SOLVENT-DEPENDENT REGRESSION EQUATIONS FOR THE PREDICTION OF RETENTION IN PLANAR CHROMATOGRAPHY

Both first- and second-order regression models are presented that rela te retention, as the Rf of individual solutes, the log k' of individua l solutes, or the average Rf of a mixture of solutes, to the propertie s of the weak solvent in each of a series of 25 binary mobile phases c onsisting of a specified concentration of ethyl acetate as a common st rong solvent The stepwise procedure is used for constructing these mod els, which are for either simulated or experimental separations on sil ica gel, Similar regression models are used to predict separation qual ity, defined by a suitable metric, A comparison of the forward and bac kward stepwise procedures finds that the former is the more reliable m ethod for constructing these models, The solutes are either steroids o r the p-nitrobenzyl esters of dansyl amino acids, and the solvent desc riptors are density, dipole moment, molar volume, polarizability, satu rated surface area, and unsaturated surface area, The quality of regre ssion fits obtained with models using computed dipole moment is compar able to that obtained with models using experimental (literature) dipo le moment. Both nonstandardized and standardized regression models are presented, The relative contribution of each descriptor to the variab ility in retention may be estimated from the latter models, A set of t hree descriptors-dipole moment, polarizability and saturated surface a rea-predicts R(f) for each of the amino acid derivatives at an ethyl a cetate mole fraction of 0.30, A set of two descriptors-dipole moment a nd saturated surface area-predicts log k' for each of these compounds at an ethyl acetate mole fraction of 0.20, Such concordance in descrip tors is not found in models predicting retention of individual steroid s.