Abstract

The technique of near-infrared (NIR) reflectance spectroscopy was used in order to test its use as a potential method for determining the rancidity of soybean oils undergoing autoxidation. Commercial soybean oil was used as a substrate and was oxidized at 75°C until the oxidized oil reached the hydroperoxide decomposition stage. Peroxide values (POVs) and acid values (AVs) of the substrate determined by the chemical method were correlated with NIR spectral absorbances. Two wavelengths (2080 and 2020 nm) were selected for the POV calibration, and the resulting coefficient of multiple determination for regression was quite high (r = 0.9970, SEC = 9.10 meq / kg oil and SEP = 9.67 meq / kg oil). The prediction of oil rancidity by NIR using POV as a reference was found to be accurate in the early stages of oxidation. However, it was undesirable to use the POV calibration equation for the determination of rancidity in the hydroperoxide decomposition stages. Three wavelengths (2008, 1442 and 1752 nm) were used for AV calibration. High correlations were achieved between chemically analyzed AVs and NIR predicted AVs (r = 0.9987, SEC = 0.127 mg KOH / g oil, and SEP = 0.137 mg KOH / g oil). The AV represented a good indicator for the determination of rancidity of oil undergoing thermal oxidation, whereas evaluation of rancidity using the POV has proven to be difficult due to hydroperoxide decomposition. The result of this study suggested that prediction of rancidity by NIR using POV calibration equation was more precise than using an AV calibration equation in hydroperoxide accumulation stages, whereas the AV calibration equation was a better predictor of oil rancidity than the POV calibration equation at higher POV ranges.

© 1998 NIR Publications