edfas.org 11 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 26 NO. 1 a gradient, then the disk does not shift (Fig. 3, Panel I, a). If the gradient is larger than the probe, it will displace the disk (Fig. 3, Panel I, b); however, if the potential is smaller than the probe, intensity within the diffraction disk will change (Fig. 3, Panel I, c). Moreover, because samples are heterogenous there are many phenomena that can redistribute intensity within the diffraction pattern. This can include sample bending and thickness variations[47] as well as abrupt changes in mean inner potential due to interfaces.[65] Meaning that multiple signals can be convoluted and it becomes necessary to identify data acquisition methods and analysis routines which can mitigate the contribution of these other signals. This concept of multiple signals is well illustrated in a polycrystalline FeGe film (Fig. 3, Panel II) where contrast from the magnetic fields, as well as from changes to the diffraction condition with respect to the incident beam (i.e., diffraction contrast), can be observed.[53] One ubiquitous phenomenon that is the origin of competing signals is dynamical diffraction, which causes redistribution of intensity in the diffraction discs as a function of thickness and sample tilt, for example. One method to mitigate this effect is to acquire data from multiple orientations and average the result, which suppresses rapid variations in the signal due to dynamical diffraction and changes in local orientation.[66] This may be implemented by tilting the sample[54] or incident beam[67] incrementally and acquiring multiple scans or by precession,[56,66] where the beam is tilted over an angular range at each position in the scan. As exemplified in Fig. 4, Panel I, the unwanted diffraction contrast is suppressed improving visualization of the electric fields.[67] However, for this particular result there was still not an agreement when compared to simulations predicting the electric field distribution. It was hypothesized that sample preparation may be responsible for the discrepancy, where Ga ion implantation (from focused ion beam sample preparation) are known to passivate dopants near surfaces.[68] This means additional steps may be needed during data analysis or sample preparation to ensure accurate and representative results. As for measuring the beam shift, the center of mass (COM), also known as the first moment, of a diffraction pattern is commonly used. Yet, as previously mentioned, factors such as dynamical diffraction introduce signals that do not reflect long-range electromagnetic fields. Different schemes have been implemented to reduce these unwanted effects. To suppress diffraction contrast in a polycrystalline FeGe film the COM was calculated using an annulus around the direct beam disk edge.[53] Wu and coworkers[47] applied a circular Hough transform filter and then refined the position using a nonlinear trustregion algorithm. This achieves a significant improvement over traditional COM measurements, as shown in Fig. 4, Panel II. Similar filtering and position refinement schemes have also reported improved performance over COM measurements.[56] A natural extension of field measurements is to measure these materials under stimuli. Temperature, electrical voltage or current, and magnetic field can all be readily applied using specialized holders or by tuning the remanent field of the objective lens. This can be used to access conditions where certain features, such as skyrmions, are stable,[53] as well as observing evolving features, such as magnetic domains as a function of temperature and field;[51] the evolution of internal fields at pn-junctions have been measured under different applied voltages.[57] These measurements could be extended further by increasing the time resolution to capture transient states or using tomography to discern the field in three-dimensions rather than measuring a projected, through-thickness value. SUMMARY This article covered techniques measuring structure (phase and orientation and SRO or MRO) as well as properties (electromagnetism). Relationships between processing, structure, and properties are frequently complex making it challenging to disentangle these interconnected relationships. Because 4D datasets contain an immense amount of information understanding these complex property-processing-structure relationships becomes more accessible. This possibility is represented nicely in Fig. 3, Panel II, where both the magnetic field and crystallography of the sample were extracted from the same dataset. One could envision a similar study where a processing condition is applied in situ and both the property and the structure could then be tracked as a function of time. Part III of this series will cover the topic of ptychography. REFERENCES 1. R.R. Keller and R.H. Geiss: “Transmission EBSD from 10 nm Domains in a Scanning Electron Microscope,” Journal of Microscopy, 2012, 245(3), p. 245-251. 2. M. Lederer, et al.: “Local Crystallographic Phase Detection and Texture Mapping in Ferroelectric Zr Doped HfO2 Films by Transmission-EBSD,” Applied Physics Letters, 2019, 115(22). 3. A. Orekhov, et al.: “Wide Field of View Crystal Orientation Mapping of Layered Materials,” arXiv, 2020. 4. Y. Meng and J.-M. Zuo: “Improvements in Electron Diffraction Pattern Automatic Indexing Algorithms,” Eur. Phys. J. Appl. Phys., 2017, 80(1), p. 10701.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==