UC Berkeley

Obtaining predictions for the energies of electronic excited states, especially charge transferstates, is still a challenge in the field of electronic structure theory. Charge transfer states often play an important role in the function of solar cells and organic semiconductors, and having an accurate computational model for their energetics that is not computationally expensive is crucial for the development of new technologies that rely on the existence of these states. In this dissertation, two contributions towards this goal are discussed. First, the formulation of a state-specific perturbative method with a relatively low computational complexity of O(N^5) is described. The approximations made to reduce the scaling from its original complexity of O(N^7) had very little impact on the results of the method and it was shown that it provided accurate energetic predictions for valence and charge transfer excitations of small molecules. Next, this perturbative method was analyzed for its accuracy using a benchmarking set of 105 singlet valence excited states. Through this study it was found that, with regularization, the method can perform even better than many higher-scaling theories and can provide a warning for when a state being studied cannot be mainly described by single excitations. The demonstrated accuracy of the method combined with its relatively low computational cost makes it a promising theory that will be useful in its own right and which can act as a springboard for the development of even more sophisticated excited-state-specific correlation methods.