With the increasing degree of mine mechanization, the concentration of dust produced by mining operation also increases. The long-term exposure of workers to the mine environment containing crystalline silica will lead to the occurrence of silicosis, and the rate and degree of silicosis are closely related to the content of crystalline silica in mine dust. To improve the accuracy and efficiency of detecting crystalline silica in mine working environments, the direct-on-filter (DoF) method based on infrared spectroscopy (IR) was studied, eliminating the sample preparation step: ashing the sample filter. This study utilized three types of mine dust samples: coal dust, coal gangue dust, and metal mine dust. According to the mass of crystalline silica contained in the samples and its infrared spectral intensity data, unary linear regression (ULR) and partial least square regression (PLSR) models were established. Compared to the ULR model of mine dust samples, the PLSR model exhibits a better fitting effect, with a determination coefficient (R²) exceeding 0.98 and a root mean square error (RMSE) below 0.35μg. Among these models, the PLSR model applied to metal dust samples exhibited the best fit, with an R² as high as 0.9996 and an RMSE as low as 0.05μg. The results indicate that the DoF method based on IR is in good agreement with the established standard method, validating its reliability for accurately analyzing the crystalline silica content in mine dust.

crystalline silica, infrared spectroscopy, direct-on-filter, unary linear regression, partial least square regression