Electronic structure of monosubstituted benzenes and X-ray emission spectroscopy: 3. Fluorobenzene

Using core-hole reference states for calculating X-ray photoelectron and emission spectra

For the calculation of core-ionization energies (IEs), X-ray photoelectron spectra (XPS), and X-ray emission spectra (XES), a commonly applied approach is to use non-Aufbau reference states with a core-hole as either final (IE and XPS) or initial (XES) state. However, such reference states can introduce numerical instabilities in post-HF methods, relating to the denominator of the energy corrections involved. This may become arbitrarily close to zero if a negative virtual orbital is present, e.g. a core-hole, leading to near-singularities. The resulting instabilities lead to severe convergence issues of the calculation schemes and, in addition, can strongly affect both energies and intensities, with oscillator strengths seen to reach values up to 4 × 10⁷. For the K-edge we propose freezing the highest-energy virtual orbitals which contribute to any denominator below a threshold of 0.1 Hartree. Stable and reliable spectra are then produced, with minimal influence due to freezing energetically high-lying virtual orbitals (typically removing <5% of the total number of MOs). The developed protocol is here tested for Møller-Plesset perturbation theory and for the algebraic diagrammatic construction scheme for the polarization propagator, and it is also relevant for coupled cluster theory and other related methods.

Quantum Chemical Calculations of X-ray Emission Spectroscopy

The calculation of X-ray emission spectroscopy with equation of motion coupled cluster theory (EOM-CCSD), time-dependent density functional theory (TDDFT), and resolution of the identity single excitation configuration interaction with second-order perturbation theory (RI-CIS(D)) is studied. These methods can be applied to calculate X-ray emission transitions by using a reference determinant with a core-hole, and they provide a convenient approach to compute the X-ray emission spectroscopy of large systems since all of the required states can be obtained within a single calculation, removing the need to perform a separate calculation for each state. For all of the methods, basis sets with the inclusion of additional basis functions to describe core orbitals are necessary, particularly when studying transitions involving the 1s orbitals of heavier nuclei. EOM-CCSD predicts accurate transition energies when compared with experiment; however, its application to larger systems is restricted by its computational cost and difficulty in converging the CCSD equations for a core-hole reference determinant, which become increasing problematic as the size of the system studied increases. While RI-CIS(D) gives accurate transition energies for small molecules containing first row nuclei, its application to larger systems is limited by the CIS states providing a poor zeroth-order reference for perturbation theory which leads to very large errors in the computed transition energies for some states. TDDFT with standard exchange-correlation functionals predicts transition energies that are much larger than experiment. Optimization of a hybrid and short-range corrected functional to predict the X-ray emission transitions results in much closer agreement with EOM-CCSD. The most accurate exchange-correlation functional identified is a modified B3LYP hybrid functional with 66% Hartree-Fock exchange, denoted B(66)LYP, which predicts X-ray emission spectra for a range of molecules including fluorobenzene, nitrobenzene, acetone, dimethyl sulfoxide, and CF3Cl in good agreement with experiment.