Inverse design for compact magneto-optics

Much of the modern world relies on communication using fibre optic cables. These are essentially long tubes of glass through which pulses of light can be sent, transferring information from one end to the other. Filtering and manipulation of the light before or after the fibre optic cable enables the pulses of light to be converted to and from human-readable information. In an analogy with how electronic devices manipulate electrons, such light-manipulating devices work with photons, so their design and characterisation is the field of photonics. In the same way as miniaturisation has dramatically improved the performance of electronics, photonic devices will become more and more commonplace as their dimensions are reduced.

One photonic component that has proved particularly difficult to shrink is the optical isolator, which allows light to propagate in one direction but not in the other. These are used extensively in fibre optical communication and are beginning to find a role in the object detection systems used in self-driving cars (LiDAR). They are typically built using a class of materials exhibiting a phenomenon known as the magneto-optic effect, which can be exploited to allow unidirectional propagation. Attempts to create smaller devices using the same materials have run into significant problems. These are mostly related to some practical issues encountered when very precisely manipulating magneto-optical materials at microscopic scales.

A route around this is to use a material more suited to use in very tiny devices. An obvious candidate is silicon, as the vast existing infrastructure for computer processors means silicon-based manufacturing is very advanced. Unfortunately, silicon has weak magneto-optical properties, so it seems unsuitable for use as an optical isolator.

This project will sidestep this difficulty using a technique known as inverse design, in which the human is removed from the design process. Instead, a computer uses efficient algorithms to determine an optimal structure to achieve a particular goal. This technique has been shown to dramatically increase the performance of all kinds of devices in various contexts. In this project, the team will apply the algorithm in such a way that designs for high-performance, miniaturised optical isolators will be the end result. These will be small enough to be built into compact photonic devices, improving for example the performance of fibre-optical communications or the technology used in automated vehicles.