-
自旋神经元是一种新兴的人工神经形态器件, 其具有超低功耗、强非线性、高集成度和存算一体等优点, 是构建新一代神经网络的强有力候选者. 本文提出了一种磁场辅助磁弹应变驱动的混合时钟自旋神经元, 利用OOMMF微磁学仿真软件建立了该神经元器件的微磁学模型, 基于LLG方程建立了其数值仿真模型, 利用所设计的自旋神经元构建了3层神经网络, 研究了不同纳磁体材料(Terfenol-D, FeGa, Ni)神经元器件的激活特性及其对MNIST手写数字数据集识别性能的影响. OOMMF仿真和数值模拟发现, 设计的混合时钟结构能够成功驱动纳磁体发生随机磁化翻转, 有效模拟生物神经元的激活行为和特性. MNIST手写数字识别结果表明: 当输入不同范围的磁场使得3种材料的自旋神经元都达到饱和识别精度时, 该自旋神经元器件具有与Sigmoid神经元器件相同的识别能力, 有望替代传统的CMOS神经元, 并且选择合适的磁致伸缩层材料能够进一步降低智能计算的整体功耗; 当输入相同范围的磁场时, Ni构成的自旋神经元的识别速度较慢. 研究结果可为新型人工神经网络和智能电路的设计及应用奠定一定的理论基础.The spin neuron is an emerging artificial neural device which has many advantages such as ultra-low power consumption, strong nonlinearity, and high integration. Besides, it has ability to remember and calculate at the same time. So it is seen as a suitable and excellent candidate for the new generation of neural network. In this paper, a spin neuron driven by magnetic field and strain is proposed. The micromagnetic model of the device is realized by using the OOMMF micromagnetic simulation software, and the numerical model of the device is also established by using the LLG equation. More importantly, a three-layer neural network is composed of spin neurons constructed respectively using three materials (Terfenol-D, FeGa, Ni). It is used to study the activation functions and the ability to recognize the MNIST handwritten datasets.c Results show that the spin neuron can successfully achieve the random magnetization switching to simulate the activation behavior of the biological neuron. Moreover, the results show that if the ranges of the inputting magnetic fields are different, the three materials' neurons can all reach the saturation accuracy. It is expected to replace the traditional CMOS neuron. And the overall power consumption of intelligent computing can be further reduced by using appropriate materials. If we input the magnetic fields in the same range, the recognition speed of the spin neuron made of Ni is the slowest in the three materials. The results can establish a theoretical foundation for the design and the applications of the new artificial neural networks and the intelligent circuits.
-
Keywords:
- nanomagnet /
- spin neuron /
- magnetization switching /
- neural network computing
[1] Aleksander I 2004 Nature 432 18
[2] Linares-Barranco B, Sanchez-Sinencio E, Rodriguez-Vazquez A, Huertas J L 1991 IEEE J. Solid-State Circuits 26 956
Google Scholar
[3] Lont J B, Guggenbuhl W 1992 IEEE Trans. Neural Networks 3 457
Google Scholar
[4] 陈怡然, 李海, 陈逸中, 陈凡, 李思成, 刘晨晨, 闻武杰, 吴春鹏, 燕博南 2018 人工智能 2 46
Google Scholar
Chen Y R, Li H, Chen Y Z, Chen F, Li S C, Liu C C, Wen W J, Wu C P, Yan B N 2018 AI-View 2 46
Google Scholar
[5] Yang R, Terabe K, Yao Y P, Tsuruoka T, Hasegawa T, Gimzewski J K, Aono M 2013 Nanotechnology 24 384003
Google Scholar
[6] Chen C, Yang M, Liu S, Liu T, Zhu K, Zhao Y, Wang H, Huang Q, Huang R 2019 Symposium on VLSI Technology (Kyoto: IEEE) p136
[7] 刘东青, 程海峰, 朱玄, 王楠楠, 张朝阳 2014 物理学报 63 187301
