Loading…

Wavelength-shift-free racetrack resonator hybrided with phase change material for photonic in-memory computing

The photonic in-memory computing architecture based on phase change materials (PCMs) is increasingly attracting widespread attention due to its high computational efficiency and low power consumption. However, PCM-based microring resonator photonic computing devices face challenges in terms of reson...

Full description

Saved in:
Bibliographic Details
Published in:Optics express 2023-06, Vol.31 (12), p.18840-18850
Main Authors: Zhu, Honghui, Lu, Yegang, Cai, Linying
Format: Article
Language:English
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The photonic in-memory computing architecture based on phase change materials (PCMs) is increasingly attracting widespread attention due to its high computational efficiency and low power consumption. However, PCM-based microring resonator photonic computing devices face challenges in terms of resonant wavelength shift (RWS) for large-scale photonic network. Here, we propose a PCM-slot-based 1 × 2 racetrack resonator with free wavelength shift for in-memory computing. The low-loss PCMs such as Sb Se and Sb S are utilized to fill the waveguide slot of the resonator for the low insertion (IL) and high extinction ratio (ER). The Sb Se -slot-based racetrack resonator has an IL of 1.3 (0.1) dB and an ER of 35.5 (8.6) dB at the drop (through) port. The corresponding IL of 0.84 (0.27) dB and ER of 18.6 (10.11) dB are obtained for the Sb S -slot-based device. The change in optical transmittance of the two devices at the resonant wavelength is more than 80%. No shift of the resonance wavelength can be achieved upon phase change among the multi-level states. Moreover, the device exhibits a high degree of fabrication tolerance. The proposed device demonstrates ultra-low RWS, high transmittance-tuning range, and low IL, which provides a new scheme for realizing an energy-efficient and large-scale in-memory computing network.
ISSN:1094-4087
1094-4087
DOI:10.1364/OE.489525