Acquired Electronics360

Building Technologies

Machine Learning Algorithms Help Jump Start the Discovery of Metallic Glass

16 April 2018

Typically, the alloy formed by combining two or three metals will look and act like a metal. The atoms of these metal combinations are arranged in rigid geometric patterns that are key to this metal-like structure. But sometimes, something unexpected happens. A new alloy called metallic glass is created by combining metals under just the right conditions. The new material’s atoms are arranged similarly to the glass in a window. It is stronger and lighter than steel and can stand up to corrosion and wear better than steel.

With new, artificial intelligence approach, scientists discovered metallic glass 200 times faster than with an Edisonian approach. (Source: SLAC National Accelerator Laboratory)With new, artificial intelligence approach, scientists discovered metallic glass 200 times faster than with an Edisonian approach. (Source: SLAC National Accelerator Laboratory)

Metallic glass has potential as a protective coating for other metals and as an alternative to steel. However, only a few thousand metallic glass combinations -- out of potentially millions -- have been found in the last 50 years. Only a few of these combinations have been proved useful.

A group from the Department of Energy’s SLAC National Accelerator Laboratory, the National Institute of Standards and Technology (NIST) and Northwestern University has created a method to solve the problem with developing metallic glass. This team created a shortcut that helps to discover and improving metallic glass, making production cheaper and faster.

To create these shortcuts, the group used a system at SLAC’s Stanford Synchrotron Radiation Lightsource (SSRL). This system uses machine learning and experiments that can screen and sample hundreds of materials in a short amount of time. With this machine, the team discovered three new blends of ingredients to form metallic glass 200 times faster.

"It typically takes a decade or two to get a material from discovery to commercial use," said Chris Wolverton, the Jerome B. Cohen Professor of Materials Science and Engineering in Northwestern's McCormick School of Engineering, who is an early pioneer in using computation and AI to predict new materials. "This is a big step in trying to squeeze that time down. You could start out with nothing more than a list of properties you want in a material and, using AI, quickly narrow the huge field of potential materials to a few good candidates."

The team’s end goal is to develop this method so a researcher or scientist can scan hundreds of sample materials and immediately get feedback on the set so they can further develop the materials and test them again that day or the next day. In the past 50 years around 6,000 combinations of metals have been tested to form glass. With the new method, researchers were able to screen over 20,000 combinations in a single year.

"The unique thing we have done is to rapidly verify our predictions with experimental measurements and then repeatedly cycle the results back into the next round of machine learning and experiments,” said Mehta.

The team says they can make the process even faster. They also say that they can develop it so humans don’t even need to be involved in the process at all.

"This will have an impact not just on synchrotron users, but on the whole materials science and chemistry community," Mehta said.

"One of the more exciting aspects of this is that we can make predictions so quickly and turn experiments around so rapidly that we can afford to investigate materials that don't follow our normal rules of thumb about whether a material will form a glass or not," said paper co-author Jason Hattrick-Simpers, a materials research engineer at NIST. "AI is going to shift the landscape of how materials science is done, and this is the first step."

During the metallic glass study, the team investigated thousands of alloys that include three cheap and nontoxic metals. They started out with material data that dates back more than 50 years. Advanced machine learning algorithms were used to go through the data. Based on what the algorithms found the first time around, the team created two sample alloy sets with two methods and then tested different manufacturing methods may affect the alloys turning into glass. The SSRL x-ray beam scanned the alloys and the results were put into a database. From there new machine learning results were generated and used to prepare the new samples to be tested.

In the third round of testing the group’s success rate for finding metallic glass was one out of two or three samples tested. The samples had three different combinations of ingredients and two of these ingredients have never been used to make metallic glass.

The paper on this research was published in Science Advances.

To contact the author of this article, email Siobhan.Treacy@ieeeglobalspec.com


Powered by CR4, the Engineering Community

Discussion – 0 comments

By posting a comment you confirm that you have read and accept our Posting Rules and Terms of Use.
Engineering Newsletter Signup
Get the Engineering360
Stay up to date on:
Features the top stories, latest news, charts, insights and more on the end-to-end electronics value chain.
Advertisement
Weekly Newsletter
Get news, research, and analysis
on the Electronics industry in your
inbox every week - for FREE
Sign up for our FREE eNewsletter
Advertisement

CALENDAR OF EVENTS

Date Event Location
23-27 Apr 2018 Oklahoma City, Oklahoma
10 May 2018 Pier 94
22-24 May 2018 Los Angeles, CA
04-07 Jun 2018 Boston, MA
06-08 Jun 2018 Los Angeles, CA
Find Free Electronics Datasheets
Advertisement