Excitation Energies of Canonical Nucleobases Computed by Multiconfigurational Perturbation Theories

Abstract

In this computational work, we assessed the performance of ab initio multireference (MR) methods for the calculation of vertical excitation energies of five nucleobases: adenine, guanine, cytosine, thymine and uracil. In total, we have studied 38 singlet and 30 triplet excited states. Where possible we used the multireference configuration interaction (MRCI) method as a reference for various flavors of multireference perturbation theory to second order. In particular, we have benchmarked CASPT2, NEVPT2 and XMCQDPT2. For CASPT2, we have analyzed the single-state, multistate (MS) and extended MS variants. In addition, we have assessed the effect of the ionization potential electron affinity (IPEA) shift. For NEVPT2, we have used the partially and the strongly contracted variants. Further, we have tested the commonly used RI-CC2, RI-ADC2 and EOM-CCSD methods. Generally, we observe the following trends for singlet excited states: NEVPT2 is the closest MR method to MRCISD+Q, closely followed by CASPT2 with the default IPEA shift. The same trend is observed for triplet states, although NEVPT2 and CASPT2-IPEA are getting closer. Interestingly, the n, π* singlet excited states were described more accurately than π, π* excited states, while for triplet states the trend is inverted except for NEVPT2. This work is an important benchmark for future photochemical investigations.

Initial excited‐state relaxation of locked retinal protonated schiff base chromophore. An insight from coupled cluster and multireference perturbation theory calculations

“…As shown in Figure the performance of XMCQDPT2 and MS‐CASPT2/0‐IPEA is close to each other with deviations in absorption energies not exceeding 0.05 eV. This is consistent with what was found by Wiebeler et al for excitation energies of canonical nucleobases . It is worth mentioning that MS‐CASPT2/0‐IPEA method provides always the lowest transition energies among multistate MRPT2 methods studied here.…”

Ultrafast and radiationless electronic excited state decay of uracil and thymine cations: computing the effects of dynamic electron correlation

“…XMS-CASPT2 is shown to be blue-shifted with respect to MS and single-state formulations, a feature previously reported for the singlet manifold. 55,75 In contrast, CASSCF places the first three vertical and first adiabatic ionisation potentials at 8.58, 9.36, 9.47 and 8.18 eV, respectively (Table 1), which deviate from the experimental evidence by ∼1 eV. Nevertheless, CASSCF still provides acceptable energies when considering solely the cationic manifold and the differences among the diverse cationic states.…”

An algorithm to correct for the CASSCF active space in multiscale QM/MM calculations based on geometry ensembles

“…The intramolecular and van der Waals parameters for all nucleobases were taken from the general amber force field (GAFF) for organic molecules. 48 The geometries for all nucleobases were retrieved from Thiel's benchmark set 35 except for guanine, whose geometry was obtained from the work of Wiebeler et al 34 These geometries were reoptimized at the MP2/6-31G* level of theory using the Gaussian16 49 software, and restrained electrostatic potential (RESP)…”

“…54 For all nucleobases, no symmetry was used in the CASSCF calculations to have a better comparison with the sampled geometries, as it is expected that the C s symmetry breaks down during the MD sampling. For most of the nucleobases, the same active spaces and number of roots as used in a previous study 34 were considered for the SA-CASSCF calculations. Specifically, for uracil we computed the first 10 roots using a (14,10) active space which included all π electrons plus the four nonbonding electrons of the oxygen atoms; for cytosine we computed the first 8 roots with a (14,10) active space.…”

“…Nucleobases represent a suitable test case set as their photophysical and photochemical properties have been thoroughly studied both in vacuum and with implicit solvation models within several different MCSCF approaches. [32][33][34][35] Thus, our goal is not to reproduce accurate vertical excitation energies, but rather to show the effectiveness of our approach in recovering the appropriate CASSCF active space of an ensemble of geometries. Moreover, we seek to stress out the importance of describing in a proper manner the wavefunction for each of these geometries by comparing the uncorrected CASSCF calculations with their corrected counterparts.…”