14 January 2022 in Optical Materials Express
The unique optical properties of phase change materials (PCMs) can be exploited to develop efficient reconfigurable photonic devices. Here, we design, model, and compare the performance of programmable 1 × 2 optical couplers based on: Ge2Sb2Te5, Ge2Sb2Se4Te1, Sb2Se3, and Sb2S3 PCMs. Once programmed, these devices are passive, which can reduce the overall energy consumed compared to thermo-optic or electro-optic reconfigurable devices. Of all the PCMs studied, our ellipsometry refractive index measurements show that Sb2S3 has the lowest absorption in the telecommunications wavelength band. Moreover, Sb2S3 -based couplers show the best overall performance, with the lowest insertion losses in both the amorphous and crystalline states. We show that by growth crystallization tuning at least four different coupling ratios can be reliably programmed into the Sb2S3 directional couplers. We used this effect to design a 2-bit tuneable Sb2S3 directional coupler with a dynamic range close to 32 dB. The bit-depth of the coupler appears to be limited by the crystallization stochasticity.
25 May 2022 in Nanophotonics
We demonstrate a passive all‐chalcogenide all‐optical perceptron scheme. The network’s nonlinear activation function (NLAF) relies on the nonlinear response of Ge 2 Sb 2 Te 5 to femtosecond laser pulses. We measured the sub‐picosecond time‐resolved optical constants of Ge 2 Sb 2 Te 5 at a wavelength of 1500 nm and used them to design a high‐speed Ge 2 Sb 2 Te 5 ‐tuned microring resonator all‐optical NLAF. The NLAF had a sigmoidal response when subjected to different laser fluence excitation and had a dynamic range of −9.7 dB. The perceptron’s waveguide material was AlN because it allowed efficient heat dissipation during laser switching. A two‐temperature analysis revealed that the operating speed of the NLAF is ns. The percepton’s nonvolatile weights were set using low‐loss Sb 2 S 3 ‐tuned Mach Zehnder interferometers (MZIs). A three‐layer deep neural network model was used to test the feasibility of the network scheme and a maximum training accuracy of 94.5% was obtained. We conclude that combining Sb 2 S 3 ‐programmed MZI weights with the nonlinear response of Ge 2 Sb 2 Te 5 to femtosecond pulses is sufficient to perform energy‐efficient all‐optical neural classifications at rates greater than 1 GHz.
1 September 2023 in ACS Photonics