Computer evaluation of gas chromatographic profiles for the correlation of quality differences in cold pressed orange oils

A technique for the quality evaluation of cold-pressed orange oils based on computer evaluation of gas chromatographic profiles has been developed. Chromatographic column parameters including column materials, stationary phases and methods of column preparation were investigated. Methods of sample introduction such as direct, whole oil injection using microsyringes, whole oil injection with sample splitting and heat desorption of sample headspace volatiles entrained on porous polymer adsorbents were evaluated. A solvent extraction procedure for removal of the hydrocarbon portion of the orange oil in order to simplify the chromatographic analysis was explored. Gas-liquid chromatography employing high efficiency capillary columns with temperature programming and direct whole oil injection was the analytical method of choice. Samples of cold-pressed orange oil were evaluated by experts involved in the daily purchase and use of essential oils and flavors. These samples, which ranged from very high quality to artificially oxidized oils unfit for use, were then analyzed by gas chromatography. The chromatographic data was fed into a computer correlation program which determined mathematical combinations of chromatographic peak areas that served to distinguish between the samples. A set of sample component ratios were ascertained which, when calculated for each sample, yielded values that ranked the samples in the same order as the experts. Scaling factors were determined for each ratio such that the sum of all ratios in the set would equal five hundred for an average quality oil with each ratio contributing equally to the sum. A quality evaluation computer program was written which will accept peak area data from a standard ASCII punched paper tape. The chromatographic data is fed into the computer data file from the punched tape, the appropriate ratios are calculated and scaled, the results summed and a number which corresponds to the oil sample's quality is printed out. Trial runs indicate the system's inherent error to be on the order of two percent for repetitive runs on the same sample. The system is capable of detecting quality changes, as judged by expert evaluations, in a single sample upon storage over a period of several months.