Search

Article

x

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

Recent advances in synthesis and memory computing of large-area α-MoO3

Shan Xin Wang Fang Hu Kai Wei Jun-Qing Lin Xin Zhao Xuan-Yu Zhou Bao-Zeng Zhang Kai-Liang

Shan Xin, Wang Fang, Hu Kai, Wei Jun-Qing, Lin Xin, Zhao Xuan-Yu, Zhou Bao-Zeng, Zhang Kai-Liang. Recent advances in synthesis and memory computing of large-area α-MoO3. Acta Phys. Sin., 2021, 70(9): 098103. doi: 10.7498/aps.70.20201813
Citation: Shan Xin, Wang Fang, Hu Kai, Wei Jun-Qing, Lin Xin, Zhao Xuan-Yu, Zhou Bao-Zeng, Zhang Kai-Liang. Recent advances in synthesis and memory computing of large-area α-MoO3. Acta Phys. Sin., 2021, 70(9): 098103. doi: 10.7498/aps.70.20201813

Recent advances in synthesis and memory computing of large-area α-MoO3

Shan Xin, Wang Fang, Hu Kai, Wei Jun-Qing, Lin Xin, Zhao Xuan-Yu, Zhou Bao-Zeng, Zhang Kai-Liang
Article Text (iFLYTEK Translation)
PDF
HTML
Get Citation
  • In recent years, α-MoO3 has received extensive attention in the research of memristor devices. The variation of valence of molybdenum will lead the resistivity to change, and the unique layer structure is beneficial to the implantation of donor ion into free space to adjust the conductance, so that it has a great influence on the study of synaptic transistors. This paper mainly summarize the properties of α-MoO3, the method and characteristics of large-scale two-dimensional α-MoO3 and analyze the recent progress of in-memory computing based on α-MoO3. Primarily, this paper introduces crystal structures, band structure and defect state of α-MoO3. The synthesis methods of large-area α-MoO3 are compared with each other, including the one-step method to directly obtain α-MoO3 nanosheets, and the combined post-annealing process of magnetron sputtering or atomic layer deposition to prepare the thin α-MoO3 films. In the one-step synthesis method we conclude that the chamber pressure influences the ratio of MoO3 to MoO3–x, and the growth temperature affects the ratio of α-MoO3 to β-MoO3. That is to say, the phase composition of molybdenum trioxide, the concentration of precursors has an important influence on the film size. The advantages of α-MoO3 synthesis by different methods in memory computing applications are discussed in detail. And then, this paper summarizes the device performance of α-MoO3 in memristor and the application progress of α-MoO3-based neuromorphic devices, and analyzes the performance of α-MoO3-based resistive random access memory such as switching ratio, endurance, and stability in detail. The synaptic functions of different structural device units are extensively studied, and various typical synapse functions are realized such as short-term plasticity, long-term plasticity, paired pulse facilitation, etc. It shows the excellent characteristics of low energy consumption in the simulation of synaptic plasticity. The use of short-term memory and long-term memory modes of the device can realize the functions of image memory and preprocessing. Through the memristor array combined with the back-propagation network, the high-precision handwriting recognition can be realized. Finally, combining the recent research progress of α-MoO3, its opportunities and challenges in the field of memory computing are prospected.
      PACS:
      81.10.-h(Methods of crystal growth; physics and chemistry of crystal growth, crystal morphology, and orientation)
      61.46.Hk(Nanocrystals)
      51.50.+v(Electrical properties)
      Corresponding author: Wang Fang, fwang75@163.com ; Zhang Kai-Liang, kailiang_zhang@163.com
    • Funds: Project supported by the Natural Science Foundation of Tianjin, China (Grant Nos. 18JCZDJC30500, 17JCYBJC16100, 17JCZDJC31700), the National Natural Science Foundation of China (Grant Nos. 61404091, 61274113, 61505144, 51502203, 51502204), the Science and Technology Planning Project of Tianjin, China (Grant No. 20ZYQCGX00070), the Open Project of State Key Laboratory of Functional Materials for Information, China (Grant No. SKL202007), and the National Key Research and Development Program of China (Grant No. 2017YFB0405600)

    近年来, 人工智能迅速崛起, 对数据的处理速度和计算速度要求越来越高, 传统冯·诺依曼计算架构的存储运算分离严重限制了计算能力的提升[1]. 1971年Chua[2]提出了第四种基本电路元件—忆阻器. 这是一种非线性电阻元件, 可以实现电导连续可调的记忆功能. 基于忆阻器阵列的感知计算凭借其高智能、低能耗等特点得到研究者们的高度关注[3]. 在不久的将来器件尺寸达到2—3 nm的节点上, 利用微处理器不断增长的处理能力来构建各种复杂的计算系统将达到极限, 器件微缩的尺寸将受到物理极限影响, 摩尔定律的延续遇到瓶颈. 因此, 二维材料的出现受到研究者们的广泛关注, 并凭借其带隙可调节、层状超薄特性、集成度高等优点在存储计算领域中成为最有潜力的材料之一[4,5]. 化学计量比的α-MoO3禁带宽度较大, 几乎没有可自由移动的电子, 绝缘性很强. 研究者们发现随着离子插层的作用, α-MoO3的费米能级向导带移动, 使得导电性不断提高. 这一优良特性使二维α-MoO3在计算存储方面得到广泛研究, 其独特的层状结构有利于将不同的施主离子注入自由空间, 在离子栅结构的突触晶体管的研究中发挥着重要作用. 由于Mo的价态可变, 使其具有独特的物理化学性质, 氧含量的改变将会引起α-MoO3电阻率的变化, 研究者们已经进行了相关研究[6], 但基于α-MoO3忆阻器阵列的存储计算器件的稳定性、突触可塑性调制以及电导调制机理等还需进一步研究. 与此同时, 为了得到均匀且稳定性好的内存单元, 大面积可控合成α-MoO3仍然是亟待解决的重要问题. 本文首先介绍了α-MoO3的基本结构与性质, 作为进一步研究的理论基础. 其次, 总结了大面积合成α-MoO3的方法, 并对其生长条件进行对比, 分析不同合成方法对α-MoO3晶体结构、结晶质量、表面形貌等的影响, 给出了适用于存储计算的α-MoO3薄膜的基本要求. 然后, 综述α-MoO3在存储计算应用方面的进展, 分析α-MoO3在存储器件中的性能对比及基于α-MoO3突触器件的图像识别等, 最后给出了简短的总结和展望, 为α-MoO3在神经形态应用方面提供参考.

    三氧化钼的存在形式通常表现为正交相的α-MoO3、亚稳态的单斜相β-MoO3和六方亚稳态的h-MoO3. 正交相为热力学稳定相, 其他两种则是非热力学稳定相, 因此在外界提供能量时, 非热力学稳定相会向热力学稳定相转变[7,8].

    图1(a)所示为α-MoO3的晶体结构图, 单层α-MoO3由[MoO6]八面体的双层网络组成, 它们沿着a轴([100])共享角, 沿着c轴([001])共享边, 这些层沿着b轴([010])通过范德瓦耳斯力堆叠. 单层α-MoO3厚度为1.4 nm, 层内部主要由共价键和离子键组成. 单层α-MoO3中有三个对称但不等价的氧位点(红球), 如图1(a)所示, 末端氧O1、双连桥氧O2和三连桥氧O3. 每个扭曲的[MoO6]八面体由一个Mo原子(蓝球)结合一个O1原子、两个O2原子和三个O3原子组成.

    图 1 (a) α-MoO3晶体结构; (b) β-MoO3晶体结构; (c) h-MoO3晶体结构\r\nFig. 1. (a) Crystal structure of α-MoO3; (b) crystal structure of β-MoO3; (c) crystal structure of h-MoO3.
    图 1  (a) α-MoO3晶体结构; (b) β-MoO3晶体结构; (c) h-MoO3晶体结构
    Fig. 1.  (a) Crystal structure of α-MoO3; (b) crystal structure of β-MoO3; (c) crystal structure of h-MoO3.

    亚稳态的β-MoO3是[MoO6]八面体在三维空间中共享角形成的单斜结构, 如图1(b)所示. 对比α-MoO3的结构可以发现β-MoO3不会沿c轴([001])共享边, 其结构和ReO3类似. β-MoO3也是相对绝缘的材料, 但由于β-MoO3的热稳定性差, 所以当温度达到350 ℃以上时, β-MoO3会向α-MoO3转变[9]. 六方亚稳态h-MoO3由[MoO6]共享边和角形成顺位连接的链状结构. 晶格常数为a ≈ 10.57 Å和c ≈ 3.72 Å, 晶体结构如图1(c)所示. Pan等[10]利用水热法合成了h-MoO3纳米棒, 发现纳米棒都是沿[001]方向生长, 在300 ℃条件下退火3 h, 晶格常数为a ≈ 10.57 Å, c ≈ 3.72 Å. 研究证明在413—436 ℃的温度范围内, h-MoO3会向α-MoO3转变.

    本征α-MoO3间接带隙半导体, 实验得到禁带宽度约为3 eV, 几乎没有可自由移动的电子, 绝缘性很强(电阻率一般大于109 Ω·cm)[11]. 密度泛函理论(density functional theory, DFT)计算表明, 块体α-MoO3的间接带隙为1.96 eV, 单层α-MoO3间接带隙较大, 为2.03 eV, 如图2所示, 是由于二维材料中的量子限制效应导致带隙的增大[6].

    图 2 (a) 块体电子结构[6]; (b) 单层电子结构[6]\r\nFig. 2. (a) Bulk electronic structure[6]; (b) monolayer electronic structure[6].
    图 2  (a) 块体电子结构[6]; (b) 单层电子结构[6]
    Fig. 2.  (a) Bulk electronic structure[6]; (b) monolayer electronic structure[6].

    MoO3的掺杂和缺陷诱导可以调整带隙及载流子浓度, 从而扩大其应用范围. 研究者们提出了各种方法在α-MoO3薄膜中制造氧空位缺陷来提高载流子迁移率. Hanson等[6]利用电子束辐照氧化钼薄膜的方法制造氧空位缺陷, 实现n型掺杂使薄膜电导率提高. 单层α-MoO3中有3种不等价的氧原子位点, 因此考虑晶体结构中不同位置的氧对氧空位形成的影响是十分重要的. 通过DFT + U计算得到晶体中不同位置的氧空位形成能EfVO1<EfVO2<EfVO3, 依次增大, 分别为1.85, 2.19和3.20 eV. Inzani等[12]研究了不同氧空位对能带结构的影响, 缺陷的引入将会减小α-MoO3的带隙, 如图3所示, 从化学计量的1.96 eV降低到O1, O2和O3缺陷的1.64, 1.88和1.68 eV.

