a, the diagram of LWIR memlogic sensor structure. b, the hysteretic IV curve under the light irradiation and without intentional radiation, respectively. c,d , the operation of optical/electrical pulse write/erase data and persist state. e,f, schematic of reconfigurable logic circuit and experimental data under different electrical biases. g, high responsivity at the 12.2-micron operating wavelength.
a, Encryption transmission. The different operation bias current represent the different key. b, Image information storage and filtering preprocessing. c, image noise reduction preprocessing.
a, The Schematic diagram of software-based Artificial Neural Network (ANN). b The schematic diagram illustrates a superconducting nanowire ANN array with independent bias currents at each pixel, allowing for the adjustment of each pixel's weight. The output voltage is dependent on the summation of voltages from each pixel and the threshold voltage. c The Schematic diagram of superconducting nanowire ANN array with the same bias current at each pixel, but different local temperature at each pixel to adjust weight. The diagram on the left portrays each pixel designed to function at an identical operational wavelength, while the diagram on the right illustrates each pixel having a distinct detection wavelength. However, by designing the normal state resistance of each device differently, it becomes possible to infer the specific detected wavelength and spatial position information from the read voltage.
Newswise — With the rapid development of Artificial Intelligence (AI) and the Internet of Things (IoT), the number of optical sensors has significantly increased. These traditional sensors generate vast amounts of unstructured and redundant data, leading to unnecessary energy consumption and information latency during data transmission to storage and computing units. In contrast, the human visual system has an efficient and low-power parallel processing capability, allowing simultaneous perception and processing of images. To address the issues associated with the von Neumann architecture, researchers have developed memory-computing systems based on various materials like 2D materials, phase-change materials, and ferroelectric materials. However, these systems, based on such materials, often suffer from weak light response and large band gaps, requiring high-power light sources for optical computing, with few reported capable of in-sensor-computing at long-wave infrared. In the infrared domain, AI perception plays a critical role in various fields, including but not limited to thermal imaging, surveillance, industrial inspection, gas detection, medical thermal imaging, defense, and space exploration.
In a new paper (doi: https://doi.org/10.1038/s41377-024-01424-2 ) published in Light Science & Applications, a team of scientists, led by Professor Zhenghua An from Fudan University, Yanru Song from ShanghaiTech University and co-workers have developed a superconducting memlogic sensor that integrates the capabilities of infrared sensitivity, memory retention, and reconfigurable logic computing, all-in-one infrared memlogic sensory. They demonstrate the simultaneous four controls of this superconducting device including optical/electrical biases, and optical/electrical pulses allow different encoding logic with increasing functional complexity. With these reconfigurable operations, they show the infrared remote encrypted communication at a single device level. Given the prevalence of bistable effects, extensive wavelength coverage, and high photo-sensitivity in superconductors, their memlogic sensing concept, utilizing superconducting phase transitions, shows potential for versatile applications across a wide electromagnetic spectrum and down to quantum-levels in various fields such as machine perception, remote sensing, secure communication, and space detection. These scientists summarize the operational principle of their sensor:
“Distinctive from existing superconducting sensors (like SNSPDs and TESs), our device works in the bistable region of the hysteretic superconductor-normal phase transition. This hysteretic IV is often considered as a hindrance to most of the superconducting devices’ performance although they can be indeed utilized as switches or memories. By contrast, here we take the advantages of the hysteresis behavior and show its favorable applications in memlogic optical sensing.”
“To overcome the well-known low system detection efficiency of superconducting detection in this long-wave infrared region, we adopt a metamaterial perfect absorber concept which consists of a resonant plasmonic cavity, with tri-layer structure of Nb-Si-Nb.” they added.
“This sensor can be integrated on a large scale and requires only two electrodes for readout circuit, which is crucial for cryogenic application. Future prospects for this technology include applications in telecommunications, superconducting computing, AI-driven image processing, and intelligent on-chip spectrometers.” the scientists forecast.
###
References
DOI
Original Source URL
https://doi.org/10.1038/s41377-024-01424-2
Funding information
This work was supported by National Natural Science Foundation of China (NSFC) (12027805, 11991060), the Shanghai Science and Technology Committee (18JC1420400, 20JC1414700 and 20DZ1100604), Shanghai Pujiang Program (No. 20PJ1410900)
About Light: Science & Applications
The Light: Science & Applications will primarily publish new research results in cutting-edge and emerging topics in optics and photonics, as well as covering traditional topics in optical engineering. The journal will publish original articles and reviews that are of high quality, high interest and far-reaching consequence.
MEDIA CONTACT
Register for reporter access to contact detailsArticle Multimedia
Credit:
Caption: a, the diagram of LWIR memlogic sensor structure. b, the hysteretic IV curve under the light irradiation and without intentional radiation, respectively. c,d , the operation of optical/electrical pulse write/erase data and persist state. e,f, schematic of reconfigurable logic circuit and experimental data under different electrical biases. g, high responsivity at the 12.2-micron operating wavelength.
Credit: Light: Science & Applications
Caption: a, Encryption transmission. The different operation bias current represent the different key. b, Image information storage and filtering preprocessing. c, image noise reduction preprocessing.
Credit: Light: Science & Applications
Caption: a, The Schematic diagram of software-based Artificial Neural Network (ANN). b The schematic diagram illustrates a superconducting nanowire ANN array with independent bias currents at each pixel, allowing for the adjustment of each pixel's weight. The output voltage is dependent on the summation of voltages from each pixel and the threshold voltage. c The Schematic diagram of superconducting nanowire ANN array with the same bias current at each pixel, but different local temperature at each pixel to adjust weight. The diagram on the left portrays each pixel designed to function at an identical operational wavelength, while the diagram on the right illustrates each pixel having a distinct detection wavelength. However, by designing the normal state resistance of each device differently, it becomes possible to infer the specific detected wavelength and spatial position information from the read voltage.
RSS