Modeling the Carbothermal Chlorination Mechanism of Titanium Dioxide in Molten Salt Using a Deep Neural Network Potential

The molten salt chlorination method is one of the two main methods for producing titanium tetrachloride, an important intermediate product in the titanium industry. To effectively improve chlorination efficiency and reduce unnecessary waste salt generation, it is necessary to understand the mechanism of the molten salt chlorination reaction, and consequently this paper conducted studies on the carbon chlorination reaction mechanism in molten salts by combining ab initio molecular dynamics (AIMD) and deep potential molecular dynamics (DeePMD) methods. The use of DeePMD allowed for simulations on a larger spatial and longer time scale, overcoming the limitations of AIMD in fully observing complex reaction processes. The results comprehensively revealed the mechanism of titanium dioxide transforming into titanium tetrachloride. In addition, the presence form and conversion pathways of chlorine in the system were elucidated, and it was observed that chloride ions derived from NaCl can chlorinate titanium dioxide to yield titanium tetrachloride, which was validated through experimental studies. Self-diffusion coefficients of chloride ions in pure NaCl which were acquired by DeePMD showed good agreement with the experimental data.