High-uniformity Memristor Arrays Based on Two-dimensional MoTe2 for Neuromorphic Computing

HE Huikai, YANG Rui, XIA Jian, WANG Tingze, DONG Dequan, MIAO Xiangshui

High-uniformity Memristor Arrays Based on Two-dimensional MoTe2for Neuromorphic Computing

HE Huikai1, YANG Rui2,3, XIA Jian3,4, WANG Tingze3,4, DONG Dequan3,4, MIAO Xiangshui2,3

(1. Nanhu Academy of Electronics and Information Technology, Jiaxing 314002, China; 2. Hubei Yangtze Memory Laboratories, Wuhan 430205, China; 3. Wuhan National Laboratory for Optoelectronics, School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan 430074, China; 4. State Key Laboratory of Material Processing and Die & Mould Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China)

Two-dimensional transition metal dichalcogenides are appealing materials for the preparation of nanoelectronic devices, and the development of memristors for information storage and neuromorphic computing using such materials is of particular interest. However, memristor arrays based on two-dimensional transition metal dichalcogenides are rarely reported due to low yield and high device-to-device variability. Herein, the 2D MoTe2film was prepared by the chemical vapor deposition method. Then the memristive devices based on 2D MoTe2film were fabricated through the polymethyl methacrylate transfer method and the lift-off process. The as-prepared MoTe2devices perform stable bipolar resistive switching, including superior retention characteristics (>500 s), fast switching (～60 ns for SET and ～280 ns for RESET), and excellent endurance (>2000 cycles). More importantly, the MoTe2devices exhibit high yield (96%), low cycle-to-cycle variability (6.6% for SET and 5.2% for RESET), and low device-to-device variability (19.9% for SET and 15.6% for RESET). In addition, a 3×3 memristor array with 1R scheme is successfully demonstrated based on 2D MoTe2film. And, high recognition accuracy (91.3%) is realized by simulation in the artificial neural network with the MoTe2devices working as synapses. It is found that the formation/rupture of metallic filaments is the dominating switching mechanism based on the investigations of the electron transport characteristics of high and low resistance states in the present MoTe2devices. This work demonstrates that large-scale two-dimensional transition metal dichalcogenides film is of great potential forfuture applications in neuromorphic computing.

two-dimensional material; MoTe2; memristor array; neuromorphic computing

The last few years have witnessed rapid progress tow­ards the realization of artificial intelligence (AI), espe­cially software AI thanks to the advances in the deve­lopment of the algorithm. Note that software AI is typi­cally demonstrated based on the digital computer with conventional Von Neumann architectures at the cost of huge power consumption and massive data through­put[1]. In contrast, hardware AI systems based on in-memory com­puting architecture can handle probabilistic and unstru­ctured problems with low power dissipation rese­m­bling biological neural networks. Recently, memristor has att­racted increasing attention as a promising candi­date for the construction of hardware neuromorphic computing systems due to its prominent advantages, including simple structure and rich switching dynamics resembling biological synapses and neurons[2-8].

Emerging two-dimensional (2D) materials, especially 2D layered transition metal dichalcogenides (TMDs), are actively studied for fabricating high-performance nanoe­le­ctronic devices[9-11]. The 2D TMDs have shown pro­spective potential for memristor applications, and these devices could be of use in both information storage and neuromorphic computing[12]. Such devices exhibit pro­perties that traditional thin film-based memristors do not have, including high thermal stability[13], high contro­llability of potentiation, depression and relaxation[14], excellent flexibility and transparency[15]. However, mem­ristors based on 2D TMDs are typically fabricated by mechanical exfoliation, which is not feasible for large-scale array preparation. Thus, the yield and device-to-device variability of 2D TMD-based devices are rarely reported.

Herein, the memristive devices were prepared based on 2D MoTe2film fabricated by chemical vapor depo­sition (CVD). This MoTe2device shows stable bipolar resistive switching with superior retention characteristics, good endurance, high yield, and excellent uniformity.Furthermore, without selector devices, a 3×3 memristor array was demonstrated based on the MoTe2film. And, a handwritten digits recognition neural network simulation was implemented using the prepared MoTe2devices as synapses. This work indicates that large-scale 2D TMDs films are promising materials for future neu­romorphic computing.

