Research

On Progress

Photonic integrated circuits for next-generation computing (optical neural network, data centers, hyper-scale computing)

The advent of parallel computing paradigm that requires large bandwidth has created a need for integrating high performance photonic systems with electronics. Also, new computing fields where photonics can play unique roles are emerging such as optical neural network and hyperscale computing. In our group, we are developing scalable and power-efficient photonic integrated circuits such as silicon photonic switches, tunable directional couplers, and optical phase shifters.

<Chip-scale Coherent Ising Machine>

Combinatorial optimization problems (COPs) are ubiquitous in our society, and play central roles in diverse applications such as drug discovery, integrated circuit design, artificial intelligence, and path planning for autonomous vehicles. Large sets of COPs are difficult to solve even with the state-of-the-art computing technologies. It is also well known that large portion of COPs can be mapped to the Ising model which is a mathematical model for ferromagnetism that describes behaviors of spin networks and their phase transitions. Problems of finding the ground state of the Ising model are called “Ising problems”. The Ising problem can be reduced to the Ising model within polynomial time. Therefore, if one can devise a dedicated hardware that emulates the behavior of Ising model, i.e. Ising machine, there will be a tremendous impact on solving COPs with reasonable amount of resources (e.g. time, energy consumption) for diverse applications. Throughout this project, we will build a chip-scale Ising machines which outperforms the state-of-the-art Ising machines in nearly all aspects.

- T. J. Seok# , N. Quack, S. Han, R. S. Muller, and M. C. Wu*, “Large-scale broadband digital silicon photonic switches with vertical adiabatic couplers,” Optica, Vol. 3, 64-70 (2016)

<Large-scale Tunable Directional Coupler Switches for Optical Computing>

Optical circuit switches (OCSs) are essential building blocks in optical communication networks. Studies have suggested that fast OCSs with microsecond switching time can significantly increase the performance and efficiency of datacenter networks. While recent silicon photonic switches have been reported exhibiting micro/nanosecond response time, exploiting some methods, their efficiency is limited due to the weak perturbation of the refractive index. Moving waveguides with MEMS (micro-electromechanical-system) actuators allow a much stronger optical effect. This opens up new switch architectures that are inherently scalable. We have reported on a new type of MEMS silicon photonic switch built on a low-loss crossbar network.

-S. Han, T. J. Seok, C. -K. Kim, R. S. Muller, and M. C. Wu, “Multicast silicon photonic MEMS switches with gap-adjustable directional couplers,” Optics Express, 2019, 27 (13), 17561-17570.

-S. Han, T. J. Seok, N. Quack, B. W. Yoo, and M. C. Wu, “Large-scale silicon photonic switches with movable directional couplers,” Optica, 2015, 2 (4), 370-375.

<Silicon Photonic Switch for Data Centers>

Cloud computing and big data applications have fueled the rapid growth of datacenters. As the data traffic continues to grow at a fast pace, the majority of the traffic (about 75%) remains within the datacenter. The legacy networks are unable to handle the growing demand of cloud computing. New networks with flatter leaf-spine architecture are increasingly used to facilitate the communication between servers and storages. However, as the per-server link speeds continue to increase, it has become challenging for the CMOS (Complementary Metal-Oxide Semiconductor) switch to provide high radix at high port-bandwidth. Several hybrid datacenter network architectures that use optical switching to augment the electronic packet switching have been proposed. We have reported a new type of silicon photonic switch with MEMS (micro-electromechanical-system) switching mechanism.

-H. Y. Hwang, J. S. Lee, T. J. Seok, A. Forencich, H. R. Grant, D. Knutson, N. Quack, S. Han, R. S. Muller, G. C. Papen, M. C. Wu, and P. O'. Brien, “Flip chip packaging of digital silicon photonics MEMS switch for cloud computing and data centre,” IEEE Photonics Journal, 2017, 9 (3), 1-10.

-M. C. Wu, T. J. Seok, and S. Han, “Silicon photonic switches for datacenters,” Frontiers in Optics, 2016, OSA, FTu1D.6.

<Multicast Switch Optical Computing>

Our group have previously demonstrated several versions of micro-electromechanical-system (MEMS)-based silicon photonic switches. They have desirable characteristics. However, those switches provide 1-to-1 connections only, like most of other optical switches. In many applications that require high bandwidth interconnects between one source to multiple destinations, such as video streaming servers and video conference systems, multicast optical switches are desirable. Some multicast silicon photonic switches have been reported. However, the scale of the switch is small or it has unnecessary high excess optical loss due to its architecture. We reported on a silicon photonic MEMS switch capable of multicasting.

Mid-infrared (mid-IR) wavelength region (2 – 20μm) is so-called “molecular fingerprint region” where vibration energies of molecules lied. By analyzing the absorption spectrum of mid-infrared region, one can extract information on existence, spatial distribution, and structure of the molecules. Therefore, the information can be used for chem/bio analysis such as breath analysis, cancer diagnosis, gas sensing, and pollution monitoring. In our group, we are developing nanophotonic mid-IR lasers for high-performance and portable spectroscopy on chip.

S. Han, D. -G. Kim, J. Hwang, I. H. Do, D. Jeong, Y. -H. Lee, D. -Y. Choi, and H. Lee, “Brillouin lasers based on 11 million-Q on-chip chalcogenide resonators without direct etch process,” International Conference on Optical MEMS and Nanophotonics (OMN), 2019, IEEE.S. Han, D. -G. Kim, J. Hwang, I. H. Do, D. Jeong, Y. -H. Lee, D. -Y. Choi, and H. Lee, “On-chip stimulated Brillouin lasers based on chalcogenide glass resonators with 10 million Q-factor,” Conference on Lasers and Electro-Optics (CLEO), 2019, OSA, paper SM4O.2.