Data-Derived Conceptual DFT Nucleophilicity Index.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Organic Chemistry

Background:

• Accurate characterization of nucleophilicity is crucial but lacks a universal descriptor.
• Existing conceptual Density Functional Theory (CDFT) indices show limitations in correlating with experimental reactivity.
• Mayr's reactivity scale is a benchmark for nucleophilicity but requires robust theoretical descriptors.

Purpose of the Study:

• To evaluate the performance of established conceptual DFT nucleophilicity indices.
• To develop a novel, data-driven empirical nucleophilicity index (N_emp).
• To investigate the impact of molecular scaffold on reactivity prediction accuracy.

Main Methods:

• Utilized semiempirical methods for analyzing nucleophilicity indices.
• Employed symbolic regression to construct the empirical nucleophilicity index (N_emp).
• Incorporated frontier orbital energies, CDFT descriptors, Fukui functions, charge information, and solvent characteristics.

Main Results:

• Tested DFT nucleophilicity indices showed weak correlations with Mayr's scale outside small sets.
• The global empirical model (N_emp) achieved R² = 0.737, outperforming conventional descriptors.
• Refining models by molecular scaffold significantly improved accuracy to R² = 0.811.

Conclusions:

• Global CDFT nucleophilicity indices have inherent limitations for broad applicability.
• Data-driven strategies and context-aware descriptors show significant potential for reactivity prediction.
• Empirical nucleophilicity indices, especially when scaffold-specific, offer superior predictive power.