You are here:
Internal Standards: A Source of Analytical Bias For Volatile Organic Analyte Determinations
Citation:
HIATT, M. H. Internal Standards: A Source of Analytical Bias For Volatile Organic Analyte Determinations. JOURNAL OF CHROMATOGRAPHY A. Elsevier Science Ltd, New York, NY, 1218:498-503, (2011).
Impact/Purpose:
Volatile organic compounds (VOCs) make up a major group of compounds routinely monitored as environmental contaminants. RCRA SW-846 Method 8260C is the determinative protocol of choice with a pre-concentration protocol such as headspace (Method 5021) or purge-and-trap (Method 5030C) (1). There are investigations addressing measurement uncertainty as analytical (2, 3), sampling and sub-sampling errors (4), and their comparative importance (5); however, for VOC determinations there is a potential bias rarely addressed. That is, the dissimilarity between analytes and their internal standards can result in quite different behavior during analyses.
Description:
The use of internal standards in the determination of volatile organic compounds as described in SW-846 Method 8260C introduces a potential for bias in results once the internal standards (ISTDs) are added to a sample for analysis. The bias is relative to the dissimilarity between the analyte and internal standard physical properties that influence how easily analytes are separated from a matrix and concentrated during analysis. Method 5032 is a vacuum distillation procedure for extracting analytes from a sample for use with Method 8260C. Vacuum distillation is also incorporated within another GC/MS analytical procedure, Method 8261A. Method 8260C/5032 and Method 8261A are experimentally identical, however Method 8261A uses internal standards differently by relating the recovery of each compound to its boiling point and relative volatility. By processing each analysis (water, soil, and biota) using both Method 8260C and Method 8261A, the two approaches are compared on the basis of analyte bias and the failure rate of the quality controls. Analytes were grouped by how similar their boiling points and natural log of their relative volatilities (lnRVs) were to their Method 8260C recommended ISTDs. For the most similar analytes, the Method 8260C determinations yielded an average bias less than 10% and a failure to meet calibration criteria less than 7%. However, as the difference between analyte and ISTD became greater the bias increased to over 40% (matrix dependent) and its calibration failure rate approached 70%. In comparison, when the Method 8260C data were reprocessed as Method 8261A determinations, this trend for groupings was minimized with biases increasing from 6 to only 20% and the calibration failure rate went from 0 to 15%.