Abstract text (incl. figure legends and references)
Organic materials remain challenging to characterize in electron microscopy due to their weak scattering from and high dose sensitivity to the electron beam. This talk will show how advances in detector technology and computational imaging have enabled four dimensional- scanning transmission electron microscopy (4D-STEM) characterization of functional organic thin films [1]. By carefully tuning the size of the converged electron probe, the distance between adjacent measurements, and the total exposure of the electron beam, we can maximize the amount of measurable information. This talk will show experimental examples characterizing the morphology of crystalline and semi-crystalline polymers and other organic thin films. It will also show several methods which can improve 4D-STEM characterization including cryogenic cooling, custom probe-forming apertures, 4D-scanning confocal electron diffraction (4D-SCED), and using machine learning methods to invert noisy and nonlinear diffraction measurements.
The primary goal of this research is to measure structural and morphological material properties from weakly-scattering and beam sensitive organic molecules. These properties include the degree of crystallinity, crystalline phase and orientation, and degree of short and medium range order present. We aim to map these properties over the functional length scale of the materials in question, where the field of view may range from nanometers to hundreds of micrometers.
Our experiments are performed using several FEI/ThemoFisher Titan and Themis platform STEM instruments. We typically use 300 kV accelerating voltage, and probe convergence semiangles from 0.1 to 4 mrads. We also make use of custom amplitude apertures in the probe-forming plane [2]. Our electron detectors range from scintillator-based cameras such as the Gatan Ultrascan, and direct electron detectors such as the Gatan K2 and K3. We also show results from the Berkeley Lab 4D Camera operating at 87 000 images/second [3]. The 4D-STEM experimental geometry and an example 4D-STEM sdataset is shown in Figure 1.
We also demonstrate alternative experimental geometries such as 4D-SCED which can minimize the electron beam damage induced by adjacent probe positions measurements [4]. Thick samples may also produce significant amounts of multiple scattering of the electron beam. In order to perform accurate measurements of these complex nonlinear diffraction signals at low electron dose, we have implemented machine learning methods [5]. Finally, all of our methods and datasets are freely available in the py4DSTEM open source analysis toolkit [6].
Figure 1 - 4D-STEM experimental geometry showing diffraction patterns measured at various probe positions for a small organic molecule sample. In each diffraction pattern, the pi-pi stacking direction(s) can be identified from the primary Bragg peak positions. The reconstructed sample morphology is shown above as an orientation map. Adapted from [7].
[C1] C Ophus, Microscopy and Microanalysis 25, 563 (2019).
[C2] S Zeltmann et al., Ultramicroscopy 209, 112890 (2020).
[C3] P Pelz et al., IEEE Signal Processing Magazine 39. 25 (2021).
[C4] M Wu et al., Nature Communications 13, 1 (2022).
[C5] J Munshi et al., arXiv:2202.00204 (2022).
[C6] B Savitzky et al., Microscopy and Microanalysis 27, 712 (2021).
[C7] K Bustillo et al., Accounts of Chemical Research 54, 2543 (2021)