Google Scholar
Liu D Q, Cheng H F, Zhu X, Wang N N, Zhang C Y 2014 Acta Phys. Sin. 63 187301
Google Scholar
[8] Tuma T, Pantazi A, Gallo M L, Sebastian A, Eleftheriou E 2016 Nat. Nanotechnol. 11 693
Google Scholar
[9] Cai J L, Fang B, Zhang L, Lv W X, Zhang B S, Zhou T J, Finocchio G, Zeng Z M 2019 Phys. Rev. Appl. 11 034015
Google Scholar
[10] Zhu J D, Zhang T, Yang Y C, Huang R 2020 Appl. Phys. Rev. 7 011312
Google Scholar
[11] Yue K, Liu Y Z, Lake R K, Parker A C 2019 Sci. Adv. 5 eaau8170
Google Scholar
[12] Fukami S, Ohno H 2018 J. Appl. Phys. 124 151904
Google Scholar
[13] Sengupta A, Choday S H, Y Kim, Roy K 2015 Appl. Phys. Lett. 106 143701
Google Scholar
[14] Fulara H, Zahedinejad M, Khymyn R, Dvornik M, Fukami S, Kanai S, Ohno H, Akerman J 2020 Nat. Commun. 11 4006
Google Scholar
[15] Dong I, Yoon G, Sik H, Park, Wanjun 2015 J. Appl. Phys. 117 17D714
Google Scholar
[16] Vincent A F, Jerome L, Locatelli N, Nesrine B R, Bichler O, Gamrat C, Zhao W S, Klein J O, Galdin-Retailleau S, Querlioz D 2015 IEEE T. Biomed. Circ. S 9 166
Google Scholar
[17] Chen Y B, Song M X, Wei B, Yang X K, Cui H Q, Liu J H, Li C 2020 IEEE Magn. Lett. 11 4504505
Google Scholar
[18] Kim Y, Fong X, Roy K 2015 IEEE Magn. Lett. 6 3001004
Google Scholar
[19] Fukushima A, Seki T, Yakushiji K, Kubota H, Imamura H, Yuasa S, Ando K 2014 Appl. Phys. Express 7 083001
Google Scholar
[20] Ostwal V, Debashis P, Faria R, Chen Z H, Appenzeller J 2018 Sci. Rep. 8 16689
Google Scholar
[21] Yang X K, Cai L, Zhang B, Cui H Q, Zhang M L 2015 J. Magn. Magn. Mater. 394 391
Google Scholar
[22] Carlton D B, Emley N C, Tuchfeldand E, Bokor J 2008 Nano Lett. 8 4173
Google Scholar
[23] Kurenkov A, DuttaGupta S, Zhang C H, Fukami S, Horio Y, Ohno H 2019 Adv. Mater. 31 1900636
Google Scholar
[24] Cai J L, Fang B, Wang C, Zeng Z M 2017 Appl. Phys. Lett. 111 182410
Google Scholar
[25] Zhang S, Luo S J, Xu N, Zou Q M, Song M, Yun J J, Luo Q, Guo Z, Li R F, Tian W C, Li X, Zhou H G, Chen H M, Zhang Y, Yang X F, Jiang W J, Shen K, Hong J M, Yuan Z, Xi L, Xia K, Salahuddin S, Dieny B, You L 2019 Adv. Electron. Mater. 5 1800782
Google Scholar
[26] Zhang S, Su Y, Li X, Li R, Tian W, Hong J, You L 2019 Appl. Phys. Lett. 114 042401
Google Scholar
[27] Sheng Y, Edmonds K W, Ma X Q, Zheng H Z, Wang K Y 2018 Adv. Electron. Mater. 4 1800224
Google Scholar
[28] Cao Y, Rushforth A W, Sheng Y, Zheng H Z, Wang K Y 2019 Adv. Funct. Mater. 29 1808104
Google Scholar
[29] 王宗巍, 杨玉超, 蔡一茂, 朱涛, 丛杨, 王志衡, 黄如 2019 中国科学基金 33 656
Google Scholar
Wang Z W, Yang Y C, Cai Y M, Zhu T, Cong Y, Wang Z H, Huang R 2019 Bulletin of National Natural Science Foundation of China 33 656
Google Scholar
[30] Liu J H, Yang X K, Cui H Q, Wei B, Li C, Chen Y B, Zhang M L, Li C, Dong D N 2019 J. Magn. Magn. Mater. 491 165607
Google Scholar
[31] Ma J, Hu J M, Li Z, Nan C W 2011 Adv. Mater. 23 1062
Google Scholar
[32] 杨娜娜, 陈轩, 汪尧进 2018 物理学报 67 157508
Google Scholar
