Solubility parameter calculations concerning

The major drawback of CO2, insofar as solvent behavior is concerned, is that it is a poor solvent for many polar and nonpolar compounds, although it can dissolve many small molecules (9). It was once believed that CO2 had solvent properties similar to those of hexane based on solubility parameter calculations but in fact, McFann et al. have shown that the quadrapole moment of CO2 serves to inflate the calculated solubility parameter by 20% (10,11). Consequently, using the solubility parameter as a sole determination of solubility could be misleading. The polarizability/volume has been suggested as a better parameter on which to estimate the solvent power of CO2, but the polarizability/volume of CO2 indicates that it is a weak solvent (11). Other properties of CO2 include low polarizability and electron accepting capacity, since CO2 is a Lewis acid. 

The selection of a solvent for a new separation problem, even today, is a matter of trial and error. The application of theory (2) with the additional application of the solubility parameters (6J-65) makes it possible to estimate the composition of appropriate solvent mixtures for the separation of relatively simple compounds. In order to calculate the necessary solvent strength, however, a set of experimental data concerning the behavior of the sample components, the adsorbent, and the elution strength of the eluents with the specific adsorbent are necessary. Others (J5) recommend a graphical method as a time-saving alternative to bi th calculation and the trial-and-error approach to obtain a first approximation of the eluent composition appropriate for the separation of a givin sample. 

Of course, all the calculations here are approximate. In particular, we have used UNIFAC, which is not meant to be applicable to liquid-liquid equilibrium, and further, —25°C is below the temperature at which the UNIFAC parameters were obtained, so. the results we have obtained are not expected to be accurate. Therefore, all the conclusions should be checked against experimental data. As examples of the uncertainty of such predictions, from experimental data it is found that the amount of ethylene glycol required to result in a mixture freezing point of —25°C is closer to 45 wt % than the - 57 wt % estimated here, and that -pentanol is actually only soluble in water to about 2 j wt %, rather than the complete miscibility found here. This should serve as a warning concerning the use of any completely predictive method, even the UNIFAC model, which is currently the best. 

---