Solvent Boundary Potentials for Hybrid QM/MM Computations Using Classical Drude Oscillators: A Fully Polarizable...

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Quantum Mechanics/Molecular Mechanics (QM/MM)

Background:

• Accurate QM/MM treatments require accounting for molecular mechanical (MM) polarization and long-range electrostatic interactions.
• Existing methods often struggle to efficiently and accurately incorporate these effects, particularly in large systems.

Purpose of the Study:

• To develop and validate a novel QM/MM approach that integrates the Drude oscillator (DO) model for MM polarizability with generalized solvent boundary potential (GSBP) and solvated macromolecule boundary potential (SMBP).
• To enable accurate treatment of electronic polarization and long-range electrostatics in complex molecular systems.

Main Methods:

• Implementation of fully polarizable three-layer QM/MM-DO/BP methods combining DO model with GSBP/SMBP and a polarizable dielectric continuum (PDC).
• Development of efficient schemes for simultaneous polarization convergence across layers, including dynamic treatment of MM polarization for MD simulations (GSBP) and a dual self-consistent field approach for geometry optimizations (SMBP).
• Evaluation using a glycine molecule in a water ball system to assess gradient accuracy and computational efficiency.

Main Results:

• The QM/MM-DO/BP methods accurately reproduce gradients for inner-region atoms and Drude oscillators.
• Inclusion of Drude oscillator effects in all boundary potential terms is crucial for accuracy.
• The polarizable dielectric continuum (PDC) with high dielectric constants does not induce polarization catastrophes in DO models.
• Significant computational speedups were observed: ~40% reduction per step for geometry optimizations (SMBP) and increasing advantages with system size/steps for MD (GSBP).

Conclusions:

• The developed QM/MM-DO/BP methods offer a robust and efficient framework for accurate polarizable molecular simulations.
• These advancements are particularly beneficial for large-scale molecular dynamics and geometry optimization studies.
• The approach provides a significant improvement over standard QM/MM-DO calculations without boundary potentials.