My research area is in Nano-Photonics and Aritificial Intelligence, and for the recent years I have been trying to find ways to enhance our control over light. Besides the fundamental investigation on photonic theories, I'm also interested in inventing things that can be applied to our daily lifes.

The ability to control light has long been a major scientific and technological goal. In electromagnetic theory, a monochromatic electromagnetic plane wave is characterized (apart from its phase and amplitude) by three fundamental properties: its frequency, its polarization, and its propagation direction. The ability to select light according to each of these separate properties would be an essential step in achieving control over light. 
My research interests during my PhD has been mainly focused on the directional and frequency control of light using nanoscaled-material systems, and their applications in solar energy harvesting, structural colors and displays. 

Recently, I have also been deeply attracted by the power of Artificial Intelligence (AI), and specifically Deep Neural Networks.

Directional Control of Light

Creating a nanoscale-material that can achieve light selection based purely on the angle of propagation is a long-standing scientific challenge. We tailored the overlap of the band gaps of multiple one-dimensional photonic crystals, each with a different periodicity, in such a way as to preserve the characteristic Brewster modes across a broadband spectrum. Our method enables transparency throughout the visible spectrum at one angle—the generalized Brewster angle—and reflection at every other viewing angle.

Here is an interview article featured on MIT Material Processing Center (MPC) about my work on this field.

Reference Publications:

Here is a video that demonstrate the effect of our angular selective work.

YouTube Video

Light Selection Based on Their Propagation Directions

   Light Selection Based on Their Propagation Directions

This work was featured on MIT homepage on March 28th, 2014.

Some nice news article written about this work:

In English:

In Chinese:

A more detailed webpage can be found here.

Spectrum Control of Light with Metasurface

My second research interest is investigating how to control the apparent color of materials using all dielectric nanostructures, or the so-called Physical Colors. Different from conventional paint, which rely on the chemical properties of the dye molecules, physical colors rely solely on the geometry of the nanoscaled-materials. Therefore, with physical colors it is possible to achieve much better reflectance, longer durability and even dynamic tuning of the material's color.

Reference Publications:
  • Y. Shen*, V. Rinnerbauer*, I. Wang, V. Stelmakh, J. D. Joannopoulos and M. Soljačić, "Structural Colors from Fano ResonanceACS Photonics Vol 2, Iss. 1, pp. 27-32 (2015) (Arxiv)
  • E. Regan*, Y. Shen*, J. J. LopezC. Hsu, B. Zhen, J. D. Joannopoulos and M. Soljačić, "Geometrically Protected Resonance Modes and Optical Fano Resonances"
Structural Colors from All-Dielectric Surface Resonator

On-Chip Optical Neural Morphic Computing

The brain, unlike the von Neumann processors found in conventional computers, is very power efficient, extremely effective at certain computing tasks, and highly adaptable to novel situations and environments. Artificial Neural Networks (ANNWs), an area of research that has recently received an explosion of interest, are algorithms that mimic the signal processing architecture of the brain. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection, and many other scientific domains such as drug discovery and genomics.
Now, both the quantity and also the size of data files are growing at a rapid speed, therefore computing speed and the power efficiency is the key on evaluating the performance of any machine learning algorithm.  With the recent advances in quantum optical devices and on-chip nanophotonic circuit fabrication, we reasoned it is possible to design a viable on-chip optical neural network (ONNW) architecture. In this work, we will propose that conventional neural networks architecture can be entirely and equivalently represented by on-chip optical components.