    图 3 (a) 在化学计量比情况下MoO3能带结构[12]; (b) 在O1空位情况下MoO3能带结构[12]; (c) 在O2 空位情况下MoO3能带结构[12]; (d) 在O3空位情况下MoO3能带结构[12]\r\nFig. 3. (a) Band structures of MoO3 in the stoichiometric case[12]; (b) band structures of MoO3 in the vacancy on O1 case[12]; (c) band structures of MoO3 in the vacancy on O2 case[12]; (d) band structures of MoO3 in the vacancy on O3 case[12].
    图 3  (a) 在化学计量比情况下MoO3能带结构[12]; (b) 在O1空位情况下MoO3能带结构[12]; (c) 在O2 空位情况下MoO3能带结构[12]; (d) 在O3空位情况下MoO3能带结构[12]
    Fig. 3.  (a) Band structures of MoO3 in the stoichiometric case[12]; (b) band structures of MoO3 in the vacancy on O1 case[12]; (c) band structures of MoO3 in the vacancy on O2 case[12]; (d) band structures of MoO3 in the vacancy on O3 case[12].

    2017年Wang等[13]将氧化钼在真空退火后, 氧空位缺陷浓度大幅度增大, 正是由于氧空位的引入使得缺陷态存在, 缺陷态会产生电子载体, 导致暗电流的增大, 从而提高了器件的光电响应特性. Crowley等[14]在还原气氛下(Ar/H2, 90%/10%)改变温度退火α-MoO3, 由于还原气氛的作用使得Mo+6被还原为Mo+5和Mo+4, 氧空位的引入让本征α-MoO3变成n型掺杂的MoO3–x, 从而提高了电导率. 氧空位的引入, 由于每个氧空位存在两个剩余电子, 因此产生n型掺杂行为.

    α-MoO3具有优异的性能, 利于各种离子(如氢、碱金属)插入其晶格间隙, 从而带来丰富的物理化学性质, 可用于许多未来的智能终端设备[15]. 然而, 大面积合成少层正交相的α-MoO3的均匀性和可控性, 仍然是一个难题. 下面总结了几种电子器件应用中常用的大面积合成α-MoO3的方法.

    物理气相传输制备α-MoO3的方法非常受欢迎, 一步实验可以直接得到α-MoO3薄膜, 具有操作简单、成本低等优势. 控制MoO3的升华环境来改变产物的质量和尺寸是目前广大研究者们关注的热点. 本文总结了大面积α-MoO3制备过程中的关键参数及结果, 如表1[13,16-21]所列. 下面分别从衬底温度、腔室压力及前驱物浓度三方面进行表述.

    表 1  一步法制备α-MoO3的生长条件(1 Torr = 1.33322 × 102 Pa)
    Table 1.  Growth conditions of α-MoO3 prepared by one-step method.
    Ref.Carrier flowPressureSubstrate temperature/℃Growth time/minLateral sizeThickness/nm
    [13]Atmospheric58012069.6 μm5.3—89.8
    [16]Atmospheric850—1050300.72 cm
    [17]ArAtmospheric75020Centimeters scale4.3—550
    [18]Ar310 mTorr53045500 µm × 500 µm8 ± 0.75
    [19]Ar & O2100 mTorr15090Wafer-scale30
    [20]Atmospheric54020Centimeters scale≥ 1.4
    [21]N2Atmospheric350—48020Several tens micron2.8—14
    下载: 导出CSV 
    | 显示表格

    图4(a)α-MoO3生长过程的示意图[13]. 该方法通常采用MoO3粉末作为原料, 在一定气压下将MoO3升华, 再通过与载气碰撞冷凝, 沉积在衬底上. 通常, MoO3会在700 ℃以上升华, 600 ℃以下再沉积, 不同的沉积温度形成的MoO3形貌是不同的. 2017年Wang等[13]利用MoO3粉末作为前驱物加热蒸发, 通过改变衬底温度得到大面积的MoO3纳米片. 随着衬底温度的升高, 越来越多的α-MoO3被沉积在衬底上, 并且尺寸变大, 厚度减小. 蒸发产生的MoOx蒸气在石英管的中间形成饱和区, 一旦温度降低, 就会出现过饱和现象, 导致MoOx成核. Diaz-Droguett等[22]在100—1200 Pa的压力范围下, 以氦气作为载气直接蒸发MoO3, 可生长出MoO3和MoO3–x共存的纳米结构. 不同的载气压强可影响生长的晶体和无定形材料的比例, 在600 Pa的氦气压力下, 源材料在763—910 ℃之间进行蒸发时, 晶体结构的产率可达100%, 反之, 压强过高将会减小蒸发气体的平均自由程, 限制材料蒸发的扩散速率, 导致单个原子或小原子簇沉积在衬底表面, 因此得到的无定形材料比例会变大. 前驱物的蒸发浓度可以调节α-MoO3的横向生长速度, 2018年Zheng等[16]同样利用管式炉法, 通过控制石英管反应室的内径, 来增大前驱物的浓度. 可实现α-MoO3单晶的横向尺寸从微米到厘米级的调控. 当管内径减小到2.1 cm时, 平均宽度可从0.02 cm增大到0.15 cm, 平均长度可从0.68 cm增大到0.72 cm. 图4(b)中厘米级的纳米片几乎是长方形的, MoO3矩形长宽分别边沿c轴和a轴生长. 图4(c)的高分辨透射电子显微镜(high resolution transmission electron microscopy, HRTEM)图进一步证明了α-MoO3纳米片择优取向的生长方式. 图4(d)为X射线光电子能谱图(X-ray photoemission spectroscopy, XPS), 计算可得到单晶α-MoO3纳米片的Mo/O的原子比为2.97. 在此基础上, Sun等[17]同样利用管式炉方法, 在常压下制备了厘米级α-MoO3, 将内径为2 cm的石英管作为蒸发源容器同时作为生长衬底, 放置在内径为8 cm炉管内, α-MoO3晶体生长在石英管的内壁上, 在没有衬底的情况下减小了相互干扰的作用, α-MoO3晶体能够沿高对称轴不断堆积.

    图 4 (a) 在两温区的物理气相沉积系统中合成α-MoO3原理图[13]; (b) 在SiO2/Si衬底上生长的α-MoO3单晶的照片[16]; (c) α-MoO3的HRTEM图像, 内嵌图为相应的透射电子显微镜图像和傅里叶变换图像[16]; (d) 生长α-MoO3的O 1s和Mo 3d的高分辨XPS能谱[16]\r\nFig. 4. (a) Schematic illustration of synthesis of α-MoO3 in a two-temperature-zone PVD system[13]; (b) photograph of as-grown α-MoO3 single crystals on SiO2/Si substrate[16]; (c) HRTEM image of the α-MoO3 sheet. Insets: corresponding TEM and FFT images[16]; (d) high-resolution XPS spectra of O 1s and Mo 3d of as-grown α-MoO3[16].
    图 4  (a) 在两温区的物理气相沉积系统中合成α-MoO3原理图[13]; (b) 在SiO2/Si衬底上生长的α-MoO3单晶的照片[16]; (c) α-MoO3的HRTEM图像, 内嵌图为相应的透射电子显微镜图像和傅里叶变换图像[16]; (d) 生长α-MoO3的O 1s和Mo 3d的高分辨XPS能谱[16]
    Fig. 4.  (a) Schematic illustration of synthesis of α-MoO3 in a two-temperature-zone PVD system[13]; (b) photograph of as-grown α-MoO3 single crystals on SiO2/Si substrate[16]; (c) HRTEM image of the α-MoO3 sheet. Insets: corresponding TEM and FFT images[16]; (d) high-resolution XPS spectra of O 1s and Mo 3d of as-grown α-MoO3[16].
    3.2.1   磁控溅射法

    由于磁控溅射方法生长的薄膜多为无定型或多晶状态, 所以为了得到单晶的α-MoO3薄膜, 研究者们通常会结合后退火工艺实现薄膜的晶相和形态可控的大面积制备. Cauduro等[23]利用直流反应溅射制备MoO3薄膜, 并把样品分别在90, 150, 200, 500 ℃下进行退火, 对于在500 ℃下退火的薄膜, 在表面形成了至少两个不同的MoOx相, 通过表征确定出表面具有两种不同的化学计量比(x ~ 3.0和x < 3.0)组成的MoOx . 通过薄膜的表面功函数进一步确定氧化钼的结晶情况, 功函数约为5.5—6.2 eV的纳米级区域, 对应高度还原MoOx (x < 3), 功函数大于6.3 eV (最高6.9 eV)的区域, 对应晶体α-MoO3. Arita等[24] 用两步法沉积出MoO3薄膜. 在相对较高的衬底温度(300 ℃)下向溅射靶施加了50 W的射频功率用来沉积Mo膜. 之后, 将薄膜在300—800 ℃的空气中热氧化10 min, 得到了α-MoO3薄膜. 如图5所示, X射线衍射(X-ray diffraction, XRD)结果表明: 在400 ℃以上氧化的MoO3薄膜转变为α相晶体. 在350—400 ℃的退火温度条件下观察到αβ相共存的MoO3; 进一步提高退火温度(> 700 ℃)会导致薄膜大量蒸发. 磁控溅射方法制备α-MoO3操作简单, 成本低, 适合大批量生产.

    图 5 在不同温度下氧化的Mo薄膜的XRD图谱[24]\r\nFig. 5. XRD patterns of Mo films oxidized at various temperatures[24].
    图 5  在不同温度下氧化的Mo薄膜的XRD图谱[24]
    Fig. 5.  XRD patterns of Mo films oxidized at various temperatures[24].
    3.2.2   原子层沉积法

    原子层沉积(atomic layer deposition, ALD)是一种先进的气相薄膜沉积方法, 它依赖于交替的自限制表面反应[25,26]. 由于ALD的循环特性, 可以精确地控制薄膜厚度. 图6(a)为ALD制备氧化钼的过程[27], 首先反应源通入腔室吸附在衬底表面, 之后通入氮气吹扫衬底表面过量的反应源, 最后通入氧等离子体和源反应得到最终的氧化钼薄膜, 此过程为一个循环, 通过控制循环次数可以精准调节MoO3的厚度. 2010年, Diskus等[28]利用Mo(CO)6, O3, H2O作为前驱物, 研究臭氧和水在氧化钼生长过程中的作用, 引入臭氧可使每个循环的生长速率增大至0.75 Å. 他们还证明了ALD方法得到的MoO3为无定形薄膜. 2012年, Diskus等[29]进一步研究了无定形的MoO3α-MoO3转变所需的条件. 在空气中对MoO3在400—600 ℃的条件下退火, 图6(b)为退火后的拉曼表征. 当退火温度达到400 ℃时, 无定形的MoO3转变为β-MoO3, 当温度进一步升高达到600 ℃时, 无定形的MoO3转变为α-MoO3. 由ALD方法实现了可控制备原子级厚度的α-MoO3薄膜, 且具有高均匀性、一致性和可重复性.

