Electrically Programmable Phase-Change Nanophotonic Circuits Designed for Edge Detection Application

All-optical integrated phase-change memories have most recently attained considerable interest not only in conventional storage field, but also in emerging neuromorphic regime. Its relatively weak scalability and complex programming scenario however posed insurmountable challenges to overwhelm its electronic rival. To combine the advantages of both photonic and electronic phase-change devices, we here proposed physically-realistic electric field and temperature dependent phase-change field and optical models to simulate programming and readout performances of previously designed phase-change nanophotonic circuit integrated with crossbar electrodes and microheaters. Results clearly revealed its performance superiorities such as ultra-low energy consumption, ultra-fast switching speed, and multi-level trait. The ability of using the proposed circuits to realize numeric digit edge detection was also demonstrated via a system level behavior model, paving its promising path towards neuromorphic application.