1 Experimental

1.1 Device fabrication

2D MoTe2film was prepared through the CVD method. Firstly, a 1 nm Mo film was deposited onto a heavily doped Si substrate with 300 nm SiO2by electron beam evaporation. Then the Mo film was fully oxidized to MoO3in air. The resulting MoO3film was placed in a ceramic crucible containing Te powder. A mixture of argon and hydrogen was used as carrier gas and formed a reducing atmosphere during the CVD growth. The MoO3film was tellurized into a MoTe2film after annealing in Te vapor at 700 ℃. The bottom conductive layer (30 nm Au/10 nm Ti) and the top conductive layer (100 nm Au/10 nm Ti) were deposited through DC sput­tering and lift-off process. And the MoTe2film was transferred onto the bottom conductive layer through the polymethyl meth­a­crylate (PMMA) transfer method. The transfer process is shown schematically in Fig. S1.

1.2 Characterization and electrical measure­ment

The component of the MoTe­2film was measured by Raman spectra with an inVia Reflex spectrometer operated under a 532 nm laser and the XRD mea­surement (Bruker D8 Advance). An optical microscope (DSX 510) was employed to check the structure of the Au/Ti/MoTe2/Au/Ti device. The thickness and the stru­cture of the memristor array were measured through an atomic force microscope (AFM, SPM9700). All elec­trical measurements were conducted in air at room temperature with a Keithley 4200 semiconductor chara­cterization system connected with a tabletop cryogenic probe station (PS-100, Lakeshore).

2 Results and discussion

2.1 Composition and structure characteri­zation of the MoTe2 device

Fig. 1(a) shows a typical photo of the MoTe2film after being transferred on the silicon substrate with bottom electrodes. The film is uniform and continuous across the whole area (～1 cm), as seen from the homo­geneous color contrast in the image.And, this thin film is pure MoTe2without other phases, which is verified by the XRD measurement (Fig. S2). Raman spectroscopy was further employed to investigate the structure of the prepared MoTe2film, as shown in Fig. 1(b). The MoTe2film shows several Raman peaks between 100 and 300 cm–1: the out-of-plane A1gmode at ～171 cm–1and the prominent peak of the in-plane E2gmode at ～233 cm–1. These Raman features coincide with those observed in few-layer MoTe2with the hexagonal (2H) phase, thus unequivocally identifying the as-grown film as 2H MoTe2[16-17].

After characterization of the MoTe2film, the Au/Ti/MoTe2/Au/Ti device was fabricated through the PMMA transfer method and lift-off process, and the detailed fabrication process is shown in Fig. S3. The optical image of 50 Au/Ti/MoTe2/Au/Ti devices is presented in Fig. 1(c).

In addition to the independent MoTe2device, a 3×3 memristor array was prepared based on the continuous MoTe2film. The optical image of the 3×3 memristor array is shown in Fig. 1(d). And the atomic force mic­roscope (AFM) image and height profiles of the prepared array (Fig. S4) demonstrate that the MoTe2film can adapt to the contour morphology of the bottom electrodes, resulting in a conformal coating.

Fig. 1 Characterization of MoTe2 film and electrical mea­surement of Au/Ti/MoTe2/Au/Ti device

(a) Photo of the centimeter-scale MoTe2film; (b) Raman spectrum of the MoTe2film; (c) Optical image of the prepared memristive devices with the structure of Au/Ti/MoTe2/Au/Ti; (d) Optical image of a 3×3 memristor array

2.2 Memristive behavior of the Au/Ti/MoTe2/ Au/Ti device

After the electroforming process (Fig. S5), stable bipolar resistive switching was obtained when the vo­ltage was swept between –0.8 and 1.0 V, as shown in Fig. 2(a). During a continuous bias sweeping from 0 V→1 V→–0.8 V→0 V, a pinched hysteresis loop was obtained, in which SET (switching from high resistance state (HRS) to low resistance state (LRS)) occurred at about 0.7 V and RESET (switching from LRS to HRS) happened at about –0.5 V. A compliance current of 3 mA was applied in the SET process to prevent hard break­down. This hysteresis behavior is reproducible in the successive 20 voltage sweeps, indicating the stability of the switching behavior in the present device. The reten­tion characteristics of the MoTe2device were mea­sured at room temperature, as shown in Fig. 2(b). The ON/OFF ratio is higher than 10 and the current levels of the HRS and LRS show no degradation for 500 s, indicating good retention of the device.