    图 6 (a) 采用ALD合成MoO3薄膜的一个循环的工艺原理图[27]; (b) 采用ALD在Al2O3上沉积得到的40 nm MoO3薄膜的拉曼光谱: a)沉积和空气中退火, b) 400 ℃退火8 min, c) 600 ℃退火4 min, d) 600 ℃退火7.5 min[29]\r\nFig. 6. (a) Schematic of one cycle ALD process for MoO3 film[27]; (b) Raman spectra of 40 nm thin films of MoO3 deposited on Al2O3 by ALD: a) as deposited, and annealed in air; b) 8 min at 400 ℃; c) 4 min at 600 ℃; d) 7.5 min at 600 ℃[29].
    图 6  (a) 采用ALD合成MoO3薄膜的一个循环的工艺原理图[27]; (b) 采用ALD在Al2O3上沉积得到的40 nm MoO3薄膜的拉曼光谱: a)沉积和空气中退火, b) 400 ℃退火8 min, c) 600 ℃退火4 min, d) 600 ℃退火7.5 min[29]
    Fig. 6.  (a) Schematic of one cycle ALD process for MoO3 film[27]; (b) Raman spectra of 40 nm thin films of MoO3 deposited on Al2O3 by ALD: a) as deposited, and annealed in air; b) 8 min at 400 ℃; c) 4 min at 600 ℃; d) 7.5 min at 600 ℃[29].

    阻变式存储器凭借优越的特性得到广泛研究, 如较长的数据保持时间、高存储密度、与当代互补金属氧化物半导体(complementary metal oxide semiconductor, CMOS)工艺兼容、器件尺寸很容易缩小至10 nm以下等[30-33]. 高介电常数氧化物在电阻开关的应用中得到极大关注, 其中多功能过渡金属氧化物MoO3凭借其高功函数、大电子亲和势、宽禁带以及高的热力学稳定性等特点, 成为理想的电阻存储器应用材料之一[34,35]. 下文将综述氧化钼基电阻随机存储器(resistive Random access memory, RRAM)的研究进展.

    4.1.1   阻变存储器件

    氧化钼基阻变存储器的性能参数对比如表2所列[24,36-45], 可以看出α-MoO3做存储材料, 具有较大的开关比、稳定的耐久性、操作电压低、功耗小等优势. 2012年Arita等[24]利用热氧化钼膜的方法制备MoO3基RRAM器件, 研究了在400—600 ℃下氧化的样品的开关特性. 利用Pt-Ir探针作为顶部电极, 发现此器件既有双极性又有单极性特点, 阻变开关比在10—102之间. 但热氧化钼膜的方法得到的阻变器件开关比较低且稳定性较差. 2015年Tan等[36]利用水热法合成单晶MoO3纳米带, 并采用Au/MoO3/Au结构制作了一维阻变器件. 器件表现出稳定的单极开关特性, 开关比约为103. 在此基础上, 2019年Wang等[37]利用气相传输法合成了高质量二维层状α-MoO3纳米片并制作了阻变器件. 如图7(a)所示, 器件具有单极非易失性电阻开关行为, 在施加正电压时器件既可以发生高阻向低阻的跳变(set), 又可以完成低阻向高阻的转变(reset), 蓝色曲线和绿色曲线分别为第一次set和reset过程. 可以看到, 蓝色曲线在3.2 V时器件阻值处于低阻状态, 即器件初始化(forming)过程. 当再次施加正向电压(绿色曲线)达到4.1 V时电阻急剧下降到高阻状态, 完成器件reset过程. 图中红色曲线在3.2 V时电阻急剧下降, 器件被set, 随着正向电压的继续施加在焦耳热的作用下器件发生reset, 阻值升高, 这个过程为典型的单极性电阻开关行为. 如图7(b)所示, 它的高阻态(high resistance state, HRS)和低阻态(low resistance state, LRS)比值可以达到105、耐久性持续时间超过3 d并且具有良好的稳定性. 2018年, Rahman等[38]用化学气相沉积(chemical vapor deposition, CVD)法合成了层状MoOx薄膜, 并在较低的电激活电压(约2.7 V)下显示出优异的电阻开关性能. 如图7(c)图7(d)所示, 器件开关比达到103, 数据的保持时间大于104 s, 耐久性可以达到6000次以上循环.

    表 2  三氧化钼基忆阻器件性能对比
    Table 2.  Comparison of performance of molybdenum trioxide memristors.
    Ref.Ron/RoffEndurenceVforming/VVset/VVreset/VRetention/sCompliance/mA
    [24]10—102601.8—31—20.5—12—5
    [36]103–2.21.73.310310-3
    [37]> 105503.2—4.13—3.94.45—5.910450
    [38]103> 6000–2.7–22.5> 1040.8
    [39]1824.51.8–1.13.6 × 1031
    [40]~3501—32.6 × 106
    [41]1.15 × 1037484.5 × 103
    [42]1.6 × 102500–33
    [43]106> 1000.8–0.2—1.62 × 104102
    [44]202.2–1.11.4 × 1030.1
    [45]Free0.4–10.1
    下载: 导出CSV 
    | 显示表格
    图 7 (a) 单极电阻开关的代表性行为, 横向面积为2 μm × 2 μm的α-MoO3横杆三明治器件中的I-V曲线, 内嵌图显示了Au/Cr/α-MoO3/Au横杆夹层结构的相应光学图像, 刻度线代表5 μm[37]; (b) α-MoO3横杆器件开关在室温下具有稳定的保持性; 在0.01 V的小偏置下, 通过测量电流来确定HRS和LRS[37]; (c) 存储器单元在重复开关周期下的特性电流-电压曲线[38]; (d) 开关周期的耐久性(读取电压为500 mV时获得的电阻值), 内嵌图为耐久性测量过程中的电压脉冲序列, 每个脉冲的持续时间为2 ms[38]\r\nFig. 7. (a) Representative I-V curves of unipolar resistive switching behavior in the α-MoO3 crossbar sandwich device with a lateral area of 2 μm × 2 μm. The inset shows the corresponding optical image of Au/Cr/α-MoO3/Au crossbar sandwich structures and the scale bar represents 5 μm[37]. (b) Time-dependent measurements of α-MoO3 crossbar device switch featuring stable retention at room temperature. The resistance of the HRS and LRS is determined by measuring the current at a small bias of 0.01 V[37]. (c) Characteristic current-voltage curves of a memory cell subjected to repetitive switching cycles[38]. (d) Performance endurance with respect to number of switching cycles (Resistance values obtained at a read voltage of 500 mV). The inset shows the voltage pulse train during endurance measurement with duration of 2 ms of each pulse[38].
    图 7  (a) 单极电阻开关的代表性行为, 横向面积为2 μm × 2 μm的α-MoO3横杆三明治器件中的I-V曲线, 内嵌图显示了Au/Cr/α-MoO3/Au横杆夹层结构的相应光学图像, 刻度线代表5 μm[37]; (b) α-MoO3横杆器件开关在室温下具有稳定的保持性; 在0.01 V的小偏置下, 通过测量电流来确定HRS和LRS[37]; (c) 存储器单元在重复开关周期下的特性电流-电压曲线[38]; (d) 开关周期的耐久性(读取电压为500 mV时获得的电阻值), 内嵌图为耐久性测量过程中的电压脉冲序列, 每个脉冲的持续时间为2 ms[38]
    Fig. 7.  (a) Representative I-V curves of unipolar resistive switching behavior in the α-MoO3 crossbar sandwich device with a lateral area of 2 μm × 2 μm. The inset shows the corresponding optical image of Au/Cr/α-MoO3/Au crossbar sandwich structures and the scale bar represents 5 μm[37]. (b) Time-dependent measurements of α-MoO3 crossbar device switch featuring stable retention at room temperature. The resistance of the HRS and LRS is determined by measuring the current at a small bias of 0.01 V[37]. (c) Characteristic current-voltage curves of a memory cell subjected to repetitive switching cycles[38]. (d) Performance endurance with respect to number of switching cycles (Resistance values obtained at a read voltage of 500 mV). The inset shows the voltage pulse train during endurance measurement with duration of 2 ms of each pulse[38].
    4.1.2   自整流及阈值开关器件

    2018年, Hsu等[42]利用射频磁控溅射方法制备了MoOx薄膜, 通过调整O2流量, 来控制氧空位含量, 制作了Al/MoOx/Pt阻变器件. 研究了O2流量对MoOx器件自整流特性的影响, 随着氧流量的增大, 器件的高低阻值都增大. 随着氧气流量增大, 对应的非晶格氧减小, 因此电阻的增大得到了很好的解释. 在MoOx/Pt界面会形成高的肖特基势垒, 约为3.3 eV, 在Al/MoOx界面势垒为1.9 eV. 这种不对称电极接触使器件具有自整流特性, 并且随着氧气流量增大, 自整流效果越来越明显. 当O2流量为9 sccm (1 sccm = 1 mL/min)时, 得到了显著的自整流特性阻变特性, 并且电阻的开关比较大(2.0 × 102), 如图8(a)8(b)所示. 2019年, Du等[34]利用一维α-MoO3纳米带制作了横向阻变器件, Au电极α-MoO3纳米带阻变器件表现为典型的非易失双极性阻变特性, 开关电流比可达到5.0 × 102, 开关电压约为3 V. 由于Ag和α-MoO3之间的接触电阻较大, 采用Ag作为电极时可以实现挥发性的双向阈值开关特性, 在3.5 V时, 电流突然增大并在电压正扫描过程中保持在低阻状态, 而在负电压扫描中I-V曲线几乎与正电压扫描曲线对称. 表明当驱动电压降到零时, 低阻状态不能保持. 该阈值开关器件在读取电压为0.5 V时, 开关电流比可达到3.0 × 105.

