Robust Atomistic Modeling of Materials, Organometallic, and Biochemical Systems

作者：Sebastian Spicher, Stefan Grimme · 发表于：Angewandte Chemie International Edition · 年份：2020 · DOI：10.1002/anie.202004239 · 被引用次数：544 · 研究领域：Machine Learning in Materials Science、Computational Drug Discovery Methods、Mass Spectrometry Techniques and Applications

Modern chemistry seems to be unlimited in molecular size and elemental composition. Metal-organic frameworks or biological macromolecules involve complex architectures and a large variety of elements. Yet, a general and broadly applicable theoretical method to describe the structures and interactions of molecules beyond the 1000-atom size regime semi-quantitatively is not self-evident. For this purpose, a generic force field named GFN-FF is presented, which is completely newly developed to enable fast structure optimizations and molecular-dynamics simulations for basically any chemical structure consisting of elements up to radon. The freely available computer program requires only starting coordinates and elemental composition as input from which, fully automatically, all potential-energy terms are constructed. GFN-FF outperforms other force fields in terms of generality and accuracy, approaching the performance of much more elaborate quantum-mechanical methods in many cases.