Photonic Inverse Design of On-Chip Microresonators

The automation of device design enabled by optimization and machine learning techniques has been transformative for photonics. While this automation has been successful for nonresonant devices, automated photonic design has remained elusive for resonant devices, key elements for on-chip communication technologies of biosensing and quantum optics, due to their highly nonconvex optimization landscapes. We propose a framework that solves this problem by mapping the design of photonic resonators to a set of nonresonant design problems. We theoretically and experimentally demonstrate this framework and show flexible dispersion engineering, a quality factor beyond 2 million on silicon-on-insulator with single-mode operation, and selective wavelength-band operation.

Automated summary

The automation of device design in photonics has been revolutionary, but the design of resonant devices has remained challenging due to their complex optimization landscapes. In this study, we propose a framework that maps the design of photonic resonators to nonresonant design problems, enabling flexible dispersion engineering and high-quality operation. The effectiveness of this framework is demonstrated both theoretically and experimentally.