PhD - Leuven | More than two weeks ago
Apply computer-aided optimization techniques to scale up the bandwidth of silicon photonics based optical links in datacenters
Silicon photonics is a rapidly developing and viable industrial platform for telecom and datacom applications. To guide and govern light at the micro-scale, the photonic integrated chip is comprised of structures with subwavelength features. To improve device specifications the geometry of such subwavelength features can be optimized according to some physical insights. However, all options and possibilities of device performance remain unraveled. Therefore, alongside with well-established semi-analytical methodologies to design photonic building blocks, new approaches are welcome to improve optical components.
Automated design approaches allow free-shape optimization of components when dielectric permittivity varies arbitrarily inside the structure as a function of position. By using automated design, a set of favorable device attributes can be assessed: improving specs, reducing footprint, merging functionalities of several components. Optimization of photonic building blocks suitable for fabrication should be based upon spatial permittivity variation bounded by minimum feature size, selection of materials, given device functionality and design area. However, these new design techniques do not always yield designs that can easily be fabricated. Imec’s state-of-the-art immersion lithography tools can resolve deep submicrometer features, but even then, the optimization process should take care to respect the constraints of the patterning process and the materials available during fabrication. These considerations, and how to address them, define some essential questions and challenges for this PhD project.
Another critical aspect of these automated design methods is that they do not necessarily give an insight in the fundamental operation principles of the resulting optimized component. And because the optimized device geometry is also very sensitive to the initial conditions of the optimization process, this lack of insight could mean that we miss out on the best performing, or most robust design. The PhD student should identify fundamental design patterns and principles that would eventually allow for optimized parametric components.
One way to enable scaling of datarates in optical links is the use of wavelength-division multiplexing (WDM), combining multiple wavelength channels in the same optical waveguide or fiber. But as more wavelength channels are used, the requirements of the multiplexer/demultipler circuits or devices become increasingly demanding. This is where we want to apply these new design methodologies. The driving ambition behind this PhD is to scale WDM mux/demux to 128 channel wavelength using automated design and considering the fabrication facilities at Imec. The designed wavelength mux/demux filters will then be prototyped and tested.
Required background: Master’s degree in EE or Physics, Nanoscience and Nanotechnology, Preferred: Python, experience with automated design
Type of work: Master’s degree in EE or Physics, Nanoscience and Nanotechnology, Preferred: Python, experience with automated design
Supervisor: Wim Bogaerts
Daily advisor: Aliaksandra Ivinskaya
The reference code for this position is 2021-143. Mention this reference code on your application form.