Improved GC/MS identification rates for untargeted waste water screening due to advanced deconvolution tools

For medium to complex matrices, it is well know that GC/MS deconvolution is probably one of the most important and critical steps to get reliable identifications for screening assays. The ultimate task is to obtain best estimated spectra for the pure components present in the sample. Besides deconvolution, the identification algorithm used to match unknown spectra with a spectral library also can have big impact on the final identification results, especially for components at low to very low levels.

New deconvolution algorithms have been developed that will take into account close elutions from interfering components. The processing for multiple peaks at the same time will in general lead to more precisely estimated spectra for all components. It will be shown that 30-40% more reliable identifications (at a certain identification threshold) can be obtained on complex samples in which a substantial part of the components show interference from nearby peaks. The high identification rate and reliability of results using above procedure will greatly reduce the time spend on data analysis. A second advantage is that the new algorithm will give improved quantitation results for a complex mixture of overlapping peaks.

Finally we will show that a new identification algorithm, developed for low quality spectra, can give improved identification rates. The algorithm is based on established algorithms but will take into account ""chromatographic"" information like signal to noise ratio's which are obtained directly from the deconvolution process.

Technical details of the algorithms will be explained and examples will be given for waste water screening assays using GC/MS and GCxGC/MS in the Netherlands and Belgium.