    图 8 (a) MoOx RRAMs的电流-电压特性[42]; (b) MoOx RRAMs的耐久性[42]\r\nFig. 8. (a) Current-voltage characteristics of the MoOx RRAMs[42]; (b) endurance characteristics of the MoOx RRAMs[42].
    图 8  (a) MoOx RRAMs的电流-电压特性[42]; (b) MoOx RRAMs的耐久性[42]
    Fig. 8.  (a) Current-voltage characteristics of the MoOx RRAMs[42]; (b) endurance characteristics of the MoOx RRAMs[42].

    电子突触器件是神经形态计算系统的重要组成部分, 可以超越冯·诺依曼体系结构的限制, 通过调节突触权重, 将信息同时存储和处理[46-48]. 二维层状结构的α-MoO3具有易于离子插层、微缩能力强、便于集成等特点, 下文将对α-MoO3基神经形态器件的研究进展给出详细的总结.

    4.2.1   三明治结构

    Xue等[45]利用磁控溅射生长MoOx薄膜, 采用图9(a)所示的垂直Ag/MoOx/氧化铟锡(ITO)结构对其突触可塑性进行研究. 图9(b)图9(c)为阻变器件在连续施加相同的直流电压下, 电导的连续增强和抑制现象, 结果表明, 该器件的阻值连续可调特性可以用来模拟突触权重的增强/抑制. 2016年Yang等[49]制作了Ag/MoOx/氟掺杂氧化锡(FTO)结构的垂直阻变器件并研究其突触可塑性, 上电极Ag相当于神经突触的突触前膜, 下电极FTO相当于突触后膜, 电脉冲相当于神经刺激. 重复施加相同的负(正)电压脉冲表现出突触的增强(抑制)特性, 此过程中的电阻变化行为类似于生物突触的传输特性, 实现了105 s的长时可塑性. 另外, 他们通过改变脉冲的时间间隔(Δt)来测试突触晶体管的时间依赖可塑性(spiking-time-dependent plasticity, STDP)行为. STDP学习函数可以很好地拟合兴奋性突触后电流(excitatory postsynaptic current, EPSC)变化(ΔW)与Δt的关系. Wang等[50]利用α-MoO3薄膜构建了突触器件, 通过增大脉宽以及幅值实现了短时可塑向长时增强的转变, 如图9(d)所示. 他们利用25个突触单元构建5 × 5的突触阵列, 依次输入代表字母“x”, “y”和“z”, 字母“x”和“z”利用短时记忆(short-term memory, STM)模式输入. 字母“y”利用长时记忆(long-term memory, LTM)模式输入. 图9(e)显示电导在第一条蓝色虚线标记的时刻自发恢复到原始状态, 即输入最后一个脉冲后8 s. 在LTM模式下输入字母“y”的信号后, 产生较高的电导, 而字母“z”的输入不干扰电导状态, 在整个训练过程中仍然是可区分的, 如图9(f)(h)所示, 字母“x”和“z”的残差变得模糊, 成功地利用突触阵列实现了图像记忆功能.

    图 9 (a) 垂直器件结构示意图及测试说明[45]; (b) 负电压I-V曲线[45]; (c) 正电压I-V曲线; 内嵌图为 ± 0.25 V读电压下随周期的增加, 电流的变化[45]; (d) 由施加的10个不同电压和宽度的脉冲所表示的STM和LTM[50]; (e) 电导在第一个像素内的变化, 蓝色虚线表示对于一个给定字母, 输入最后一个脉冲后的读出时间[50]; (f), (g), (h)在STM模式下(e)图中蓝线标记的时刻所记忆的字母“x”和“z”, 而字母“y”是在LTM模式下被记忆[50]\r\nFig. 9. (a) Schematic illustration of vertical device structure and measurement; (b) I-V curves of negative voltage[45]; (c) I-V curves of positive voltage, the insets show the variation of the current read at ±0.25 V with increasing cycle[45]; (d) STM and LTM demonstrated by applying 10 pulses with different amplitudes and widths[50]; (e) conductance change at the first pixel; the blue dashed lines indicate the read-out moment after inputting the last pulse for a given letter[50]; (f), (g), (h) images memorized at the moments marked by blue lines in (e) letters “x” and “z” are memorized in the STM mode, while the letter “y” is memorized in the LTM mode[50].
    图 9  (a) 垂直器件结构示意图及测试说明[45]; (b) 负电压I-V曲线[45]; (c) 正电压I-V曲线; 内嵌图为 ± 0.25 V读电压下随周期的增加, 电流的变化[45]; (d) 由施加的10个不同电压和宽度的脉冲所表示的STM和LTM[50]; (e) 电导在第一个像素内的变化, 蓝色虚线表示对于一个给定字母, 输入最后一个脉冲后的读出时间[50]; (f), (g), (h)在STM模式下(e)图中蓝线标记的时刻所记忆的字母“x”和“z”, 而字母“y”是在LTM模式下被记忆[50]
    Fig. 9.  (a) Schematic illustration of vertical device structure and measurement; (b) I-V curves of negative voltage[45]; (c) I-V curves of positive voltage, the insets show the variation of the current read at ±0.25 V with increasing cycle[45]; (d) STM and LTM demonstrated by applying 10 pulses with different amplitudes and widths[50]; (e) conductance change at the first pixel; the blue dashed lines indicate the read-out moment after inputting the last pulse for a given letter[50]; (f), (g), (h) images memorized at the moments marked by blue lines in (e) letters “x” and “z” are memorized in the STM mode, while the letter “y” is memorized in the LTM mode[50].
    4.2.2   晶体管结构

    由于神经突触在传递信息的同时会发生信息反馈, 使得突触权重发生改变, 即在突触学习过程中, 信息传递与学习是同时发生的. 在三端阻变器件中, 信息可以通过源-漏间的沟道进行传递, 通过栅极进行反馈信息从而对突触权重进行调节, 使得信息传递和突触学习是独立的, 并且由于栅极电流相比于源漏电流小得多, 因此在突触晶体管中更多关注读取功耗, 使得能耗大大降低[51-53]. α-MoO3的特殊结构, 有利于电子-离子的可逆注入和抽出, 在锂离子电池[54]、光致变色[55]等的研究中已被证实, 基于以上对α-MoO3性质的研究基础, 研究者们将其应用在离子栅结构的突触器件中, 发现α-MoO3可以很好地与离子栅进行匹配. 下面给出了不同功能层材料与α-MoO3为沟道的突触晶体管的性能对比, 从表3可以直观看出α-MoO3在低功耗及快速响应等方面具有明显优势[56-60].

    表 3  突触晶体管性能对比
    Table 3.  Performance comparison of synaptic transistors.
    Ref.Channel materialElectrolyteOperation votage/VPulse width/ msEnergy consumption
    [56]α-MoO3EMIM(CF3SO2)2N2.519.6 pJ
    [57]α-MoO3LiClO4/PEO2.51≈0.16 pJ
    [58]IGZOChitosan2251.0 nJ
    [59]IZOMethyl cellulose0.2—115—6016 nJ
    [60]MoS2chitosan1.5105 pJ
    下载: 导出CSV 
    | 显示表格

    Yang等[56]制作了离子液体栅的突触晶体管, 图10(a)为一种化学递质在生物突触中传递的示意图, 它是两个相邻神经元细胞(突触前和突触后神经元)的连接物. 图10(b)所示为α-MoO3基突触晶体管, 沟道的电导变化代表突触权重的改变, 栅极作为前神经元, 源极和漏极用作后神经元输出端. 在此基础上, Yang等[57]又进一步研究了α-MoO3纳米片为沟道的突触晶体管制作的阵列结构, 原理如图10(c), 基于α-MoO3的突触晶体管的阵列结构作为记忆元件, 其沟道电导变化用作执行反向传播算法的权重更新. 实现了手写体的识别, 图10(d)记录了每个训练周期后模拟网络的识别精度, 可以看出, 该精度高达94.1%.

    图 10 (a) 生物突触示意图[56]; (b) 三端突触晶体管结构原理图[56]; (c) 由电压编程锂离子突触晶体管crossbar阵列和接入器件组成的突触权重层示意图[57]; (d) 对于8 × 8像素手写数字图像, 其识别精度随训练epochs的变化[57]\r\nFig. 10. (a) Schematic illustration of a biological synapse[56]; (b) schematic diagram of the three-terminal synaptic transistor[56]; (c) schematics of a synaptic weight layer composed of voltage programmed Li-ion synaptic transistor crossbar array and access devices[57]; (d) the recognition accuracy evolution with training epochs for 8 × 8 pixel handwritten digit image[57].
    图 10  (a) 生物突触示意图[56]; (b) 三端突触晶体管结构原理图[56]; (c) 由电压编程锂离子突触晶体管crossbar阵列和接入器件组成的突触权重层示意图[57]; (d) 对于8 × 8像素手写数字图像, 其识别精度随训练epochs的变化[57]
    Fig. 10.  (a) Schematic illustration of a biological synapse[56]; (b) schematic diagram of the three-terminal synaptic transistor[56]; (c) schematics of a synaptic weight layer composed of voltage programmed Li-ion synaptic transistor crossbar array and access devices[57]; (d) the recognition accuracy evolution with training epochs for 8 × 8 pixel handwritten digit image[57].

    光调制的神经形态器件在模仿人类视觉系统的基本功能方面具有相当大的潜力, 可以超出可见光区域, 并且光电突触单元可以将图像传感和信息处理结合在一起, 大大提高了数据处理的效率[61,62]. Zhou等[35]利用垂直结构光电电阻随机存储器(optoelectronic resistive random access memory, ORRAM)作突触器件, 实现了图像感知和记忆功能. 在神经形态视觉系统中, ORRAM阵列被用来模拟人视网膜中的传感和预处理功能, 预处理后的图像被传输到三层人工神经网络, 以完成图像训练和识别功能. 阵列中的一个存储单元对应于一个像素单元, 如图11(a)所示, 每个存储单元初始为高阻, set过程利用365 nm波长的光脉冲照射使器件切换到低阻状态, reset过程则是利用负电压脉冲使器件恢复高阻状态. 在神经网络中, ORRAM直接连接到输入层, 如图11(b), 在训练过程中, 图像被ORRAM阵列进行存储, 一个接一个地传递到输入层. 人工神经形态视觉系统采用了一个由6 × 7像素的字母p, u和c的图像组成的图像数据库. 从数据库中提取每个字母的两个图像示例(1000幅图像)进行演示. 图11(c)比较了预处理前后的图像, 经过处理的图像突出显示字母本身特征, 平滑背景噪声信号, 然后将预处理后的图像数据库输入到人工网络中, 进行图像训练和识别. 图11(d)对比了有无ORRAM的视觉系统的识别率, 发现使用ORRAM进行图像预处理后在识别率和效率方面有了明显的提高, 直接证明了ORRAM阵列的图像存储及预处理能力.

