Determination of the enantiomeric composition of guest molecules by chemometric analysis of the UV-visible spectra of cyclodextrin guest-host complexes.

Multivariate statistical techniques were applied to the UV spectra of a series of solutions at pH 12 containing a fixed concentration (30 mM) of beta-cyclodextrin (beta-CD) and a fixed concentration (15 mM) of 2-phenylglycine (phi-Gly) with various known enantiomeric compositions. Multivariate correlation of the spectral data for the solutions containing the phi-Gly/beta-CD guest-host complexes with the known enantiomeric composition of the phi-Gly samples was accomplished by partial-least-squares regression. When the multivariate model was used to predict the enantiomeric purity of a test set of samples over the mol fraction range of 0.5-0.9 R-phi-Gly, the average magnitude of the relative errors in the mol fraction determination of enantiomeric composition was 3%. A plot of the enantiomeric composition predicted by the model versus the known enantiomeric composition of the calibration set gave a straight line with a correlation coefficient of 0.955, a slope of 1.05, and an offset of 5.61 x 10-4.