Date Published: July 31, 2017
Publisher: Public Library of Science
Author(s): Julie Wroble, Timothy Frederick, Alicia Frame, Daniel Vallero, Yiu Fai Tsang.
Established soil sampling methods for asbestos are inadequate to support risk assessment and risk-based decision making at Superfund sites due to difficulties in detecting asbestos at low concentrations and difficulty in extrapolating soil concentrations to air concentrations. Environmental Protection Agency (EPA)’s Office of Land and Emergency Management (OLEM) currently recommends the rigorous process of Activity Based Sampling (ABS) to characterize site exposures. The purpose of this study was to compare three soil analytical methods and two soil sampling methods to determine whether one method, or combination of methods, would yield more reliable soil asbestos data than other methods. Samples were collected using both traditional discrete (“grab”) samples and incremental sampling methodology (ISM). Analyses were conducted using polarized light microscopy (PLM), transmission electron microscopy (TEM) methods or a combination of these two methods. Data show that the fluidized bed asbestos segregator (FBAS) followed by TEM analysis could detect asbestos at locations that were not detected using other analytical methods; however, this method exhibited high relative standard deviations, indicating the results may be more variable than other soil asbestos methods. The comparison of samples collected using ISM versus discrete techniques for asbestos resulted in no clear conclusions regarding preferred sampling method. However, analytical results for metals clearly showed that measured concentrations in ISM samples were less variable than discrete samples.
As part of the Superfund program, the Environmental Protection Agency (EPA) continues to clean up legacy sites contaminated with asbestos. Asbestos is consistently listed as a top 10 contaminant of concern for time-critical removals at Superfund removal sites (Information from the Superfund Enterprise Management System (SEMS) as of May 2, 2016). Many sites have asbestos contamination from building materials that were not removed or disposed in accordance with EPA regulations. Other sites, such as the Sumas Mountain Asbestos site, have naturally-occurring asbestos that has been disturbed by human activity and/or natural processes.
This study was designed to meet the following research objectives:
We found that there was no significant difference in variability in asbestos concentrations between ISM & discrete samples, although ISM resulted in significantly lower relative standard deviations (RSDs) for metals. We found that FBAS with TEM was the most sensitive analytical method, but concluded that different analytical methods cannot easily be compared. Finally, we found that metal concentrations did correlate with asbestos concentrations, which supports the use of metals as indicators on a site-specific basis.
This project supports the need for better, more accurate methods for determining asbestos levels in soils. Variability occurs during sample collection and analysis, depending on the methodologies chosen. Collection of triplicate ISM samples for each DU allowed for calculation of RSDs for metals and asbestos. Relatively low RSDs were calculated for ISM metals data compared to the RSDs for asbestos samples by either sampling method (ISM or discrete). While the low variability of the metals in discrete samples suggests a relatively homogenous underlying soil matrix, the variability of metals in ISM samples was even lower. The high variability of asbestos in soil samples regardless of sampling method suggests that either asbestos behaves differently from metals in the soil, or that the available laboratory preparation and analytical methods provide less reliable and more variable estimates of soil concentrations.
This study did not achieve its goal of identifying a combination of sampling, preparation, and analytical methods that would result in reproducible sample data for asbestos in soil, although we did find that FBAS/ISO 10312 was the most sensitive method available. Future study is needed to ascertain whether statistical error is due to field sampling and/or preparation, analytical preparation and/or analysis. Field sampling error could be addressed by increasing the number of increments (50–100) which may adequately address heterogeneity of asbestos in soils and allow for more reproducible sample results. Field preparation error could be addressed by having subsampling performed more consistently, perhaps in a laboratory instead of in the field. Analytical preparation error could be addressed by improving preparation of samples such as using a Turbula mixer to more thoroughly mix soils in the laboratory. Analytical error might be addressed through analysis of a larger sample aliquot or through improvements in analytical methods . Gy discusses systematic methods for resolving sampling error [30, 31]. Additional lines of inquiry should also include determining whether there is a difference in data reproducibility between fiber types (chrysotile vs amphibole fibers) and continued exploration of metals ratios as a predictor of asbestos content in soil.