Abstract: In order to avoid environmental pollution and health hazards caused by the use of organic solvents in Soxhlet extraction, and improve the detection efficiency of fat content in pork, a method for determining fat content in pork based on low-field nuclear magnetic resonance (LF-NMR) was proposed. Signal acquisition parameters (echo time and number of scanning cycles), sample mass, drying time and detection temperature were optimized, and the repeatability and precision of the method were verified. When 4 g of minced pork was dried for 6 h in an oven and detected under the following conditions: temperature, 50 ℃; echo time, 0.3 ms; and number scanning cycles, 64, good results were obtained using lard as a standard sample. The correlation coefficient (R2) of the standard calibration curve established was 0.999 9. The repeatability relative standard deviation (RSD) of the proposed method was 1.69%–2.72%, and the intra-day and intra-day precision (RSDs) were 3.07% and 2.57%, respectively. The correlation coefficient between the LF-NMR method and the Soxhlet extraction method specified in the national standard (GB 5009.6–2016) was 0.999 5. Therefore, the LF-NMR method can accurately determine fat content in pork and other meats.

Key words: low-field nuclear magnetic resonance; fat content; drying; parameter optimization; relaxation characteristics