Yang N N, Chen X, Wang Y J 2018 Acta Phys. Sin. 67 157508
Google Scholar
[33] Cowburn R P, Welland M E 2000 Science 287 1466
Google Scholar
[34] Locatelli N, Cros V, Grollier J 2013 Nat. Mater. 13 11
Google Scholar
[35] Chen Y B, Wei B, Yang X K, Liu J H, Cui H Q, Li C, Song M X 2020 J. Magn. Magn. Mater. 514 167216
Google Scholar
[36] Li X, Carka D, Liang C Y, Sepulveda A E, Keller S M, Amiri P K, Carman G P, Lynch C S 2015 J. Appl.Phys. 118 014101
Google Scholar
[37] 王庆伟, 张晶晶, 马天宇, 严密 2009 稀有金属材料与工程 38 1234
Google Scholar
Wang Q W, Zhang J J, Ma T Y, Yan M 2009 Rare. Metal. Mat. Eng. 38 1234
Google Scholar
[38] Bertotti G, Serpico C, Mayergoyz I D 2009 Nonlinear Magnetization Dynamics in Nanosystems (Oxford: Elsevier) pp401−445
[39] Beleggia M, Graef M D, Millev Y T, Goode D A, Rowlands G 2005 J. Phys. D. Appl. Phys. 38 3333
Google Scholar
[40] Liyanagedera C M, Sengupta A, Jaiswal A, Roy K 2017 Phys. Rev. Appl. 8 064017
Google Scholar
[41] Glorot X, Bengio Y 2010 J. Mach. Learn. Res. 9 249
[42] Fashami M S, Atulasimha J, Bandyopadhyay S 2012 Nanotechnology 23 105201
Google Scholar
[43] Vacca M, Graziano M, Crescenzo L D, Chiolerio A, Lamberti A, Balma D, Canavese G, Celegato F, Enrico E, Tiberto P, Boarino L, Zamboni M 2014 IEEE Trans. Nanotechnol. 13 963
Google Scholar
[44] Liu J H, Yang X K, Zhang M L, Wei B, Li C, Dong D N, Li C 2018 IEEE Electron Device Lett. 40 220
Google Scholar
[45] Das J, Alam S M, Bhanja S 2011 IEEE J. Emerg. Sel. Top. Circuits Syst. 1 267
Google Scholar
-
图 3 动态磁化过程 (a) 磁化初始方向为–y; (b) 施加30 MPa应力, 磁化翻转90°; 撤去应力, 施加10 mT磁场; (c)方向相反, 磁化翻转180°; (d)方向相同, 磁化翻转0°
Fig. 3. Magnetization process: (a) Initial direction of magnetization is –y; (b) a 30 MPa strain is applied and then 90° switching is achieved, removing strain and applying a 10 mT magnetic field; (c) 180° magnetization switching; (d) 0° magnetization switching.
图 4 磁化矢量随时间变化曲线图 (a) 1 ns时, 施加磁场沿y轴正方向时的磁化矢量变化图; (b) 1 ns时, 施加磁场沿y轴负方向的磁化矢量变化图
Fig. 4. Magnetization vector with time: (a) When t = 1 ns, a magnetic field is applied in the direction along +y; (b) when t = 1 ns, a magnetic field is applied in the direction along –y.
图 5 室温下随机磁化翻转的动态磁化过程 (a) 180°磁化翻转; (b) 0°磁化翻转
Fig. 5. Magnetization dynamics at room temperature: (a) 180° magnetization switching; (b) 0° magnetization switching.
图 6 180°磁化翻转概率与输入磁场的关系, 磁致伸缩层材料为(a) Terfenol-D, (b) FeGa, (c) Ni
Fig. 6. 180° magnetization switching probability versus magnetic field, the magnetostrictive layer material is (a) Terfenol-D, (b) FeGa, (c) Ni.
图 7 基于磁场辅助磁弹时钟的自旋神经元的三层神经网络结构示意图
Fig. 7. Three-layer neural network based on magnetic field + strain spin neurons.
图 8 不同材料神经元器件的识别精度
Fig. 8. Recognition rate obtained from the neural network with different materials.
图 9 输入磁场为0—15 mT时, 不同材料神经元器件的识别精度
Fig. 9. When the magnetic field is 0–15 mT, the recognition rate obtained from the neural network with different materials.
表 1 材料参数表
Table 1. Parameters of different materials.