In addition to stable bipolar switching behavior and good retention characteristic, the MoTe2device exhibits fast switching and good endurance. Current responses of the MoTe2device under the SET and RESET pulses are presented in Fig. 3(a–b). It was found that the device could be switched to LRS in about 60 ns and switched back to HRS in about 280 ns. Furthermore, over 2000 switching cycles can be obtained under SET pulse of 1.7 V/700 ns and RESET pulse of –1.2 V/7 μs (Fig. 3(c)), indicating good endurance of the present MoTe2device.

Fig. 2 Stable bipolar resistive switching behavior and retention characteristics of the MoTe2 device

(a) 20 cycles ofcurves with a compliance current of 3 mA; (b) Re­tention characteristics of the HRS and the LRS read at 0.1 V

2.3 High yield and excellent uniformity of the MoTe2 device

The yield and uniformity of the prepared device are systematically studied, since they are crucial for the construction of large-scale memristor arrays. Among 25 prepared devices, 24 devices show stable bipolar resistive switching behavior similar to that shown in Fig. 2(a), indicating a yield of 96%. We statistically analyzed 480curves collected in 24 devices (detailed results are shown in Fig. S6), and quantified the cycle-to-cycle and the device-to-device variability of the SET voltage (SET), RESET voltage (RESET), HRS and LRS by calculating the coefficient of variation (V) as the standard deviation () divided by the mean value (), in absolute value[18]. The cycle-to-cyclevariabilityin a single device results from the stochastic nature of the switching process, and the device-to-device variability is attributed to inhomogeneity in the samples derived from the fabrication process[19], such as device area and thickness fluctuations, and so on.

The minimum cycle-to-cycle variabilities ofSETandRESETare 6.6% and 5.2% for a given device, respe­ctively, and the device-to-device variabilities ofSETandRESETrise to 19.9% and 15.6%, respectively, when considering all 24 devices (Fig. 4). In addition, the device-to-device variabilities of HRS and LRS are 16.8% and 12.7%, respectively (Fig. S7). Note that the device-to-device variabilities ofSETandRESETare 6.06% and 29.07%, respectively, for the CVD h-BN device reported recently[20], indicating the uniformity of the present device is comparable to that of the h-BN device. Such excellent uniformity makes the MoTe2film promising for the construction of large-scale memristor arrays.

2.4 Realization of a 3×3 memristor array based on the MoTe2 film

According to the estimation using the one-bit pull-up scheme (detailed results are shown in Fig. S8 and Fig. S9), the maximum array sizes with the 1R scheme and the 1S1R scheme are 4×4 and 870×870 (740 kb), respe­c­tively. Therefore, a 3×3 array was fabricated to verify the feasibility of the memristor array based on the MoTe­2film. After an electroforming process, stable resi­stive switching was achieved in the MoTe2array device, as shown in Fig. 5(a–b).Notably, due to the inevitable issue of leakage current, the resistance values of the MoTe2array devices are lower than those of the inde­pendent MoTe2devices, especially for the HRS, leading to a smaller switching ratio (detailed results are pre­sented in Fig. S10). These results demonstrate that the MoTe2film is promising for the construction of me­mristor array, and the array size can be further enlarged through combing with proper selectors.

Fig. 3 Fast switching and good endurance of the MoTe2 device

(a) SET speed under the pulse with the amplitude 1.3 V; (b) RESET speed of the MoTe2device under the pulse with the amplitude of –1.0 V; (c) Over 2000 switching cycles obtained by applying SET pulse of 1.7 V/700 ns and RESET pulse of –1.2 V/7 μs

Colorful figures are available on website

Fig. 4 Cumulative distribution of (a)VSET and (b) VRESET of 24 devices

2.5 Consecutive conductance modulation and neural network simulation of the MoTe2 device

Continuous pulse stimulations are applied on the device to mimic the long-term potentiation (LTP) and depression (LTD) of synapses, which are essential synaptic functions for neuromorphic computing. The conductance triggered by alternating 50 positive (1 V, 500 ns) and 50 negative (–1 V, 5 µs) input pulses exhibits repeatable and stable response of LTP and LTD of biological synapses (Fig. S11).