    图 11 (a) 8 × 8 ORRAM阵列原理图[35]; (b) 带有ORRAM阵列的三层人工神经网络连接结构[35]; (c) 经ORRAM阵列预处理前(左列)和后(右列)的图像示例[35]; (d) 有无ORRAM阵列图像预处理情况下图像识别率的对比[35]\r\nFig. 11. (a) Schematic structure of an 8 × 8 ORRAM array[35]; (b) structure of the three-layered artificial neural network with ORRAM[35]; (c) examples of images before (left columns) and after (right columns) ORRAM-based pre-processing[35]; (d) comparisons of the image recognition rate with and without ORRAM-based image preprocessing[35].
    图 11  (a) 8 × 8 ORRAM阵列原理图[35]; (b) 带有ORRAM阵列的三层人工神经网络连接结构[35]; (c) 经ORRAM阵列预处理前(左列)和后(右列)的图像示例[35]; (d) 有无ORRAM阵列图像预处理情况下图像识别率的对比[35]
    Fig. 11.  (a) Schematic structure of an 8 × 8 ORRAM array[35]; (b) structure of the three-layered artificial neural network with ORRAM[35]; (c) examples of images before (left columns) and after (right columns) ORRAM-based pre-processing[35]; (d) comparisons of the image recognition rate with and without ORRAM-based image preprocessing[35].

    本文首先系统综述了α-MoO3基本结构及性质, 着重对比分析了MoO3的3个相: 正交晶系(α)、单斜晶系(β)和六方晶系(h). 其次对大面积α-MoO3的合成方法进行了归纳整理, 针对准二维材料α-MoO3, 总结了管式炉方法中生长温度、腔室压力以及前驱物浓度对生长薄膜的质量影响. 最后, 本文对α-MoO3基阻变器件开关比、耐久性以及稳定性等关键性能指标进行对比分析, 总结了α-MoO3基突触器件基于STDP学习规则的模拟特性, 并且展现出低功耗的优异性能, 实现图像记忆和预处理的功能以及手写体识别等.

    二维α-MoO3单晶纳米片的突触器件中在尺寸微缩、计算能耗和CMOS兼容性上展现明显优势, 并且大面积的单晶α-MoO3的制备已被证实可行, 为矩阵计算系统的构建提供了高质量且均一的功能材料. 然而, 面对日益增长的数据处理需求, 下一代存储计算系统亟待解决的问题有: 大面积高质量的二维材料的制备; 内存单元的稳定性; 矩阵计算能力的增强及能耗的降低. 因此, α-MoO3基忆阻器件的性能稳定性提升是接下来研究的重点, 其次, 基于α-MoO3的忆阻器件机理研究还需进一步挖掘, 最重要的就是从α-MoO3的器件到矩阵计算系统的性能研究.

    [1]

    Yang J J, Strukov D B, Stewart D R 2013 Nat. Nanotechnol. 8 13Google Scholar

    [2]

    Chua L O 1971 IEEE Trans. Circuit Theory 18 507Google Scholar

    [3]

    Ielmini D, Wong H S P 2018 Nat. Electron. 1 333Google Scholar

    [4]

    Liu Y, Huang Y, Duan X 2019 Nature 567 323Google Scholar

    [5]

    张宝军, 王芳, 沈稼强, 单欣, 邸希超, 胡凯, 张楷亮 2020 物理学报 69 048101Google Scholar

    Zhang B J, Wang F, Shen J Q, Shan X, Di X C, Hu K, Zhang K L 2020 Acta Phys. Sin. 69 048101Google Scholar

    [6]

    Hanson E D, Lajaunie L, Hao S, Myers B D, Shi F, Murthy A A, Wolverton C, Arenal R, Dravid V P 2017 Adv. Funct. Mater. 27 1605380Google Scholar

    [7]

    Carcia P F, Mccarron E M 1987 Thin Solid Films 155 53Google Scholar

    [8]

    De Castro I A, Datta R S, Ou J Z, Castellanos-Gomez A, Sriram S, Daeneke T, Kalantar-zadeh K 2017 Adv. Mater. 29 1701619Google Scholar

    [9]

    Yao D D, Ou J Z, Latham K, Zhuiykov S, O’Mullane A P, Kalantar-zadeh K 2012 Cryst. Growth Des. 12 1865Google Scholar

    [10]

    Pan W, Tian R, Jin H, Guo Y, Zhang L, Wu X, Zhang L, Han Z, Liu G, Li J, Rao G, Wang H, Chu W 2010 Chem. Mater. 22 6202Google Scholar

    [11]

    Lei Y H, Chen Z X 2016 Appl. Surf. Sci. 361 107Google Scholar

    [12]

    Inzani K, Grande T, Vullum-Bruer F, Selbach S M 2016 J. Phys. Chem. C 120 8959Google Scholar

    [13]

    Wang Y, Du X, Wang J, Su M, Wan X, Meng H, Xie W, Xu J, Liu P 2017 ACS Appl. Mater. Interfaces 9 5543Google Scholar

    [14]

    Crowley K, Ye G, He R, Abbasi K, Gao X P A 2018 ACS Appl. Nano Mater. 1 6407Google Scholar

    [15]

    Zhang C, Pudasaini P R, Oyedele A D, Ievlev A V, Xu L, Haglund A V, Noh J H, Wong A T, Xiao K, Ward T Z, Mandrus D G, Xu H, Ovchinnikova O S, Rack P D 2018 ACS Appl. Mater. Interfaces 10 22623Google Scholar

    [16]

    Zheng B, Wang Z, Chen Y, Zhang W, Li X 2018 2D Mater. 5 045011Google Scholar

    [17]

    Sun H, Zhang H, Jing X, Hu J, Shen K, Liang Z, Hu J, Tian Q, Luo M, Zhu Z, Jiang Z, Huang H, Song F 2019 Appl. Surf. Sci. 476 789Google Scholar

    [18]

    Arash A, Ahmed T, Rajan A G, Walia S, Rahman F, Mazumder A, Ramanathan R, Sriram S, Bhaskaran M, Mayes E 2019 2D Mater. 6 035031Google Scholar

    [19]

    Kim H U, Son J, Kulkarni A, Ahn C, Kim K S, Shin D, Yeom G Y, Kim T 2017 Nanotechnology 28 175601Google Scholar

    [20]

    Molina-Mendoza A J, Lado J L, Island J O, Niño M A, Aballe L, Foerster M, Bruno F Y, López-Moreno A, Vaquero-Garzon L, van der Zant H S J, Rubio-Bollinger G, Agraït N, Pérez E M, Fernández-Rossier J, Castellanos-Gomez A 2016 Chem. Mater. 28 4042Google Scholar

    [21]

    Wang D, Li J N, Zhou Y, Xu D H, Xiong X, Peng R W, Wang M 2016 Appl. Phys. Lett. 108 053107Google Scholar

    [22]

    Diaz-Droguett D E, Fuenzalida V M 2010 J. Nanosci. Nanotechnol. 10 6694Google Scholar

    [23]

    Cauduro A L F, dos Reis R, Chen G, Schmid A K, Méthivier C, Rubahn H G, Bossard-Giannesini L, Cruguel H, Witkowski N, Madsen M 2017 ACS Appl. Mater. Interfaces 9 7717Google Scholar

    [24]

    Arita M, Kaji H, Fujii T, Takahashi Y 2012 Thin Solid Films 520 4762Google Scholar

    [25]

    George S M 2010 Chem. Rev. 110 111Google Scholar

    [26]

    Leskelä M , Ritala M 2002 Thin Solid Films 1 138Google Scholar

    [27]

    Shi M L, Chen L, Zhang T B, Xu J, Zhu H, Sun Q Q, Zhang D W 2017 Small 13 1603157Google Scholar

    [28]

    Diskus M, Nilsen O, Fjellvåg H 2011 J. Mater. Chem. 21 705Google Scholar

    [29]

    Diskus M, Nilsen O, Fjellvåg H, et al. 2012 J. Vac. Sci. Technol. 30 01A107Google Scholar

    [30]

    Zhou G, Ren Z, Wang L, Sun B, Duan S, Song Q 2019 Mater. Horiz. 6 1877Google Scholar

    [31]

    Zhou G, Wu J, Wang L, et al. 2019 Nanoscale 11 17222Google Scholar

    [32]

    Sun L, Zhang Y, Han G, Hwang G, Jiang J, Joo B, Watanabe K, Taniguchi T, Kim Y M, Yu W J, Kong B S, Zhao R, Yang H 2019 Nat. Commun. 10 3161Google Scholar

    [33]

    张志超, 王芳, 吴仕剑, 李毅, 弥伟, 赵金石, 张楷亮 2018 物理学报 67 057301Google Scholar

    Zhang Z C, Wang F, Wu S J, Li Y, Mi W, Zhao J S, Zhang K L 2018 Acta Phys. Sin. 67 057301Google Scholar

    [34]

    Du H, Chen J, Tu M, Luo S, Li S, Yuan S, Gong T, Huang W, Jie W, Hao J 2019 J. Mater. Chem. C 7 12160Google Scholar

    [35]

    Zhou F, Zhou Z, Chen J, Choy T H, Wang J, Zhang N, Lin Z, Yu S, Kang J, Wong H S P, Chai Y 2019 Nat. Nanotechnol. 14 776Google Scholar

    [36]

    Tan Z H, Yin X B, Guo X 2015 Appl. Phys. Lett. 106 023503Google Scholar

    [37]

    Wang J, Wang F, Yin L, Sendeku M G, Zhang Y, Cheng R, Wang Z, Li N, Huang W, He J 2019 Nanoscale 11 20497Google Scholar

    [38]

    Rahman F, Ahmed T, Walia S, Mayes E, Sriram S, Bhaskaran M, Balendhran S 2018 Nanoscale 10 19711Google Scholar

    [39]

    Yang C S, Shang D S, Chai Y S, Yan L Q, Shen B G, Sun Y 2016 Phys. Chem. Chem. Phys. 18 12466Google Scholar

    [40]

    Hasegawa J, Nagase T, Kobayashi T, Naito H 2016 Jpn. J. Appl. Phys. 55 03DC05Google Scholar

    [41]

    Abhijith T, Kumar T V, Reddy V S 2017 Nanotechnology 28 095203Google Scholar

    [42]

