Department of Chemistry and Biochemistry, University of Mississippi, University, Mississippi 38677-1848, USA.
J Chem Phys. 2012 Jan 7;136(1):014103. doi: 10.1063/1.3671950.
The basis set dependence of higher-order correlation effects on π-type interaction energies was examined by scanning the potential energy surfaces of five dimer systems. The dimers of acetylene (H-C≡C-H), diacetylene (H-C≡C-C≡C-H), cyanogen (N≡C-C≡N), diphosphorous (P≡P), and 1,4-diphosphabutadiyne (P≡C-C≡P) were studied in three different configurations: cross, parallel-displaced, and t-shaped. More than 800 potential energy curves (PECs) were generated by computing the interaction energies for all 15 dimer configurations over a range of intermolecular distances with the MP2, coupled-cluster single double (CCSD), and coupled-cluster single double triple (CCSD(T)) methods in conjunction with 21 basis sets ranging from a small 6-31G*(0.25) split-valence basis set to a large aug-cc-pVQZ correlation consistent basis set. Standard extrapolation techniques were also used to construct MP2, CCSD, and CCSD(T) complete basis set (CBS) limit PECs as well as CBS limit higher-order correlation corrections based on the differences between CCSD(T) and MP2 interaction energies, denoted δ(MP2)(CCSD(T)), and the corresponding differences between CCSD(T) and CCSD interactions energies, denoted δ(CCSD)(CCSD(T)). Double-ζ basis sets struggled to reproduce the former but provided quite reasonable descriptions of the latter as long as diffuse functions were included. The aug-cc-pVDZ basis deviated from the δ(CCSD)(CCSD(T)) CBS limit by only 0.06 kcal mol(-1) on average and never by more than 0.24 kcal mol(-1), whereas the corresponding deviations were approximately twice that for the δ(MP2)(CCSD(T)) term. While triple-ζ basis sets typically improved results, only aug-cc-pVTZ provided appreciable improvement over utilizing the aug-cc-pVDZ basis set to compute δ(CCSD)(CCSD(T)). Counterpoise (CP) corrections were also applied to all double- and triple-ζ basis sets, but they rarely yielded a better description of these higher-order correlation effects. CP corrections only consistently improved results when the aug-cc-pVDZ basis set was used to compute δ(MP2)(CCSD(T)), yielding mean and maximum absolute deviations from the CBS values of 0.10 and 0.39 kcal mol(-1), respectively, for all five dimer systems.
通过扫描五个二聚体系统的势能表面，研究了更高阶相关效应对于π型相互作用能的基组依赖性。研究了乙炔（H-C≡C-H）、二乙炔（H-C≡C-C≡C-H）、氰（N≡C-C≡N）、膦（P≡P）和 1,4-二磷杂丁二炔（P≡C-C≡P）的三种不同构型：交叉、平行位移和 T 形的二聚体。使用 MP2、耦合簇单双（CCSD）和耦合簇单双三（CCSD(T)）方法与 21 个基组相结合，在 0.25 分裂价基组到大型 aug-cc-pVQZ 相关一致基组的范围内，计算了所有 15 种二聚体构型在各种分子间距离下的相互作用能，生成了超过 800 条势能曲线（PECs）。还使用标准外推技术构建了 MP2、CCSD 和 CCSD(T)完全基组（CBS）极限 PEC 以及基于 CCSD(T)和 MP2 相互作用能之间的差异的 CBS 极限高阶相关校正，记为 δ(MP2)(CCSD(T))，以及 CCSD(T)和 CCSD 相互作用能之间的相应差异，记为 δ(CCSD)(CCSD(T))。双 ζ 基组很难再现前者，但只要包含弥散函数，就可以对后者进行相当合理的描述。aug-cc-pVDZ 基组平均仅偏离 δ(CCSD)(CCSD(T))CBS 极限 0.06 kcal mol(-1)，且从未超过 0.24 kcal mol(-1)，而 δ(MP2)(CCSD(T))项的相应偏差约为两倍。虽然三 ζ 基组通常会改善结果，但只有 aug-cc-pVTZ 能够显著改进使用 aug-cc-pVDZ 基组计算 δ(CCSD)(CCSD(T))的结果。还对所有双 ζ 和三 ζ 基组应用了平衡核赝势（CP）校正，但它们很少能更好地描述这些高阶相关效应。只有在使用 aug-cc-pVDZ 基组计算 δ(MP2)(CCSD(T))时，CP 校正才能始终改善结果，对于所有五个二聚体系统，其平均和最大绝对偏差分别为 CBS 值的 0.10 和 0.39 kcal mol(-1)。