Fig. 6 presents one cycle of LTP and LTD realized in the present device. The LTP and LTD processes can be fitted by the equation below[21-22]:

where G is conductance, N is the number of pulses, a, c and β are constants. Here, the exponential factor β can reflect the linearity of conductance modulation. The smaller the β, the better the linearities. As shown in Fig. 6, the β for LTP is 0.143 and the β for LTD is 0.203,which is comparable to the previously reported work[21].

(a) Electroforming process of the MoTe2array device; (b) 20 cyclescurves of the MoTe2array device with a compliance current of 5 mA

Fig. 6 Potentiation and depression processes of the present device

With the present device working as synapses, a fully connected network with one hidden layer has been simulated using the CrossSim platform[23]. The neural network is with 784 input neurons, 300 hidden neurons, and 10 output neurons for 28×28 input pixels and 10 output classifications for recognition handwritten digits from Modified National Institute of Standards and Technology (MNIST) dataset. The MoTe2devices were used as storage of the weight in the network and the change of the conductance of the artificial synapses was adopted as the weight update in the process of the backpropagation algorithm. In addition, another network with the size of 64×36×10 was trained using another dataset of small digits with 8×8 pixels from the “Optical Recognition of Handwritten Digits” dataset. The test recognition accuracies of the simulation with two datasetsusing the present device are benchmarked with ideal floating-point numeric precision that represents the neuromorphic algorithm limit, as shown in Fig. 7. The recognition accuracies of the small and large digits can reach ～91.3% and 88.2% after 40 epochs, lower than the ideal numeric accuracies of ～96.7% and 98.1%. This accuracy difference between the experiment derived and the ideal numeric results from the nonlinear and asymmetrical conductance change in LTP and LTD pro­cesses, which can be further improved through inter­face engineering of the device.

Fig. 7 Recognition accuracy for small and large hand­written digit images with experimental devices and ideal numeric

Colorful figures are available on website

2.6 Mechanism analysis

The electron transport processes at LRS and HRS were analyzed to explain the underlying mechanism of resistive switching behavior. As shown in Fig. 8(a), non­linearcharacteristics are observed at the HRS, with the current increasing as the temperature increases. And it was found that the HRS was best-fitted by the Schottky emission model[24]as shown in the inset of Fig. 8(a);

where A* is the effective Richardson constant, m0 is the free electron mass, T is the absolute temperature, q is the electron charge, φB is the Schottky barrier height, E is the electric field across the dielectric, k is Boltzmann’s constant, ε0 is the permittivity in vacuum, and εr is the optical dielectric constant.

(a)characteristics at HRS under different temperatures, with current increasing as the temperature increasesFitted data using the Schottky emission model for HRS is shown in the inset; (b)characteristics at LRS at different temperatures,with current decreasing as the temperature increase, indicating a metallic characteristic

In contrast, the LRS performs Ohmic conduction due to the linear characteristic of thecurve (Fig. 8(b)), indicating the formation/rupture of conductive filaments might be the dominating mechanism of the resistive switching. Furthermore, the current of the LRS decreases as the temperature increases, implying the metallic filaments are formed in the LRS. These metallic fila­ments might be composed of the monoclinic (1T′) phase (1T′-MoTe2) resulting from phase transition of MoTe2[25]or metallic Ti filaments induced by Ti ions drifting into native defects in MoTe2[26]. The detailed atomic structure of these filaments need further investigations.

3 Conclusions

Through CVD growth of the MoTe2film, the PMMA transfer method and lift-off process, the memristive devices based on 2D MoTe2film were successfully pre­pared. The devices exhibit stable bipolar resistive swit­ching with superior retention characteristics and good endurance. More importantly, the devices perform high yield, low cycle-to-cycle variability, and low device-to-device variability. Furthermore, a 3×3 memristor array based on the MoTe2device with 1R scheme was successfully demonstrated. And, the simulation of neural network with the prepared MoTe2devices as synapses was implemented for handwritten digits recognition, and the recognition accuracy of around 90% was realized. Our demonstration of memristor array based on centimeter-scale 2D MoTe2film provides an avenue for future neuromorphic circuits using large-scale 2D TMDs film.

Supporting materials related to this article can be found at https://doi.org/10.15541/jim20210658.

