Multiple-column RP-HPLC retention modelling based on solvatochromic or theoretical solute descriptors

Area of Science:

• Analytical Chemistry
• Chromatography
• Computational Chemistry

Background:

• Accurate prediction of retention times in reversed-phase high-performance liquid chromatography (RP-HPLC) is crucial for method development and optimization.
• Existing models often struggle with multi-column applicability and varying mobile phase compositions.
• Artificial neural networks (ANNs) offer a powerful framework for complex non-linear modeling in chromatography.

Purpose of the Study:

• To rigorously evaluate the reliability of a previously proposed ANN-based approach for multi-column RP-HPLC retention modeling.
• To assess the model's performance using independent experimental data from Reta et al. across diverse stationary phases.
• To compare the predictive power of solvatochromic versus theoretical molecular descriptors within the ANN framework.

Main Methods:

• Utilized an ANN regression model incorporating five molecular descriptors, mobile phase composition, and a novel column descriptor.
• The column descriptor was defined as the extrapolated average retention in pure water.
• Validated the model using retention data for 17 aromatic compounds on eight different columns with water-methanol mobile phases at varying compositions (45-60% v/v methanol).

Main Results:

• The ANN model demonstrated high accuracy in predicting retention times across multiple columns and mobile phase compositions.
• The approach proved reliable when tested against independent experimental data.
• Both solvatochromic and theoretical descriptors showed good explanatory capability, with theoretical descriptors offering an alternative for model development.

Conclusions:

• The ANN-based retention modeling approach is robust and applicable to multi-column RP-HPLC under isocratic conditions.
• The model's accuracy extends to various stationary phases and mobile phase compositions.
• This method provides a valuable tool for predicting and optimizing chromatographic separations.