Chanda Shamik, Saha Subhasish, Sen Sangita
Department of Chemical Sciences, Indian Institute of Science Education and Research (IISER) Kolkata, Nadia, Mohanpur 741246, WB, India.
MLR Institute of Technology, Dundigal, Hyderabad 500043, Telangana, India.
J Chem Phys. 2025 Jan 14;162(2). doi: 10.1063/5.0238105.
In this paper, we demonstrate the performance of several density-based methods in predicting the inversion of S1 and T1 states of a few N-heterocyclic triangulene based fused ring molecules (popularly known as INVEST molecules) with an eye to identify a well performing but cost-effective preliminary screening method. Both conventional linear-response time-dependent density functional theory (LR-TDDFT) and ΔSCF methods (namely maximum overlap method, square-gradient minimization method, and restricted open-shell Kohn-Sham) are considered for excited state computations using exchange-correlation (XC) functionals from different rungs of Jacob's ladder. A well-justified systematism is observed in the performance of the functionals when compared against fully internally contracted multireference configuration interaction singles and doubles and/or equation of motion coupled-cluster singles and doubles (EOM-CCSD), with the most important feature being the capture of spin-polarization in the presence of correlation. A set of functionals with the least mean absolute error is proposed for both the approaches, LR-TDDFT and ΔSCF, which can be more cost-effective alternatives for computations on synthesizable larger derivatives of the templates studied here. We have based our findings on extensive studies of three cyclazine-based molecular templates, with additional studies on a set of six related templates. Previous benchmark studies for subsets of the functionals were conducted against the domain-based local pair natural orbital-similarity transformed EOM-CCSD (STEOM-CCSD), which resulted in an inadequate evaluation due to deficiencies in the benchmark theory. The role of exact-exchange, spin-contamination, and spin-polarization in the context of DFT comes to the forefront in our studies and supports the numerical evaluation of XC functionals for these applications. Suitable connections are drawn to two and three state exciton models, which identify the minimal physics governing the interactions in these molecules.
在本文中，我们展示了几种基于密度的方法在预测一些基于N-杂环三角烯的稠环分子（通常称为INVEST分子）的S1和T1态反转方面的性能，旨在确定一种性能良好但成本效益高的初步筛选方法。使用来自雅各布天梯不同梯级的交换相关（XC）泛函，考虑了传统的线性响应含时密度泛函理论（LR-TDDFT）和ΔSCF方法（即最大重叠方法、平方梯度最小化方法和受限开壳Kohn-Sham方法）进行激发态计算。与完全内部收缩的多参考组态相互作用单双激发和/或运动方程耦合簇单双激发（EOM-CCSD）相比，泛函的性能呈现出合理的系统性，最重要的特征是在存在相关性的情况下捕获自旋极化。针对LR-TDDFT和ΔSCF这两种方法，都提出了一组平均绝对误差最小的泛函，它们对于此处研究的可合成更大模板衍生物的计算可能是更具成本效益的替代方案。我们的研究结果基于对三种基于环嗪的分子模板的广泛研究，并对一组六个相关模板进行了额外研究。之前针对这些泛函子集的基准研究是针对基于域的局部对自然轨道相似性变换的EOM-CCSD（STEOM-CCSD）进行的，由于基准理论的缺陷，导致评估不充分。在我们的研究中，精确交换、自旋污染和自旋极化在DFT背景下的作用凸显出来，并支持对这些应用中XC泛函的数值评估。与二态和三态激子模型建立了适当的联系，确定了控制这些分子中相互作用的最小物理机制。
