Abstract: A novel method to predict retention parameters of liquid chromatography is developed. By determining their lowest energy conformation, molecules are aligned according to one specific rule and the scope of the molecular field is defined around the aligned molecules. A positively charged sp3 carbon probe atom is used to calculate the characters of the molecular field such as the electrostatic and steric interaction energies. Then partial least square (PLS) method is exploited to analyze relationships between the a, c parameters of compounds and characters of the molecular field and finally comparative molecular field analysis (CoMFA) models are built, which are exploited to study three-dimensional quantitative structure-retention relationship (QSRR) between the a, c parameters of compounds in methanol/water on a reversed-phase C18 column and the characters of the molecular field. The models are applied to four different forms of arenes: alkylation-substituted benzenes, multi-ring arenes, chlorobenzenes and dinitro-substituted arenes. The results of PLS analyses show that three-dimensional models are obtained from the first three forms. The cross-validated q2 of the first three models are all above 0.5 and the non-cross-validated r2 of the first two forms are even above 0.995, so the predictive abilities of these models are considerable. The model of dinitro-substituted arenes is not so reasonable. The possible reason is that the nitro-substitutions are on different benzene rings. The novel method explored can be utilized to predict the retention parameters of the compounds, which are unknown but have the same kind of structure with the known compounds, and also provide information about the class of the compounds. The results of this project can be applied to direct the further study of the relationship between three-dimensional quantitative structure and retention parameters of liquid chromatography.

Key words: comparative molecular field analysis (CoMFA), compounds , quantitative structure-retention relationship (QSRR), liquid chromatography