Accurate, Exascale-Ready Methods for Theoretical Core-Level Spectroscopy of Complex Materials

• Applicant: Dr. Dorothea Golze
• Subject Area: Theoretical Chemistry: Molecules, Materials, Surfaces ; Theoretical Chemistry: Electronic Structure, Dynamics, Simulation
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 453275048

The CoreXL project will develop highly accurate theoretical methods for core-level spectroscopy of complex materials based on fully relativistic Green's function theory in the GW approximation. X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) are powerful tools for materials characterization. However, the experimental spectra are generally difficult to interpret without aid from predictive theoretical methods. GW has become the method of choice for the computation of addition and removal energies of valence electrons. Its extension to the Bethe-Salpeter equation (BSE) formalism, GW+BSE, is nowadays also frequently applied to compute optical absorption spectra. Even though XPS is often more powerful for chemical analysis than its valence equivalent, it was completely uncharted in GW. I recently unlocked the potential of GW and GW+BSE for molecular 1s core excitations and molecular K-edge transitions. I will use this initial success as seed to advance GW for large-scale application to complex materials, including light and heavy elements. I will extend the formalism to a fully relativistic description, which enables the computation of both s-type and spin-orbit coupled p, d and f excitations. The CoreXL project will also push GW-based methods, which are computationally very expensive, to previously inaccessible system sizes. This will be achieved by reducing the scaling with respect to system size N from O(N^5) to effectively O(N^2) and by exploiting the performance boost provided by the emerging generation of exascale supercomputers, which are capable of 10^18 operations per second - a factor of 100-1000 more than current supercomputers. The developed methods will be made exascale-ready by investing in massively parallel, memory-efficient and GPU-supported algorithm design. Equipped with the developed tools, I can go far beyond what is currently possible in theoretical core-level spectroscopy: accurate computation of XPS and XAS spectra will become feasible for complex two-dimensional nanomaterials, surface and solid-state systems. In collaboration with theoretical and experimental partners, I will apply my developments to the following systems: i) two-dimensional framework materials, which are candidates for magnetic semiconductors, ii) noble metal (phospho)chalcogenides, whose properties can be tuned by defect engineering, and iii) amorphous carbon, which shows promise as electrode material for biomedical devices. These are challenging applications where XPS and XAS can, in synergy with theory, resolve details of the local atomic structures, but where standard core-level simulation techniques fail. The method developments proposed in the CoreXL project are the missing link to fully realize the potential of XPS and XAS for materials characterization.

• DFG Programme: Independent Junior Research Groups
• International Connection: Czech Republic, Finland, USA
• Cooperation Partners: Miguel Caro, Ph.D.; Professor Dr. Patrick Rinke; Dr. Sami Sainio; Professor Dr.-Ing. Zdenek Sofer