Yeo Christina, Nguyen Minh, Wang Lee-Ping
Department of Physics and Astronomy, University of California, Davis. 1 Shields Avenue, Davis, California 95616, United States.
Department of Chemistry, University of California, Davis. 1 Shields Avenue, Davis, California 95616, United States.
J Phys Chem A. 2022 Oct 27;126(42):7566-7577. doi: 10.1021/acs.jpca.2c03072. Epub 2022 Oct 17.
Many renewable energy technologies, such as hydrogen gas synthesis and carbon dioxide reduction, rely on chemical reactions involving hydride anions (H). When selecting molecules to be used in such applications, an important quantity to consider is the thermodynamic hydricity, which is the free energy required for a species to donate a hydride anion. Theoretical calculations of thermodynamic hydricity depend on several parameters, mainly the density functional, basis set, and solvent model. In order to assess the effects of the above three parameters, we carry out hydricity calculations with different combinations of density functionals, basis sets, and solvent models for a set of organic molecules with known experimental hydricity values. The data are analyzed by comparing the and root-mean-squared error (RMSE) of linear fits with a fixed slope of 1 and using the Akaike Information Criterion to determine statistical significance of the RMSE rank ordering. Based on these results, we quantified the accuracy of theoretical predictions of hydricity and found that the best compromise between accuracy and computational cost was obtained by using the B3LYP-D3 density functional for the geometry optimization and free-energy corrections, either ωB97X-D3 or M06-2X-D3 for single-point energy corrections, combined with a basis set no larger than def-TZVP and the C-PCM ISWIG solvation model. At this level of theory, the RMSEs of hydricity calculations for organic molecules in acetonitrile and dimethyl sulfoxide were found to be <4 and <10 kcal/mol, respectively, for an experimental data set with a dynamic range of 20-150 kcal/mol.
许多可再生能源技术，如氢气合成和二氧化碳还原，都依赖于涉及氢负离子（H⁻）的化学反应。在选择用于此类应用的分子时，一个需要考虑的重要量是热力学氢负离子给予能，即一个物种捐赠一个氢负离子所需的自由能。热力学氢负离子给予能的理论计算取决于几个参数，主要是密度泛函、基组和溶剂模型。为了评估上述三个参数的影响，我们对一组具有已知实验氢负离子给予能值的有机分子，使用不同的密度泛函、基组和溶剂模型组合进行氢负离子给予能计算。通过比较固定斜率为1的线性拟合的决定系数（R²）和均方根误差（RMSE），并使用赤池信息准则来确定RMSE排名顺序的统计显著性，对数据进行分析。基于这些结果，我们量化了氢负离子给予能理论预测的准确性，发现通过使用B3LYP-D3密度泛函进行几何优化和自由能校正，使用ωB97X-D3或M06-2X-D3进行单点能量校正，结合不大于def-TZVP的基组和C-PCM ISWIG溶剂化模型，可以在准确性和计算成本之间取得最佳平衡。在这个理论水平上，对于动态范围为20 - 150 kcal/mol的实验数据集，在乙腈和二甲基亚砜中有机分子氢负离子给予能计算的RMSE分别小于4和小于10 kcal/mol。
