90% Accurate Machine Learning Approach Developed by MIT to Screen Topological Materials for Next-Generation Computer Chips and Quantum Devices
Researchers from MIT have developed a new machine learning-based approach with 90 percent accuracy to screen candidate materials if they are topological for next-generation computer chips or quantum devices
Topological materials, a type of special material, have different properties for their surfaces and interiors. Electrical properties are one of them. These materials could be used to improve the efficiency of electronic and optical devices or as components in quantum computers. Recent theories and calculations show that thousands of compounds can have topological characteristics. Testing all of these compounds to determine their properties will require years of analysis and work. There is an urgent need for methods that can be used to study and test topological materials faster.
A team of researchers at MIT, Harvard University and Princeton University as well as Argonne National Lab proposed a new method that can screen candidate materials faster and predict topological properties with greater than 90% accuracy. This problem has been solved in a complicated way.
The new method proposed is based upon how the material absorbs X rays. This is different than the old methods which were based upon photoemissions and tunneling electrons. The use of X-ray data has several advantages. The experiment is set up using X-ray spectrometers, which can be used in any environment and are easily available. Second, these measurements have been made in chemistry or biology for many other applications. This means that the data for numerous materials is already available.
Researchers From MIT Have Developed A New Machine Learning Based Approach With 90 Percent Accuracy To Screen Candidate Materials If They Are Topological For Next-Generation Computer Chips or Quantum Devices