dc.description.abstract | The PSRK EOS and PR+COSMOSAC (PRCS) EOS were used to predict the solubility of drug-like and dye-like molecules in supercritical carbon dioxide (SCCO2). The required input of critical properties of the studied solute for group-contribution model PSRK were obtained by either NRR model (based on the group contribution concept) or PRCS EOS (based on the first-principles solvation model). In PRCS EOS, the molecular structure of the studied solute was the only input for QM/COSMO solvation calculation to generate the key information for phase equilibrium calculations. In the solid-fluid equilibrium calculation, the solid phase fugacity of the solute can be obtained from its melting point and heat of fusion, but most of the newly synthesized solutes are lack of experimental melting properties. Besides, the solid phase fugacity can also be determined with the solid-vapor equilibrium line as the reference state. In order to overcome the lack of required experimental melting properties, the accuracy of using sublimation pressure estimated from the Lee-Kesler (LK) equation in predicting solid solute solubility was investigated. The required critical properties for the LK equation were estimated by the aforementioned NRR or PRCS methods. Thus, the proposed approach requires the only input of molecular structure of the studied solute.
A total of 99 drug-like and 58 dye-like solutes were investigated in this study. When using PRCS for phase equilibrium calculation, if it is combined with the LK equation of regression sublimation pressure to optimize the parameters, the predicted result of solubility will be systematically overestimated, so the calculation is changed to the LK equation with regression solubility to optimize the parameters. Because the group contribution method faces the issue of lacking interaction parameters and functional group definitions, PSRK can only predict the solubility of 57 drug-like solutes, the average logarithmic deviation in solubility (ALD-x) of PSRK using critical properties from NRR and PRCS are 0.914 and 0.842, respectively. The ALD-x of PRCS using critical properties from NRR and PRCS are 1.597 and 0.911, respectively, for the same 57 solutes. PR+COSMOSAC requires the molecular structure as the only input, and thus can be applied to describe all solid solutes. The ALD-x for predicting solubility of 99 drug-like and 58 dye-like solutes is 0.974 and 1.138, respectively. | en_US |