FOOD COMPOSITION AND ADDITIVES Improved Determination of Isolated trans Isomers in Edible Oils by Fourier Transform Infrared Spectroscopy Using Spectral Reconstitution FREDERIK R. VAN DE VOORT and JACQUELINE SEDMAN McGill IR Group, McGill University, Department of Food Science and Agricultural Chemistry, 21,111 Lakeshore Rd, Ste. Anne de Bellevue, QC H9X 3V9, Canada SYED TUFAIL HUSSAIN SHERAZI University of Sindh, National Center of Excellence in Analytical Chemistry, Jamshoro, 76080 Pakistan A substantially more sensitive and accurate alternative to the single-bounce attenuated total reflectance (SB-ATR) Fourier transform infrared spectroscopic method of AOAC/American Oil Chemists’ Society (AOCS) was developed for determination of isolated trans isomers, based on transmission measurements using a technique called spectral reconstitution (SR). The method involves the 1:1.5 dilution of an oil with odorless mineral spirits (OMS) containing a spectral marker. The resulting reduction in sample viscosity facilitates the use of a transmission flow cell, with the spectral marker serving to determine the precise dilution ratio. This allows the spectral contributions of the OMS to be eliminated and a facsimile of the neat oil spectrum to be mathematically reconstituted. The transmission- SR (T-SR) procedure was initially evaluated relative to SB-ATR to track changes in the trans content of mixtures of unhydrogenated canola and a highly hydrogenated sunflower oil (0–30% trans). The results indicated that the T-SR procedure had the potential to serve as the basis of an accurate quantitative method. A subsequent T-SR calibration based on the spectral ratioing principle of the SB-ATR AOAC/AOCS method was developed by gravimetrically adding trielaidin (0–4%) to extra virgin olive oil (EVO), producing an excellent linear response with a standard deviation (SD) of <0.04% trans. Subsequent comparison of SB-ATR and T-SR calibrations developed for 5 oils of different types, each spiked with low levels of trielaidin (0–1.2% trans), clearly indicated that SB-ATR was signal-limited, whereas the T-SR procedure performed well. The EVO calibration was subsequently used to predict the added trans content of these spiked oils, after the spectrum of the corresponding unspiked oil had been ratioed out. The resulting plot of predicted versus added trans was linear, with a slope of 1.02 and an overall SD of <0.05% trans. When the spectra of these oils were ratioed against the spectrum of EVO, the trans predictions for some of the oils were offset by 2–3 percentage points, emphasizing the need for the appropriate trans-free reference oil to perform accurate analyses. If the latter condition is met, then T-SR provides a very simple technique, with the potential for automation, for analysis of oils by transmission spectroscopy, with approximately 20´ the sensitivity of the AOAC/AOCS SB-ATR method. N utritional labeling legislation in North America and elsewhere requiring declaration of the trans fat content of foods has resulted in the need for a simple, accurate, and routine analytical method to determine trans content. Among the officially approved methods, the simplest is the direct analysis of neat oils by single-bounce attenuated total reflectance (SB-ATR) Fourier transform infrared (FTIR) spectroscopy (1, 2). Like earlier IR transmission-based methods for trans analysis (3), this methodology is based on measurement of the area of the CH out-of-plane deformation absorption of isolated trans double bonds at 966 cm –1 , but dissolution of the sample in CS 2 is no longer required. Furthermore, the need for conversion of triacylglycerols (TAGs) to fatty acid methyl esters (FAMEs) before analysis is eliminated; this advantage is predicated on the availability of a trans-free reference oil of the oil type analyzed so that interfering TAG absorption bands can be ratioed out (4). However, even when the latter condition is met, the limit of quantitation of this method, specified by the American Oil Chemists’ Society (AOCS; 1) as 1% of total fat and by AOAC (2) as 5% of total fat, is inadequate for measuring low trans levels for compliance with nutrition labeling regulations. A major factor contributing to this lack of sensitivity is the inherently short effective pathlength associated with the SB-ATR technique (approximately 4.4 mm at 966 cm –1 ). As demonstrated in our earlier work (5), this limitation can readily be overcome if one uses a transmission cell; the optimal pathlength for measurement of the trans band is 50–75 mm (depending on trans content). However, the inherently high VAN DE VOORT ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 90, NO. 2, 2007 1 Received July 13, 2006. Accepted by SG September 14, 2006. Corresponding author's e-mail: frederik.vandevoort@mcgill.ca