STUDY OF CHEMICAL AND MOLECULAR INFORMATION RELATED TO NIR AND IR SPECTROSCOPIC DATA FOR DAIRY SECTOR

In the last decades, the spectroscopic techniques have acquired reliability, since they are sufficiently accurate and precise for analysis of the macro-composition of food. In this work, NIR and IR techniques were applied to study some aspects of cow milk casein and fat globules. In dairy field, casein amount and quality have great influence on milk rennet properties and cheese yield. Caseins from milk of ruminants have been extensively studied, but the exact structure of the casein micelle is still debated. The research activity was addressed to verify the ability of spectroscopic techniques in the evaluation of modifications of casein fractions and sub-fractions as a function of pH and temperature. The NIRS ability in predicting casein fractions content and in detecting bonds involved in the micelle complex were also evaluated. The study was carried out on both commercial preparations of casein fractions and reconstituted casein samples. These were obtained by ultracentrifugation (native casein) and by precipitation at the iso-electric pH (acid casein) of individual milk samples, collected during two months periods in the Austria’s region. The IR spectra of commercial caseins showed the phosphate band at 1100 cm-1, confirming its role in the stabilization of casein micelle structure. When NIR casein spectra were measured as a function of temperature, exclusively changes in water bands were detected, while regarding pH, Absorbance differences from the mean spectrum evidenced some modifications of linearity due to the number of negative charged amino acid residues at pH > 6.80 in the casein sub-fractions. Casein fractions content of reconstituted samples was determined by Capillary Zone Electrophoresis analyses. PLS analyses, performed with electrophoretic and NIR data, revealed the NIRS ability to determine and quantify casein genetic variants useful for milk selection for its final purpose. Moreover, the PCA analysis on the same samples proved the NIRS ability also to discriminate between samples obtained by physical and chemical treatments and to detect bonds involved in the micelle structure, especially phosphate group and its binding to calcium. FT-NIR spectroscopy was also applied to study the size distribution of fat globules: an aspect influencing the technological and sensorial milk characteristics. In this contest, the variability in the distribution of fat globules within cow breedings in Lombardy was studied during two years period. The reference particle size analyses of fat globules were performed using a granulometer. The Sauter Mean Diameter (SMD) was chosen as the best descriptor of particle size distribution. This parameter resulted to be more influenced by genetic factors than seasonal aspects. The differences among farms could be determinant in planning the milk collection for the technological destination, while the differences among the breeding bulls can be used for the animals’ selection. Despite the importance of this parameter in several dairy processes, instrumentation for particle size analysis are not available in dairy laboratories. NIR instrumentation, instead, is largely used in dairy labs. NIR spectrum of whole milk arises from absorbance due to both molecular vibrations and elastic scattering related to the presence of fat globules in emulsion. Moreover, the amount of scattered photons depends on their size and wavelength. A rapid and economic method for estimating the distribution of fat globules in milk through a physical-mathematical model based on the study of the scattering component in the NIR spectrum was developed. The model, working in Visual Basic for Excel, calculates the optical density produced by milk fat globules, given the fat concentration. On the basis of the Weibull distribution, the model calculates the amount of globules with a certain diameter range, returning a first distribution curve. After the generation of a theoretical NIR spectrum, the model inversion was performed by minimizing the sum of squared differences between measured and theoretical spectra. At the end of the process, the new distribution curve was given. The performances of the model was tested by analyzing an external data set with both NIR and reference diffractometric data. For the SMD a very high coefficient of determination in prediction was found. In order to improve the applicability of the model, the use of a portable spectrometer for estimating the distribution of milk fat globules was evaluated trough standardization to bench-top instrument spectra. The calculation of milk fat globules diameter by a portable instrument standardized spectra through the mathematical model, gave comparable results with those calculated on master spectra.