Raw milk is hard to analyze by NIR because of light scattering caused by large particles in the fat component; however, improvements to sample preparation and presentation could improve analysis. In this study, sample presentation and particle size reduction using homogenization was investigated for 160 diverse samples to predict component quantities in raw milk by FT-NIR. Excellent standard errors were obtained for all measured components following no-homogenization. Sonication and tube-dispersion methods were not different than no-homogenization, but were better compared to two-stage homogenization for moisture and casein due to whey protein denaturation during processing. High sample temperatures during analysis likely contributed to positive results for all component predictions compared to references. For sample presentation, static and dynamic flow-cell methods produced the best standard errors for components. Meanwhile, petri-dish presentation was accurate but may have been limited by the method design, which allowed for sample dehydration and loss of reflected light.
University of Minnesota M.S. thesis. April 2013. Major: Food science. Advisors: Tonya Schoenfuss, Ph.D., Leonard Marquart, Ph.D., R.D. 1 computer file (PDF); vii, 143 pages, appendix p. 104-143.
Reuter, Anthony Joseph.
Optimizing sample preparation and scanning methods for component analysis of raw milk by fourier transform near infrared spectroscopy.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.