Photonic chips have revolutionized technology for handling large amounts of data. Alone or in conjunction with conventional electronic circuits, these laser-driven devices transmit and process information at the speed of light, making them a promising solution for data-intensive applications of artificial intelligence.
Light creates a computational network on an unpatterned semiconductor wafer. The Feng Lab team’s work enables on-chip processing without the use of lithographic etching, demonstrating cheaper and easier manufacturing and superior precision for AI applications. (Image: Penn Engineering Today)
In addition to unmatched speed, photonic circuits consume significantly less energy than electronic circuits. Electrons move relatively slowly through the hardware, colliding with other particles and generating heat, while photons flow without losing energy and thus generate no heat at all. Integrated photonics, free of the energy loss burden inherent in electronics, is poised to play a leading role in sustainable computing.
Photonics and electronics rely on a stencil-like technique known as lithography. In this technique, intense light is passed through a patterned mask to transfer a circuit design onto a semiconductor wafer by selectively exposing and removing photosensitive material to define circuit elements and connect them sequentially. To do.
Photonic chips do not use the ever-shrinking layered trench-forming transistors of electronic chips, but their complex lithographic patterning guides laser beams into coherent circuits, forming photonic networks that can run computational algorithms. .
Now, researchers from the Faculty of Engineering and Applied Sciences offer programmable on-chip information processing without the use of lithography, delivering the speed of photonics enhanced with superior accuracy and flexibility for AI applications. I created a new photonics device.
Offering unparalleled optical control, the device consists of spatially distributed optical gain and loss. Lasers irradiate semiconductor wafers directly without requiring a defined lithography path.
Mr. Liang Feng is a professor in the Faculty of Materials Science and Engineering and the Faculty of Electrical System Engineering and holds a Ph.D. Student Tianwei Wu and postdoctoral researchers Zihe Gao and Marco Menarini introduced microchips in a study published in 2008. natural photonics.
“However, photonic chips for machine learning applications face the obstacles of complex manufacturing processes that have fixed lithographic patterning, limited reprogrammability, are prone to error and damage, and are expensive.” says Feng. “By eliminating the need for lithography, we are creating a new paradigm. and provide ultimate reconfigurability.”
This story is by Devora Fischler. For more information, see Penn Engineering Today.
