Robust and Efficient Implicit Solvation Model for Fast Semiempirical Methods.

Area of Science:

• Computational Chemistry
• Theoretical Chemistry
• Physical Chemistry

Background:

• Implicit solvation models are crucial for accurately simulating molecular behavior in solution.
• Modern semiempirical quantum mechanics (QM) and general force fields (FF) require efficient and accurate solvation models.
• Existing methods may lack the accuracy or efficiency needed for diverse chemical applications.

Purpose of the Study:

• To present a robust and efficient implicit solvation model for semiempirical QM and FF methods.
• To parameterize the analytical linearized Poisson-Boltzmann (ALPB) model for extended tight binding (xTB), density functional tight binding (DFTB), and GFN-FF.
• To validate the performance of the new solvation model across various chemical systems and properties.

Main Methods:

• Parameterization of the analytical linearized Poisson-Boltzmann (ALPB) model.
• Integration of ALPB with extended tight binding (xTB), density functional tight binding (DFTB), and GFN-FF.
• Testing across diverse applications including conformational energies, transition-metal complexes, and supramolecular associations.

Main Results:

• GFN1-xTB(ALPB) achieves high accuracy for hydration free energies (1.4 kcal/mol MAD) comparable to explicit solvation methods.
• GFN2-xTB(ALPB) accurately predicts octanol-water partition coefficients (0.65 log Kow MAD).
• The model demonstrates consistent performance across over twenty solvents and six semiempirical methods.

Conclusions:

• The ALPB model provides a computationally efficient and accurate approach for implicit solvation in QM/MM.
• The parameterized methods are readily available in xtb and dftb+ programs for broad computational use.
• This work significantly advances the capability of simulating solvated molecules with semiempirical methods.