{"title":"基于超晶格Sb/ sb2s3的神经形态计算突触装置的长期可塑性和抑制作用","authors":"Li Guo, Yegang Lu","doi":"10.1016/j.jallcom.2025.179415","DOIUrl":null,"url":null,"abstract":"Thermal stability and resistance drift are the two main issues for phase change memory (PCM) in application of neuromorphic computing. Though monatomic Sb thin films possess rapid crystallization and mitigate the risk of component segregation, they suffer from poor thermal stability. In this paper, Sb/Sb<sub>2</sub>S<sub>3</sub> with a superlattice-like (SLL) arrangement were fabricated by employing pure Sb and thermally stable Sb<sub>2</sub>S<sub>3</sub>. The crystallization temperature of the Sb/Sb<sub>2</sub>S<sub>3</sub> SLL film reaches approximately 226<!-- --> <sup>o</sup>C. The data retention temperature for 10 years achieves about 137<!-- --> <sup>o</sup>C and the crystallization activation energy is approximately 3.44<!-- --> <!-- -->eV. These findings highlight the remarkable thermal stability of these films. The presence of tension at the interface of SLL thin films hinders the growth of Sb crystals. By analyzing the crystallization mechanism, it was found that the crystallization kinetic exponent (<span><span style=\"\"></span><span data-mathml='<math xmlns=\"http://www.w3.org/1998/Math/MathML\"><mi is=\"true\">n</mi></math>' role=\"presentation\" style=\"font-size: 90%; display: inline-block; position: relative;\" tabindex=\"0\"><svg aria-hidden=\"true\" focusable=\"false\" height=\"1.394ex\" role=\"img\" style=\"vertical-align: -0.235ex;\" viewbox=\"0 -498.8 600.5 600.2\" width=\"1.395ex\" xmlns:xlink=\"http://www.w3.org/1999/xlink\"><g fill=\"currentColor\" stroke=\"currentColor\" stroke-width=\"0\" transform=\"matrix(1 0 0 -1 0 0)\"><g is=\"true\"><use xlink:href=\"#MJMATHI-6E\"></use></g></g></svg><span role=\"presentation\"><math xmlns=\"http://www.w3.org/1998/Math/MathML\"><mi is=\"true\">n</mi></math></span></span><script type=\"math/mml\"><math><mi is=\"true\">n</mi></math></script></span>) is 0.481, indicating one-dimensional growth of Sb/Sb<sub>2</sub>S<sub>3</sub> SLL thin films. This helps reduce nucleation randomness and minimize resistance drift. In addition, we simulated synapses using PCM units based on [Sb(2<!-- --> <!-- -->nm)Sb<sub>2</sub>S<sub>3</sub>(4<!-- --> <!-- -->nm)]<sub>16</sub> by applying specific pulse schemes, achieving long-term potentiation (LTP) and long-term depression (LTD) of synaptic weights. The three-layer perceptron device achieved recognition accuracy of 96% in a handwritten digit identification task, suggesting that our approach has the potential to enhance the efficiency of neuromorphic computing networks.","PeriodicalId":344,"journal":{"name":"Journal of Alloys and Compounds","volume":"14 1","pages":""},"PeriodicalIF":6.3000,"publicationDate":"2025-02-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Long-term plasticity and inhibition of superlattice-like Sb/Sb2S3-based synaptic devices for neuromorphic computing\",\"authors\":\"Li Guo, Yegang Lu\",\"doi\":\"10.1016/j.jallcom.2025.179415\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Thermal stability and resistance drift are the two main issues for phase change memory (PCM) in application of neuromorphic computing. Though monatomic Sb thin films possess rapid crystallization and mitigate the risk of component segregation, they suffer from poor thermal stability. In this paper, Sb/Sb<sub>2</sub>S<sub>3</sub> with a superlattice-like (SLL) arrangement were fabricated by employing pure Sb and thermally stable Sb<sub>2</sub>S<sub>3</sub>. The crystallization temperature of the Sb/Sb<sub>2</sub>S<sub>3</sub> SLL film reaches approximately 226<!-- --> <sup>o</sup>C. The data retention temperature for 10 years achieves about 137<!-- --> <sup>o</sup>C and the crystallization activation energy is approximately 3.44<!-- --> <!