Quantitative structure/eluent-retention relationships in reversed-phase high-performance liquid chromatography based...

Area of Science:

• Analytical Chemistry
• Chromatography

Background:

• Predictive modeling in reversed-phase high-performance liquid chromatography (RP-HPLC) typically focuses on single binary eluents.
• Existing models often use volume fraction of organic modifiers, limiting their applicability.

Purpose of the Study:

• To develop a unified predictive model for solute retention in RP-HPLC applicable to both acetonitrile-water and methanol-water mobile phases simultaneously.
• To assess the model's predictive performance and its ability to generalize to unseen eluents and solutes.

Main Methods:

• Utilized a multi-layer artificial neural network (ANN) incorporating Kamlet-Taft solvatochromic descriptors for both eluents and analytes.
• Employed the Kennard-Stones algorithm for selecting training data from a set of 31 molecules across five different columns.
• Evaluated predictive performance using independent test sets and cross-eluent prediction scenarios.

Main Results:

• The ANN-based model accurately predicted solute retention across the explored composition range (20-70%) for both acetonitrile- and methanol-water mobile phases.
• The model demonstrated a promising capability for predicting retention of external solutes in mobile phases not used during training.
• High accuracy was achieved for predicting the behavior of external solutes in novel eluent compositions.

Conclusions:

• The proposed ANN model offers a robust and versatile approach for predicting solute retention in RP-HPLC, overcoming limitations of single-eluent models.
• This method enhances the predictability of chromatographic separations by considering molecular and eluent properties comprehensively.
• The model shows significant potential for optimizing RP-HPLC method development and predicting solute behavior in diverse mobile phase conditions.