Arsenic and thallium data in environmental samples: Fact or fiction?

Abstract

AbstractMatrix effects may increasingly lead to erroneous environmental decisions as regulatory limits or risk‐based concentrations of concern for trace metals move lower toward the limits of analytical detection. A U.S. Environmental Protection Agency Office of Technical Standards Alert estimated that environmental data reported using inductively coupled plasma spectrometry (ICP‐AES) has a false‐positive rate for thallium of 99.9 percent and for arsenic of 25 to 50 percent. Although this does not seem to be widely known in the environmental community, using three case studies, this article presents data in environmental samples that demonstrate severe matrix effects on the accuracy of arsenic and thallium results. Case Study 1 involves soil results with concentrations that approached or exceeded the applicable regulatory soil cleanup objectives of 13 mg/kg for arsenic and 2 mg/kg for thallium. Reanalysis using ICP coupled with a mass spectrometer (ICP‐MS) confirmed all thallium results were false positives and all arsenic results were biased high, concluding no action was required for soil remediation. Case Study 2 involves groundwater results for thallium at a Superfund site, where thallium was detected in groundwater up to 21.6 μ g/L using ICP‐AES. Reanalysis by ICP‐MS reported thallium as nondetect below the applicable regulatory level in all samples. ICP‐MS is usually a more definitive and accurate method of analysis compared to ICP‐AES; however, this is not always the case, as we demonstrate in Case Study 3, using data from groundwater samples at an industrial site. Through a weight‐of‐evidence approach, it is demonstrated that although method quality control results were acceptable, interferences in some groundwater samples caused biased high results for arsenic using ICP‐MS, which were significantly lower when reanalyzed using hydride generation atomic fluorescence spectrometry. Causes of these interference effects and conclusions from the three case studies to obtain accurate metal data for site assessment, risk characterization, and remedy selection are discussed. © 2010 Wiley Periodicals, Inc.