Related Experiment Video
Updated: May 5, 2026

09:36
Characterization of Anisotropic Leaky Mode Modulators for Holovideo
Published on: March 19, 2016
7.9K
Ultra-broadband all-optical nonlinear activation function enabled by MoTe2/optical waveguide integrated devices
Chenduan Chen1, Zhan Yang2,3, Tao Wang1
1Zhejiang Lab, Hangzhou, Zhejiang, 311121, China.
Nature Communications
|October 19, 2024
Summary
Researchers developed a novel all-optical nonlinear activation function using 2D MoTe2 and optical waveguides. This breakthrough enables faster, more efficient optical neural networks with high accuracy in image recognition tasks.
Area of Science:
- Photonics and Materials Science
- Integrated Optics
- Artificial Intelligence Hardware
Background:
- All-optical nonlinear activation functions (NAFs) are essential for optical neural networks (ONNs).
- Current on-chip NAFs struggle with integration, latency, power consumption, and high activation thresholds.
- Advancements in integrated ONNs necessitate improved NAFs for efficient computation.
Purpose of the Study:
- To develop an integrated nonlinear optical activator for advanced ONNs.
- To overcome limitations of existing all-optical NAFs.
- To demonstrate the potential of 2D materials in photonic devices.
Main Methods:
- Butt-coupling integration of 2D Molybdenum Ditelluride (MoTe2) with optical waveguides (OWGs).
- Characterization of the activator's ultra-broadband response (visible to near-infrared).
- Application of the developed NAFs in simulating MNIST and CIFAR-10 image recognition.
Main Results:
- Achieved an ultra-broadband response from visible to near-infrared wavelengths.
- Demonstrated a low activation threshold (0.94 μW) and ultra-fast response rate (2.08 THz).
- Attained high accuracy in image recognition: 97.6% for MNIST and 94.6% for CIFAR-10.
Conclusions:
- The developed nonlinear optical activator shows significant potential for high-density, low-threshold, ultra-broadband 3D ONNs.
- Integration of 2D materials offers a promising route for next-generation optical computing.
- This work paves the way for practical implementation of efficient ONNs.

