Researchers at Seoul National University and the University of Seoul have demonstrated a programmable photonic chip that delays light on demand, according to a study published July 21, 2026, in the journal Advanced Science (DOI: 10.1002/advs.76378). The work targets a documented bottleneck in optical computing.

What does the chip actually do?

The device controls optical signal speed and shape through coupled-resonator-induced transparency, or CRIT. CRIT is an interference effect among several optical resonators. Conventional CRIT designs lock in their delay characteristics at fabrication time, the researchers noted.

Professor Namkyoo Park of the SNU Department of Electrical and Computer Engineering stated: "This research is significant in that it proposes a new design principle that allows the flow of light within photonic integrated circuits to be reconfigured as needed, greatly enhancing design flexibility." Park led the study alongside Professor Sunkyu Yu of SNU and Professor Xianji Piao of the University of Seoul.

How does the design differ from prior CRIT work?

The group treated two CRIT optical states, called the bright mode and the dark mode, as one unified parameter. The structure added two controllable loop couplers to the resonator chain. That formulation lets engineers adjust bandwidth, delay length, and pulse shape after fabrication has finished. Earlier devices could not be reconfigured once manufactured.

Co-first author Dr. Seungkyun Park, affiliated with the InnoCORE PICORE Center at KAIST, said: "Through this study, we realized that reinterpreting conventional photonic resonator physics from a different perspective can serve as a starting point for discovering new functionalities in photonic integrated circuits." Co-first author and Ph.D. student Beomjoon Chae added that experimental validation is the next step.

What did the simulations show, and what did they not?

The team modeled the device on a silicon nitride (Si3N4) photonic integrated circuit platform using three-dimensional electromagnetic simulations. They included material losses, resonator quality differences, backscattering, coupling fluctuations, loop coupler phase errors, and thermal crosstalk.

The simulations indicate that delay times can shift without reducing processing performance. Frequency conversion also occurred inside the same structure, removing one external component. Real silicon parts have not yet been fabricated and measured, a gap the authors acknowledge in the paper.

Where does the work sit commercially?

The chip is framed by the authors as a software-defined optical building block. A single programmable device could replace separate delay lines, buffers, and frequency converters inside AI-server and data-center hardware. Energy savings claims rest on simulation only, not on bench data.

The team also names autonomous driving, next-generation communications, and quantum technologies as longer-term targets. Each sits beyond the current experimental stage, per the team's own roadmap. Funding came from South Korea's Ministry of Science and ICT through three programs: the Innovative Research Center (IRC), the Basic Research Laboratory (BRL), and the Young Researcher Program.

What comes next for the program?

Park said the group plans to scale the design into large-scale programmable photonic integrated circuits based on silicon photonics and photonic AI technologies. Co-first authors Park and Chae will pursue practical device fabrication and experimental validation in parallel.

Whether the CRIT approach moves from simulation to measured silicon will determine its commercial weight inside next-generation AI-server optics. The evidence so far supports design flexibility, not yet volume manufacturability.