Pattern recognition for analysis of cigarette smoke by capillary gas chromatography. Part 1: Total particulate matter (TPM)

Abstract

AbstractComplex smoke profiles from fused silica capillary gas chromatography of different cigarette types produce rich information requiring computer and sophisticated statistical analysis to reveal the hidden correlation patterns. In our technique, a Cambridge pad containing total particulate matter (TPM) is extracted with Freon‐11 in a micro‐Soxhlet apparatus. The condenser is maintained at −15°C. After extraction, Freon‐11 is allowed to evaporate under mild conditions. Bis(trimethylsilyl)trifuoroacetamide (BSTFA) is subsequently added to the extract. Trimethylsilylation is completed by maintaining the reaction vessel at 60°C for two hours. A microprocessor‐controlled capillary gas chromatograph with an automatic sampler is used to generate profiles. A link to a host computer provides for transmission of profile reports and access to various statistical and graphic packages. Factor and discriminant analyses are applied to gas chromatographic data from TPM extracts because of their ability in data reduction, pattern extraction and ranking the importance of the gas chromatographic peaks. Some preliminary results from these statistical analyses are discussed.