Scientists are using machine learning to accelerate materials discovery

The end product of the machine learning algorithm: the metastable phase diagrams for carbon. Colored regions indicate the conditions under which carbon exists in certain metastable states that can yield useful material properties. Credit: Argonne National Laboratory

A new computational approach will improve understanding of the different states of carbon and guide the search for yet-to-be-discovered materials.

Materials – we use them, wear them, eat them and create them. Sometimes we invent them by accident, like with Silly Putty. But far more often, making useful materials is a tedious and expensive process of trial and error.

Scientists at the US Department of Energy’s (DOE) Argonne National Laboratory recently demonstrated an automated process for identifying and exploring promising new materials by combining machine learning (ML) – a type of artificial intelligence – and high performance computing. The new approach could help accelerate the discovery and design of useful materials.

Using the single-element carbon as a prototype, the algorithm predicted how atoms order over a wide range of temperatures and pressures to form different substances. From there, he constructed a series of what scientists call phase diagrams, a kind of map that helps guide their search for new and useful states of matter.

“We trained a computer to probe, interrogate, and learn how carbon atoms might be organized to create phases that we might not find on earth or fully understand, thereby automating a whole step in the process of developing materials,” said Pierre Darancet, scientist from Argon and author of the study. “The more a computer can handle this process on its own, the more materials science we can do.”

Balance and beyond

When the atomic structure of a material changes, its electronic, thermal and mechanical properties also change. Scientists want to find new ways to arrange atoms to develop useful materials. One of the main ways to change the atomic structure of a material is to vary the surrounding pressure and temperature.

We see this kind of structural change commonly in water. At room temperature and normal atmospheric pressure, water is more stable in liquid form. If you lower the temperature enough, the same water molecules will organize themselves to form solid ice.

Similarly, diamond and graphite are very different materials, but they both consist exclusively of carbon atoms-just arranged in different ways. Under normal conditions, graphite is a much more stable form of carbon than diamond. Under conditions of extreme pressure and heat, however, graphite slowly crystallizes into diamond. When removed from these extreme conditions, the diamond persists, persisting in what scientists call a metastable state.

The ML algorithm constructed phase diagrams that mapped hundreds of these metastable states of carbon, some known and some new.

“It is experimentally difficult to predict and produce states of matter that are not close to equilibrium conditions,” said Jianguo Wen, an Argonne experimenter on the study. “This new computing approach allows us to explore those little-known regions on maps that aren’t otherwise accessible, or that we don’t even know exist yet.”

Les scientifiques utilisent l'apprentissage automatique pour accélérer la découverte de matériaux

High resolution TEM images of the metastable phases of carbon. an orthorhombic graphite with an AB’ stacking pattern and a rhombohedral graphite with an ABC stacking along with the experimental and simulated diffraction patterns (blue circles). b Hexagonal-diaphite and cubic-diaphite with experimental and simulated diffraction patterns (blue circles). c Different combinations of stacking patterns resulting from the simultaneous inter-growth of hexagon and cubic-diamond. Credit: Nature Communication (2022). DOI: 10.1038/s41467-022-30820-8

Algorithm verification

The scientists trained the ML algorithm with synthetic data, which is produced by simulation and approximates the results the scientists would obtain from the experiment. They generated the dataset using molecular dynamics and density functional theory, two common computational chemistry tools.

The training data was produced using Carbon, a high-performance computing cluster from the Center for Nanoscale Materials (CNM) at Argonne, a DOE Office of Science user facility. Two other DOE user facilities were also used: the Argonne Leadership Computing Facility and the National Energy Research Scientific Computing Center at Lawrence Berkeley National Laboratory.

Using the algorithm’s predictions as a guide, the team verified its effectiveness by synthesizing real samples and characterizing them using a transmission electron microscope at the CNM.

The algorithm successfully predicted well-known phase diagrams for carbon, and the computer-generated phase diagrams affirmed and shed light on several yet unexplained experimental observations.

In particular, the algorithm identified the previously ambiguous structure of n-diamond (means “new diamond”), a state of carbon that has mystified scientists since it was theorized more than 30 years ago. “The algorithm made new and surprising predictions about the structural characteristics of n-diamond that we verified by experiment, demonstrating that the algorithm holds up even with high-profile phases,” Wen said.

The team also synthesized several phases predicted by the algorithm that have not yet been reported in the scientific literature. The sample structures matched the predictions, further verifying the algorithm.

“Synthesizing materials, especially those with exotic properties, can often take multiple experimental trials and years of effort,” said Argonne scientist Subramanian Sankaranarayanan, lead author of the study. “Our machine learning algorithms allow us to identify the conditions for the synthesis of exotic materials, potentially reducing the time for their experimental realization.”

In this study, the algorithm was applied to carbon only. In the future, the scientists hope to apply the same approach to systems with more than one element. Application of machine learning algorithm to more complex systems could have a broad impact on the discovery and design of useful materials.

An article about the study, “Machine learning the metastable phase diagram of covalently bonded carbon”, was published in Nature Communication.

A model trained to predict spectroscopic profiles helps decipher the structure of materials

More information:
Srilok Srinivasan et al, Machine learning the metastable phase diagram of covalently bonded carbon, Nature Communication (2022). DOI: 10.1038/s41467-022-30820-8

Quote: Scientists Use Machine Learning to Accelerate Materials Discovery (2022, October 5) Retrieved October 5, 2022, from

This document is subject to copyright. Except for fair use for purposes of private study or research, no part may be reproduced without written permission. The content is provided for information only.

#Scientists #machine #learning #accelerate #materials #discovery

Leave a Comment

Your email address will not be published. Required fields are marked *