    Hsu C C, Wang S Y, Lin Y S, Chen Y T 2019 J. Alloys Compd. 779 609Google Scholar

    [43]

    Fang S L, Liu W H, Li X, Wang X L, Geng L, Wu M S, Huang X D, Han C Y 2019 Appl. Phys. Lett. 115 244102Google Scholar

    [44]

    Dai T, Qian L, Ren Y, Liu X 2017 International Conference on Electron Devices and Solid-State Circuits (EDSSC) Hsinchu, Taiwan China, October 18–20, 2017 p1

    [45]

    Xue Q, Wang Y C, Wei X H 2019 Appl. Surf. Sci. 479 469Google Scholar

    [46]

    Kim G H, Lee, J H, Ahn Y, Jeon W, Song S J, Seok J Y, Yoon J H, Yoon K J, Park T J, Hwang C S 2013 Adv. Funct. Mater. 23 1440Google Scholar

    [47]

    Rehman M M, Rehman H M M U, Gul J Z, Kim W Y, Karimov K S, Ahmed N 2020 Sci. Technol. Adv. Mater. 21 147Google Scholar

    [48]

    Wang X, Xie W, Xu J B 2014 Adv. Mater. 26 5496Google Scholar

    [49]

    Yang C S, Shang D S, Chai Y S, Yan L Q, Shen B G, Sun Y 2017 Phys. Chem. Chem. Phys. 19 4190Google Scholar

    [50]

    Wang Z, Yang R, Huang H M, He H K, Shaibo J, Guo X 2020 Adv. Electron. Mater. 6 1901290Google Scholar

    [51]

    Danesh C D, Shaffer C M, Nathan D, Shenoy R, Tudor A, Tadayon M, Lin Y, Chen Y 2019 Adv. Mater. 31 e1808032Google Scholar

    [52]

    Lenz J, Del Giudice F, Geisenhof F R, Winterer F, Weitz R T 2019 Nat. Nanotechnol. 14 579Google Scholar

    [53]

    Nie S, He Y, Liu R, Shi Y, Wan Q 2019 IEEE Electron Device Lett. 40 459Google Scholar

    [54]

    Chernova N A, Roppolo M, Dillon A C, Whittingham M S 2009 J. Mater. Chem. 19 2526Google Scholar

    [55]

    李琦, 章勇 2018 物理学报 67 067201Google Scholar

    Li Q, Zhang Y 2018 Acta Phys. Sin. 67 067201Google Scholar

    [56]

    Yang C S, Shang D S, Liu N, Shi G, Shen X, Yu R C, Li Y Q, Sun Y 2017 Adv. Mater. 29 1700906Google Scholar

    [57]

    Yang C S, Shang D S, Liu N, et al. 2018 Adv. Funct. Mater. 28 1804170Google Scholar

    [58]

    He Y, Nie S, Liu R, Jiang S, Shi Y, Wan Q 2019 Adv. Mater. 31 e1900903Google Scholar

    [59]

    Guo L, Wen J, Ding J, Wan C, Cheng G 2016 Sci. Rep. 6 38578Google Scholar

    [60]

    Hu W, Jiang J, Xie D, Liu B, Yang J, He J 2019 J. Mater. Chem. C 7 682Google Scholar

    [61]

    Shao L, Wang H, Yang Y, et al. 2019 ACS Appl. Mater. Interfaces 11 12161Google Scholar

    [62]

    Wang K, Dai S, Zhao Y, Wang Y, Liu C, Huang J 2019 Small 15 e1900010Google Scholar

    期刊类型引用(1)

    1. 单欣,刘平,王芳,谢杨杨,魏俊青,马泽夏,石瑶,孙翰,鲁世豪,宋志棠,闫小兵,张楷亮. 基于二维α-MoO_3的多值存储特性及其双重导电机制研究(英文). Science China(Materials). 2023(12): 4773-4781 . 百度学术

    其他类型引用(1)

  • 图 1  (a) α-MoO3晶体结构; (b) β-MoO3晶体结构; (c) h-MoO3晶体结构

    Figure 1.  (a) Crystal structure of α-MoO3; (b) crystal structure of β-MoO3; (c) crystal structure of h-MoO3.

    图 2  (a) 块体电子结构[6]; (b) 单层电子结构[6]

    Figure 2.  (a) Bulk electronic structure[6]; (b) monolayer electronic structure[6].

    图 3  (a) 在化学计量比情况下MoO3能带结构[12]; (b) 在O1空位情况下MoO3能带结构[12]; (c) 在O2 空位情况下MoO3能带结构[12]; (d) 在O3空位情况下MoO3能带结构[12]

    Figure 3.  (a) Band structures of MoO3 in the stoichiometric case[12]; (b) band structures of MoO3 in the vacancy on O1 case[12]; (c) band structures of MoO3 in the vacancy on O2 case[12]; (d) band structures of MoO3 in the vacancy on O3 case[12].

    图 4  (a) 在两温区的物理气相沉积系统中合成α-MoO3原理图[13]; (b) 在SiO2/Si衬底上生长的α-MoO3单晶的照片[16]; (c) α-MoO3的HRTEM图像, 内嵌图为相应的透射电子显微镜图像和傅里叶变换图像[16]; (d) 生长α-MoO3的O 1s和Mo 3d的高分辨XPS能谱[16]

    Figure 4.  (a) Schematic illustration of synthesis of α-MoO3 in a two-temperature-zone PVD system[13]; (b) photograph of as-grown α-MoO3 single crystals on SiO2/Si substrate[16]; (c) HRTEM image of the α-MoO3 sheet. Insets: corresponding TEM and FFT images[16]; (d) high-resolution XPS spectra of O 1s and Mo 3d of as-grown α-MoO3[16].

    图 5  在不同温度下氧化的Mo薄膜的XRD图谱[24]

    Figure 5.  XRD patterns of Mo films oxidized at various temperatures[24].

    图 6  (a) 采用ALD合成MoO3薄膜的一个循环的工艺原理图[27]; (b) 采用ALD在Al2O3上沉积得到的40 nm MoO3薄膜的拉曼光谱: a)沉积和空气中退火, b) 400 ℃退火8 min, c) 600 ℃退火4 min, d) 600 ℃退火7.5 min[29]

    Figure 6.  (a) Schematic of one cycle ALD process for MoO3 film[27]; (b) Raman spectra of 40 nm thin films of MoO3 deposited on Al2O3 by ALD: a) as deposited, and annealed in air; b) 8 min at 400 ℃; c) 4 min at 600 ℃; d) 7.5 min at 600 ℃[29].

    图 7  (a) 单极电阻开关的代表性行为, 横向面积为2 μm × 2 μm的α-MoO3横杆三明治器件中的I-V曲线, 内嵌图显示了Au/Cr/α-MoO3/Au横杆夹层结构的相应光学图像, 刻度线代表5 μm[37]; (b) α-MoO3横杆器件开关在室温下具有稳定的保持性; 在0.01 V的小偏置下, 通过测量电流来确定HRS和LRS[37]; (c) 存储器单元在重复开关周期下的特性电流-电压曲线[38]; (d) 开关周期的耐久性(读取电压为500 mV时获得的电阻值), 内嵌图为耐久性测量过程中的电压脉冲序列, 每个脉冲的持续时间为2 ms[38]

    Figure 7.  (a) Representative I-V curves of unipolar resistive switching behavior in the α-MoO3 crossbar sandwich device with a lateral area of 2 μm × 2 μm. The inset shows the corresponding optical image of Au/Cr/α-MoO3/Au crossbar sandwich structures and the scale bar represents 5 μm[37]. (b) Time-dependent measurements of α-MoO3 crossbar device switch featuring stable retention at room temperature. The resistance of the HRS and LRS is determined by measuring the current at a small bias of 0.01 V[37]. (c) Characteristic current-voltage curves of a memory cell subjected to repetitive switching cycles[38]. (d) Performance endurance with respect to number of switching cycles (Resistance values obtained at a read voltage of 500 mV). The inset shows the voltage pulse train during endurance measurement with duration of 2 ms of each pulse[38].

    图 8  (a) MoOx RRAMs的电流-电压特性[42]; (b) MoOx RRAMs的耐久性[42]

    Figure 8.  (a) Current-voltage characteristics of the MoOx RRAMs[42]; (b) endurance characteristics of the MoOx RRAMs[42].

    图 9  (a) 垂直器件结构示意图及测试说明[45]; (b) 负电压I-V曲线[45]; (c) 正电压I-V曲线; 内嵌图为 ± 0.25 V读电压下随周期的增加, 电流的变化[45]; (d) 由施加的10个不同电压和宽度的脉冲所表示的STM和LTM[50]; (e) 电导在第一个像素内的变化, 蓝色虚线表示对于一个给定字母, 输入最后一个脉冲后的读出时间[50]; (f), (g), (h)在STM模式下(e)图中蓝线标记的时刻所记忆的字母“x”和“z”, 而字母“y”是在LTM模式下被记忆[50]

    Figure 9.  (a) Schematic illustration of vertical device structure and measurement; (b) I-V curves of negative voltage[45]; (c) I-V curves of positive voltage, the insets show the variation of the current read at ±0.25 V with increasing cycle[45]; (d) STM and LTM demonstrated by applying 10 pulses with different amplitudes and widths[50]; (e) conductance change at the first pixel; the blue dashed lines indicate the read-out moment after inputting the last pulse for a given letter[50]; (f), (g), (h) images memorized at the moments marked by blue lines in (e) letters “x” and “z” are memorized in the STM mode, while the letter “y” is memorized in the LTM mode[50].

    图 10  (a) 生物突触示意图[56]; (b) 三端突触晶体管结构原理图[56]; (c) 由电压编程锂离子突触晶体管crossbar阵列和接入器件组成的突触权重层示意图[57]; (d) 对于8 × 8像素手写数字图像, 其识别精度随训练epochs的变化[57]

    Figure 10.  (a) Schematic illustration of a biological synapse[56]; (b) schematic diagram of the three-terminal synaptic transistor[56]; (c) schematics of a synaptic weight layer composed of voltage programmed Li-ion synaptic transistor crossbar array and access devices[57]; (d) the recognition accuracy evolution with training epochs for 8 × 8 pixel handwritten digit image[57].

