Reparameterization of an Accurate, Few-Parameter Implicit Solvation Model for Quantum Chemistry: Composite Method for Implicit Representation of Solvent, CMIRS v. 1.1.
Reparameterization of an Accurate, Few-Parameter Implicit Solvation Model for Quantum Chemistry: Composite Method for Implicit Representation of Solvent, CMIRS v. 1.1.
复制标题
量子化学的精确、少参数隐式溶剂化模型的重新参数化：溶剂隐式表示的复合方法，CMIRS v. 1.1。
DOI：
10.1021/acs.jctc.6b00644
复制
发表时间：
2016
影响因子：
5.5
通讯作者：
J. Herbert
中科院分区：
文献类型：
--
作者：
Zhi;J. Herbert
CMIRS（composite method for implicit representation of solvent）是一种相对较新的隐式溶剂化模型，它将溶质-溶剂分散、泡利排斥和氢键作用加入到静电学的连续处理中。在色散项的原始实现中的一个小误差，但在某些情况下可以修改色散能量高达8 kcal/mol，需要重新调整模型中的参数，我们在这里这样做。我们将修改后的实现和参数集称为CMIRS v.1.1。虽然色散能以非平凡的方式变化，但在新的实施中吸引力色散项的增加在很大程度上被拟合过程中泡利排斥的增加所抵消，使得对于分子和离子的实验溶剂化自由能的大型数据库，相对于模型的v.1.0，总体统计误差几乎不变。总的来说，当溶剂为环己烷或苯时，我们获得的平均无符号误差<0.7 kcal/mol，水<1.5 kcal/mol，二甲亚砜和乙腈<2.8 kcal/mol，尽管每个溶剂使用的经验参数不超过5个。对于重要但困难的水中离子溶质的情况，平均无符号误差<2.9 kcal/mol。
CMIRS (composite method for implicit representation of solvent) is a relatively new implicit solvation model that adds terms representing solute-solvent dispersion, Pauli repulsion, and hydrogen bonding to a continuum treatment of electrostatics. A small error in the original implementation of the dispersion term, but one that can modify dispersion energies by up to 8 kcal/mol in some cases, necessitates refitting the parameters in the model, which we do here. We refer to the modified implementation and parameter set as CMIRS v. 1.1. While the dispersion energies change in nontrivial ways, an increase in the attractive dispersion term in the new implementation is largely offset by an increase in the Pauli repulsion during the fitting process, such that overall statistical errors are virtually unchanged with respect to v. 1.0 of the model, for a large database of experimental solvation free energies for molecules and ions. Overall, we obtain mean unsigned errors of <0.7 kcal/mol when the solvent is cyclohexane or benzene, <1.5 kcal/mol for water, and <2.8 kcal/mol for dimethyl sulfoxide and acetonitrile, despite using no more than five empirical parameters per solvent. For the important but difficult case of ionic solutes in water, mean unsigned errors are <2.9 kcal/mol.