Polyolefin Production Enabled By Quantum Machine Learning
Entos Inc., a Pasadena-based startup revolutionizing molecular discovery through physics-based machine learning, and Dow Inc. (NYSE:DOW), a world-leading materials science company, have signed a new agreement that paves the way for Dow to use Entos proprietary technology to improve the design and manufacture of these materials.
During the three-year arrangement, Entos will develop machine-learning tools specifically focused on the discovery of catalysts for the manufacture of polyolefins, a common type of polymer that comprises the plastics found in everything from food packaging to bulk containers, sports equipment, automotive parts, cable jacketing, among many other applications.
Because each polyolefin has a unique molecular structure, they demand the design of highly specific catalysts, which are molecules that facilitate a chemical reaction without being consumed by it. The development of new catalysts at Dow has enabled the more efficient manufacture of many new polymer materials.
Entos has developed a software platform for the accurate prediction of catalyst properties, which uses the OrbNet machine-learning approach basedon quantum mechanical descriptors. OrbNet has been shown to provide the accuracy of standard quantum methods like density functional theory, but with a speed that is 1,000 times faster on the same computer hardware. That increase in speed means that Dow researchers can perform virtual experiments on possible catalysts with high throughput.
Entos was launched in April 2020 and has built a team of over 20 scientists and engineers, focused on developing new technologies for molecular discovery. Entos unifies quantum mechanics and machine learning to vastly improve prediction accuracy and data efficiency for accelerating molecular discovery across a host of industrial applications.