SBIR Phase II: Automating the Design of Photonic Devices using Cloud-Based Computational Electromagnetics

点击翻译按钮获取中文摘要

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project is to provide quick and cost-effective access to anyone seeking cutting-edge, electromagnetic, simulation tools for research and development. This will involve automating the most technically intensive tasks required to accurately, reliably, and efficiently design and prototype photonic and optoelectronic devices. Current simulation tools require everything to be set up by hand which, even for trained experts, is time consuming and error prone. Also, the computational design and prototyping of photonic devices involves too much trial and error. Our turn-key, on-demand, cloud platform will: (1) reduce design time by 50%, (2) lower the unit cost of simulations by an order of magnitude, and (3) enable first-time users with no prior training in electromagnetic simulations to get up and running with manufacturing-ready designs in a matter of weeks not months. Our advanced simulation tools will facilitate enterprises and entrepreneurs with bringing products to market in a variety of industries critical to national security, health, and education. Easy-to-use and affordable design tools will also reduce the need to fabricate and test as many iterations thereby conserving environmental resources. This is the next generation of computer-aided design (CAD) tools to accelerate photonics innovation and discovery. The proposed project involves the development of a turn-key, electromagnetic, simulation platform which automates complex, multi-step tasks involved in the design and prototyping of photonic and optoelectronic devices. Photonics, the science of light, underlies critical technologies in telecommunications, networking, photovoltaics, biomedicine, photolithography, imaging, displays, and solid-state lighting. Phase II involves automating the deployment of finite difference time-domain (FDTD) simulations in four key areas: (1)large-scale shape optimization for devices involving tens to hundreds of degrees of freedom, (2) sensitivity analysis to assess the impact on device performance of manufacturing errors and predicting manufacturing yields, (3) launching simulation jobs using on-demand, scalable, high-performance computing (HPC) in the public cloud, and (4) seamlessly importing and exporting planar device geometries based on the standard Graphic Database System (GDSII) file format widely supported by electronic design automation (EDA) tools and semiconductor foundries. This will be made possible by combining advances in machine learning, nonlinear optimization, and computational electromagnetics.