Discrimination of adulterated milk using temperature-perturbed two-dimensional infrared correlation spectroscopy and multivariate analysis.

The discrimination method for adulterated milk is proposed based on temperature-perturbed two-dimensional (2D) infrared correlation spectroscopy and N-way partial least squares discriminant analysis (NPLS-DA). Two brands of pure and adulterated milk samples were prepared. The mid-infrared spectra of all samples were obtained from 30 â„ƒ to 55 â„ƒ with an interval of 5 â„ƒ. Under the perturbation of temperature, synchronous 2D correlation spectra were calculated to build discrimination models of pure milk and adulterated milk. In comparison, the NPLS-DA models were built based on three-dimensional (3D) stacked map (sample × temperature × wavenumber variable). For the NPLS-DA models of two brands of milk, the discrimination accuracy of unknown samples in the prediction set is 100% using temperature-perturbed 2D infrared correlation spectra, versus 77.8% using conventional 3D stacked map. The proposed method can be used as an alternative way for classifying pure and adulterated milk.