To clarify the sources of undesirable flavors, pure hydroperoxides from autoxidized and photosensitized oxidized fatty esters were thermally decomposed in the injector port of a gas chromatograph-mass spectrometer system. Major volatile products were identified from the hydroperoxides of methyl oleate, linoleate and linolenate. Although the hydroperoxides from autoxidized esters are isomerically different in position and concentration than those from photosensitized oxidized esters, the same major volatile products were formed but in different relative amounts. Distinguishing volatiles were, however, produced from each type of hydroperoxide. The 9-and 10-hydroperoxides of photosensitized oxidized methyl oleate were thermally isomerized in the injector port into a mixture of 8-, 9-, 10-and ll-hydroperoxides similar to that of autoxidized methyl oleate. Under the same conditions, the hydroperoxides from autoxidized linoleate and linolenate did not undergo significant interconversion with those from the corresponding photosensitized oxidized esters. The compositions of the major volatile decomposition products are explained by the classical scheme involving carboncarbon scission on either side of alkoxy radical intermediates. Secondary reactions of hydroperoxides are also postulated, and the hydroperoxy cyclic peroxides from methyl linoleate (photosensitized oxidized) and methyl linolenate (both autoxidized and photosensitized oxidized) are suggested as important precursors ofvolatiles.

Atmospheric pressure chemical ionization-mass spectrometry (APCI-MS) was used for quantitative analysis of triglycerides (TG) separated by reverse-phase high-performance liquid chromatography. APCI-MS was used for analysis of mono-acid TG standards containing deuterated internal standard, of a synthetic mixture of heterogeneous TG, of randomized and normal soybean oils and of randomized and normal lard samples. Quantitation of the TG by four approaches based on APCI-MS were compared, and these were compared to quantitation obtained using liquid chromatography (LC) with flame-ionization detection (FID). The APCI-MS methods were based on (i) calibration curves from data for mono-acid TG standards, (ii) response factors obtained from a synthetic mixture of TG, (iii) response factors calculated from comparison of randomized samples to their statistically expected compositions, and (iv) response factors calculated from comparison of fatty acid (FA) compositions calculated from TG compositions to FA compositions obtained by calibrated gas chromatography (GC) with FID. Response factors derived from a synthetic mixture were not widely applicable to samples of disparate composition. The TG compositions obtained using APCI-MS data without application of response factors had average relative errors very similar to those obtained using LC-FID. Numerous TG species were identified using LC/APCI-MS which were undetected using LC-FID. Two quantitation methods, based on response factors calculated from randomized samples or on response factors calculated from FA compositions, both gave similar results for all samples. The TG compositions obtained using response factors calculated from FA compositions showed less average relative error than was obtained from LC-FID data, and were in good agreement with predicted compositions for the synthetic mixture and for randomized soybean oil and lard samples.

Canola oil triacylglycerols from genetically modified canola lines {InterMountain Canola Co., Cinnamlnson, NJ) have been evaluated for their photooxidative and autoxidative stabilities, as influenced by their fatty acid compositions and their triaeylglycerol compositions and structures. Purified canola oil triacylglycerols were oxidized in duplicate in fluorescent light at 25~ and in the dark at 60~ under oxygen, and their oxidative deterioration with time was monitored by determining eolorimetric peroxide values. Also monitored with time, oxidation products were determined by reversed-phase high-performance liquid chromatography with ultraviolet absorbance detection. Total volatiles, generated by thermal decomposition of the oxidized triacylglycerols, were quantitated by static-headspace gas chromatography. These experimental parameters were statistically correlated with predicted oxidizability, fatty acid composition, position of fatty acids on glycerol carbons and triaeylglyeerol composition. Oxidative de terioration of canola triacylglycerols correlated negative ly with oleic acid composition, with oleie add content at carbon 2 and with trioleoylglycerol content of the oil. Deterioration was positively correlated with the amount of linolenic acid on nonspecific locations on glycerol cap bons 1,2 and 3, the amount of linoleic acid on glycerol cap bon 2 and with sn~)leoyllinoleoyllinolenoyl glycerol content. Differences in character or quantity of volatile product and triacylglycerol hydroperoxides were low, whether generated during autoxidation or photooxidation of the canola triacylglycerols.

The oxidative stability of soybean oil triacylglycerols was studied with respect to composition and structure. Crude soybean oils of various fatty acid and triacylglycerol composition, hexane~extracted from ground beans, were chromatographed to remove non-triacylglycerol components. Purified triacylglycerols were oxidized at 60~ in air, in the dark. The oxidative stability or resistance of the substrate to reaction with oxygen was measured by determination of peroxide value and headspace analysis of volatiles of the oxidized triacylglycerols (at less than 1% oxidation). The correlation coefficients (r) for rates of peroxide formation (r = 0.85) and total headspace volatiles (r = 0.87) were related positively to oxidizability. Rate of peroxide formation showed a positive correlation with average number of double bonds (r = 0.81), linoleic acid (r = 0.63), linolenic acid (r = 0.85). Rate of peroxide formation also showed a positive correlation with linoleic acid (r = 0.72) at the 2-position of the glycerol moiety. A negative correlation was observed between rate of peroxide formation and oleic acid (r = --0.82). Resistance of soybean triacylglycerols to reaction with oxygen was decreased by iinolenic (r = 0.87) and increased by oleic acid (r = --0.76)-eontalning triacylglycerols. Volatile formation was increased by increased concentration of linolenic acid at exterior glycerol carbons 1,3 and by linoleic acid at the interior carbon 2. Headspace analysis of volatiles and highperformance liquid chromatography of hydroperoxides indicated that as oxidation proceeded there was a slight decrease in the linolenic acid-derived hydroperoxides and an increase in the linoleic acid~ierived hydroperoxides. The oxidative stability of soybean oil triacylglycerols with respect to composition and structure is of interest to the development of soybean varieties with oils of improved odor and flavor stability.