by Jennifer Evans
Aron Ahmadia, originally from Hilo, Hawaii, is working on his Ph.D. under Prof. David Keyes, specializing in computational science with applications in high performance computing. Working with IBM’s computational scaling team for a 3-month term, he tackled the task of redesigning lithography optimization algorithms and code to run on the world’s fourth fastest supercomputer, Blue Gene W at the Thomas J. Watson research facility in Yorktown.
“I worked on computational techniques to improve the mask design process for semiconductor chips,” he says, “taking current technology levels and advancing them so IBM can pack denser and denser chips and make a more complicated and faster processor.”
Lithography, as Ahmadia explained, is similar to a photographic process. You have a light source and a mask which blocks or transmits certain parts of the light source. A photo sensitive material then captures the image of what the mask produces.
“One way to think about what I did is if you have finger puppets on the wall and you’re using a flashlight to cast a finger puppet. Your hand is the mask and the shadow on the wall is the final image being printed on the computer chip.”
Ahmadia explained that just as you can vary what your hand is doing to get the shadow you want on the wall, he and his team are trying figure out what the optimal mask is to get the ideal chip. “Of course, in computational lithography, there are several trillion fingers to wiggle.”