Improved peak selection strategy for automatically determining minute compositional changes in fuels by gas chromatography-mass spectrometry

During the development of automated computational methods to detect minute compositional changes in fuels, it became apparent that peak selection through the spectral deconvolution of gas chromatography-mass spectrometry (GC-MS) data is limited by the complexity and noise levels inherent in the data. Specifically, current techniques are not capable of detecting minute, chemically relevant compositional differences with sufficient sensitivity. Therefore, an alternative peak selection strategy was developed based on spectral interpretation through interval-oriented parallel factor analysis (PARAFAC). It will be shown that this strategy outperforms the deconvolution-based peak selection strategy as well as two control strategies. Successful application of the PARAFAC-based method to detect minute chemical changes produced during microbiological growth in four different inoculated diesel fuels will be discussed.