The accuracy rate is 95%, machine learning predicts the synthesis of complex new materials

According to a study published in the journal "Science Advances" on the 22nd, researchers from Northwestern University and Toyota Research Institute have successfully applied machine learning to guide the synthesis of new nanomaterials and eliminate obstacles related to material discovery

The corresponding author of the paper and an expert on nanotechnology at Northwestern University, Chad Milkin, invented the data generation tool "Huge Library", which greatly expanded the horizons of researchers

The research team developed the "Giku" by using polymer pen lithography technology, which is a massively parallel nanolithography tool capable of depositing hundreds of thousands of features at specific locations per second

When mapping the human genome, the task of scientists is to identify combinations of four bases

Milkin said that even for things similar to material genomes, determining how to use or label them requires different tools

In this research, the Mirkin team compiled the previously generated "big library" structure data composed of nanoparticles with complex compositions, structures, sizes, and morphologies

The model is based on a large data set established by Northwestern University to find polymetallic nanoparticles with parameters set around phase, size, and other structural features, and these parameters will change the properties and functions of the nanoparticles

The researchers said that the technology may promote discoveries in many areas that are critical to the future, including plastic upgrades and recycling, solar cells, superconductors, and qubits