LAUSR

Abstract This study aimed to evaluate the efficacy of time-domain nuclear magnetic resonance (TD-NMR) to determine the formaldehyde in milk samples (from 1.75 to 21% v/v) using chemometric methods (principal component analysis, partial least squares regression, and soft independent modeling of class analogy, PCA, PLS, and SIMCA), when compared with the official colorimetric method. The analysis was performed at different times of refrigerated storage (0 h and 48 h). Increased formaldehyde concentration (17.5 and 21%) led to a decrease in the whiteness index [65.3–72.8] and the lightness values [70.5–85.3] and an increase in the yellowing index [36.31–38.73] compared to the control [77.8–78.7, 85.2–90.3, and 27.40–29.92, respectively]. Regarding the TD-NMR values, higher T2 values were observed, with the formation of T2,1 and T2,2 throughout the refrigerated storage. PCA and PLS proportionated adequate models while SIMCA was able to discriminate the samples regards the level of adulteration with adequate discrimination and predictive index. Overall, TD-NMR has proven to be an effective method for the dairy industry to check the formaldehyde levels in raw milk.