Programmable phase-change metasurfaces on waveguides for multimode photonic convolutional neural network

Nature Communications, Journal Year: 2021, Volume and Issue: 12(1)

Published: Jan. 4, 2021

Neuromorphic photonics has recently emerged as a promising hardware accelerator, with significant potential speed and energy advantages over digital electronics, for machine learning algorithms such neural networks of various types. Integrated photonic are particularly powerful in performing analog computing matrix-vector multiplication (MVM) they afford unparalleled bandwidth density data transmission. Incorporating nonvolatile phase-change materials integrated devices enables indispensable programming in-memory capabilities on-chip optical computing. Here, we demonstrate multimode core consisting an array programable mode converters based on metasurface made materials. The programmable utilize the refractive index change material Ge-Sb-Te during phase transition to control waveguide spatial modes very high precision up 64 levels modal contrast. This contrast is used represent matrix elements, 6-bit resolution both positive negative values, perform MVM computation network algorithms. We convolutional that can image processing classification tasks accuracy. With broad operation compact device footprint, demonstrated toward large-scale processor high-throughput networks.

Language: Английский

---

In-memory photonic dot-product engine with electrically programmable weight banks

Nature Communications, Journal Year: 2023, Volume and Issue: 14(1)

Published: May 20, 2023

Abstract Electronically reprogrammable photonic circuits based on phase-change chalcogenides present an avenue to resolve the von-Neumann bottleneck; however, implementation of such hybrid photonic–electronic processing has not achieved computational success. Here, we achieve this milestone by demonstrating in-memory dot-product engine, one that decouples electronic programming materials (PCMs) and computation. Specifically, develop non-volatile electronically PCM memory cells with a record-high 4-bit weight encoding, lowest energy consumption per unit modulation depth (1.7 nJ/dB) for Erase operation (crystallization), high switching contrast (158.5%) using non-resonant silicon-on-insulator waveguide microheater devices. This enables us perform parallel multiplications image superior contrast-to-noise ratio (≥87.36) leads enhanced computing accuracy (standard deviation σ ≤ 0.007). An system is developed in hardware convolutional recognizing images from MNIST database inferencing accuracies 86% 87%.

Language: Английский

---

Non-Volatile Reconfigurable Silicon Photonics Based on Phase-Change Materials

IEEE Journal of Selected Topics in Quantum Electronics, Journal Year: 2021, Volume and Issue: 28(3), P. 1 - 17

Published: Oct. 20, 2021

The traditional ways of tuning a silicon photonic network are mainly based on the thermo-optic effect or free carrier dispersion. drawbacks these methods volatile nature and extremely small change in complex refractive index (Δn<0.001). In order to achieve low energy consumption smaller footprint for applications such as memories, optical computing, programmable gate array, neural network, it is essential that two states system exhibit high contrast remain non-volatile. Phase materials (PCMs) Ge ₂ Sb Te xmlns:xlink="http://www.w3.org/1999/xlink">5 provide an excellent solution, thanks drastic between which can be switched reversibly non-volatile fashion. Here, we review recent progress field reconfigurable photonics PCMs. We start with general introduction material properties PCMs have been exploited integrated discuss their operating wavelengths. various switches built upon reviewed. Lastly, PCM-based circuits potential future directions this field.

Language: Английский

---

A Review of Capabilities and Scope for Hybrid Integration Offered by Silicon-Nitride-Based Photonic Integrated Circuits

Sensors, Journal Year: 2022, Volume and Issue: 22(11), P. 4227 - 4227

Published: June 1, 2022

In this review we present some of the recent advances in field silicon nitride photonic integrated circuits. The focuses on material deposition techniques currently available, illustrating capabilities each technique. then expands functionalisation platform to achieve nonlinear processing, optical modulation, nonvolatile memories and integration with III-V materials obtain lasing or gain capabilities.

Language: Английский

---

Non‐Volatile Optical Switch Element Enabled by Low‐Loss Phase Change Material

Advanced Functional Materials, Journal Year: 2023, Volume and Issue: 33(42)

Published: June 17, 2023

Abstract Mach–Zehnder interferometers (MZIs) integrated with phase‐change materials have attracted great interest due to their low power consumption and ultra‐compact size, which are favored for reconfigurable photonic processors. However, they suffer from a optical extinction ratio limited switching cycles high material loss poor reversible repeatability caused by degradation. Here non‐volatile electrically 2 × MZI low‐loss Sb Se 3 encapsulated in Al O layers is demonstrated. The phase change actuated forward‐biased silicon p‐i‐n diode. switch more than 20 dB the ‐based shifter. By dividing patch into small sub‐cells restrict reflow, 10 000 6‐bit multilevel states achieved programming electrical pulses. Its non‐volatility, endurance, fine‐tuning capability makes device promising large‐scale low‐power

