Understanding excited-state properties of molecules in solution using embedded, accurate ab-initio wave functions

对溶剂化现象的充分描述是计算模拟分子溶液性质成功的关键，该项目旨在通过将体系划分为多个子系统并使用从头算波函数方法来理解复杂环境中分子的激发态性质。在用从头算量子化学研究非平凡环境中的分子络合物时，人们同时面临着不同的挑战。一个明显的问题是现有(波函数)方法的急剧缩放，因此在实践中只能处理非常有限的原子。随着系统规模的增大，一个更微妙的问题是由于所涉及的分子的数量而产生的状态和自由度的数量，因此，与所使用的方法无关，在超分子计算中对所有分子的显式处理通常使分析化学激发的亚基变得过于困难。一个有用的方法是通过嵌入方法给出的，该方法将超系统分成更小的子单元，从而留下根据定义分配给特定分子的非常有限的状态，并且标度问题显著减少。冷冻密度嵌入(FDE)已被证明是一种有效的方法来分离由几个分子组成的络合物，所有的子系统都使用从头算方法处理。这种方法的主要特点之一是允许用波函数方法对最常见的分子性质进行系统、一致和严格的推导，避免了对某些性质的特殊处理，也不需要重新参数化半经验参数。在这个项目中，我们将为基态和激发态的近似耦合团簇单双(RI-CC2)FDE建立解析的核梯度。该方法可用于研究具有溶剂化壳层的络合物中激发的来源及其对(激发态)几何结构的影响，同时可以区分显著移位的局部激发和“真实”的超系统效应，如片间电荷转移激发。这个项目的主要应用是在被显式溶剂化分子包围的小分子络合物中，RI-CC2方法对此是适用的。这产生了许多有趣的络合物，从脱氧核糖核酸(DNA)二聚体到苯二聚体，即从氢键到范德华相互作用。特别是，感兴趣的系统是解中的激发剂，如果需要考虑多个自由度，则激发态几何优化明显比单点扫描更有效。

The adequate description of solvation phenomena is the key to successful computational modeling of molecular properties in solution.The project aims at understanding excited-state properties of molecules in complex environments by dividing the system into subsystems and using ab-initio wave-function methods.When investigating a molecular complex in a non-trivial environment with ab-initio quantum chemistry, one faces different challenges at the same time. An obvious problem is the steep scaling of the available (wave-function) methods so that in practice only a very limited number of atoms can be treated. A more subtle problem with increasing system size is the amount of states and degrees of freedom arising due to the number of molecules involved, so that, independently of the method used, an explicit treatment of all molecules in a supermolecular calculation makes the analysis of chemically motivated subunits often too hard.One useful ansatz is given by embedding methods, which divide the supersystem into smaller subunits, so that a very limited number of states are left which are by definition assigned to a certain molecule, and the scaling problem is significantly reduced. Frozen-density embedding (FDE) has proven to be an efficient approach to divide a complex consisting of several molecules, with all subsystems treated using ab-initio methods. One of the main features of this method is to allow for a systematic, consistent and rigorous derivation for the most common molecular properties with wave-function methods, avoiding special-case treatments for some properties, or the need to reparameterize semi-empirical parameters.In this project, we shall develop analytical nuclear gradients for an approximated coupled-cluster singles and doubles (RI-CC2) FDE for both ground and excited states. The new method can be used to investigate the origin of excitations and their influence on the (excited-state) geometry in complexes with solvation shells, while it is possible to discriminate between significantly shifted local excitations and ''true'' super-system effects, such as inter-fragment charge-transfer excitations. The main applications of this project are seen in small molecular complexes surrounded by explicit solvation molecules for which the RI-CC2 method is applicable. This yields a large variety of interesting complexes, ranging from for instance deoxyribonucleic acid (DNA) dimers to the benzene dimer, i.e. from hydrogen bonds to van-der Waals interactions. Particularly, systems of interest are excimers in solution, where an excited-state geometry optimization is significantly more efficient than a single-point scan if more than one degree of freedom needs to be taken into account.