    图 11  (a) 8 × 8 ORRAM阵列原理图[35]; (b) 带有ORRAM阵列的三层人工神经网络连接结构[35]; (c) 经ORRAM阵列预处理前(左列)和后(右列)的图像示例[35]; (d) 有无ORRAM阵列图像预处理情况下图像识别率的对比[35]

    Figure 11.  (a) Schematic structure of an 8 × 8 ORRAM array[35]; (b) structure of the three-layered artificial neural network with ORRAM[35]; (c) examples of images before (left columns) and after (right columns) ORRAM-based pre-processing[35]; (d) comparisons of the image recognition rate with and without ORRAM-based image preprocessing[35].

    表 1  一步法制备α-MoO3的生长条件(1 Torr = 1.33322 × 102 Pa)

    Table 1.  Growth conditions of α-MoO3 prepared by one-step method.

    Ref.Carrier flowPressureSubstrate temperature/℃Growth time/minLateral sizeThickness/nm
    [13]Atmospheric58012069.6 μm5.3—89.8
    [16]Atmospheric850—1050300.72 cm
    [17]ArAtmospheric75020Centimeters scale4.3—550
    [18]Ar310 mTorr53045500 µm × 500 µm8 ± 0.75
    [19]Ar & O2100 mTorr15090Wafer-scale30
    [20]Atmospheric54020Centimeters scale≥ 1.4
    [21]N2Atmospheric350—48020Several tens micron2.8—14
    DownLoad: CSV

    表 2  三氧化钼基忆阻器件性能对比

    Table 2.  Comparison of performance of molybdenum trioxide memristors.

    Ref.Ron/RoffEndurenceVforming/VVset/VVreset/VRetention/sCompliance/mA
    [24]10—102601.8—31—20.5—12—5
    [36]103–2.21.73.310310-3
    [37]> 105503.2—4.13—3.94.45—5.910450
    [38]103> 6000–2.7–22.5> 1040.8
    [39]1824.51.8–1.13.6 × 1031
    [40]~3501—32.6 × 106
    [41]1.15 × 1037484.5 × 103
    [42]1.6 × 102500–33
    [43]106> 1000.8–0.2—1.62 × 104102
    [44]202.2–1.11.4 × 1030.1
    [45]Free0.4–10.1
    DownLoad: CSV

    表 3  突触晶体管性能对比

    Table 3.  Performance comparison of synaptic transistors.

    Ref.Channel materialElectrolyteOperation votage/VPulse width/ msEnergy consumption
    [56]α-MoO3EMIM(CF3SO2)2N2.519.6 pJ
    [57]α-MoO3LiClO4/PEO2.51≈0.16 pJ
    [58]IGZOChitosan2251.0 nJ
    [59]IZOMethyl cellulose0.2—115—6016 nJ
    [60]MoS2chitosan1.5105 pJ
    DownLoad: CSV
  • [1]

    Yang J J, Strukov D B, Stewart D R 2013 Nat. Nanotechnol. 8 13Google Scholar

    [2]

    Chua L O 1971 IEEE Trans. Circuit Theory 18 507Google Scholar

    [3]

    Ielmini D, Wong H S P 2018 Nat. Electron. 1 333Google Scholar

    [4]

    Liu Y, Huang Y, Duan X 2019 Nature 567 323Google Scholar

    [5]

    张宝军, 王芳, 沈稼强, 单欣, 邸希超, 胡凯, 张楷亮 2020 物理学报 69 048101Google Scholar

    Zhang B J, Wang F, Shen J Q, Shan X, Di X C, Hu K, Zhang K L 2020 Acta Phys. Sin. 69 048101Google Scholar

    [6]

    Hanson E D, Lajaunie L, Hao S, Myers B D, Shi F, Murthy A A, Wolverton C, Arenal R, Dravid V P 2017 Adv. Funct. Mater. 27 1605380Google Scholar

    [7]

    Carcia P F, Mccarron E M 1987 Thin Solid Films 155 53Google Scholar

    [8]

    De Castro I A, Datta R S, Ou J Z, Castellanos-Gomez A, Sriram S, Daeneke T, Kalantar-zadeh K 2017 Adv. Mater. 29 1701619Google Scholar

    [9]

    Yao D D, Ou J Z, Latham K, Zhuiykov S, O’Mullane A P, Kalantar-zadeh K 2012 Cryst. Growth Des. 12 1865Google Scholar

    [10]

    Pan W, Tian R, Jin H, Guo Y, Zhang L, Wu X, Zhang L, Han Z, Liu G, Li J, Rao G, Wang H, Chu W 2010 Chem. Mater. 22 6202Google Scholar

    [11]

    Lei Y H, Chen Z X 2016 Appl. Surf. Sci. 361 107Google Scholar

    [12]

    Inzani K, Grande T, Vullum-Bruer F, Selbach S M 2016 J. Phys. Chem. C 120 8959Google Scholar

    [13]

    Wang Y, Du X, Wang J, Su M, Wan X, Meng H, Xie W, Xu J, Liu P 2017 ACS Appl. Mater. Interfaces 9 5543Google Scholar

    [14]

    Crowley K, Ye G, He R, Abbasi K, Gao X P A 2018 ACS Appl. Nano Mater. 1 6407Google Scholar

    [15]

    Zhang C, Pudasaini P R, Oyedele A D, Ievlev A V, Xu L, Haglund A V, Noh J H, Wong A T, Xiao K, Ward T Z, Mandrus D G, Xu H, Ovchinnikova O S, Rack P D 2018 ACS Appl. Mater. Interfaces 10 22623Google Scholar

    [16]

    Zheng B, Wang Z, Chen Y, Zhang W, Li X 2018 2D Mater. 5 045011Google Scholar

    [17]

    Sun H, Zhang H, Jing X, Hu J, Shen K, Liang Z, Hu J, Tian Q, Luo M, Zhu Z, Jiang Z, Huang H, Song F 2019 Appl. Surf. Sci. 476 789Google Scholar

    [18]

    Arash A, Ahmed T, Rajan A G, Walia S, Rahman F, Mazumder A, Ramanathan R, Sriram S, Bhaskaran M, Mayes E 2019 2D Mater. 6 035031Google Scholar

    [19]

    Kim H U, Son J, Kulkarni A, Ahn C, Kim K S, Shin D, Yeom G Y, Kim T 2017 Nanotechnology 28 175601Google Scholar

    [20]

    Molina-Mendoza A J, Lado J L, Island J O, Niño M A, Aballe L, Foerster M, Bruno F Y, López-Moreno A, Vaquero-Garzon L, van der Zant H S J, Rubio-Bollinger G, Agraït N, Pérez E M, Fernández-Rossier J, Castellanos-Gomez A 2016 Chem. Mater. 28 4042Google Scholar

    [21]

    Wang D, Li J N, Zhou Y, Xu D H, Xiong X, Peng R W, Wang M 2016 Appl. Phys. Lett. 108 053107Google Scholar

    [22]

    Diaz-Droguett D E, Fuenzalida V M 2010 J. Nanosci. Nanotechnol. 10 6694Google Scholar

    [23]

    Cauduro A L F, dos Reis R, Chen G, Schmid A K, Méthivier C, Rubahn H G, Bossard-Giannesini L, Cruguel H, Witkowski N, Madsen M 2017 ACS Appl. Mater. Interfaces 9 7717Google Scholar

    [24]

    Arita M, Kaji H, Fujii T, Takahashi Y 2012 Thin Solid Films 520 4762Google Scholar

    [25]

    George S M 2010 Chem. Rev. 110 111Google Scholar

    [26]

    Leskelä M , Ritala M 2002 Thin Solid Films 1 138Google Scholar

    [27]

    Shi M L, Chen L, Zhang T B, Xu J, Zhu H, Sun Q Q, Zhang D W 2017 Small 13 1603157Google Scholar

    [28]

    Diskus M, Nilsen O, Fjellvåg H 2011 J. Mater. Chem. 21 705Google Scholar

    [29]

    Diskus M, Nilsen O, Fjellvåg H, et al. 2012 J. Vac. Sci. Technol. 30 01A107Google Scholar

    [30]

    Zhou G, Ren Z, Wang L, Sun B, Duan S, Song Q 2019 Mater. Horiz. 6 1877Google Scholar

    [31]

    Zhou G, Wu J, Wang L, et al. 2019 Nanoscale 11 17222Google Scholar

    [32]

    Sun L, Zhang Y, Han G, Hwang G, Jiang J, Joo B, Watanabe K, Taniguchi T, Kim Y M, Yu W J, Kong B S, Zhao R, Yang H 2019 Nat. Commun. 10 3161Google Scholar

    [33]

    张志超, 王芳, 吴仕剑, 李毅, 弥伟, 赵金石, 张楷亮 2018 物理学报 67 057301Google Scholar

    Zhang Z C, Wang F, Wu S J, Li Y, Mi W, Zhao J S, Zhang K L 2018 Acta Phys. Sin. 67 057301Google Scholar

    [34]

    Du H, Chen J, Tu M, Luo S, Li S, Yuan S, Gong T, Huang W, Jie W, Hao J 2019 J. Mater. Chem. C 7 12160Google Scholar

    [35]

    Zhou F, Zhou Z, Chen J, Choy T H, Wang J, Zhang N, Lin Z, Yu S, Kang J, Wong H S P, Chai Y 2019 Nat. Nanotechnol. 14 776Google Scholar

    [36]

    Tan Z H, Yin X B, Guo X 2015 Appl. Phys. Lett. 106 023503Google Scholar

    [37]

    Wang J, Wang F, Yin L, Sendeku M G, Zhang Y, Cheng R, Wang Z, Li N, Huang W, He J 2019 Nanoscale 11 20497Google Scholar

    [38]

    Rahman F, Ahmed T, Walia S, Mayes E, Sriram S, Bhaskaran M, Balendhran S 2018 Nanoscale 10 19711Google Scholar

    [39]

    Yang C S, Shang D S, Chai Y S, Yan L Q, Shen B G, Sun Y 2016 Phys. Chem. Chem. Phys. 18 12466Google Scholar

    [40]

    Hasegawa J, Nagase T, Kobayashi T, Naito H 2016 Jpn. J. Appl. Phys. 55 03DC05Google Scholar

    [41]

    Abhijith T, Kumar T V, Reddy V S 2017 Nanotechnology 28 095203Google Scholar

    [42]

    Hsu C C, Wang S Y, Lin Y S, Chen Y T 2019 J. Alloys Compd. 779 609Google Scholar

    [43]

    Fang S L, Liu W H, Li X, Wang X L, Geng L, Wu M S, Huang X D, Han C Y 2019 Appl. Phys. Lett. 115 244102Google Scholar

    [44]

    Dai T, Qian L, Ren Y, Liu X 2017 International Conference on Electron Devices and Solid-State Circuits (EDSSC) Hsinchu, Taiwan China, October 18–20, 2017 p1