参数 Terfenol-D FeGa Ni 杨氏模量Y/(1010 Pa) 8.0 2.5 21.4 磁致伸缩系数λs/10–4 6.0 4.0 –0.2 吉尔伯特阻尼系数α 0.100 0.100 0.045 回磁比γ/(105 rad·s–1·T–1) 2.21 2.21 2.21 饱和磁化率Ms/(105 A·m–1) 8.00 13.20 4.84 交换作用常数A/(10–11 J·m–1) 0.90 1.60 1.05 
下载: 导出CSV
-
[1] Aleksander I 2004 Nature 432 18
[2] Linares-Barranco B, Sanchez-Sinencio E, Rodriguez-Vazquez A, Huertas J L 1991 IEEE J. Solid-State Circuits 26 956
Google Scholar
[3] Lont J B, Guggenbuhl W 1992 IEEE Trans. Neural Networks 3 457
Google Scholar
[4] 陈怡然, 李海, 陈逸中, 陈凡, 李思成, 刘晨晨, 闻武杰, 吴春鹏, 燕博南 2018 人工智能 2 46
Google Scholar
Chen Y R, Li H, Chen Y Z, Chen F, Li S C, Liu C C, Wen W J, Wu C P, Yan B N 2018 AI-View 2 46
Google Scholar
[5] Yang R, Terabe K, Yao Y P, Tsuruoka T, Hasegawa T, Gimzewski J K, Aono M 2013 Nanotechnology 24 384003
Google Scholar
[6] Chen C, Yang M, Liu S, Liu T, Zhu K, Zhao Y, Wang H, Huang Q, Huang R 2019 Symposium on VLSI Technology (Kyoto: IEEE) p136
[7] 刘东青, 程海峰, 朱玄, 王楠楠, 张朝阳 2014 物理学报 63 187301
Google Scholar
Liu D Q, Cheng H F, Zhu X, Wang N N, Zhang C Y 2014 Acta Phys. Sin. 63 187301
Google Scholar
[8] Tuma T, Pantazi A, Gallo M L, Sebastian A, Eleftheriou E 2016 Nat. Nanotechnol. 11 693
Google Scholar
[9] Cai J L, Fang B, Zhang L, Lv W X, Zhang B S, Zhou T J, Finocchio G, Zeng Z M 2019 Phys. Rev. Appl. 11 034015
Google Scholar
[10] Zhu J D, Zhang T, Yang Y C, Huang R 2020 Appl. Phys. Rev. 7 011312
Google Scholar
[11] Yue K, Liu Y Z, Lake R K, Parker A C 2019 Sci. Adv. 5 eaau8170
Google Scholar
[12] Fukami S, Ohno H 2018 J. Appl. Phys. 124 151904
Google Scholar
[13] Sengupta A, Choday S H, Y Kim, Roy K 2015 Appl. Phys. Lett. 106 143701
Google Scholar
[14] Fulara H, Zahedinejad M, Khymyn R, Dvornik M, Fukami S, Kanai S, Ohno H, Akerman J 2020 Nat. Commun. 11 4006
Google Scholar
[15] Dong I, Yoon G, Sik H, Park, Wanjun 2015 J. Appl. Phys. 117 17D714
Google Scholar
[16] Vincent A F, Jerome L, Locatelli N, Nesrine B R, Bichler O, Gamrat C, Zhao W S, Klein J O, Galdin-Retailleau S, Querlioz D 2015 IEEE T. Biomed. Circ. S 9 166
Google Scholar
[17] Chen Y B, Song M X, Wei B, Yang X K, Cui H Q, Liu J H, Li C 2020 IEEE Magn. Lett. 11 4504505
Google Scholar
[18] Kim Y, Fong X, Roy K 2015 IEEE Magn. Lett. 6 3001004
Google Scholar
[19] Fukushima A, Seki T, Yakushiji K, Kubota H, Imamura H, Yuasa S, Ando K 2014 Appl. Phys. Express 7 083001
Google Scholar
[20] Ostwal V, Debashis P, Faria R, Chen Z H, Appenzeller J 2018 Sci. Rep. 8 16689
Google Scholar
[21] Yang X K, Cai L, Zhang B, Cui H Q, Zhang M L 2015 J. Magn. Magn. Mater. 394 391
Google Scholar
[22] Carlton D B, Emley N C, Tuchfeldand E, Bokor J 2008 Nano Lett. 8 4173
Google Scholar
[23] Kurenkov A, DuttaGupta S, Zhang C H, Fukami S, Horio Y, Ohno H 2019 Adv. Mater. 31 1900636
