Topology Automated Force-Field Interactions (TAFFI): A Framework for Developing Transferable Force Fields.

Area of Science:

• Computational Chemistry
• Materials Science
• Chemical Physics

Background:

• Recent advances in computational chemistry include quantum mechanics-based parametrizations and machine learning for potential energy surfaces.
• However, transferable force fields are crucial for high-throughput materials and chemical discovery due to cost and consistency requirements.

Purpose of the Study:

• To introduce Topology Automated Force-Field Interactions (TAFFI), a novel framework for developing transferable force fields.
• To demonstrate TAFFI's capability by creating a new fixed-charge force field, TAFFI-gen, for organic functional groups.

Main Methods:

• TAFFI formalizes atom typing to generate systematic training data, ensuring a direct link to force field terms.
• This approach allows for arbitrary extensibility to new chemical spaces while maintaining consistency and transferability.
• A fixed-charge force field, TAFFI-gen, was developed from scratch using the TAFFI framework.

Main Results:

• TAFFI-gen demonstrates coverage of common organic functional groups comparable to established transferable force fields.
• Benchmarking against OPLS and GAFF for 87 organic liquids showed consistent performance of TAFFI-gen in reproducing experimental properties.
• The study validates the TAFFI framework and highlights the representational limits of fixed-charge force fields.

Conclusions:

• The TAFFI framework provides a robust and extensible method for developing transferable force fields.
• TAFFI-gen represents a viable new force field for applications in organic chemistry and materials science.
• The consistent performance across different force fields underscores the importance of further development in accurately modeling diverse chemical systems.