Abstract: To data, various supercritical solubility correlation models have been developed, most of which are semi-empirical models and models based on equations of state. This work focused on density-based semi-empirical models, aiming to evaluate their correlation ability for the solubility of solids in supercritical ethane/carbon dioxide （CO₂）. For this purpose, 17 supercritical binary systems were selected; the solid solutes used included n-alkanes （C24-C36）, phenanthrene, naphthalene, and triphenylmethane, and all relevant solubility data were obtained from the literature. Each system had 1-3 isotherms, totaling 37 isotherms. By calculating the average error of these isotherms, common semi-empirical models such as the Chrastil model, del Valle-Aguilera model, A-L model, S-S model, K-J model, and Tang model were verified. Subsequently, the vitrification model was introduced into the description of critical phenomena, and the form of the VFT equation was used for reference to improve the existing solubility model modification methods, thereby developing a new solubility correlation model. The correlation performance of this new model was also verified. The results showed that the above 6 semi-empirical models all belonged to different modified types of the Chrastil model, each with certain limitations. Their overall average errors ranged from 13.86% to 23.28%, indicating poor correlation accuracy. In contrast, the new model achieved the lowest overall average error of only 6.74%. Especially when dealing with solutes with high solubility and large molecular volume, the new model exhibited excellent correlation performance. The research conclusions provide a reference for the accurate correlation of solid solubility in supercritical fluids and the optimization and improvement of related models.