2024-02-16 07:30:45
Discovery of new drugs, chemicals, materials, etc., artificial intelligence algorithms have multiple applications in research and development. Indeed, new working methods based on AI can considerably accelerate the work of researchers. Here are some applications presented in recent research in materials science.
Far from the controversies linked to the use of generative AI like ChatGPT, AI is already used by the world of research, in particular to exploit the immense volumes of data already present in previous work. The publisher Elsevier just launched Scopus AIa new tool that assists scientists in their documentary research and allows them to obtain an overview of a subject.
AI: an accelerator of research and innovation
New AI-based tools seem to be real boosters of research and innovation and the list of areas of application seems endless, because it is above all a question of developing new working methods!
This theme has already been addressed several times in the News Magazine, for multiple uses.
In the rest of this article, we will explore some applications of AI in materials science.
Azure Quantum Elements: Microsoft’s AI used to develop a new solid electrolyte for batteries
Researchers at the Pacific Northwest National Laboratory (PNNL) in Richland, US, turned to Microsoft’s Azure Quantum Elements solution to accelerate the search for the ideal solid electrolyte.
This new tool from Microsoft, launched in June 2023, puts artificial intelligence and cloud computing at the service of research, with the aim of obtaining rapid results, in a few weeks, instead of several years.
As we explained to you in this other article, thanks to this revolutionary tool, PNNL researchers were able to select 18 promising candidates from 32 million possibilities, a sorting process which took barely 80 hours, thanks to quantum calculation.
Furthermore, the Microsoft tools being trained in chemistry in general, they can be used to accelerate any type of materials research.
A machine learning model to develop permanent magnets without critical elements
Still in the USA, a team of researchers from Ames National Laboratory used a machine learning algorithm to predict the crystal structure and magnetic properties of millions of hypothetical compounds, based on their chemical formula and rules of thumb.
The idea was to train the AI using experimental and theoretical data, with the aim of finding compounds with a high Curie temperature, a first step in discovering materials capable of retaining magnetic properties at high temperatures. .
To test the model, the team used compounds based on cerium, zirconium and iron, abundant earth elements, with the aim of focusing research on both magnetic materials and efficient and economically viable.
An AI-guided working method to accelerate the discovery of more efficient thermoelectric materials
For their part, researchers from the NOMAD Laboratory at the Fritz Haber Institute of the Max Planck Society, in Germany, have just proposed a AI-guided workflow. They used this approach to identify more than 50 promising thermal insulators, essential materials for manufacturing high-performance thermoelectric elements.
A more precise way to accelerate the design of corrosion-resistant alloys with machine learning
Of the researchers from another institute of the Max Planck Society use AI models to predict corrosion behavior and suggest optimal alloy formulas. They have just developed a machine learning model that improves predictive accuracy by 15%.
The model is distinguished by the fusion of numerical and textual data. Furthermore, it is versatile and can be extended to all alloy properties.
Understanding dislocations in polycrystalline materials using AI
Finally, in Japanresearchers at Nagoya University are using AI to study small defects in polycrystalline materials, widely used in electronics and solar cells.
These small defects, called dislocations, are problematic because their presence disrupts the regular arrangement of atoms in the lattice, affecting electrical conduction and overall performance.
So they used AI to create a virtual 3D model, which helped them identify areas where groups of dislocations were affecting the material’s performance. Ultimately, this method might therefore help improve the properties of polycrystalline materials with a potential impact in many areas.
These few works are only examples, but they have one thing in common: they were all published in 2023. Given the challenges and the possibilities offered by AI, it is certain that studies of this type will multiply. Equipped with such tools, it is likely that research into materials will accelerate significantly in the years to come!
1708208213
#research #develop #materials