How well can B3LYP heats of formation be improvedby dispersion correction models?
How well can B3LYP heats of formation be improvedby dispersion correction models?
复制标题
通过色散校正模型可以在多大程度上改善 B3LYP 的形成热？
DOI：
10.1007/s00214-015-1801-9
复制
发表时间：
2016
影响因子：
1.7
通讯作者：
Xu Xin
中科院分区：
文献类型：
--
作者：
Zhou Yuwei;Wu Jianming;Xu Xin
本工作编制了六个不同化学组成烃类的生成热数据集，包括21个正构烷烃到n-C_（32）H_（66）的生成热数据集，n-C_（7 H_（16）及其支化异构体的生成热数据集，36个多环饱和烃的生成热数据集，41个C_6H_（18）环、烯烃、炔和累积多烯的生成热数据集，组5用于41种基于苯的化合物，组6用于66种自由基。我们进行了密集的高水平G4计算，当实验数据不可用或不可靠时，提供参考数据。分散校正的方法B3 LYP-D2、B3 LYP-D3和B3 LYP-D3 BJ如何改进广泛使用的密度泛函方法B3 LYP，一直受到关注。结果发现，虽然色散校正的方法可以消除原来的B3 LYP错误的显着量，他们是无法治愈所有的B3 LYP缺陷，除了缺乏货车的范德华相互作用。虽然D3仍然低估了分子的稳定性，如B3 LYP，D3 BJ高估了它们的稳定性。更先进的D3和D3 BJ模型不一定优于更简单的D2模型。除了PSH集，B3 LYP-D3 BJ总是给出比B3 LYP-D2更大的误差，显示出进一步改进的空间。
In the present work, we have compiled six datasets for heats of formation (HOFs) of hydrocarbons of different chemistries, involving Set 1 for 21 n-alkanes up to n-C32H66, Set 2 for n-C7H16and its branched isomers, Set 3 for 36 polycyclic saturated hydrocarbons, Set 4 for 41 C6H8isomers of rings, alkenes, alkynes and cumulenes, Set 5 for 41 benzene-based compounds and Set 6 for 66 radicals. We have performed intensive high-level G4 calculations, which provide the reference data when the experimental data are not available or not reliable. Attention has been paid on how the dispersion-corrected methods, B3LYP-D2, B3LYP-D3 and B3LYP-D3BJ, can improve over the widely used density functional, B3LYP. It was found that, although the dispersion-corrected methods can eliminate the original B3LYP errors to a significant amount, they are unable to cure all B3LYP deficiencies besides the lack of van der Waals interactions. While D3 still underestimates the molecular stabilities as does B3LYP, D3BJ overestimates their stabilities. The more advanced D3 and D3BJ models do not necessarily outperform the simpler D2 model. Except for the PSH set, B3LYP-D3BJ always gives larger errors than does B3LYP-D2 for the other five datasets, showing room for further improvement.