Optoelectronics Integration Research Group

Research Group Introduction

We conduct research on novel optical signal and information processing technologies that treat spatial, temporal, and quantum information in an integrated manner, centered on optical circuits such as planar lightwave circuits (PLC) and optical metasurface technology. By treating light as a medium for both information and computation, we aim to create next-generation foundations for communication, sensing, and information processing from the perspectives of information and mathematical sciences.

・Optical metasurfaces and imaging applications
・Optical circuit technology for photonic computing

Theme

R & D results

Picked‑up Theme:Optical metasurfaces and imaging applications

・Technology

This technology uses tiny structures at the nanometer scale to carefully control the phase of light over distances smaller than the light's wavelength. This allows us to freely shape the light's wavefront and perform advanced light control that traditional lenses and optical devices cannot achieve. These "metasurfaces" are passive devices--they don't emit light or use electricity to work-- and that's exactly what makes them special. By using only their structure, they take advantage of natural behaviors of light like interference, diffraction, and refraction. This makes it possible to create new types of imaging that were not possible before. We regard this technology as a kind of physical pre-processor for light, where nature itself controls the light before the information is processed.

image

image

                     Optical metasurfaces and optical wavefront control


・Features

Manufacturing Compatibility: Highly compatible with semiconductor fabrication processes. Phase control is achieved simply by varying the width of nanopillars.
Ultra-Precise Optical Manipulation: In addition to lens-like functions, it is capable of phase modulation on a subwavelength scale, theoretically enabling phenomena such as bending light at right angles.


・Application

High-Efficiency Imaging: Successfully demonstrated focusing of RGB light onto separate detectors using metasurfaces instead of conventional lenses and color filters.
Sensitivity Enhancement: Achieved approximately 3 times higher brightness compared to conventional color filter methods, as all light is utilized rather than absorbed.
Expansion to Information Processing: Contributes to the creation of new imaging technologies that perform optical processing directly on light, enabling higher precision and lower latency. This also lays the groundwork for light-based information processing systems.

image

image

New Metasurface‑Based Image Sensor & Spectral Camera

Picked‑up Theme:Optical circuit technology for photonic computing

・Technology

The optical circuits we mainly work on are passive devices--they don't emit light, apply modulation, or generate electrical signals on their own. Instead, they function by allowing light to pass through them. Because of this, the value of an optical circuit depends on how we design the system: what kind of light we use, what information it carries, and how we manipulate it. Being passive also brings important advantages. Since the light simply passes through, it's possible to perform information processing with very low latency and minimal energy consumption. In other words, we aim to create a new kind of information processing that relies on the natural behavior of light itself to carry out computations. Building on technologies we have developed in optical communications--such as waveguides ※1 and interference control--we're working to establish what we call "optical computing." This approach goes beyond just speeding up processing or saving energy. It's about creating new technologies that connect the way information is processed with the fundamental nature of light.

image

・Features

Our optical circuits enable precise control of interference using silica-based Planar Lightwave Circuit technologies and can implement arbitrary unitary operations.※3 By fully exploiting the multiplexing capabilities of light--across wavelength, space, and time--we are developing large-scale photonic computing※2 hardware capable of over 100 trillion operations per second with the ultra-low energy consumption of less than 10⁻¹² joules per operation.



・Application

This technology allows extremely fast and efficient execution of large-scale nonlinear matrix operations, which are bottlenecks in AI ※4 processing. Currently, AI model training can consume hundreds of MWh--more than the lifetime CO₂ emissions of a car. Our approach seeks to fundamentally resolve this challenge by using light itself. Furthermore, by enabling direct computation within optical networks, we envision a future of "optically native" information processing, where communication and computation are seamlessly integrated.
We are currently focused on building core technologies, such as the construction of general-purpose optical circuits for unitary operations and the design of data input/output.



 ※1 Optical waveguide: Structure that confines and guides light (e.g., optical fibers)
 ※2 Photonic computing: Technology that leverages the physical properties of light for information processing
 ※3 Unitary operation: Linear transformation that preserves information (used in quantum and optical computing)
 ※4 AI: Artificial intelligence

image