Assessing the Interplay between Functional-Driven and Density-Driven Errors in DFT Models of Water.

Area of Science:

• Computational chemistry
• Materials science
• Quantum mechanics

Background:

• Density functional theory (DFT) is crucial for simulating molecular interactions.
• Errors in DFT approximations can arise from both the functional form and the electron density.
• Accurate simulation of water requires understanding these error sources.

Purpose of the Study:

• To investigate functional-driven and density-driven errors in DFT approximations for water simulations.
• To assess the impact of these errors on interaction energies and molecular dynamics.
• To guide the development of accurate DFT-based data-driven potentials.

Main Methods:

• Quantified density-driven errors in GGA (BLYP-D3, revPBE-D3) and meta-GGA (SCAN, B97M-rV) functionals.
• Assessed errors on water clusters using the BEGDB dataset.
• Employed absolutely localized molecular orbital energy decomposition analysis (ALMO-EDA) to analyze functional-driven errors.
• Performed molecular dynamics simulations using DFT and density-corrected DFT (DC-DFT) data-driven potentials.

Main Results:

• Identified and quantified density-driven errors in various DFT functionals.
• Demonstrated that functional-driven errors correlate with interaction types.
• Showcased instances where DC-DFT improves accuracy and where reducing density errors can be detrimental due to functional errors.
• Evaluated the impact of minimizing density-driven errors on liquid water simulations.

Conclusions:

• Functional-driven and density-driven errors have distinct impacts on DFT simulations of water.
• Density-corrected DFT models can offer higher accuracy than standard DFT.
• Understanding these errors is vital for developing reliable DFT-based data-driven and machine-learned potentials for condensed-phase systems.