Language: Английский

---

Nonvolatile Multilevel Switching of Silicon Photonic Devices with In₂O₃/GST Segmented Structures

Advanced Optical Materials, Journal Year: 2023, Volume and Issue: 11(8)

Published: Feb. 13, 2023

Abstract Reconfigurable silicon photonic devices are widely used in numerous emerging fields such as optical interconnects, neural networks, quantum computing, and microwave photonics. Currently, phase change materials (PCMs) have been extensively investigated promising candidates for building switching units due to their strong refractive index modulation. Here, nonvolatile multilevel of with Ge 2 Sb Te 5 (GST) is demonstrated In O 3 transparent microheaters that compatible diverse material platforms. With GST integrated on the waveguides Mach‐Zehnder interferometers (MZIs), repeatable reversible modulation achieved by electro‐thermally induced transitions. Particularly, segmented unit proposed be capable producing about one order magnitude larger temperature gradient than nonsegmented unit, resulting up 64 distinguishable levels 6‐bit precision, fine‐tuning voltage pulses push precision even further, 7‐bit, or 128 levels. The capability precise phase‐change crucial further facilitate development reconfigurable switches variable attenuation blocks large‐scale programmable optoelectronic systems.

Language: Английский

---

A Survey on Optical Phase-Change Memory: The Promise and Challenges

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 11781 - 11803

Published: Jan. 1, 2023

Silicon photonics (SiPh) technology has facilitated the deployment of integrated across different application domains, from ultra-fast communication in Datacom applications to energy-efficient optical computation emerging hardware accelerators for machine learning. More recently, integration SiPh and phase change materials created a unique opportunity realize adaptable, reconfigurable, programmable photonic platforms. In particular, nonvolatile programmability made them promising candidate implementing memory cells architectures. Accordingly, systems even in-memory computing paradigms are on rise, especially given their potential improving data access electronic processors. However, there still many challenges design fabrication phase-change circuits, which need be addressed. This article presents comprehensive survey recent advances with contemporary devices while focusing application. we explore material level architecture by presenting an overview material-level characteristics optical, electrical, thermal properties as well into architectures computing. We also present comparison discuss open research that must addressed further advance towards successful systems.

Language: Английский

---

Unveiling cutting-edge developments: architectures and nanostructured materials for application in optoelectronic artificial synapses

Nanoscale, Journal Year: 2024, Volume and Issue: 16(31), P. 14589 - 14620

Published: Jan. 1, 2024

The significant achievements and progress in the field of biomimetic devices, including breakthroughs research on neural function, neuron designs, prosthetic synapses, artificial programming.

Language: Английский

---

Ultrahigh Endurance and Extinction Ratio in Programmable Silicon Photonics Based on a Phase Change Material with ITO Heater

Laser & Photonics Review, Journal Year: 2024, Volume and Issue: 18(4)

Published: Jan. 8, 2024

Abstract Phase‐change‐materials (PCMs) integrated photonics have attracted extensive attention in the field of optical neural networks. However, present PCMs‐integrated are still far from meeting requirements real‐world applications due to its unsatisfactory endurance (<3000 cycles) and extinction ratio (ER<20 dB). Here, ultrahigh (>30 000 large ER (≥50 dB) achieved photonic device by introducing a trench structure indium tin oxide (ITO) heater utilizing tin‐doped Ge 2 Sb Te 5 (Sn‐GST) as PCMs. This performance represents landmark photonics, since it solves problem poor ITO heaters can be comparable state‐of the‐art devices terms ER. excellent stem high contrast PCM Sn‐GST. Trench improves heating efficiency effectively alleviates thermal stress imposed on heater, resulting device. Sn‐doping is also beneficial improvement, because reduce crystallization temperature increase absorption coefficient layer. work provides new technology for development programmable with endurance.

Language: Английский

---

Overview of Phase-Change Materials Based Photonic Devices

IEEE Access, Journal Year: 2020, Volume and Issue: 8, P. 121211 - 121245

Published: Jan. 1, 2020

Non-volatile storage memory is widely considered to be one of the most promising candidates replace dynamic random access and even static memory. It has recently received particular attention because its great potential for brain-like neuromorphic applications. Phase-change materials, also known as Chalcogenide alloys, exhibit several especially advantageous traits non-volatile These include scalability, fast switching speeds, low energy outstanding thermal stability. As a result, research date sought identify electrical applications phase-change materials in relation memory, memristors neuro networks, while overlooking their photonic To address this issue, we provide comprehensive review that examines remarkable physical properties applications, together with emerging devices. The begins by presenting atomic structure followed an elaboration issues are currently facing strategies being developed overcome them. current state-of-the-art technical challenges confronting such neuro-networks, metasurfaces color displays then considered. concludes discussing outlook successfully implementing domains.

Language: Английский

---