Mineral dust serves as a crucial source of ice-nucleating particles (INPs), influencing various cloud properties relevant to climate. The nucleation activity of desert dust is influenced by its mineralogy. There are significant differences in dust mineralogy among various source regions. Moreover, the representation of INPs in cloud and climate models exhibits significant uncertainty. We described the immersion freezing properties of seven dust samples from the Inner Mongolia deserts. In order to elucidate the differences in ice nucleation properties among the samples, we developed a classification method at the single-particle level to describe the mineral compositions of dust samples. Based on singular and stochastic hypothesis, we parameterized the ice-active surface site density and heterogeneous ice nucleation rate coefficient over the temperature range of 245 to 271 K. In addition, according to the water activity theory, the water activity criterion from 0.04 to 0.23 only for heterogeneous ice nucleation rate coefficient. We observed a strong correlation between the ice nucleation performance of droplets and the surface area, mass, and number of particles they contain. Samples with higher content of feldspar typically nucleate at higher temperatures. Conversely, a higher content of clay minerals is associated with a lower ice nucleation activity. All three parameterization schemes effectively describe the ice nucleation activity of the studied samples. Our results affirm the influence of mineral composition on the ice nucleation performance of natural dust, emphasizing the necessity to develop INP parameterizations to provide more accurate for models.