CRII: SHF: Ultra-fast Simulation and Automated Design of Silicon Photonics Devices

University Of Southern California, Los Angeles CA

Investigators

Abstract

Integrated circuits are responsible for most of modern technology today. While electronic circuits have been sufficient for advancing technology until this point, the challenging computational, energy-efficiency, and communication demands of the 21st century require a radically new paradigm. This can be found in silicon photonics, an emerging technology which allows the manipulation of light by integrating optical devices on the same chips as electrical circuits. The ability to control light on a chip has already led to important technological advancements, including ultra-high-speed wired and wireless communications, lens-free imaging, Light Detection and Ranging (LiDAR), gyroscopes without moving parts, gas sensors, and label-free biosensors. This project seeks to solve device simulation and design difficulties which currently severely limit the potential of silicon photonics, ushering in a new generation of chips with unprecedented levels of performance and design complexity, and paving the way for on-chip optical computation and ultra-fast, efficient communication. Leveraging mathematical and computational modeling, this research will engage and train graduate and undergraduate students in multi-disciplinary fields spanning from mathematics to computer engineering and applied physics. Another objective is to engage high school students from local schools interested in STEM, with a specific focus on women and underrepresented groups. This project will spur the growth of powerful new design tools for silicon photonics which will allow human designers to maximize their efforts designing at the system level, rather than at the device level, allowing them to develop highly intricate systems. Due to their large size, small features, and complexity, photonic devices are very challenging to numerically simulate, and the lack of analytical or even approximate solutions further complicates the design of new devices. Advanced numerical techniques based on boundary integral equation methods will be developed in this project and applied to model photonic devices. Unlike present approaches which necessitate volumetric meshing of devices to be simulated, the project's approach only requires meshing the surfaces or boundaries of devices, leading to a significant reduction of problem size and dramatic increases in speed and CPU and memory efficiency. High-order polynomial functions will be used to approximate the electromagnetic unknowns, leading to an almost exponential rate of convergence to the exact solution with respect to the mesh size, in contrast to present methods which can only achieve linear or quadratic convergence. An optimization framework will be designed and implemented which will be capable of automatically designing new, ready-to-fabricate photonic devices without any human intervention other than specification of desired functionality and performance. The efficacy of the new simulation and optimization platform will then be demonstrated by utilizing the framework to design and evaluate real silicon photonic devices such as wavelength demultiplexers, power splitters, and grating couplers. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.