Towards reliable modeling of excited states of uranium compunds Theoretical prediction of electronic spectra of uranium-containing complexes

Theoretical prediction of electronic spectra of uranium-containing complexes remains quite a challenge for quantum chemistry since it requires an accurate treatment of both correlation and relativistic effects at the same time. While reliable electronic excitation energies can be obtained from correlated wave function approaches that take relativistic effects into account, they are, however, limited to small system sizes. This includes small model complexes but prevents the treatment of full-fledged ligands. Large complexes that are of interest in nuclear waste reprocessing or in catalysis can only be investigated within a DFT framework based on the Kohn–Sham formalism. Since in Kohn–Sham DFT the exact exchange–correlation functional remains unknown, the accuracy of DFT calculations strongly depends on the applied approximations thereof and on which class of molecules or properties the density functional has been designed for. However, currently available approximate exchange–correlation functionals do not contain heavy elements like actinides in their parametrization sets, which may result in severe drawbacks when actinide-containing compounds are described by means of DFT. Hence, a thorough knowledge about the performance of DFT in studying such complexes is desirable, in particular, since actinides posses active and chemically different f -orbitals compared to lighter elements for which common density functionals are optimized. Although DFT or its time-dependent extension TD-DFT represent the methods of choice when large molecules and their properties are to be investigated, a systematic study of the excited-states of uranium compounds with different approximate exchange–correlation functionals has not yet been performed. This raises the question about the reliability and accuracy of the excitation energies obtained by TD-DFT. In order to gain first insights into the performance of TD-DFT in a systematic way, we gradually defined a suitable benchmark set of small and medium-sized model complexes which represent essential building blocks in uranium-containing complexes and possess interesting electronic features. The excitation spectra of these benchmark molecules have been tested against highly accurate ab initio data obtained by CASPT2, CR-EOMCCSD(T) or IHFSCC-SD. In a preliminary study (chapter 9), we chose three small-sized and isoelectronic uranium containing model complexes as the starting benchmark set: UO2+ 2 , NUO + and NUN. This choice was deliberately made, since all complexes represent the building blocks found in larger uranium-containing complexes which are, e.g., important in the processing of nuclear waste. The accuracy of different classes of exchange–correlation energy functionals, i.e., local (LDA), GGA (PBE and BLYP), meta-GGA (M06-L), hybrid (B3LYP and PBE0), meta-hybrid (M06