Silicon photonics has grown over the last decade into a well-established platform, with state-of-the-art dedicated foundries in the United States, Europe and Asia. The transparency of silicon at telecom wavelengths, combined with the already refined processes and tools for silicon microfabrication have led to a fast growth of applications utilizing integrated silicon photonics.
Silicon photonic circuits can already be found in datacenters in the form of footprint-efficient optical interconnects. Data giants such as Google, Facebook and Amazon are investing a lot into silicon photonics as a key technology to keep up with the increase in data, relying in the process on major chip companies such as Intel. The technology is also expected to become critically important in the industrial development of quantum computing, A.I. dedicated chips, and high-performance computing.
The number of functions, or computing power increase, requires however photonic circuits with larger complexity. The high confinement of silicon waveguides allows for densely integrated components, but their active reconfiguration relies on thermal tuning, which results in high power consumption and limits the scalability. Ultra-low power components are required for such an upscaling, and MEMS technology fills that gap very well. MEMS-based silicon photonics devices have already shown promise in research, but there is no photonic MEMS platform offered by foundries yet.
At the same time, silicon photonic circuits are expensive for research or product development purposes, with long fabrication times (~ 6-8 months). To facilitate the entry of new actors into the field, and hence increase the growth of the silicon photonics market, reprogrammable chips are needed. Generic reprogrammable chips could be manufactured in large amounts and configured as needed by the customer, which would drastically reduce both the time and price invested in new ideas. This step is the optical counterpart of the successful FPGAs, which resulted in a blooming product development era.
In this context, the European project MORPHIC was born, with the objective of creating a first photonic MEMS platform, demonstrating its potential with a first FPGA-like photonic chip.
MEMS actuation offers much lower power consumption than current thermal tuning mechanism, and MEMS latching can push that even further in reconfigurable chips. No power would be consumed to maintain a different chip configuration, leading to zero-power versatile chips. Power would then only be required when reconfiguring the circuit.
To demonstrate the viability of MEMS for FPGA-like photonic chips, three functions will be tested: an optical switch matrix, a programmable microwave photonics filter; and an optical beam forming network.
The project involves six actors: IMEC (project leader), Tyndall National Institute of technology, Ecole Polytechnique Federale de Lausanne, KTH; and two companies, COMMSCOPE, and VLC Photonics.
Here at MST, our involvement in the project is two-fold. First, we develop and design the silicon photonic MEMS components for the reconfigurable network, with a focus on analog actuation and latching. Second, we bring our knowledge on MEMS hermetic packaging to seal the MEMS building blocks and to ensure stable device performance over extended time periods and under various operating environments.