Join us for the next episode of Ansys’ photonics webinar series for an overview of photonic inverse design (PID), along with new features available with 2020 R2. PID is an optimization method that empowers designers to automatically attain design improvements quickly and reliably. It works by simultaneously optimizing design parameters for a set of manufacturing and operating conditions to improve performance, reduce the product’s footprint and make designs more robust for manufacturing.
Ansys Lumerical’s implementations of PID are based on the adjoint method, an efficient, gradient-based optimization in which runtime is independent of the number of design parameters. This makes it well-suited for even the most challenging designs having hundreds to thousands of design parameters.
Photonic Inverse Design Variants Meet Different Design Challenges
Ansys Lumerical’s popular finite difference time domain (FDTD) and varFDTD products work with two variants of photonic inverse design, both accessible via Ansys Lumerical’s Python application programming interface, which is packaged with all products. The parameterized variant is based on an open source implementation originally developed at the University of California, Berkeley. It is well-suited to designs whose general shape is known with a moderate set (less than 100) of parameters to optimize.
The topological variant treats the design as a pixel map, considering every pixel as a design parameter. Each pixel is first optimized to a greyscale mapping and then binarized to become either guiding or cladding. This variant is computationally more complex but enables users to tackle the most challenging problems with nonintuitive solutions.
Tackle Grating Coupler Design with Photonic Inverse Design
The grating coupler is potentially the most complex component to design for photonics designers. Grating couplers play an important role in the coupling between silicon photonics devices and fiber optics. Coupling loss is a key pain point engineers face when developing optical transceivers, and tradeoffs are often constrained by process parameters such as etch depth and minimum feature size.
A leading photonics foundry recently developed SiPh grating couplers using the photonic inverse design method with Lumerical’s FDTD and varFDTD. PID was configured to automatically optimize a parameterized 2D model of the grating coupler in FDTD with over 50 design parameters. The optimization goal was set to minimize insertion loss over the entire band while adhering to minimum feature constraints enforced to ensure manufacturability. The result was a manufacturing-robust design with a footprint 20X smaller than the original, traditionally designed, grating coupler. The optimization reduced manufacturing costs and the risk of respins.
For a detailed description and demonstration of Ansys Lumerical’s Photonic Inverse Design workflows, register for the upcoming webinar Ansys 2020 R2: Improve Productivity with Photonic Inverse Design.