Publication: ------------ Input data, extracted results and evaluation scripts for Phys. Rev. B 104, 165135 Charge disproportionation and Hund's insulating behavior in a five-orbital Hubbard model applicable to d^4 perovskites Maximilian E. Merkel, Claude Ederer https://doi.org/10.1103/PhysRevB.104.165135 Abstract: --------- We explore the transition to a charge-disproportionated insulating phase in a five-orbital cubic tight-binding model applicable to transition-metal perovskites with a formal d^4 occupation of the transition-metal cation, such as ferrates or manganites. We use dynamical mean-field theory to obtain the phase diagram as a function of the average local Coulomb repulsion U and the Hund’s coupling J. The main structure of the phase diagram follows from the zero bandwidth (atomic) limit and represents the competition between high-spin and low- spin homogeneous and an inhomogeneous charge-disproportionated state. This results in two distinct insulating phases: the standard homogeneous Mott insulator and the inhomogeneous charge-disproportionated insulator, recently also termed Hund’s insulator. We characterize the unconventional nature of this Hund’s insulating state. Our results are consistent with previous studies of two- and three-orbital models applicable to isolated t2g and eg subshells, respectively, with the added complexity of the low-spin/high-spin transition. We also test the applicability of an effective two-orbital (eg-only) model with disordered S=3/2 t2g core spins. Our results show that the overall features of the phase diagram in the high-spin region are well described by this simplified two- orbital model, but also that the spectral features exhibit pronounced differences compared to the full five-orbital description. Description of uploaded tar.gz archive and pythtb_to_Hk_file.py: ---------------------------------------------------------------- The archive contains the scripts and data needed to run the DMFT calculations on the tight-binding model from the paper, mainly as jupyter notebooks and python scripts. - Atomic-limit considerations: Computed with the notebook schematic_atomic_limits.ipynb - Generating tight-binding (TB) model/DMFT input: The folder Generate-models contains the jupyter notebooks to generate the TB models (create with pythTB). This notebook imports the pythtb_to_Hk_file script, which we also uploaded on the materials cloud. The readme inside this folder describes the workflow to obtain the correct h5 archives to put into solid-dmft. These h5 archives as well as the files directly used to generate them are also in this directory. - Running DMFT: The folders DMFT-TB-R1+_* contain the calculations for the phase diagrams. Each of the folders inside represents one data point and contains the dmft_config_*.ini as input for solid-dmft as well as the main results in the observables_imp*.dat and the computation's prints in the stdout.*. The complete results, the h5 archives, are too large and are therefore excluded. However, all the important properties for the phase diagrams of every data point are saved in the results.pkl in the DMFT-TB-R1+_* folders. These are python pickle files and were generated and can be read by phase_diagram.ipynb. Besides that, the folders also contain different jupyter notebooks used for the analysis of other plots. - Code versions: We used triqs version 3.0.x, integrated in the wrapper program solid_dmft (https://github.com/flatironinstitute/solid_dmft), mainly in the version 3.0.x and its non-public predecessors. The DFT calculations for the wannier band structure were performed in Vasp 5.4.4 and with wannier90 v3.1. For the TB model, we use pythTB v1.7 (https://www.physics.rutgers.edu/pythtb/index.html).