Principal component analysis of Raman spectra from phosphorus-poisoned automotive exhaust-gas catalysts

Ramanspectrafromreal-world,phosphorus-poisonedautomotivecatalystscanberemarkablycomplicated, owing to the fact that factors such as variable driving conditions, catalyst formulations, and oil and fuel additives can contribute to the formation of a huge variety of compounds. To classify the Raman spectra from many catalysts without necessarily identifying all phosphorus contaminants, principal component analysis was performed. Examination of the scores in the PC1-PC2 plane showed that samples with similar spectra are grouped together while dissimilar ones are spaced apart from one another. The score plot also confirmed that the strongest peaks of orthophosphates generally lie close to one another. Examination of the loadings revealed that CePO4, and possibly Ca10(OH)2(PO4)6and BaSO4 ,a re among the major components present in the spectra. CePO4 is readily observed in many of the spectra, but Ca10(OH)2(PO4)6 is not, showing that principal component analysis was useful in uncovering possible spectral constituents difficult to unambiguously assign from the original spectra. The spectra can be adequately reproduced using eight principal components, which means that a much smaller set of numbers can be used to represent the spectra instead of typical wavenumber-intensity data. Copyright  2005 John Wiley & Sons, Ltd.