NEO-PGA: Nonvolatile Electro-Optically Programmable Gate Array
Abstract
Large-scale, electronically reconfigurable photonic integrated circuits (PICs) can enable programmable gate array (PGA) to realize extremely fast, arbitrary linear operations, with potential applications in classical and quantum optical information technology. The basic building blocks of existing PGAs are thermally tunable broadband Mach-Zehnder-Interferometers, which pose several limitations in terms of size, power, and scalability. Phase change materials (PCMs), exhibiting large nonvolatile change in the refractive index, can potentially transform these devices, providing at least one order of magnitude reduction in the device size, zero static energy consumption, and minimal cross-talk. Unfortunately, the most well-characterized PCM, namely GeSbTe (GST), has a small band-gap, leading to a large absorptive loss at 1550nm. Moreover, the phase transition generally happens in a binary fashion, i.e., the state is either amorphous or crystalline, precluding continuous tunability. We aim to solve these problems via material and device engineering. By introducing an additional GST-clad silicon waveguide section between two waveguides in a directional coupler, we recently reported only 1dB optical loss, with greater than 10dB rejection. We will use this new structure with other wide band-gap PCMs, including germanium selenide and antimony trisulfide, to further reduce the optical loss. Additionally, we will achieve quasi-continuous tuning by using multiple stripes of PCMs, instead of a continuous film. By selecting number of PCM-stripes to change the phase, the length of the coupling section can be tuned, leading to a tunable beam-splitter. Finally,we will actuate the phase transitions electronically. In phase change memories, PCMs are generally part of the electronic circuit, and during the phase transition, the resistivity changes significantly, limiting the changed volume of the PCM. The problem becomes more severe for wide bandgap PCMs as they have higher resistivity. To circumvent this, we will employ an external heater wrapped around the PCM film and avoid the PCMs in the electronic circuit. We have analyzed the performance of three different heaters: graphene, ITO and p-i-n heaters in silicon. We will initially employ a p-i-n heater due to the simpler fabrication, and later switch to graphene or ITO based heaters, which provide better performance. With these developed low-loss quasi-analog switches,we will implement a PGA. This PGA can be used as a vector-matrix multiplier to build an optical artificial neural network (ANN) that outperforms its electronic counterpart, in terms of speed and energy. Optical ANNs will have far-reaching impact on real time signal processing, with applications in autonomous navigation and brain-machine interface. Additionally, such a nonvolatile optical PGA can be used to route single photons on a chip to connect different functional quantum devices. Thus, the PGA will add redundancy to the PIC and make the photonic system-on-chip robustections in device fabrication and device failures. The resulting PGA will enable novel computing modalities with profound impact on a wide range of Navy-related electromagnetic warfare applications. A successful completion of the proposed research will advance the Electromagnetic Materials and EO/IR Sensors and Sensor Processing Program of the ONR. Results of this ONR YIP will be transitioned to DoD research labs, including NRL and AFRL (with whom the PI already has active collaborations), and companies, such as Intel inoD forums, in addition to the venues of peer reviewed journal publications, conference presentations, and workshops.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 31, 2020
- Source ID
- N000142012657
Entities
People
- Arka Majumdar
Organizations
- Office of Naval Research
- United States Navy
- University of Washington