Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Hydrogen evolution electrocatalysts in high-fold degenerate topological semimetals with chiral structures.

Communications chemistry·2026
Same author

Super-moiré spin textures in twisted two-dimensional antiferromagnets.

Nature nanotechnology·2026
Same author

Deformation-Driven Enhancement of Spin Defect Emission in Hexagonal Boron Nitride.

ACS nano·2026
Same author

Non-volatile tuning of cryogenic silicon photonic micro-ring modulators.

Nature communications·2025
Same author

Probing Vortex Dynamics in 2D Superconductors with Scanning Quantum Microscope.

Physical review letters·2025
Same author

Intermodal microwave-to-optical transduction using silicon-on-sapphire optomechanical ring resonator.

Science advances·2025

Related Experiment Video

Updated: Nov 23, 2025

Demonstration of Equal-Intensity Beam Generation by Dielectric Metasurfaces
09:33

Demonstration of Equal-Intensity Beam Generation by Dielectric Metasurfaces

Published on: June 7, 2019

6.5K

Programmable phase-change metasurfaces on waveguides for multimode photonic convolutional neural network.

Changming Wu1, Heshan Yu2, Seokhyeong Lee1

  • 1Department of Electrical and Computer Engineering, University of Washington, Seattle, WA, 98195, USA.

Nature Communications
|January 5, 2021
PubMed
Summary

Neuromorphic photonics offers faster, more energy-efficient computing for AI. This study introduces a programmable photonic chip using phase-change materials for high-precision matrix-vector multiplication in neural networks.

More Related Videos

Transmission of Multiple Signals through an Optical Fiber Using Wavefront Shaping
09:43

Transmission of Multiple Signals through an Optical Fiber Using Wavefront Shaping

Published on: March 20, 2017

10.1K
Demonstration of Spin-Multiplexed and Direction-Multiplexed All-Dielectric Visible Metaholograms
08:48

Demonstration of Spin-Multiplexed and Direction-Multiplexed All-Dielectric Visible Metaholograms

Published on: September 25, 2020

6.0K

Related Experiment Videos

Last Updated: Nov 23, 2025

Demonstration of Equal-Intensity Beam Generation by Dielectric Metasurfaces
09:33

Demonstration of Equal-Intensity Beam Generation by Dielectric Metasurfaces

Published on: June 7, 2019

6.5K
Transmission of Multiple Signals through an Optical Fiber Using Wavefront Shaping
09:43

Transmission of Multiple Signals through an Optical Fiber Using Wavefront Shaping

Published on: March 20, 2017

10.1K
Demonstration of Spin-Multiplexed and Direction-Multiplexed All-Dielectric Visible Metaholograms
08:48

Demonstration of Spin-Multiplexed and Direction-Multiplexed All-Dielectric Visible Metaholograms

Published on: September 25, 2020

6.0K

Area of Science:

  • Optoelectronics
  • Materials Science
  • Artificial Intelligence

Background:

  • Neuromorphic photonics presents a significant advancement in hardware acceleration, offering superior speed and energy efficiency for machine learning algorithms compared to traditional digital electronics.
  • Integrated photonic networks excel at analog matrix-vector multiplication (MVM), providing high speed and bandwidth density crucial for data-intensive tasks.
  • Phase-change materials are key for nonvolatile programming and in-memory computing in integrated photonic devices, enabling on-chip optical computing.

Purpose of the Study:

  • To demonstrate a novel multimode photonic computing core utilizing programmable mode converters.
  • To leverage phase-change materials for precise control of waveguide spatial modes and high-resolution MVM computation.
  • To showcase the potential of this photonic core for developing efficient optical neural networks for image processing.

Main Methods:

  • Development of an array of programmable mode converters based on on-waveguide metasurfaces composed of the phase-change material Ge$_{2}$Sb$_{2}$Te$_{5}$.
  • Utilizing the refractive index modulation of Ge$_{2}$Sb$_{2}$Te$_{5}$ during phase transition to achieve up to 64 levels of modal contrast for precise control of waveguide spatial modes.
  • Representing matrix elements with 6-bit resolution (positive and negative values) to perform MVM computations essential for neural network algorithms.

Main Results:

  • Demonstration of a photonic computing core capable of precise MVM computation with high-resolution (6-bit) representation of matrix elements.
  • Successful implementation of a prototypical optical convolutional neural network utilizing the photonic core for image processing and recognition tasks with high accuracy.
  • Validation of the device's broad operation bandwidth and compact footprint, indicating suitability for large-scale photonic neural networks.

Conclusions:

  • The developed multimode photonic computing core, based on programmable phase-change material metasurfaces, offers a promising platform for high-throughput optical computing.
  • This technology enables precise, high-resolution matrix-vector multiplication, crucial for advancing neural network hardware.
  • The compact and efficient design paves the way for scalable, next-generation neuromorphic photonic systems with ultrahigh computation capabilities.