Benchmarking semiempirical quantum chemical methods on liquid water

Area of Science:

• Computational chemistry
• Physical chemistry
• Materials science

Background:

• Renewed interest in semiempirical quantum chemical (SQC) methods for noncovalent interactions.
• SQC methods offer a computationally efficient alternative to ab initio methods.
• Accurate simulation of liquid water properties is crucial for various scientific disciplines.

Purpose of the Study:

• To assess the performance of various SQC methods in describing liquid water properties under ambient conditions.
• To compare SQC methods with experimental data and ab initio density functional theory (DFT) based simulations.
• To identify reliable and computationally efficient SQC methods for simulating liquid water.

Main Methods:

• Molecular dynamics (MD) simulations using conventional SQC methods (AM1, PM6) and DFTB-type methods (DFTB2, GFN-xTB).
• Inclusion of reparameterized SQC methods (AM1-W, PM6-fm, DFTB2-iBi) for water systems.
• Validation against experimental data and BLYP-D3 ab initio MD simulations.

Main Results:

• Original SQC parameter sets poorly describe static and dynamic properties of bulk water, predicting overly fluid water with distorted hydrogen bond kinetics.
• Reparameterized PM6-fm method quantitatively reproduces static and dynamic features of liquid water.
• DFTB2-iBi predicts slightly overstructured water, while AM1-W results in an amorphous ice-like structure.

Conclusions:

• Most SQC methods with original parameters are inadequate for simulating liquid water.
• The reparameterized PM6-fm method serves as an efficient and accurate alternative for large-scale liquid water simulations.
• Specific reparameterizations are crucial for improving the accuracy of SQC methods for condensed-phase systems.