Data Analysis Pipelines for Low Dose In Situ SPLEEM and TEM Experiments

Electron microscopy is a powerful technique for studying dynamic nanoscale processes in materials due to its unmatched spatial resolution and many flexible imaging modes. These include techniques such as transmission electron microscopy (TEM) and spin-polarized low energy electron microscopy (SPLEEM). However, time-resolved TEM and SPLEEM measurement accuracy is always limited by the electron dose, either because of source brightness limitations or the tolerance of the sample to the electron beam. We can boost the spatial resolution of our measurements by averaging in the time dimension, but this comes at the cost of time resolution. Denoising algorithms can improve both time and spatial resolution, but we must take care to over-regularize the data and reduce or even remove the desired signals. Therefore, in order to extract the maximum information possible from each measurement, we must develop custom computer analysis workflows which optimize the tradeoffs between space and time resolution, and the degree of regularization which can be safely applied. In this talk, I will demonstrate computational analysis pipelines for two types of low dose <i>in situ</i> experiments. The first is SPLEEM measurements of reversible field-free writing/deleting of skyrmions at room temperature using hydrogen-induced magnetic anisotropy changes. The second example I will show is TEM measurements of disorder induced in complex oxides by ion irradiation. In both cases, the low electron dose required us to use both denoising methods and knowledge of the sample configuration in order to efficiently measure time-resolved properties.