Google Scholar
[24] Cai J L, Fang B, Wang C, Zeng Z M 2017 Appl. Phys. Lett. 111 182410
Google Scholar
[25] Zhang S, Luo S J, Xu N, Zou Q M, Song M, Yun J J, Luo Q, Guo Z, Li R F, Tian W C, Li X, Zhou H G, Chen H M, Zhang Y, Yang X F, Jiang W J, Shen K, Hong J M, Yuan Z, Xi L, Xia K, Salahuddin S, Dieny B, You L 2019 Adv. Electron. Mater. 5 1800782
Google Scholar
[26] Zhang S, Su Y, Li X, Li R, Tian W, Hong J, You L 2019 Appl. Phys. Lett. 114 042401
Google Scholar
[27] Sheng Y, Edmonds K W, Ma X Q, Zheng H Z, Wang K Y 2018 Adv. Electron. Mater. 4 1800224
Google Scholar
[28] Cao Y, Rushforth A W, Sheng Y, Zheng H Z, Wang K Y 2019 Adv. Funct. Mater. 29 1808104
Google Scholar
[29] 王宗巍, 杨玉超, 蔡一茂, 朱涛, 丛杨, 王志衡, 黄如 2019 中国科学基金 33 656
Google Scholar
Wang Z W, Yang Y C, Cai Y M, Zhu T, Cong Y, Wang Z H, Huang R 2019 Bulletin of National Natural Science Foundation of China 33 656
Google Scholar
[30] Liu J H, Yang X K, Cui H Q, Wei B, Li C, Chen Y B, Zhang M L, Li C, Dong D N 2019 J. Magn. Magn. Mater. 491 165607
Google Scholar
[31] Ma J, Hu J M, Li Z, Nan C W 2011 Adv. Mater. 23 1062
Google Scholar
[32] 杨娜娜, 陈轩, 汪尧进 2018 物理学报 67 157508
Google Scholar
Yang N N, Chen X, Wang Y J 2018 Acta Phys. Sin. 67 157508
Google Scholar
[33] Cowburn R P, Welland M E 2000 Science 287 1466
Google Scholar
[34] Locatelli N, Cros V, Grollier J 2013 Nat. Mater. 13 11
Google Scholar
[35] Chen Y B, Wei B, Yang X K, Liu J H, Cui H Q, Li C, Song M X 2020 J. Magn. Magn. Mater. 514 167216
Google Scholar
[36] Li X, Carka D, Liang C Y, Sepulveda A E, Keller S M, Amiri P K, Carman G P, Lynch C S 2015 J. Appl.Phys. 118 014101
Google Scholar
[37] 王庆伟, 张晶晶, 马天宇, 严密 2009 稀有金属材料与工程 38 1234
Google Scholar
Wang Q W, Zhang J J, Ma T Y, Yan M 2009 Rare. Metal. Mat. Eng. 38 1234
Google Scholar
[38] Bertotti G, Serpico C, Mayergoyz I D 2009 Nonlinear Magnetization Dynamics in Nanosystems (Oxford: Elsevier) pp401−445
[39] Beleggia M, Graef M D, Millev Y T, Goode D A, Rowlands G 2005 J. Phys. D. Appl. Phys. 38 3333
Google Scholar
[40] Liyanagedera C M, Sengupta A, Jaiswal A, Roy K 2017 Phys. Rev. Appl. 8 064017
Google Scholar
[41] Glorot X, Bengio Y 2010 J. Mach. Learn. Res. 9 249
[42] Fashami M S, Atulasimha J, Bandyopadhyay S 2012 Nanotechnology 23 105201
Google Scholar
[43] Vacca M, Graziano M, Crescenzo L D, Chiolerio A, Lamberti A, Balma D, Canavese G, Celegato F, Enrico E, Tiberto P, Boarino L, Zamboni M 2014 IEEE Trans. Nanotechnol. 13 963
Google Scholar
[44] Liu J H, Yang X K, Zhang M L, Wei B, Li C, Dong D N, Li C 2018 IEEE Electron Device Lett. 40 220
Google Scholar
[45] Das J, Alam S M, Bhanja S 2011 IEEE J. Emerg. Sel. Top. Circuits Syst. 1 267
Google Scholar
下载:
计量
- 文章访问数: 4595
- PDF下载量: 81
- 被引次数: 0
