实验测定了链甾醇在正己烷、丁酮、正丁醇、DMSO、乙酸等五种有机溶剂中的溶解度,填补溶解度数据空白。采用理想溶液、Apelblat和COSMO-RS模型对链甾醇的溶解度进行了关联和预测,通过平均相对偏差(ARD)评价各模型的关联或预测效果。理想溶液和Apelblat模型对实验数据关联的ARD均小于11.0%,但都没有预测新系统溶解度的能力。COSMO-RS预测结果误差大。通过改进,预测时输入一个实验数据,结果预测值与实验值吻合得较好,总的ARD从304.30%降为26.98%。上述结果表明只需一个实验数据的COSMO-RS (ref)可用于链甾醇溶解度的预测,为工业分离纯化过程中的溶剂快速筛选提供重要的理论依据。
The solubilities of desmosterol in n-hexane, 2-butanone, 1-butanol, DMSO, acetic acid were determined. The results were correlated and predicted with ideal solution, Apelblat equation and COSMO-RS models, respectively, and average relative deviations (ARD) were used to evaluate these models. The ARDs of experimental data correlated by ideal solution and Apelblat equation are less than 11.0%, however which have no prediction capability for determining the solubility of desmosterol in these systems. COSMO-RS prediction results were extremely dissatisfying, but the predicted values can be in agreement with the experimental values by using one experimental value. The total ARD for desmosterol in above five pure solvents can be reduced to 26.98% for COSMO-RS (ref) from 304.30%. The results show that COSMO-RS (ref) with only one experimental data can be used for the prediction of desmosterol solubility in organic solvents, which is useful in rapid selecting suitable solvents for industrial separation and purification.