[1] SILVER D, SCHRITTWIESER J, SIMONYAN K,. Mastering the game of Go without human knowledge., 2017, 550(7676): 354–359.

[2] CHUA L O. Memristor-the missing circuit element., 1971, 18(5): 507–519.

[3] STRUKOV D B, SNIDER G S, STEWART D R,. The missing memristor found., 2009, 459(7250): 1154.

[4] YANG R, HUANG H M, GUO X. Memristive synapses and neurons for bio-inspired computing., 2019, 5(9): 1900287.

[5] HE H K, YANG R, ZHOU W,Photonic potentiation and electric habituation in ultrathin memristive synapses based on monolayer MoS2., 2018, 14(15): 1800079.

[6] YANG R, HUANG H M, HONG Q H,. Synaptic suppression triplet-STDP learning rule realized in second-order memristors., 2018, 28(5): 1704455.

[7] HE H K, YANG R, HUANG H M,. Multi-gate memristive synapses realized with the lateral heterostructure of 2D WSe2and WO3., 2019, 12(1): 380–387.

[8] HUANG H M, YANG R, TAN Z H,. Quasi-hodgkin-huxley neurons with leaky integrate-and-fire functions physically realized with memristive devices., 2019, 31(3): 1803849.

[9] XU M, LIANG T, SHI M,Graphene-like two-dimensional materials., 2013, 113(5): 3766–3798.

[10] FIORI G, BONACCORSO F, IANNACCONE G,. Electronics based on two-dimensional materials., 2014, 9(10): 768–779.

[11] CHANG C, CHEN W, CHEN Y,. Recent progress on two-dimensional materials., 2021, 37(12): 2108017.

[12] WANG C Y, WANG C, MENG F,. 2D layered materials for memristive and neuromorphic applications., 2020, 6(2): 1901107.

[13] WANG M, CAI S, PAN C,. Robust memristors based on layered two-dimensional materials., 2018, 1(3): 203.

[14] SHI Y, LIANG X, YUAN B,. Electronic synapses made of layered two-dimensional materials., 2018, 1(8): 458–465.

[15] HE H K, YANG F F, YANG R. Flexible full two-dimensional memristive synapses of graphene/WSe2xO/graphene., 2020, 22(36): 20658.

[16] YAMAMOTO M, WANG S T, NI M,. Strong enhancement of raman scattering from a bulk-inactive vibrational mode in few-layer MoTe2., 2014, 8(4): 3895–3903.

[17] GUO H, TENG Y, YAMAMOTO M,Double resonance Raman modes in mono- and few-layer MoTe2., 2015, 91(20): 205415-205415.

[18] ADAM G C, HOSKINS B D, PREZIOSO M,3-D memristor crossbars for analog and neuromorphic computing applications., 2016, 64(99): 312–318.

[19] SANGWAN V K, LEE H S, BERGERON H,. Multi-terminal memtransistors from polycrystalline monolayer molybdenum disulfide., 2018, 554(7693): 500–504.

[20] CHEN S, MAHMOODI M R, SHI Y,. Wafer-scale integration of two-dimensional materials in high-density memristive crossbar arrays for artificial neural networks., 2020, 3(10): 638–645.

[21] WANG Z, YIN M, TENG Z,. Engineering incremental resistive switching in TaObased memristors for brain-inspired computing., 2016, 8(29): 14015–14022.

[22] BURGT Y, LUBBERMAN E, FULLER E J,A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing., 2017, 16(4): 414–418.

[23] CrossSim platform. https://cross-sim.sandia.gov/ (accessed: July 2019).

[24] ALAMGIR Z, BECKMANN K, HOLT J,. Pulse width and height modulation for multi-level resistance in bi-layer TaObased RRAM., 2017, 111(6): 063111.

[25] WANG Y, XIAO J, ZHU H,. Structural phase transition in monolayer MoTe2driven by electrostatic doping.,2017, 550(7677): 487.

[26] SHI Y, LIANG X, YUAN B,. Electronic synapses made of layered two-dimensional materials.,2018, 1(8): 458–465.

