Solheim Harald, Ruud Kenneth, Astrand Per-Olof
Department of Chemistry, University of Tromsø, N-9037 Tromsø, Norway.
J Chem Phys. 2004 Jun 8;120(22):10368-78. doi: 10.1063/1.1738631.
We have implemented analytical second-moment gradients for Hartree-Fock and multiconfigurational self-consistent-field wave functions. The code is used to calculate atomic dipole moments based on the generalized atomic polar tensor (GAPT) formalism [Phys. Rev. Lett. 62, 1469 (1989)], and the proposal of Dinur and Hagler (DH) for the calculation of atomic multipoles [J. Chem. Phys. 91, 2949 (1989)]. Both approaches display smooth basis-set convergence toward a well-defined basis-set limit and give reasonable electron correlation effects on the calculated atomic properties. However, the atomic charges and atomic dipole moments obtained from the GAPT partitioning scheme are unable to provide even qualitatively meaningful molecular quadrupole moments for some molecules, and thus the atomic multipole moments calculated in this scheme cannot be considered well suited for analyzing the electron density in molecules and for calculating intermolecular interaction energies. In contrast, the DH approach gives atomic charges and dipole moments that by definition exactly reproduce the molecular quadrupole moments. The approach of DH is, however, restricted to planar molecules and thus suffers from not being applicable to molecules of arbitrary shape. Both the GAPT and DH approaches give rather poor results for octupole and hexadecapole moments, indicating that at least atomic quadrupole moments are required for an accurate representation of the molecular charge distribution in terms of atomic electric moments.
我们已为哈特里-福克和多组态自洽场波函数实现了分析性二阶矩梯度。该代码用于基于广义原子极化张量(GAPT)形式体系[《物理评论快报》62, 1469 (1989)]以及迪努尔和哈格勒(DH)关于计算原子多极矩的提议[《化学物理杂志》91, 2949 (1989)]来计算原子偶极矩。两种方法在朝着明确的基组极限方向上都呈现出平滑的基组收敛性,并且在计算得到的原子性质上给出了合理的电子相关效应。然而,从GAPT划分方案获得的原子电荷和原子偶极矩,对于某些分子甚至无法提供定性上有意义的分子四极矩,因此该方案中计算得到的原子多极矩不能被认为非常适合用于分析分子中的电子密度以及计算分子间相互作用能。相比之下,DH方法给出的原子电荷和偶极矩,根据定义能精确重现分子四极矩。然而,DH方法仅限于平面分子,因此不适用于任意形状的分子。GAPT和DH方法对于八极矩和十六极矩的结果都相当差,这表明就原子电矩而言,至少需要原子四极矩才能准确表示分子电荷分布。