxides would be expected to enhance arsenic sequestration. Experiments examining possible strategies for augmentation of the natural attenuation of arsenic demonstrated that the oxidative capacity of the native sediments could be increased by amendment with synthetic birnessite (nominally an Mn(IV) oxide). Birnessite amendment substantially increased arsenic sequestration but only in the presence of Fe(II), suggesting that the in situ formation of Fe(III) oxyhydroxides is important for the effective sequestration of arsenic. Arsenic sorption is more favorable at the Devens site than at the Tyndall site due to the higher iron content of the native sediment. This difference would be even more pronounced if fresh Fe(III) oxyhydroxide surfaces are formed by the in situ precipitation of Fe(II).

In the one-year extension project, precipitation experiments used starting solutions with As³⁺ (as arsenite, AsO₃^3-), sulfide (S^2--), and either Fe²⁺ or Fe³⁺, and As-only or Fe-only end-members at pH 3 or 4, 6, and 8 and aged for up to 210 days. Products were characterized by synchrotron XAS and by synchrotron and laboratory X-ray diffraction (XRD). Results from precipitation experiments under strict anoxic conditions showed that the primary solid products at low pH (3, 4) were mixtures of amorphous-to-crystalline Fe and As sulfides, with no evidence for extensive solid solution. With starting solutions containing As³⁺ + Fe²⁺ + S^2--, mixtures of Fe or As sulfides, a green rust-type phase (Fe²⁺, Fe³⁺-hydroxide), and a fraction of As³⁺ bonded to oxygen formed at higher pH (6, 8). With starting solutions containing As³⁺ + Fe³⁺+ S^2--, no As sulfide was precipitated and all arsenic was identified as the arsenite species, probably present mostly as a sorbed complex. Iron precipitated as a mixture of Fe sulfide and an Fe(III)-oxide phase that with aging formed poorly crystalline hematite. Precipitates were initially amorphous to XRD and became more crystalline with aging for end members, but the rate of transformation of Fe sulfides and oxides from amorphous to crystalline was generally much slower (weeks to months) when arsenic was present in solution. Characterization of precipitation products by XAS at aging times up to one month indicated that the local structure around iron or arsenic was indicative of the final solid products after 210 days of aging. These data suggest that small (perhaps nano-to-micro-meter sized) particles are nucleating on short time scales consistent with overall thermodynamic stability but that ripening to long-range crystallographic ordering are inhibited when both iron and arsenic are present in solution.

The comparison of the two sites provides insight into the conditions under which MNA can be relied upon to protect downgradient receptors from exposure to arsenic contamination.

The results of the Tyndall study do not support reliance on MNA despite the observed persistence of arsenic in the source area and the apparent stability of the plume. Further investigations to determine the integrity of the JBF as a confining layer are needed.

The Devens study suggests that MNA may be an effective remedial option for sites where naturally occurring arsenic has been mobilized due to localized introduction of organic carbon. The zero-order question relevant to these sites is whether arsenic was immobile under ambient conditions (i.e., before the introduction of organic carbon). This question can be answered affirmatively if dissolved arsenic concentrations in groundwater are low upgradient of the organic carbon inputs and/or in the far-field downgradient of the influence of the organic carbon plume. The first-order questions that must be subsequently addressed include: (1) Is arsenic in the plume undergoing attenuation? (2) What is the capacity for arsenic sequestration in the field? and (3) What is the long-term stability of arsenic sequestered in the far-field?  Ultimately, the capacity for arsenic sequestration must be assessed in the context of the organic carbon loading to the site.