    [45]

    Xue Q, Wang Y C, Wei X H 2019 Appl. Surf. Sci. 479 469Google Scholar

    [46]

    Kim G H, Lee, J H, Ahn Y, Jeon W, Song S J, Seok J Y, Yoon J H, Yoon K J, Park T J, Hwang C S 2013 Adv. Funct. Mater. 23 1440Google Scholar

    [47]

    Rehman M M, Rehman H M M U, Gul J Z, Kim W Y, Karimov K S, Ahmed N 2020 Sci. Technol. Adv. Mater. 21 147Google Scholar

    [48]

    Wang X, Xie W, Xu J B 2014 Adv. Mater. 26 5496Google Scholar

    [49]

    Yang C S, Shang D S, Chai Y S, Yan L Q, Shen B G, Sun Y 2017 Phys. Chem. Chem. Phys. 19 4190Google Scholar

    [50]

    Wang Z, Yang R, Huang H M, He H K, Shaibo J, Guo X 2020 Adv. Electron. Mater. 6 1901290Google Scholar

    [51]

    Danesh C D, Shaffer C M, Nathan D, Shenoy R, Tudor A, Tadayon M, Lin Y, Chen Y 2019 Adv. Mater. 31 e1808032Google Scholar

    [52]

    Lenz J, Del Giudice F, Geisenhof F R, Winterer F, Weitz R T 2019 Nat. Nanotechnol. 14 579Google Scholar

    [53]

    Nie S, He Y, Liu R, Shi Y, Wan Q 2019 IEEE Electron Device Lett. 40 459Google Scholar

    [54]

    Chernova N A, Roppolo M, Dillon A C, Whittingham M S 2009 J. Mater. Chem. 19 2526Google Scholar

    [55]

    李琦, 章勇 2018 物理学报 67 067201Google Scholar

    Li Q, Zhang Y 2018 Acta Phys. Sin. 67 067201Google Scholar

    [56]

    Yang C S, Shang D S, Liu N, Shi G, Shen X, Yu R C, Li Y Q, Sun Y 2017 Adv. Mater. 29 1700906Google Scholar

    [57]

    Yang C S, Shang D S, Liu N, et al. 2018 Adv. Funct. Mater. 28 1804170Google Scholar

    [58]

    He Y, Nie S, Liu R, Jiang S, Shi Y, Wan Q 2019 Adv. Mater. 31 e1900903Google Scholar

    [59]

    Guo L, Wen J, Ding J, Wan C, Cheng G 2016 Sci. Rep. 6 38578Google Scholar

    [60]

    Hu W, Jiang J, Xie D, Liu B, Yang J, He J 2019 J. Mater. Chem. C 7 682Google Scholar

    [61]

    Shao L, Wang H, Yang Y, et al. 2019 ACS Appl. Mater. Interfaces 11 12161Google Scholar

    [62]

    Wang K, Dai S, Zhao Y, Wang Y, Liu C, Huang J 2019 Small 15 e1900010Google Scholar

  • [1] Wu Peng, Tan Lun, Li Wei, Cao Li-Wei, Zhao Jun-Bo, Qu Yao, Li Ang. Preparation and photoelectric property of large scale monolayer MoS2. Acta Physica Sinica, 2023, 72(11): 118101. doi: 10.7498/aps.72.20230273
    [2] Bi Wen-Jie, Yang Shuang, Zhou Jing, Jin Wei, Chen Wen. Research on synthesis of Cu3Mo2O9/MoO3 nanocomposite and trimethylamine gas sensing properties. Acta Physica Sinica, 2023, 72(16): 168103. doi: 10.7498/aps.72.20230720
    [3] Zhu Jia-Xue, Zhang Xu-Meng, Wang Rui, Liu Qi. Flexible memristive spiking neuron for neuromorphic sensing and computing. Acta Physica Sinica, 2022, 71(14): 148503. doi: 10.7498/aps.71.20212323
    [4] Zhang Xiao, Lü Jia-Yu, Guan Yan-Qiu, Li Hui, Wang Xi-Ming, Zhang La-Bao, Wang Hao, Tu Xue-Cou, Kang Lin, Jia Xiao-Qing, Zhao Qing-Yuan, Chen Jian, Wu Pei-Heng. Design and fabrication of single photon detector with ultra-large area superconducting nanowire array. Acta Physica Sinica, 2022, 71(24): 248501. doi: 10.7498/aps.71.20221569
    [5] Shen Liu-Feng, Hu Ling-Xiang, Kang Feng-Wen, Ye Yu-Min, Zhuge Fei. Optoelectronic neuromorphic devices and their applications. Acta Physica Sinica, 2022, 71(14): 148505. doi: 10.7498/aps.71.20220111
    [6] Chen Yang-Yang, He Yu-Hui, Miao Xiang-Shui, Yang Dao-Hong. 3D-NAND flash memory based neuromorphic computing. Acta Physica Sinica, 2022, 71(21): 210702. doi: 10.7498/aps.71.20220974
    [7] Li Ce, Yang Dong-Liang, Sun Lin-Feng. Research progress of neuromorphic devices based on two-dimensional layered materials. Acta Physica Sinica, 2022, 71(21): 218504. doi: 10.7498/aps.71.20221424
    [8] Wang Shuo, Wang Wen-Hui, Lü Jun-Peng, Ni Zhen-Hua. Chemical vapor deposition growth of large-areas two dimensional materials: Approaches and mechanisms. Acta Physica Sinica, 2021, 70(2): 026802. doi: 10.7498/aps.70.20201398
    [9] Wang Hai-Yu, Liu Ying-Jie, Xun Lu-Lu, Li Jing, Yang Qing, Tian Qi-Yun, Nie Tian-Xiao, Zhao Wei-Sheng. Research progress of preparation of large-scale two-dimensional magnetic materials and manipulation of Curie temperature. Acta Physica Sinica, 2021, 70(12): 127301. doi: 10.7498/aps.70.20210223
    [10] Ren Kuan, Zhang Ke-Jia, Qin Xi-Zi, Ren Huan-Xin, Zhu Shou-Hui, Yang Feng, Sun Bai, Zhao Yong, Zhang Yong. Research progress of neuromorphic computation based on memcapacitors. Acta Physica Sinica, 2021, 70(7): 078701. doi: 10.7498/aps.70.20201632
    [11] Wei Chen-Wei, Cao Tun. α-MoO3 based tunable Fabry-Pérot cavity colorimetric biosensor. Acta Physica Sinica, 2021, 70(4): 048701. doi: 10.7498/aps.70.20201548
    [12] Gao Bo-Wen, Meng Jing. Large area and flexible CH3NH3PbI3 perovskite solar cell fabricated by all ink jet printing. Acta Physica Sinica, 2021, 70(20): 208801. doi: 10.7498/aps.70.20210788
    [13] Li Rui-Dong, Deng Jin-Xiang, Zhang Hao, Xu Zhi-Yang, Pan Zhi-Wei, Sun Jun-Jie, Wang Gui-Sheng. Preparation, optical, and electrical properties of rubrene∶MoO3 films. Acta Physica Sinica, 2019, 68(17): 178101. doi: 10.7498/aps.68.20190035
    [14] Wang Wen-Rong, Zhou Yu-Xiu, Li Tie, Wang Yue-Lin, Xie Xiao-Ming. Research on synthesis of high-quality and large-scale graphene films by chemical vapor deposition. Acta Physica Sinica, 2012, 61(3): 038702. doi: 10.7498/aps.61.038702
    [15] You Feng, Ji Lu, Xie Qing-Lian, Wang Zheng, Yue Hong-Wei, Zhao Xin-Jie, Fang Lan, Yan Shao-Lin. Fabrication and properties of large area Tl2Ba2CaCu2O8 superconducting thin film on sapphire substrate. Acta Physica Sinica, 2010, 59(7): 5035-5043. doi: 10.7498/aps.59.5035
    [16] Cai Ya-Ping, Li Wei, Feng Liang-Huan, Li Bing, Cai Wei, Lei Zhi, Zhang Jing-Quan, Wu Li-Li, Zheng Jia-Gui. Preparation of large area CdS thin film and its application in photovoltaic cells. Acta Physica Sinica, 2009, 58(1): 438-443. doi: 10.7498/aps.58.438
    [17] Hai Kuo, Tang Dong-Sheng, Yuan Hua-Jun, Peng Yue-Hua, Luo Zhi-Hua, Liu Hong-Xia, Chen Ya-Qi, Yu Fang, Yang Yi. Synthesis of large-area, vertically aligned α-Fe2O3 nanowire and nanobelt arrays. Acta Physica Sinica, 2009, 58(2): 1120-1125. doi: 10.7498/aps.58.1120
    [18] Zhang Hong-Jun, Wang Dong, Chen Zhi-Quan, Wang Shao-Jie, Xu You-Ming, Luo Xi-Hui. Positron annihilation study of Mo dispersion in MoO3/Al2O3 catalysts. Acta Physica Sinica, 2008, 57(11): 7333-7337. doi: 10.7498/aps.57.7333
    [19] San Hai-Sheng, Chen Chong, He Yu-Yang, Wang Jun, Feng Bo-Xue. Study on electrical properties of n-type transparent and conductiveCdIn22O44 thin film and the optimum preparation parameters for large- area film. Acta Physica Sinica, 2005, 54(4): 1736-1741. doi: 10.7498/aps.54.1736
    [20] Oyangxiaoping, Li Zhen-Fu, Zhang Guo-Guang, Huo Yu-Kun, Zhang Qian-Mei, Zhang Xian-Peng, Song Xian-Cai, Jia Huan-Yi, Lei Jian-Hua, Sun Yuan-Cheng. . Acta Physica Sinica, 2002, 51(7): 1502-1505. doi: 10.7498/aps.51.1502
  • 期刊类型引用(1)

    1. 单欣,刘平,王芳,谢杨杨,魏俊青,马泽夏,石瑶,孙翰,鲁世豪,宋志棠,闫小兵,张楷亮. 基于二维α-MoO_3的多值存储特性及其双重导电机制研究(英文). Science China(Materials). 2023(12): 4773-4781 . 百度学术

    其他类型引用(1)

Metrics
  • Abstract views:  11939
  • PDF Downloads:  249
  • Cited By: 2
Publishing process
  • Received Date:  31 October 2020
  • Accepted Date:  09 December 2020
  • Available Online:  22 April 2021
  • Published Online:  05 May 2021

/

返回文章
返回