Sampling procedures for intrinsically valid volatile organic compound measurements

Abstract

The concept of an intrinsically valid volatile organic compound (VOC) measurement is described, whereby the final data contain all the information needed to calibrate and validate the determination. To realise such a system, all modes of failure within the measurement system need to be tested as the analysis proceeds, and the results of the tests incorporated into the final analytical output. The sources of error and uncertainty associated with the seven principal stages of a VOC analysis sequence are considered. In particular, the effect of competitive adsorption, arising from a transient concentration excursion, on the determination of other analytes is highlighted with an example showing a 97.5% reduction in sensitivity towards benzene in air that, in many recommended methods, would remain undetected. An experimental approach is presented which indicates how the problem of validating adsorbent-based VOC measurements may be addressed by adopting an intrinsically valid measurement approach. A sampling manifold is used that enables two internal standards, fluoropentane and fluorononane, to be added to sampled air in a reproducible and traceable manner before the sample is presented to six adsorbent traps. Thus, these internal standards are simultaneously trapped on the adsorbent alongside the airborne VOC. Upon analysis, the internal standards provide a reference within the chromatogram, allowing the validity of the sampling and subsequent steps in the measurement sequence to be assessed. Fluoropentane and fluorononane were chosen because, amongst other reasons, they bracket the adsorbent capacities of Tenax towards benzene, toluene and xylene. Thus, by comparing the behavour of the fluorinated standards within the analysis, it was possible to identify the range of faults that may arise during active sampling, thermal desorption, chromatographic separation and detection. The chromatogram obtained from this approach had associated with it all the features of an intrinsically valid VOC measurement. Furthermore, the system described resulted in 21–51% reductions in the relative standard deviations for the compounds studied, with the exception of 1-xylene.