Predicting p Ka values from EEM atomic charges.

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Physical Organic Chemistry

Background:

• The acid dissociation constant (p Ka) is a critical molecular property.
• Accurate and rapid p Ka prediction methods are highly sought after in chemistry.

Purpose of the Study:

• To evaluate the p Ka prediction performance of Quantitative Structure-Property Relationship (QSPR) models utilizing empirical atomic charges derived from the Electronegativity Equalization Method (EEM).
• To compare the efficiency and accuracy of EEM-based QSPR models against those based on quantum mechanical (QM) charge calculations.

Main Methods:

• Collected 18 EEM parameter sets and 8 quantum mechanical (QM) charge calculation schemes.
• Prepared a training set of 74 substituted phenols and their dissociated forms.
• Calculated EEM and QM charges for all molecules.
• Developed QSPR models using non-dissociated molecule charges (3 descriptors) and combined charges (5 descriptors).
• Evaluated model performance using quality criteria and cross-validation.

Main Results:

• QSPR models using EEM charges demonstrated good p Ka prediction capabilities, with 63% achieving R2 > 0.9 and a best R2 of 0.924.
• QM QSPR models showed slightly higher accuracy, but the difference was not statistically significant.
• EEM charge calculation is markedly faster than QM charge calculation.
• EEM QSPR models exhibited greater robustness and less sensitivity to the charge calculation approach compared to QM QSPR models.
• Applicability was confirmed for carboxylic acids.

Conclusions:

• EEM-based QSPR models provide a computationally efficient and accurate alternative for p Ka prediction.
• These models are suitable for applications like virtual screening due to their speed and reliability.
• The findings support the use of EEM for predicting p Ka across different chemical classes.