Three-dimensional dopant depth determination from momentum-resolved core-loss 4D-STEM data
Moving beyond specimen-projected micrographs in scanning transmission electron microscopy to determine a three-dimensional distribution of dopants is of vital importance for materials characterization and development. Within the Born approximation, Fourier transformation over probe position of a momentum-resolved core-loss 4D-STEM dataset reveals fringes whose frequency encodes the dopant's distance from the focal plane. We address two principal challenges which arise under experimental circumstances. First, fringe contrast is degraded by dynamical scattering, but can be partially recovered by acquiring off major zone axes. Second, the excessive doses typical of core-loss scattering can be alleviated using a Bayesian framework which, on a simulated Fe-doped strontium titanate, achieves nanometre to sub-nanometre depth precision from 100 to 1000 core-loss counts, depending on the signal-to-background.