-- -->eV. These findings highlight the remarkable thermal stability of these films. The presence of tension at the interface of SLL thin films hinders the growth of Sb crystals. By analyzing the crystallization mechanism, it was found that the crystallization kinetic exponent (<span><span style=\\\"\\\"></span><span data-mathml='<math xmlns=\\\"http://www.w3.org/1998/Math/MathML\\\"><mi is=\\\"true\\\">n</mi></math>' role=\\\"presentation\\\" style=\\\"font-size: 90%; display: inline-block; position: relative;\\\" tabindex=\\\"0\\\"><svg aria-hidden=\\\"true\\\" focusable=\\\"false\\\" height=\\\"1.394ex\\\" role=\\\"img\\\" style=\\\"vertical-align: -0.235ex;\\\" viewbox=\\\"0 -498.8 600.5 600.2\\\" width=\\\"1.395ex\\\" xmlns:xlink=\\\"http://www.w3.org/1999/xlink\\\"><g fill=\\\"currentColor\\\" stroke=\\\"currentColor\\\" stroke-width=\\\"0\\\" transform=\\\"matrix(1 0 0 -1 0 0)\\\"><g is=\\\"true\\\"><use xlink:href=\\\"#MJMATHI-6E\\\"></use></g></g></svg><span role=\\\"presentation\\\"><math xmlns=\\\"http://www.w3.org/1998/Math/MathML\\\"><mi is=\\\"true\\\">n</mi></math></span></span><script type=\\\"math/mml\\\"><math><mi is=\\\"true\\\">n</mi></math></script></span>) is 0.481, indicating one-dimensional growth of Sb/Sb<sub>2</sub>S<sub>3</sub> SLL thin films. This helps reduce nucleation randomness and minimize resistance drift. In addition, we simulated synapses using PCM units based on [Sb(2<!-- --> <!-- -->nm)Sb<sub>2</sub>S<sub>3</sub>(4<!-- --> <!-- -->nm)]<sub>16</sub> by applying specific pulse schemes, achieving long-term potentiation (LTP) and long-term depression (LTD) of synaptic weights. The three-layer perceptron device achieved recognition accuracy of 96% in a handwritten digit identification task, suggesting that our approach has the potential to enhance the efficiency of neuromorphic computing networks.\",\"PeriodicalId\":344,\"journal\":{\"name\":\"Journal of Alloys and Compounds\",\"volume\":\"14 1\",\"pages\":\"\"},\"PeriodicalIF\":6.3000,\"publicationDate\":\"2025-02-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Alloys and Compounds\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://doi.org/10.1016/j.jallcom.2025.179415\",\"RegionNum\":2,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"CHEMISTRY, PHYSICAL\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Alloys and Compounds","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1016/j.jallcom.2025.179415","RegionNum":2,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CHEMISTRY, PHYSICAL","Score":null,"Total":0}
Long-term plasticity and inhibition of superlattice-like Sb/Sb2S3-based synaptic devices for neuromorphic computing
Thermal stability and resistance drift are the two main issues for phase change memory (PCM) in application of neuromorphic computing. Though monatomic Sb thin films possess rapid crystallization and mitigate the risk of component segregation, they suffer from poor thermal stability. In this paper, Sb/Sb2S3 with a superlattice-like (SLL) arrangement were fabricated by employing pure Sb and thermally stable Sb2S3. The crystallization temperature of the Sb/Sb2S3 SLL film reaches approximately 226 oC. The data retention temperature for 10 years achieves about 137 oC and the crystallization activation energy is approximately 3.44 eV. These findings highlight the remarkable thermal stability of these films. The presence of tension at the interface of SLL thin films hinders the growth of Sb crystals. By analyzing the crystallization mechanism, it was found that the crystallization kinetic exponent () is 0.481, indicating one-dimensional growth of Sb/Sb2S3 SLL thin films. This helps reduce nucleation randomness and minimize resistance drift. In addition, we simulated synapses using PCM units based on [Sb(2 nm)Sb2S3(4 nm)]16 by applying specific pulse schemes, achieving long-term potentiation (LTP) and long-term depression (LTD) of synaptic weights. The three-layer perceptron device achieved recognition accuracy of 96% in a handwritten digit identification task, suggesting that our approach has the potential to enhance the efficiency of neuromorphic computing networks.
期刊介绍:
The Journal of Alloys and Compounds is intended to serve as an international medium for the publication of work on solid materials comprising compounds as well as alloys. Its great strength lies in the diversity of discipline which it encompasses, drawing together results from materials science, solid-state chemistry and physics.