Predicting caking behaviors in powdered foods using a low-field nuclear magnetic resonance (NMR) technique

This study is concerned with caking of powder mixes. Our previous study tied the caking process to the nuclear magnetic resonance (NMR) spin-spin (T₂) relaxation characteristics exhibited by powder ingredients subjected to temperature scan from -20°C to 110°C (Journal of Food Science 65 (1) (2000) 134). Four temperature-T₂ curve patterns were identified, each of which represents a group of powder ingredients having a unique tendency to cake. The present study showed that which curve pattern a mixture of two or more powder ingredients would exhibit was governed by and could be predicted from the NMR relaxation characteristics of the individual powder ingredients that made up the mixture. The study also showed that caking susceptibility of a mixture could be regulated by varying the mixing ratio, which provides a basis for formulation of caking-resistant products. The close relationship between the curve patterns and caking was verified by the storage tests conducted at 37°C, 45°C, 55°C, and 65°C for 60 days, and described by a multiple regression model, which can be used for prediction of day until caking of the mixed systems at different temperatures based on their NMR relaxation behaviors.