Photonic devices based on phase-change materials for neuromorphic computing

Abstract

This paper reviews recent advances in the field of photonic devices integrating phase-change materials (PCMs). In particular, certain PCMs showcase interesting properties such as non-volatile behaviour and ultra-low optical losses that can be leveraged in computing architectures. For example, they can operate as complex-valued weights enabling ultra-low power consumption matrix-vector multiplications. Here, we first review the optical properties of multiple PCM compounds as interesting candidates for photonic computing. Then, we discuss about GeSe-based electrically activated photonic devices that exploit such properties in novel device concepts. Specifically, we present a hybrid-mode interferometer and an asymmetric heater geometry along the propagation direction for electrical PCM actuation. These approaches lead to a higher number of encoded states as well as ultra-low optical insertion losses compared to previous implementations, while retaining a compact footprint and practical driving conditions. Finally, we present experimental results for optically actuated Ge 2 Sb 2 Te 5 (GST225) based waveguides in terms of encoding capabilities and reliability for all-optical reconfigurability and processing.

Multifunctional synaptic learning and neuromorphic computing using crystalline TiO_x/NiO_x heterojunction-based memory devices
Dip Manna, Koushik Mondal, Subhajit Karmakar +5
arXiv8 Oct 2026PreprintarXiv licenceFull textCondensed matterarXiv
Non-Volatile Memory Devices for Neuromorphic Computing
Debasis Das
arXiv29 Oct 2026PreprintCC BY-NC-SA 4.0Full textCondensed matterarXiv
Resistive Switching and Neuromorphic Computing in Metal/Nb:SrTiO₃: Mechanisms, Interface Physics, and Charge Transport
Christopher Broyles, Elizabeth Krenkel, Frank Barrows +2
arXiv14 Sep 2026PreprintarXiv licenceFull textCondensed matterarXiv
A van der Waals Floating-Gate Platform for Broadband Optoelectronic Neuromorphic Computing
Thi Phuong Anh Bach, Sangeun Cho
arXiv26 Oct 2026PreprintCC BY 4.0Full textCondensed matterarXiv
Controlling Switching Evolution in Lead-Free Perovskite-Inspired Chalcogenide Memristors for Neuromorphic Computing
Emmanuel Joseph Shaji, Zhiyuan Li, Srikanth Doddapaneni +9
arXiv30 Sep 2026PreprintarXiv licenceFull textCondensed matterarXiv

Cite

BibTeX RIS

BibTeX

@article{dwivedi2026photonic,
  title = {{Photonic devices based on phase-change materials for neuromorphic computing}},
  author = {Ranjeet Dwivedi and Charles Pourny and Anthony Albanese and Agraj Yadav and Régis Orobtchouk and Gaya Benane and Rajath Sawant and Martina Tomelleri and Yoann Brûlé and Arnaud Rogemont and Aurélien Coillet and Benoit Charbonnier and Benoît Charbonnier and Pierre Noé and Benoit Cluzel and Fabio Pavanello},
  journal = {Frontiers in Photonics},
  year = {2026},
  month = sep,
  doi = {10.3389/fphot.2026.1918762},
  url = {https://hal.science/hal-05782170v1},
  note = {CC BY 4.0},
}

RIS

TY  - JOUR
TI  - Photonic devices based on phase-change materials for neuromorphic computing
AU  - Ranjeet Dwivedi
AU  - Charles Pourny
AU  - Anthony Albanese
AU  - Agraj Yadav
AU  - Régis Orobtchouk
AU  - Gaya Benane
AU  - Rajath Sawant
AU  - Martina Tomelleri
AU  - Yoann Brûlé
AU  - Arnaud Rogemont
AU  - Aurélien Coillet
AU  - Benoit Charbonnier
AU  - Benoît Charbonnier
AU  - Pierre Noé
AU  - Benoit Cluzel
AU  - Fabio Pavanello
PY  - 2026
DA  - 2026/09/30
JO  - Frontiers in Photonics
DO  - 10.3389/fphot.2026.1918762
UR  - https://hal.science/hal-05782170v1
AB  - This paper reviews recent advances in the field of photonic devices integrating phase-change materials (PCMs). In particular, certain PCMs showcase interesting properties such as non-volatile behaviour and ultra-low optical losses that can be leveraged in computing architectures. For example, they can operate as complex-valued weights enabling ultra-low power consumption matrix-vector multiplications. Here, we first review the optical properties of multiple PCM compounds as interesting candidates for photonic computing. Then, we discuss about GeSe-based electrically activated photonic devices that exploit such properties in novel device concepts. Specifically, we present a hybrid-mode interferometer and an asymmetric heater geometry along the propagation direction for electrical PCM actuation. These approaches lead to a higher number of encoded states as well as ultra-low optical insertion losses compared to previous implementations, while retaining a compact footprint and practical driving conditions. Finally, we present experimental results for optically actuated Ge 2 Sb 2 Te 5 (GST225) based waveguides in terms of encoding capabilities and reliability for all-optical reconfigurability and processing.
SN  - 2673-6853
ER  -

Abstract

Related

Cite