A group of researchers at University of Michigan developed a machine-learning algorithm to study the composition of stiff glass. The disordered nature of glass particles makes it difficult to understand glass properties from looking at its composition. Therefore, the researchers developed a distinct algorithm that uses computer simulations to study the properties of glass. This algorithm can help in studying the composition of extremely stiff glass materials. Stiff glass is used to manufacture components of wind turbines and next-generation vehicles. The findings of the researchers are published in a new paper titled ‘npj Computational Materials’.
Professor of Material Sciences at U-M, Liang Qi answered important questions about the research.
Importance of Studying Glass Materials
Liang revealed the importance of elastic stiffness in understanding the action of glass materials. Glass materials can recover their original shape depending on the value of the material’s elastic stiffness. Moreover, structural applications of various materials also rely on its elastic stiffness. Higher stiffness means that thin sheets of glass can be imparted greater shock resistance. Structural glass used in smart phones, windshields, and touchscreen devices are also made from stiff glass materials. In addition to this, glass fiber composites are also used in manufacturing lightweight items for trucks, cars, and wind turbines. Liang believes that the new algorithm could play a crucial role in manufacturing light vehicles.
The disordered and amorphous nature of glass particles makes it difficult to ascertain their atomistic structure. In addition to this, dearth of data about glass compositions poses challenges for machine learning algorithms. To resolve these problems, the researchers used machine learning standards that are suited for small amount of data. Using small data sets helped researchers in making more trustable predictions for glass compositions.