Saturday , November 27 2021

AI contracts extensive material properties to dismantle the wall that previously could not be crossed


Tokyo – If the properties of the materials can be reliably predicted, the process of developing new products for a variety of industries can be streamlined and accelerated. In a study published in XXX in Advanced Smart Systems, researchers from the Tokyo University of Industrial Sciences used a core loss spectroscopy to determine the properties of organic molecules using machine learning.

Spectroscopy techniques Energy loss near-edge structure (ELNES) and near-edge X-ray structure (XANES) are used to determine information about the electrons, and through them the atoms, in materials. They have high sensitivity and high resolution and have been used to study a variety of materials, from electronic devices to drug delivery systems.

However, the connection of spectral data to the properties of matter — things like optical properties, electron conductivity, density, and stability — remains ambiguous. Machine learning (ML) methods have been used to extract information for large complex data sets. Such approaches use artificial neural networks, based on how our brain works, to constantly learn to solve problems. Although the group has previously used the ELNES / XANES and ML spectrum to find out information about substances, what it found did not address the properties of the substance itself. Therefore, the information could not be easily translated into developments.

The team now used ML to uncover hidden information in the simulated ELNES / XANES spectrum of 22,155 organic molecules. “The ELNES / XANES spectrum of the molecules, or their ‘outlines’ in this scenario, were then introduced into the system,” explains lead author Kakeru Kikumasa. “This outline is something that can be measured directly in experiments and therefore can be determined with sensitivity and high resolution. This method is very useful for material development because it has the potential to reveal where, when and how certain material properties result.”

A model created only from the spectrum was able to successfully predict what are known as intense features. However, it has not been able to predict extensive properties, depending on the molecular size. Therefore, to improve the prediction, the new model was constructed by including the ratio of three elements with respect to carbon (found in all organic molecules) as additional parameters in order to allow to correctly predict the extensive properties such as molecular weight.

“Our ML – learning on the core loss spectrum provides accurate prediction of extensive material properties, such as internal energy and molecular weight. The connection between the core loss spectrum and extensive properties has never been made; however artificial intelligence has been able to reveal the hidden connections. Our approach may also be applied. To predict the properties of new materials and functions, “says senior author Teruyasu Mizoguchi. “We believe our model will be a very useful tool for developing high output materials in a wide range of industries.”

The study, “Quantification of the properties of organic molecules using a core-loss spectrum as described by neural networks,” was published in Advanced Intelligence Systems at DOI: 10.1002 / aisy.202100103.

Information on the Institute of Industrial Sciences (IIS), University of Tokyo

The Institute of Industrial Sciences (IIS), University of Tokyo is one of the research institutes affiliated with the University of Japan.

More than 120 research labs, each led by a faculty member, include IIS, with more than 1,200 members, including about 400 staff and 800 students who are actively engaged in education and research. Our activities cover almost all areas of the engineering disciplines. Since its inception in 1949, IIS has worked to bridge the huge gaps that exist between academic disciplines and real-world applications.

Public Release. This material comes from the organization / authors (source) and may have a specific nature, edited for clarity, style and length. The opinions and opinions expressed are those of the authors. Watch it in full here.

Source link