Optical artificial neural networks (ONNs)-analog computing hardware tailored for machine learning-have significant potential for achieving ultrahigh computing speed and energy efficiency. A new approach to architectures for ONNs based on integrated Kerr microcomb sources that is programmable, highly scalable, and capable of reaching ultra-high speeds is proposed here. The building block of the ONN-a single neuron perceptron-is experimentally demonstrated that reaches a high single-unit throughput speed of 11.9 Giga-FLOPS at 8 bits per FLOP, corresponding to 95.2 Gbps, achieved by mapping synapses onto 49 wavelengths of a microcomb. The perceptron is tested on simple standard benchmark datasets-handwritten-digit recognition and cancer-cell detection-achieving over 90% and 85% accuracy, respectively. This performance is a direct result of the record low wavelength spacing (49 GHz) for a coherent integrated microcomb source, which results in an unprecedented number of wavelengths for neuromorphic optics. Finally, an approach to scaling the perceptron to a deep learning network is proposed using the same single microcomb device and standard off-the-shelf telecommunications technology, for high-throughput operation involving full matrix multiplication for applications such as real-time massive data processing for unmanned vehicles and aircraft tracking.
Atomically thin materials face an ongoing challenge of scalability, hampering practical deployment despite their fascinating properties. Tin monosulfide (SnS), a low‐cost, naturally abundant layered material with a tunable bandgap, displays properties of superior carrier mobility and large absorption coefficient at atomic thicknesses, making it attractive for electronics and optoelectronics. However, the lack of successful synthesis techniques to prepare large‐area and stoichiometric atomically thin SnS layers (mainly due to the strong interlayer interactions) has prevented exploration of these properties for versatile applications. Here, SnS layers are printed with thicknesses varying from a single unit cell (0.8 nm) to multiple stacked unit cells (≈1.8 nm) synthesized from metallic liquid tin, with lateral dimensions on the millimeter scale. It is reveal that these large‐area SnS layers exhibit a broadband spectral response ranging from deep‐ultraviolet (UV) to near‐infrared (NIR) wavelengths (i.e., 280–850 nm) with fast photodetection capabilities. For single‐unit‐cell‐thick layered SnS, the photodetectors show upto three orders of magnitude higher responsivity (927 A W−1) than commercial photodetectors at a room‐temperature operating wavelength of 660 nm. This study opens a new pathway to synthesize reproduceable nanosheets of large lateral sizes for broadband, high‐performance photodetectors. It also provides important technological implications for scalable applications in integrated optoelectronic circuits, sensing, and biomedical imaging.
Mode and polarization-division multiplexing technologies (MDM and PDM)can offer considerable parallelism for optical multiplexing biosensors, complex optical neural networks, and high-capacity optical interconnects, while requiring only a single-wavelength laser source. Thanks to the mature fabrication processes of silicon nitride and superior material properties of lithium niobate, the silicon nitride loaded lithium niobate on insulator (LNOI) platform allows the integration of high-speed optical modulators and optical (de)multiplexing devices to achieve high-capacity and low-cost photonic integrated circuits suitable for data communication applications. In this contribution, MDM and PDM are investigated in a silicon nitride loaded LNOI (X-cut) platform. As a proof of concept, an asymmetrical directional coupler-based mode ( de)multiplexer (MMUX) and polarization splitter-rotator (PSR) are designed, fabricated, and experimentally demonstrated. The measured insertion losses are lower than 1.46 and 1.49 dB, while the inter-modal crosstalk is lower than −13.03 and −17.75 dB for the MMUX and PSR, respectively, for a wavelength range of 1525-1565 nm. A 40 Gbps data transmission experiment demonstrates the data transmission capabilities of the fabricated devices. The measured eye diagrams are clear and wide-open, and the bit error rate measurements show reasonable power penalties, indicating good device performance.
(De)Multiplexing Technologies In article number 2100529, Yonghui Tian, Arnan Mitchell, Yikai Su, and co‐workers experimentally demonstrated mode and polarization‐division multiplexing on a thin‐film lithium niobate on insulator (LNOI) platform. By introducing silicon nitride as a loading material atop the LNOI photonics chip, the devices can be integrated with high‐speed electro‐optic modulators to achieve high‐capacity and low‐cost photonic integrated circuits suitable for data communication applications, while avoiding the direct etching of lithium niobate.
Lithium niobate on insulator (LNOI) has emerged as a promising platform for photonic integrated circuits, with a fast‐growing toolbox of components. This paper proposes, designs, and experimentally demonstrates compact subwavelength grating (SWG) waveguides on an LNOI platform for on‐chip mode and polarization manipulation. To overcome the limitation of waveguide fabrication, the SWGs are designed and formed on a silicon nitride thin film deposited onto the surface of LNOI chip. As proof‐of‐concept devices, the SWG‐based spatial mode filters (including a TE1‐mode‐pass filter and a TE2‐mode‐pass filter) and a TM‐pass polarizer are fabricated successfully on the same chip, with the device lengths of only ≈50 μm. The measured insertion losses for the devices are lower than 3.1 dB, with high extinction ratio larger than 30 dB, at a wavelength of 1550 nm. The proposed and demonstrated SWGs can serve as important building blocks in a series of mode and polarization handling devices for LNOI integrated photonics.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.