Quantifying phase magnitudes of open-source focused-probe 4D-STEM ptychography reconstructions.

Abstract

Accurate computational ptychographic phase reconstructions are enabled by fast direct-electron cameras with high dynamic ranges used for four-dimensional scanning transmission electron microscopy (4D-STEM). The availability of open software packages is making such analyses widely accessible, and especially when implemented in Python, easy to compare in terms of computational efficiency and reconstruction quality. In this contribution, I reconstruct atomic phase shifts from convergent-beam electron diffraction maps of pristine monolayer graphene, which is an ideal dose-robust uniform phase object, acquired on a Dectris ARINA detector installed in a Nion UltraSTEM 100 operated at 60 keV with a focused-probe convergence semi-angle of 34 mrad. For two different recorded maximum scattering angle settings, I compare a range of direct and iterative open-source phase reconstruction algorithms, evaluating their computational efficiency and tolerance to reciprocal-space binning and real-space thinning of the data. The quality of the phase images is assessed by quantifying the variation of atomic phase shifts using a robust parameter-based method, revealing an overall agreement with some notable differences in the absolute magnitudes and the variation of the phases. Although such variation is not a major issue when analysing data with many identical atoms, it does put limits on what level of precision can be relied upon for unique sites such as defects or dopants, which also tend to be more dose-sensitive. Overall, these findings and the accompanying open data and code provide useful guidance for the sampling required for desired levels of phase precision, and suggest particular care is required when relying on electron ptychography for quantitative analyses of atomic-scale electromagnetic properties.