Graduate college students Eunice Aissi and Alexander Siemenn, SM ’21, who reported on the work with colleagues together with professor of mechanical engineering Tonio Buonassisi, used the approach to research perovskites, supplies which have nice promise for photo voltaic cells however are inclined to degrade rapidly. About 70 samples—every with a barely completely different composition—have been deposited on a single slide that was then scanned with a hyperspectral digicam, which captures a lot richer visible info than a human can course of. With this information, one of many algorithms they developed was capable of compute the band hole for 3 slides of samples in a complete of six minutes—a course of that might take a human skilled a number of days.
To check for stability, the group positioned the slide in a chamber wherein they different situations equivalent to humidity, temperature, and light-weight publicity. They photographed the samples with a typical digicam each 30 seconds for 2 hours and used a second algorithm to estimate how they modified shade over time, indicating the diploma to which they degraded within the completely different environments. It took 20 minutes to research 48,000 photos.
The last word objective is an autonomous lab, says Aissi: “The entire system would enable us to present a pc a supplies downside, have it predict potential compounds, after which run 24-7 making and characterizing these predicted supplies till it arrives on the desired resolution.”