-
Sub-milliwatt threshold power and tunable-bias all-optical nonlinear activation function using vanadium dioxide for wavelength-division multiplexing photonic neural networks
Authors:
Jorge Parra,
Juan Navarro-Arenas,
Pablo Sanchis
Abstract:
The increasing demand for efficient hardware in neural computation highlights the limitations of electronic-based systems in terms of speed, energy efficiency, and scalability. Wavelength-division multiplexing (WDM) photonic neural networks offer a high-bandwidth, low-latency alternative but require effective photonic activation functions. Here, we propose a power-efficient and tunable-bias all-op…
▽ More
The increasing demand for efficient hardware in neural computation highlights the limitations of electronic-based systems in terms of speed, energy efficiency, and scalability. Wavelength-division multiplexing (WDM) photonic neural networks offer a high-bandwidth, low-latency alternative but require effective photonic activation functions. Here, we propose a power-efficient and tunable-bias all-optical nonlinear activation function using vanadium dioxide (VO2) for WDM photonic neural networks. We engineered a SiN/BTO waveguide with a VO2 patch to exploit the phase-change material's reversible insulator-to-metal transition (IMT) for nonlinear activation. We conducted numerical simulations to optimize the waveguide geometry and VO2 parameters, minimizing propagation and coupling losses while achieving a strong nonlinear response and low-threshold activation power. Our proposed device features a sub-milliwatt threshold power, a footprint of 5 $μ$m, and an ELU-like activation function. Temporal dynamics show a rise time as low as 5 $μ$s. Moreover, the bias of our device could be thermally tuned, improving the speed and power efficiency. On the other hand, performance evaluations using the CIFAR-10 dataset confirmed the device's potential for convolutional neural networks (CNN). Our results show that a hybrid VO2/SiN/BTO platform could play a prominent role in the path towards the development of high-performance photonic neural networks.
△ Less
Submitted 23 July, 2024;
originally announced July 2024.
-
Nonvolatile reconfigurable polarization rotator at datacom wavelengths based on a Sb2Se3/Si waveguide
Authors:
Jorge Parra,
Miroslavna Kovylina,
Amadeu Griol,
Pablo Sanchis
Abstract:
Silicon photonics has become a key platform for photonic integrated circuits (PICs) due to its high refractive index and compatibility with complementary metal-oxide-semiconductor manufacturing. However, the inherent birefringence in silicon waveguides requires efficient polarization management. Here, we report a reconfigurable polarization rotator (PR) using a Sb2Se3/Si waveguide operating at dat…
▽ More
Silicon photonics has become a key platform for photonic integrated circuits (PICs) due to its high refractive index and compatibility with complementary metal-oxide-semiconductor manufacturing. However, the inherent birefringence in silicon waveguides requires efficient polarization management. Here, we report a reconfigurable polarization rotator (PR) using a Sb2Se3/Si waveguide operating at datacom wavelengths (1310 nm), providing nonvolatile switching with zero static power consumption. The polarization conversion relies on the interference of hybrid electric-magnetic (EH) modes, which can be reconfigured by changing the Sb2Se3 state between amorphous and crystalline. Our experimental device exhibits a polarization conversion efficiency (PCE) and a polarization extinction ratio (PER) as high as -0.08 dB and 17.65 dB, respectively, in a compact footprint of just 21$μ$m length. Therefore, the proposed reconfigurable PR offers a compact and energy-efficient solution for polarization management in silicon photonics, with potential applications in data communication networks and emerging applications benefiting from polarization information encodings, such as optical neural networks and quantum computing.
△ Less
Submitted 3 February, 2025; v1 submitted 5 July, 2024;
originally announced July 2024.
-
Ultra-high endurance silicon photonic memory using vanadium dioxide
Authors:
Juan José Seoane,
Jorge Parra,
Juan Navarro-Arenas,
María Recaman,
Koen Schouteden,
Jean Pierre Locquet,
Pablo Sanchis
Abstract:
Silicon photonics arises as a viable solution to address the stringent resource demands of emergent technologies, such as neural networks. Within this framework, photonic memories are fundamental building blocks of photonic integrated circuits that have not yet found a standardized solution due to several trade-off among different metrics such as energy consumption, speed, footprint, or fabricatio…
▽ More
Silicon photonics arises as a viable solution to address the stringent resource demands of emergent technologies, such as neural networks. Within this framework, photonic memories are fundamental building blocks of photonic integrated circuits that have not yet found a standardized solution due to several trade-off among different metrics such as energy consumption, speed, footprint, or fabrication complexity, to name a few. In particular, a photonic memory exhibiting ultra-high endurance performance (> 10^6 cycles) has been elusive to date. Here, we report an ultra-high endurance silicon photonic memory using vanadium dioxide (VO_2) exhibiting a record cyclability of up to 10^7 cycles without degradation. Moreover, our memory features an ultra-compact footprint below 5 μm with potential for nanosecond and picojoule programming performance. Our silicon photonic memory could find application in emerging photonic applications demanding high number of memory updates such as photonic neural networks with in-situ training.
△ Less
Submitted 3 February, 2025; v1 submitted 15 March, 2024;
originally announced March 2024.
-
Temporal dynamics of all-optical switching in hybrid VO2/Si waveguides
Authors:
Jorge Parra,
Todora Ivanova,
Mariela Menghini,
Pia Homm,
Jean-Pierre Locquet,
Pablo Sanchis
Abstract:
Vanadium dioxide (VO2) is one of the most promising materials for developing hybrid photonic integrated devices (PICs). However, despite switching times as low as a few femtoseconds have been reported, the all-optical temporal dynamics of VO2 embedded in a waveguide using an in-plane optical signal remain still hidden. Here, we experimentally investigate this behavior in hybrid VO2/Si waveguides b…
▽ More
Vanadium dioxide (VO2) is one of the most promising materials for developing hybrid photonic integrated devices (PICs). However, despite switching times as low as a few femtoseconds have been reported, the all-optical temporal dynamics of VO2 embedded in a waveguide using an in-plane optical signal remain still hidden. Here, we experimentally investigate this behavior in hybrid VO2/Si waveguides by using pump-probe measurements at telecom wavelengths. Our results show switching times in the micro and nanosecond range, suggesting that the phase transition is triggered thermally from the light absorbed by the VO2 and the temporal response is governed by thermal conductive dynamics. By properly engineering the optical pulse, we prospect switching times of nanoseconds with an energy consumption of a few nanojoules. Our results unveil a new temporal dynamic that would be useful for developing future all-optical VO2 photonic integrated devices.
△ Less
Submitted 22 July, 2020;
originally announced July 2020.
