Pembuatan Model Kalibrasi Sampel Recovery Produk Susu dengan Metode Spektroskopi Inframerah di PT Frisian Flag Indonesia
Abstract
Milk is a kind of food that has nutritional contents and compositions almost perfect compared to other foods. Manufacture of milk products result by-products called recovery samples. Infrared spectroscopy methods such as Near Infrared Reflectance (NIR) and Fourier Transform Infrared (FTIR) can be used as a secondary method for evaluating quality of dairy products. PT Frisian Flag Indonesia will use infrared spectroscopy methods to analyze quality of recovery samples. FTIRS and NIRS instruments must have a calibration model for recovery sample that includes parameters of composition i.e total solids, fat content, sucrose, and proteins so that can be used to evaluate quality of recovery sample. The purpose of this study was to create a calibration model from recovery samples of dairy products by infrared spectroscopy methods. In general, the development stages of calibration models using NIR and FTIR methods consist of sample preparation, analysis of composition parameters, calibrations modeling, and verification of the models. Sample preparation was canducted by using two variants of A product recovery sample (NIR method) and B product recovery sample (FTIR method). Instrument used for analyzing NIR method is FOSS NIRSystems 5000 and for FTIR method is FOSS MilkoScan FT120. The conventional analysis methods were also used to analyze parameters of composition. Calibration modeling that conducted by software WinISI and models was verificated by t-test. The results of spectrum NIR shows A product recovery sample for variant X and Y have similar spectrums. Model calibration of total solid is the best model in NIR methods (R2 = 0.9998) and FTIR method (R2 = 0.9993). The results of t-test with 5% level of significance shows that the parameters of composition in both calibration models of infrared spectroscopy methods were not significant different (P (two-tail) > 0.05). This results showed that both of calibration models can be used.