Content area
Programmable nano-photonics have the potential to completely transform a range of emerging applications, including optical computing, optical signal processing, light detecting and ranging, and quantum applications. However, implementing energy-efficient and large-scale systems remains elusive because commonly used programmable photonic approaches are volatile and energy-hungry. Recent results on non-volatile phase-change material integrated photonics present a promising opportunity to create truly programmable nano-photonics. The ability to drastically change the refractive index of the PCMs in a non-volatile fashion allows creating programmable units with zero-static energy. By taking advantage of the electrical control, non-volatile reconfiguration, and zero crosstalk between each unit, PCMs can enable both extremely large-scale integrated photonics and metasurfaces.
In this dissertation, we present our main progress in PCM nano-photonics and discuss the challenges and limitations of this emerging technology. We first demonstrated 2 × 2 electrically programmable units using a well-studied, prototypical PCM Ge2Sb2Te5 and scalable doped silicon PIN heaters. These components exhibit low insertion loss (~2 dB), high extinction ratio (8 dB), and large endurance (> 2,800 cycles), and are critical for large-scale photonic networks and photonic field programmable arrays. We also designed a three-waveguide directional coupler multiple operation levels by adding a detuning parameter between the center and the other waveguides. To further reduce the loss, we explored an emerging PCM Sb2S3 with lower absorption loss and demonstrated phase shifters in both micro-ring resonators and Mach-Zehnder interferometers. An asymmetric directional coupler with two waveguides is also experimentally demonstrated, showing better loss (~1 dB) and extinction ratio (~ 15 dB) than the GST one. Interestingly, the device also supported a stepwise multi-level switching behavior, which offered unique opportunity for deterministic tuning, and we showed 32 levels with excellent repeatability. With this multilevel operation, we demonstrate an application of post-fabrication trimming to correct the phase error in a balanced MZI. Beyond the university level, in-house fabrication, which relies on electron beam lithography and has limited scalability, we also cooperated with Intel and showed a scalable process to put PCMs on commercial 300-mm silicon photonic wafers.
Active metasurfaces are another emerging field in nanophotonics. However, current electrically controlled PCM-based metasurfaces are limited to global amplitude modulation, insufficient for SLMs. We demonstrated an individual-pixel addressable, transmissive metasurface using the low-loss PCM Sb2Se3 and doped silicon nanowire heaters. The nanowires simultaneously form a diatomic metasurface, supporting a high quality-factor (~406) quasi-bound-state-in-the-continuum mode. Global phase-only modulation of ~0.25π (~0.2π) in simulation (experiment) is achieved, showing ten times enhancement. 2π phase shift is further obtained using a guided-mode resonance with enhanced light-Sb2Se3 interaction. Finally, individual-pixel addressability and SLM functionality are demonstrated through deterministic multilevel switching (ten levels) and tunable far-field beam shaping. Our work presents zero-static power transmissive phase-only SLMs, enabled by electrically controlled low-loss PCMs and individual meta-molecule addressable metasurfaces.
As a perspective, we argue that energy efficiency is a more critical parameter than the operating speed for programmable photonics, making PCMs an ideal candidate. This has the potential for a disruptive paradigm shift in the reconfigurable photonics research philosophy, as slow but energy-efficient and large index modulation can provide a better solution for extremely large-scale integrated photonics than fast but power-hungry, small index tuning methods. We also highlight the exciting opportunities to leverage wide bandgap PCMs for visible-wavelength applications, such as quantum photonics and optogenetics, and for rewritable photonic integrated circuits using nanosecond pulsed lasers. The latter can dramatically reduce the fabrication cost of PICs and democratize the PIC manufacturing process for rapid prototyping.