The past few decades have seen an increasing demand for miniaturized photonic technologies for guiding, manipulating, and analyzing light in various integrated research and commercial applications, such as telecommunications, astronomy and on-chip biosensing. Spectral analysis has been crucial to many breakthrough innovations and discoveries, leading to significant recent advances in developing miniaturized spectroscopy platforms. However, challenges remain in fully realizing these platforms as integrated, compact, and cost-effective analysis solutions. I will discuss the development of an integrated photonic spectrometer based on top-down imaging of a multi-mode interference (MMI) waveguide combined with convolutional neural network (CNN) analysis. High-performance narrowband and broadband operations in the visible and near-infrared wavelength ranges are demonstrated, achieving a spectral resolution of 0.05 nm. The potential applications of this compact spectrometer are further demonstrated by a 4x4 arrayed spectrometer on the same chip and the spectral analysis of the solar spectrum to detect gas absorption dips using the MMI spectrometer. I will also explore the design improvements for spectroscopy using low-light images, as enhancing spectrometer signals for sub-nanowatt input power levels can reduce the need for expensive photodetectors in applications such as astronomy or molecular spectroscopy. In this study, the sensitivity and dynamic range of the MMI spectrometer are enhanced by 15 dB, enabling the analysis of input light as low as 300 picowatts through selective roughening of the waveguide surface via plasma etching. These results open new avenues for high-performance spectroscopy on a chip.

Event Host: Md Nafiz Amin, Ph.D. Candidate, Electrical & Computer Engineering

Advisor: Holger Schmidt