Multivariate prediction of clarified butter composition using Raman spectroscopy

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Food Science

Background:

• Dietary fat composition, particularly fatty acids (FA), influences dairy product quality.
• Accurate and rapid methods are needed for quantifying FA in butterfat for quality control.
• Traditional methods like Gas Chromatography (GC) can be time-consuming and require sample preparation.

Purpose of the Study:

• To evaluate Raman spectroscopy as a rapid, non-destructive method for predicting FA abundance in butterfat.
• To assess the accuracy of Raman spectroscopy for predicting bulk parameters such as iodine value and solid fat content.
• To determine the feasibility of on-line, high-throughput analysis of butter samples using Raman spectroscopy.

Main Methods:

• Raman spectra were collected from clarified butterfat from cows fed varying levels of rapeseed oil.
• Partial Least Squares (PLS) regression was used to correlate Raman spectra with FA compositions determined by GC.
• Prediction accuracy was assessed using R-squared (R2) and root mean square error of prediction (RMSEP).

Main Results:

• Raman spectroscopy achieved good prediction for major FA (R2 = 0.74-0.92) and excellent prediction for unsaturated FA (R2 = 0.85-0.92), especially trans unsaturated FA.
• Accurate predictions were obtained for iodine value (R2 = 0.80) and low-temperature solid fat content (R2 = 0.87).
• While prediction errors are larger than GC, Raman offers high accuracy for quality control with rapid (60-s) and sample-prep-free analysis.

Conclusions:

• Raman spectroscopy is a viable tool for rapid, on-line prediction of FA composition and key quality parameters in butterfat.
• The method's effectiveness is particularly strong for unsaturated fatty acids due to distinct spectral signatures.
• High-throughput Raman analysis of butter samples is feasible, enabling efficient quality control in the dairy industry.