We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat.mtrl-sci

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Condensed Matter > Materials Science

Title: Mapping Structural Heterogeneity at the Nanoscale with Scanning Nano-structure Electron Microscopy (SNEM)

Abstract: Here we explore the use of scanning diffraction electron microscopy (4D-STEM) coupled with electron atomic pair distribution function analysis (ePDF) to understand the local structure and chemistry in a complex multicomponent system, a hot rolled, Ni-encapsulated, Zr$_{65}$Cu$_{17.5}$Ni$_{10}$Al$_{7.5}$ bulk metallic glass, as a function of position with 3~nm spatial resolution. We show that it is possible to gain insight into the chemistry and chemical clustering/ordering tendency in different regions of the sample, including in the vicinity of nano-scale crystallites that are identified from virtual dark-field images and in heavily deformed regions at the edge of the BMG. This gives critical insights into the formation mechanism of these features. Unsupervised machine learning was used to extract partial PDFs from the material, modeled as quasi-binary alloy, and map them in space allowing key insights not only into the local average composition but also any chemical short-range ordering tendency in that region of the sample. The experiments are straightforward and rapid and unlike spectroscopic measurements don't require energy filters on the instrument. We spatially map different quantities of interest (QoI's), defined as scalars that can be computed directly from an analysis of the data such as positions and widths of PDF peaks or parameters refined from fits to the patterns, and have developed a flexible and rapid data analysis software framework that allows experimenters to rapidly explore images of the sample on the basis of different QoI's. The power and flexibility of this approach is explored and described in detail. Sicne we are getting spatially resolved images of the nanoscale structure obtained from PDFs we call this approach scanning nano-structure electron microscopy (SNEM). We believe that it will be powerful and useful extension of current 4D-STEM methods.
Subjects: Materials Science (cond-mat.mtrl-sci); Disordered Systems and Neural Networks (cond-mat.dis-nn)
Cite as: arXiv:2110.03589 [cond-mat.mtrl-sci]
  (or arXiv:2110.03589v3 [cond-mat.mtrl-sci] for this version)

Submission history

From: Yevgeny Rakita [view email]
[v1] Thu, 7 Oct 2021 16:14:52 GMT (40103kb,D)
[v2] Mon, 18 Oct 2021 02:02:32 GMT (40100kb,D)
[v3] Sun, 7 Nov 2021 03:40:59 GMT (36015kb,D)

Link back to: arXiv, form interface, contact.