Evaluation of Commercially Available Milk Powders for Calibration of Mid-Infrared Analyzers
Lynch, J.M.; Barbano, D.M.; Fleming, J.R.chard
Journal - Association of Official Analytical Chemists 78(5): 1219-1224
The performance of commercially available shelf-stable milk powders that upon reconstitution can be used as calibration samples for mid-infrared (IR) analyzers was evaluated. Twelve reconstituted powders and 7 raw milk samples were sent to each of 7 laboratories. Each laboratory analyzed the reconstituted powders and milk samples by IR and obtained both corrected and uncorrected data. The corrected data reflected the current calibrations (based on calibration with fresh, raw milk) of each instrument. The uncorrected data from the reconstituted powders were used to derive separate calibration equations for each instrument. Two of the 12 reconstituted powers were discarded from the analysis because of poor quality. The abilities of the powder-based and current calibrations to predict accurately the total protein and fat contents of the 7 raw milk samples were compared. For the current and powder-based calibrations, respectively, the mean differences from chemical analysis for total protein were -0.006 and 0.031%, and the standard deviations of the differences (SDD) were 0.029 and 0.029%. For fat, the mean difference for the current calibration was -0.001%, and the SDD was 0.034%. For the powder-based fat calibration, the mean differences were 0.034% (fat A only), -0.047% (fat B only), and -4.061% (fat A and fat B), and SDDs were 0.046% (fat A only), 0.036% (fat B only), and 0.056% (fat A and fat B). The reconstituted powders did not provide as accurate an IR fat calibration for testing raw milk samples as can be obtained with raw milk calibration samples. This is attributed primarily to differences in the characteristics of the fat in the reconstituted powders and in raw milk. For protein, analytical precisions for both types of calibrations were comparable, and the positive bias for the powder-based calibration is relatively easy to correct by basing calibration on true protein nitrogen instead of total nitrogen.