Episode 20: Resistance of analog deep learning device responds in ~5 nanoseconds

In this podcast episode, MRS Bulletin’s Sophia Chen interviews Murat Onen, a postdoctoral researcher at the Massachusetts Institute of Technology, about analog deep learning that could help lower the cost of training artificial intelligence (AI). The programmable analog device stores information in the same place where the information is processed. The resistor’s main material is tungsten oxide, which can be reversibly doped with protons from an electrolyte material known as phosphosilicate glass, or PSG, layered on top of the tungsten oxide. Palladium is above the PSG layer, which is a reservoir for the protons when they are shuttled out of the tungsten oxide to make it more resistive. “When protons get in, it becomes more conductive. When the protons go out, it becomes less conductive,” says Onen. The resistance of this device responds in about 5 ns. This work was published in a recent issue of Science (doi:10.1126/science.abp8064).