Davidson School of Chemical Engineering, Purdue University, West Lafayette, Indiana 47906, United States.
J Chem Inf Model. 2020 Apr 27;60(4):2199-2207. doi: 10.1021/acs.jcim.0c00092. Epub 2020 Mar 20.
The gas-phase enthalpy of formation (Δ) plays a fundamental role in predicting reaction thermodynamics and constructing kinetic models. With advances in computational power and method development, chemically accurate quantum chemistry methods that can predict Δ values for small molecules are available; however, large molecules are still out of reach. Increment theories provide a means of extending the prediction capability of high-level methods by decomposing the molecular Δ into the additive contributions from individual atoms, bonds, groups, or components. Here, we introduce a novel component increment theory, topology-automated force-field interaction component increment theory (TCIT), in which all component contributions are derived exclusively from Gaussian-4 (G4) results for algorithmically generated model compounds. In a benchmark evaluation of noncyclic compounds from the Pedley, Naylor, and Kline experimental Δ dataset, TCIT exhibits consistently lower signed and absolute errors compared with the conventional Benson group increment theory (BGIT). These results pave the way for future extensions of TCIT to ring-containing, ionic, and radical species for which experimental data scarcity currently limits the application of BGIT.
气相生成焓（Δ）在预测反应热力学和构建动力学模型方面起着至关重要的作用。随着计算能力和方法的发展，现在已经有了可以预测小分子Δ值的化学精确量子化学方法；然而，对于大分子，这些方法仍然无法企及。增量理论通过将分子的Δ分解为单个原子、键、基团或组件的加和贡献，提供了一种扩展高精度方法预测能力的手段。在这里，我们引入了一种新的组件增量理论，拓扑自动力场相互作用组件增量理论（TCIT），其中所有组件贡献都仅从算法生成的模型化合物的高斯-4（G4）结果中推导出来。在对来自 Pedley、Naylor 和 Kline 实验Δ数据集的非环化合物进行的基准评估中，TCIT 与传统的 Benson 基团增量理论（BGIT）相比，具有更小的符号和绝对误差。这些结果为 TCIT 未来扩展到包含环、离子和自由基的物种铺平了道路，目前 BGIT 的应用受到这些物种实验数据缺乏的限制。
