Simple Model for Challenging Excited States: Low-Lying, Core-Level and Conical-Intersection Regimes

Time-dependent density functional theory (TDDFT) has been widely used to model electronic excitations but remains unreliable for charge-transfer and doubly excited states, core excitations, and the complex topology near conical intersections (CI). We introduce a simple exciton model to arbitrary open-shell singlet excited (OSE) states by incorporating nonlocal singlet correlation energy, derived from the corresponding triplet excited state, thereby bypassing the direct optimization of singlet states and approximate spin-projection schemes. This model achieves near-quantitative accuracy for low-lying valence and Rydberg excitations and successfully reproduces the absorption spectrum of Chlorophyll a and zinc phthalocyanine. For core excitations, we found an excellent agreement across K and L-edge excitations of first and second group elements and also successfully capture the distinct experimental X-ray absorption fingerprints of imidazole-imidazolium system in water without any empirical shifting. More importantly, this model also captures near-degeneracy behavior in the vicinity of the CI for photoisomerization of cis-trans azobenzene, where standard TDDFT fails. These findings, collectively, position this simple model as a reliable and alternative ΔSCF-like approach for modeling nontrivial excited-state phenomena.