JOURNAL OF THE AMERICAN OIL CHEMISTS SOCIETY, vol.92, no.2, pp.175-184, 2015 (SCI-Expanded)
The performance of a portable infrared system combined with pattern recognition to discriminate between organically and conventionally produced bovine butter samples as well as to predict the levels of conjugated linoleic acid (CLA) were evaluated. Sixty butter (27 organic and 33 conventional) samples were used in this study. Bovine butter–fat were applied onto an attenuated total reflectance infrared (ATR-IR) accessory equipped with a five-bounce ZnSe crystal set at 65 °C for spectral collection. In addition, ATR-IR spectra of bovine butter were directly collected at room temperature to avoid phase separation. The fatty acid profile and the levels of CLA were determined using reference FAME-GC-FID analysis. SIMCA models showed well separated clusters that discriminated between organic and conventional bovine butters due to C=C trans bending out of the plane vibration modes band at 967 cm−1. Additionally, strong PLSR models were developed to predict CLA levels using butter–fat and bovine butter spectra with SEP of 0.05 % and RPD of 4.7, indicating that the models are suitable for quality control applications. Portable IR technology offers the ability for “in situ” analysis of butters that is much less time consuming than current analytical practices for authentication and quality control efforts by the industry.