Search

Abstract: The assumption of an instrument response that is linear with carbon number is frequently used to quantify atmospheric non-methane hydrocarbons (NMHCs) when using gas chromatography (GC) and detection by flame ionisation detector (FID). In order to assess the validity of this widely used method the results of intercomparison measurements by 14 laboratories across Europe were evaluated. The intercomparison measurements were made on synthetic, gravimetrically-prepared, gas mixtures containing 30 hydrocarbons (C2–C9) in the low ppbv range, using various different GC–FID systems. The response per carbon atom of GC–FID systems to individual NMHCs, relative to that of butane, were found to differ by more than 25% across different systems. The differences were mostly caused by analytical errors within particular GC–FID systems and to a more minor degree by systematic deviations related to the molecular structure. (Correction factors due to the molecular structure would lessen the differences, e.g. by about 5% for olefin compounds.) The differences were larger than 10% even after elimination of obvious outliers. Thus, calibration of GC–FID systems with multicomponent NMHC mixtures is found to be essential whenever the accuracy of NMHC measurements is required to be better than about 10%. If calibration by multicomponent gas mixtures is not possible and effective carbon atom response factors are used to quantify the individual NMHC compounds then the particular analytical system should be carefully characterised and its responses to individual compounds be verified. [Copyright &y& Elsevier]