Learning motives and their hierarchies in atomic resolution microscopy

Learning motives and their hierarchies in atomic resolution microscopy

Learning motifs and their hierarchies in atomic-resolution microscopy Basic pentagonal motifs follow in a hierarchy of three levels to form increasingly complex larger motifs. The figure illustrates how a seemingly “random” and disordered collection of atoms can be organized as a hierarchy of motifs listed as 1st, 2nd and 3rd levels. At the first level, each of the “disks” in the images represents a column of atoms. The blue motifs are pentagonal arrangements of atoms that are “hollow” (five-headed), while the orange motifs are “filled” (six-headed). The framework proposed by the researchers learned the rules for creating motifs at a higher level (eg 2nd level) from the lower levels (eg 1st level). The numbers in the diagrams show the relative amount of motifs at a higher level. Credit: The progress of science (2022). DOI: 10.1126 / sciadv.abk1005

Researchers from the National University of Singapore have developed a machine learning scheme that quickly identifies previously invisible new structures in disordered materials without human supervision. Their study was published in The progress of science.

To understand very disordered complex materials is a long-term challenge. A research team with Assistant Professor Duane Loh from the Department of Physics and Biological Sciences and Professor Stephen Pennycook from the Department of Materials Science, both from the National University of Singapore, created a framework for machine learning that can teach the universal “vocabulary” and “grammar” to describe disturbed systems. Using this framework, they discovered that a wide range of disordered materials can be logically divided into recurring motifs and their compositional rules. These motifs are the building blocks that can significantly simplify how we understand and even classify complex disturbances in real materials.

Many forms of atomic resolution microscopy allow us to look into the secret world of atoms whose arrangement creates the wealth of material that modern civilization relies on. How atoms in these materials arrange themselves under duress, however, is still a mystery. Unsurprisingly, these atomarrangemang are always imperfect and full of disorder. In fact, many materials are sought after due to properties resulting from this disordered arrangement.

Despite advances in machine vision technology, it is still far from automated to learn the rules of atomic arrangement from many atomic resolution micrographs. This is due to the challenges due to the fragility of atomic arrangements for probing: limited signals from each atom; subtle and non-linear differences between atomic types; and inevitable and empirical artifacts from experimental photomicrographs. Technically, these challenges make it difficult to build machine models for automated analyzes. An unsupervised statistical learning technique would be needed, and this is a particularly cumbersome form of machine learning.

The research team used a sequence of mathematical expressions, known as the Zernike polynomial, to quantify the subtle structural and chemical properties of atomic arrangements. These special mathematical expressions can effectively model the functions despite different atomic orientations. To overcome the limited signal from each atom, the team generalized a single-particle imaging method that automatically reveals distinct building blocks (ie motifs) in disordered materials. This approach is also robust enough that imaging artifacts do not affect the result.

After learning the motifs from tens of thousands of atoms in an automated way, the team was now able to discover how these motifs are self-assembled into complex but disordered hierarchies. They found that some disordered materials can be described with only a handful of motives; yet these few motives create different structures due to complex motive-motive hierarchies. In comparison, some materials begin with a continuous range of motifs, which blurs the boundaries between their motifs and hierarchies.

Prof. Loh said: “These motifs form a vocabulary for a disordered material, and motive-motives hierarchy its grammar. This motif-plus-hierarchy description can meaningfully simplify our descriptions of disordered material. Fortunately, this description has led to the discovery of a new structure hidden in a very disordered catalytic material. “

“The first author of this work, Dr. Jiadong Dan, looks at the photomicrographs from our staff and discovers many structural insights previously lacking in the disordered materials,” added Prof. Loh.

Dr Dan said, “I think this motive-plus hierarchy can be used to quantitatively classify the degree of disorder in materials and will open the door to massive machine learning from atomic resolution micrographs.”

The team hopes to transform this framework into a complementary application of artificial intelligence next to a microscope to quickly understand disorder material.


The crucial role of functional motifs – microstructural units that control material functions – in materials research


More information:
Jiadong Dan et al, Learning Motives and Their Hierarchies in Atomic Resolution Microscopy, The progress of science (2022). DOI: 10.1126 / sciadv.abk1005

Quote: Learning motives and their hierarchies in atomic resolution microscopy (2022, 22 June) retrieved 22 June 2022 from https://techxplore.com/news/2022-06-motifs-hierarchies-atomic-resolution-microscopy.html

This document is subject to copyright. Apart from any fair business for private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.


#Learning #motives #hierarchies #atomic #resolution #microscopy

Leave a Comment

Your email address will not be published.