2002 Volume 45 Issue 5 Pages 271-278
Field measurement of volatile petroleum hydrocarbon in soil usually precedes the submission of samples for more rigorous analysis in the laboratory. Hand-held photoionization detectors (HHPIDs) are commonly used for field measurements, although there is growing concern about their reliability under field conditions. This concern is accentuated by an increasing tendency to rely on such field measurements alone. Three different case studies, where commercial HHPID field data and the corresponding laboratory data have been made available to us, were critically examined. Disturbingly, there is little correlation between the HHPID readings obtained in the field and the corresponding measurements in the laboratory, with a large number of false positives being recorded. Also, the measurements obtained from the HHPID were invariably much higher than the corresponding laboratory measurements. In an attempt to identify some of the underlying problems that could influence the response of HHPIDs in the field, a series of laboratory experiments was conducted under controlled conditions in order to monitor, by Gas Chromatography with Mass Selective Detection (GCMSD), the linearity and accuracy of HHPID response to spiked BTEX levels (benzene, toluene, ethyl benzene and xylenes) in a variety of matrices under varying moisture content conditions (namely; in pure water, dry sandy loam, dry silt and moist silt). These studies reveal that, under controlled conditions, the HHPID response is linear over a spiked BTEX range of approximately 0-500mg/kg for a variety of matrices. However, the sensitivity (as measured by the slope of the response line) depends very much on the specific combination of conditions, such as the matrix and moisture content. These results suggest that for linearity of response to be established in the field, corrections must be made for variations in moisture content and soil type, and perhaps for other parameters. The slope of the response line could then be used as an "instrument factor" to quantify the level of contaminant from the HHPID response. Several other factors that could influence HHPID readings were also investigated including the HHPID response over time and the effect on response of component interactions in contaminant mixtures. The former investigation shows that there is no statistical difference between HHPID measurements taken up to 1.5h apart from identical spiked samples. The latter study suggests that multiple components in a sample may affect the HHPID response.