Abstract
Solid-state color center defects are attracting an increasing amount of attention for applications in quantum information science, due to their potential to be integrated with existing architectures for on-chip photonic circuits and fiber optic networks. An isolated defect in a semiconductor forms localized states that act as a spin qubit, allowing for the storage and transfer of quantum information. In this work we have performed high-throughput calculations of over 50,000 point defects in various semiconductors including diamond, silicon carbide, and silicon. Focusing on quantum applications, we characterize the relevant optical and electronic properties of these defects, including formation energies, spin characteristics, transition dipole moments, zero-phonon lines. However, this dataset constitutes only a small fraction of the trillions of combinatorically possible defects. To reduce the computational load and accelerate defect discovery, we explore using a graph convolutional neural network to take features of defect structures and predict hard-to-compute defect properties. This AI based approach can be used to quickly iterate through novel defect structures and discover defects that are ideally tailored to specific applications.
*This work was supported by the Office of Science, Office of Fusion Energy Sciences, of the U.S. Department of Energy, under Contract No. DE-AC02-05CH11231. L.Z.T. and V.I. were supported by the Molecular Foundry, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231 using NERSC award NERSC DDR-ERCAP0025754. B.K. acknowledges support from the NSF QLCI programme through grant number OMA-2016245, and the NSF QuIC-TAQS award 2137645.
