Formatted Title

Overlapping Aroclors and the Difficulty of Demonstrating Cleanup Standards: An Example

Background/Objectives

This case study illustrates an unanticipated analytical issue encountered during confirmation sampling of PCB-contaminated soils at the former Bannister Federal Complex (BFC) in Kansas City, Missouri.  An unexpected conflict occurred where sample dilution was necessary to achieve analytical precision in some Aroclors but also prevented confirmation that the non-detect values were achieved at the required reporting level for overlapping Aroclors in the same samples. This article describes the analytical conflict and resolution of the substantial regulatory issues that arose. 

Approach/Activities

As part of RCRA Corrective Measures, approximately 10,000 cubic yards of PCB-contaminated soil were targeted for excavation from the former BFC in Kansas City Missouri.  For all contaminants on site, the extent of contamination and waste classification had been developed based upon an excess cancer risk of 1x10^-5 or a Hazard Index (HI) of 0.1 – the HI<1 was mandated by regulators to account for mixtures of multiple contaminants.  Aroclors 1260 and 1242/1248 are the dominant PCB mixtures on site.  (Weathering of the PCBs likely contributed to the latter mixture being identified as either Aroclor 1242 or Aroclor 1248, depending on the laboratory.)  Aroclor-1254 was also identified as a contaminant of concern on site, and required a SSCL based on its non-cancer risk.  The HI-based shallow-soil SSCL for Aroclor 1254 (0.7 mg/kg) was an order of magnitude lower than the cancer-risk-based SSCL for Aroclors 1242/1248 and 1260 (10 mg/kg). 

Because there is substantial overlap in chromatographic range and signature of these Arolcors, the reporting limits (RLs) for Aroclor-1254 often exceeded its 0.7 mg/kg when Aroclors 1242/1248 and 1260 were reported in the same sample, even if below their own SSCLs.  We examined the overlap of Aroclor peaks for key samples, aligning by internal standard to facilitate.  Many of both the early- and late-eluting peaks from the Aroclor-1254 standard overlap those of the Aroclor-1248 and 1260 standards.  There is overlap between the late eluting peaks of the Aroclor-1248 standard and the early eluting peaks of the Aroclor-1260 standard.  Importantly, in its identification and quantification, the laboratory did not sequentially subtract peak heights after a certain Aroclor is identified.  Therefore, for the individual peaks noted above, the full peak height is considered during pattern matching to Aroclor standards, even if the same peak is used in quantifying more than one Aroclor. 

Consequently, the elevated RLs for Aroclor-1254 made it impossible to demonstrate achievement of all cleanup targets, even if chromatograms suggested Aroclor 1254 was not actually present in these samples. Because this appeared to be primarily an analytical issue, additional excavation would not be an effective approach, and re-sampling and analyzing for relevant congeners was not a practical option.  In consultation with the regulator, it was agreed that closure of PCB-impacted excavations could be based on the 95th percentile Upper Confidence Limit (UCL) concentration of Aroclor-1254 across each excavation area. Using non-parametric statistics that can account for left-censored data we were able to demonstrate that even with the elevated reporting limits for Aroclor-1254, the overall UCL for the sidewall confirmatory samples met even the Aroclor-1254 criteria in addition to those for Aroclors 1242/1248 and 1260.

Results/Lessons Learned

Under certain circumstances, unanticipated analytical uncertainty becomes a roadblock to achieving previously-approved goals.  In this case, the use of statistics that can take into account uncertainty in reported concentrations made it possible to complete a high profile and critical excavation.