Interlaboratory comparison of HPLC-fluorescence detection and GC/MS: analysis of PAH compounds present in diesel exhaust

For laboratories involved in polycyclic aromatic hydrocarbon (PAH) analyses in environmental samples, it is very useful to participate in interlaborat ory comparison studies which provide a mechanism for comparing analytical m ethods. This is particularly important when PAH analyses are routinely done using a single technique. The results are reported for such an interlabora tory comparison study, in which the four selected participating laboratorie s quantitatively analyzed several PAH compounds in diesel exhaust samples. The samples included particle and vapor phase extracts collected and prepar ed at Michigan Technological University (MTU PE and MTU VE, respectively), a diesel particle extract prepared by the National Institute for Standards and Technology (NIST, SRM 1975), and a fully characterized diesel particle sample (NIST SRM 1650), One of the laboratories used only HPLC-FLD, one use d only GC/MS and two laboratories used both methods for the routine analysi s of PAH in environmental samples. Data were obtained for five PAH compounds: fluoranthene, pyrene, benz[a]ant hracene, benzo[a]pyrene, and benzo[g,h,i]perylene. The mean PAH levels foun d for SRM 1650 were outside the range reported by NIST. The range in the re ported means was from 24% lower than certified for benz[a]anthracene to 41% higher for benzo[g,h,i]perylene. For the previously uncharacrerized samples in this study (SRM 1975. MTU PE and MTU VE), two-thirds of the reported results were higher for the HPLC-FL D method than for the GC/MS. The range in differences between methods was f rom-54 to + 31% calculated as the difference in GC/MS value relative to the HPLC value for each or the compared compounds. Coefficients of variation for the uncharacterized samples appeared to be hi gher, in most (but not all) cases, for the HPLC-FLD than for the GC/MS. The resolution of certain PAH isomers (e.g. benz[a]anthracene and chrysene, or the benzofluoranthenes), was better. as expected, for HPLC than for GC. Ge nerally lower detection limits (by an order of magnitude or more) were repo rted for GC/MS than for HPLC-FLD. On the basis of this limited study, it se ems as though significant differences may exist between laboratories. if no t between methods, in the analysis of certain PAH compounds in real diesel samples by HPLC-FLD compared to GC/MS. If possible. measurements should be made using both methods. This is particularly important where potential int erferences are undefined or subject to change, as is frequently the case wi th real environmental samples. (C) 2000 Elsevier Science B.V. All rights re served.