Deep learning to make nanoscale designs more robust against defects

Optical metasurfaces, ultrathin interfaces made up of uniform nanoscale structures, change the behavior of light waves hitting them to produce effects ranging from unique reflection and transmission properties to lens distortion removal. However, due to the small size of metasurface features, manufacturing defects can significantly reduce performance -- and they are hard to anticipate. A team of Penn State researchers developed a method to account for the effect of small defects before they've occurred to enable designs that can withstand these performance reductions. They published their approach in Nanophotonics in November. "With modern nanofabrication technology, superfine features -- or small structures inside metasurface components -- can be made consistently, but this can affect the processing time," said Ronald Jenkins, an electrical engineering doctoral candidate and first author on the paper.