Accurate thermochemistry for molecules containing heavy p-block elements using local explicitly correlated ab initio methods

在过去的几年中，已经证明精确的从头算方法可以预测不确定度为1-4 kJ/mol的热化学数据，这通常比实验值更准确。然而，到目前为止，这仅限于只包含第一排和第二排原子的非常小的分子。本建议的目的是将这种方法扩展到剩余的较重的p块元素和较大的分子。这项工作的新颖之处有四个方面：(i)将利用新开发的显式相关方法（MP2-F12和CCSD(T)-F12）来最小化由于1粒子基集的缓慢收敛而可能产生的实质性误差。（ii）在这些系统中至关重要的标量和自旋轨道相对论效应，将使用仔细标定的相对论有效核心势（ECPs）加以恢复。作为工作的一部分，将开发F12方法与ecp相结合的新的系统收敛基集。局部相关方法将用于较大的系统。（iv）将对MP2-F12和CCSD(T)-F12方法实施分析能量梯度。

Over the last years it has been demonstrated that accurate ab initio methods can predict thermochemical data with an uncertainty of 1-4 kJ/mol, which is often more accurate than experimental values. However, so far this is restricted to very small molecules containing first and second row atoms only. The aim of the present proposal is to extend this methodology to the remainder of the heavier p-block elements and larger molecules. The novel aspects of this work are fourfold: (i) newly developed explicitly correlated methods (MP2-F12 and CCSD(T)-F12) will be utilitized to minimize the substantial errors that can arise from the slow convergence of the 1-particle basis sets. (ii) Scalar and spin-orbit relativistic effects, which are of crucial importance in these systems, will be recovered using carefully benchmarked relativistic effective core potentials (ECPs). New systematically convergent basis sets for F12 methods in conjunction with ECPs will be developed as part of the work. (iii) Local correlation methods will be used for larger systems. (iv) Analytical energy gradients will be implemented for MP2-F12 and CCSD(T)-F12 methods.