Predicting the Aqueous Solubility of Neutral Organic Compounds by Reversed-Phase Liquid Chromatography

Area of Science:

• Analytical Chemistry
• Physical Chemistry
• Computational Chemistry

Background:

• Accurate water solubility prediction is vital for chemical screening.
• Traditional methods like the General Solubility Equation (GSE) have limitations, especially for weak electrolytes, requiring melting point and partition coefficient data.
• There is a need for efficient, high-throughput methods for solubility estimation.

Purpose of the Study:

• To develop and evaluate two reversed-phase liquid chromatography (RPLC) correlation strategies for predicting water solubility.
• To compare the predictive accuracy of RPLC-based methods against traditional approaches like GSE.
• To offer descriptor-free and combined retention/melting point RPLC models for diverse chemical screening needs.

Main Methods:

• Exploration of isocratic RPLC models using retention data as the sole input for neutral organic compounds.
• Development of RPLC models incorporating both retention and melting point data.
• Evaluation across 49 binary solvent systems and three stationary phases, focusing on organic volume fractions from 50%-70% v/v.

Main Results:

• Descriptor-free RPLC models achieved a predictive accuracy of 0.307 log units, outperforming GSE for non-electrolytes and weak electrolytes.
• RPLC models combining retention and melting point data showed improved accuracy, particularly at higher organic volume fractions.
• A specific model (70% acetonitrile on XTerra MS C18) yielded a coefficient of determination of 0.962 and an average absolute error of 0.222 log units.

Conclusions:

• Two complementary RPLC strategies provide powerful, high-throughput alternatives for rapid water solubility screening.
• The descriptor-free RPLC approach is suitable for compounds lacking structural information.
• The combined RPLC retention and melting point model offers high precision, minimizing prediction errors below 0.3 log units.