Characterization of the optical properties of soymilk with different pH values using excitation-emission matrix and...

Fluorescence spectroscopy was evaluated for non-destructive pH measurement in soymilk using excitation-emission matrix (EEM). EEMs of 110 soymilk samples with different pH levels were measured, showing two characteristic fluorescence peaks at excitation/emission wavelengths of 280/340 and 365/460 nm, attributed to aromatic amino acids and other fluorescent compounds, respectively. Both peaks showed increased intensity with elevated pH levels. Partial Least Squares Regression (PLSR) and Support Vector Machine (SVM) models were then constructed using various preprocessing methods. The SVM model using raw EEM achieved the highest accuracy with R²cv of 0.972 and RMSECV of 0.095, while the PLSR model performed best with logarithmically transformed EEM showing R²cv of 0.951 and RMSECV of 0.127. Complementary visible-near infrared transmission measurements suggested that the fluorescence changes were related to the changed in light scattering properties under alkaline conditions. These results demonstrate the potential of fluorescence spectroscopy as a tool for non-destructive pH measurement in soymilk.