Monitored natural attenuation (MNA) is an important groundwater remediation technology based on a carefully controlled and monitored demonstration of contaminant attenuation. However, demonstrating contaminant mass destruction can be challenging. Compound-specific isotope analysis (CSIA) is a specialized laboratory method that can provide a direct signal of biological or abiotic degradation and support assessment of the strength of physical attenuation processes. The popularity of CSIA has risen rapidly among project managers as one line of evidence supporting MNA remedies. While CSIA results help to refine conceptual site models (CSM), CSIA data can be difficult to interpret, especially at sites with complex hydrogeology or with competing degradation pathways. The overall goal of the project was to present methods for quantitative assessment of natural attenuation processes, including mass destruction, for chlorinated solvents, using a combination of CSIA with modeling-assisted data interpretation. 

Many elements, such as carbon or chlorine, occur as different isotope species, differing in their atomic mass. CSIA permits to determine the isotopic makeup of the contaminants present in environmental samples and the information obtained can be used as a line of evidence in contaminant studies. The majority of CSIA applications concern the assessment of volatile organic compound (VOC) contaminants degradation in groundwater. The principle of the approach is that isotope ratios of a contaminant, for example ¹³C/¹²C, remain constant as the groundwater is diluted, while the fraction of the heavy isotope, ¹³C, typically increases with degradation. The difference between ¹²C and ¹³C behavior originates from energetically favored reactions for the molecules containing the lower atomic mass isotope (e.g., ¹²C). On the other hand, the rates of non-degradative processes tend to have no or little selectivity in respect to the isotope composition of the contaminant. The benefit of the CSIA approach for contaminant studies lies in its ability to distinguish mass destruction (by biodegradation and/or abiotic degradation) from other types of mass attenuation. However, interpretation of field CSIA data can be difficult due to competing degradation pathways and/or complex transport conditions in the aquifer. The value added to contaminated site assessment by the use of CSIA ultimately depends on the specificity of the interpretation. This study centers on demonstration of numerical modeling to improve the capabilities for attenuation pathway identification and quantitative assessment of CSIA data.

Reactive transport modeling (RTM) simulates transport and contaminant degradation, using a simplified numerical representation of the features of the modeled site. RTMs enables to simulate complex reaction networks (e.g., sequential reductive dechlorination together with oxidative degradation) together with isotope fractionation (C, H, Cl), while accounting for physical processes that may influence isotope ratios such as hydrodynamic dispersion. However, as discussed below, RTMs also enable sound data interpretation through simulating fewer dimensions like 2-D cross-sections, 1-D flow paths or even 0-D batch degradation.

The demonstration followed two main tracks, a development and initial calibration of the modeling software and a demonstration of the combined CSIA/RTM approach through an assessment of a contaminated site (Hill AFB, Operable Unit 10, consisting of Shallow TCE Plume and Deep TCE Plume). The success of the technology demonstration was defined in terms of producing site assessment results that are useful for development/improvement of Conceptual Site Model and are superior to those obtained by the “classic” CSIA alone.

The performance objectives for the software development and validation were met successfully: (i) 0/1-D PHREEQC model templates for simulation of isotope fractionation in reductive dechlorination, for carbon, chlorine and hydrogen were developed and calibrated using a data set from a microcosm experiment (dechlorination of trichloroethylene by a Dehalococcoides culture); (ii) two 2/3-D model platforms, PHAST and PHT3D were then adopted to simulate the same set of reactions as PHREEQC.

The performance objectives for evaluation of the Demonstration Site had to be revised, after initial evaluation of the data collected at the Demonstration Site. Observed trends of isotope enrichment did not correlate to the distance from the plume source, the distance across the plume fringe or to the groundwater age. Instead, degradation in the Shallow Plume was localized in disconnected zones of the plume. Degradation in the Deep plume was occurring primarily in an irregular area in the proximity to the plume source zone. Since no meaningful trends of isotope fractionation could be identified along 1-D flow lines, the exercise of 1/2/3-D modeling would be meaningless. Instead, the modeling was conducted using the batch (0-D) mode of the 0/1-D PHREEQC software. Spatial and temporal dimensions were thus not explicitly simulated. Even so, in comparison with the “classical” CSIA evidence, the combined CSIA/modeling approach (using the model to test alternative attenuation scenarios; the scenarios are defined using the “classical” CSIA evidence) permitted: (i) a reduction of uncertainties in identification of specific degradation pathways; and (ii) more accurate identification of the range of isotope enrichment factors, leading to more accurate quantitation of contaminant mass destruction. 

The proposed methodology is cost-effective: (i) the cost added by basic 0-D modeling of CSIA data is low in comparison to the complete cost of sample collection and CSIA analytical work; and (ii) the only requirement for implementation of CSIA/modeling is a reasonable CSIA and contaminant concentrations data completeness. End users can choose that line data interpretation the “classical” CSIA indicates a need, eliminating question marks by modeling various attenuation scenarios.

As indicated above, site heterogeneity complicates implementation of the modeling in data interpretation. Even after the model dimensions downgrade to 0-D, simulation of the degradation processes in the Deep Plume using the existing model template was difficult and less successful than in the case of the Shallow Plume. In the Deep Plume, it is likely that the problems were caused by the proportionally more significant role of diffusion/back-diffusion.