Supporting materials:

High-uniformity Memristor Arrays Based on Two-dimensional MoTe2for Neuromorphic Computing

HE Huikai1, YANG Rui2,3, XIA Jian3,4, WANG Tingze3,4, DONG Dequan3,4, MIAO Xiangshui2,3

(1. Nanhu Academy of Electronics and Information Technology, Jiaxing 314002, China; 2. Hubei Yangtze Memory Labora­tories, Wuhan 430205, China; 3. Wuhan National Laboratory for Optoelectronics, School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan 430074, China; 4. State Key Laboratory of Material Processing and Die & Mould Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China)

1 PMMA transfer method

2 XRD analysis of the MoTe2film

3 Fabrication process of the Au/Ti/MoTe2/Au/ Ti device

4 AFM image of the 3×3 memristor array

5 Electroforming process of the device

Before achieving a stable resistive switching, an elec­troforming process was required, as shown in Fig. S5. At the beginning, the current of the CVD-MoTe2device was low due to its high resistance. After the application of a relatively large positive voltage (about 1.15 V), the device switched to the LRS.

6curves of 24 Au/Ti/MoTe2/Au/Ti devices

7 Cumulative distribution of device HRS and LRS of 24 devices

8 Estimation of the array size

In case of memristor crossbar array, the cross talk between the adjacent memory cells restricts the maxi­mum possible size of the array. Especially when all memory cells in the array are in low resistance state, the sneak path leakage problem will be highly predominant in the array. Therefore, to obtain the maximum possible crossbar array size, the worst case read scheme is utilized to measure the number of possible word lines with read margin of more than 10%, which is called the one-bit pull-up scheme[1-3].

In this model, the bit line of the selected cell is biased to the read voltage, the word line of the selected cell is grounded and all the other word and bit lines are left floated. Regarding an×square crossbar array with the most challenging data pattern (, all unselected cells in LRS) and negligible line resistance, the crossbar array can be simplified into three regions as shown in the left panel of Fig. S8(a). And the corresponding equi­valent circuit is present in the right panel of Fig. S8(a). The resistance value of the selected cell will be found by measuring the output voltage across the pull up resistor,pu. To obtain the best result during the measurement, the pull up resistor value will be set to the resistance during the low resistance state (pu=LRS). The read margin normalized to the pull up voltage is calculated by solving the Krichhoff equation:

Fig. S8(b) shows the calculated read margin for both 1R and 1S1R schemes for different number of word lines. From this figure, it is clearly seen that the read margin reduced drastically for 1R scheme and the number of word lines with at least 10% read margin is found to be only4. In case of 1S1R scheme, the non-linearity of the device resulted in increased number of word lines to 870 for 10% read margin, making a 740 kb possible crossbar array fabrication with good working possibility. Enhan­cing the selectivity of the selector can result in further increase of the crossbar array size to get a high density and large size crossbar array.

Table 1 Key resistance values and the corresponding maximum number of word lines/bit lines (N) with read margin >10%

9 Memristive behavior of the array devices

10 Consecutive conductance modulation of the device

Fig. S1 PMMA transfer method

Fig. S2 XRD pattern of the MoTe2film

Fig. S3 Fabrication process of the Au/Ti/MoTe2/Au/Ti device

Fig. S4 AFM image (a) of 3×3 memristor array, and the height profile (b) along the horizontal line and the vertical line

Fig. S5 Electroforming process of the device

Fig. S6curves of 24 Au/Ti/MoTe2/Au/Ti devices

All devices show stable resistive switching with low cycle-to-cycle and device-to-device variability

Fig. S7 Cumulative distribution of device (a) HRS and (b) LRS of 24 devices

Fig. S8 Estimation of the array size for the prepared CVD-MoTe2device

(a) Sneak current path at read in a square crossbar array where all bits except the selected one are at LRS, and the equivalent circuit can be repre­sented by three resistors (region 1, region 2 and region 3); (b) Dependence of the read margin on the crossbar line number for both 1R and 1T1R schemes

Fig. S9characteristics of the Pt/TaO/TiO2/TaO/Pt selector[4]

Fig. S10curves of 9 devices in 3×3 array

Fig. S11 Long-term potentiation and depression of the device using 50 potentiation (1 V, 500 ns) and depression (–1 V, 5 µs) presynaptic pulses

[1] HUANG J J, TSENG Y M, HSU C W,. Bipolar nonlinear Ni/TiO2/Ni selector for 1S1R crossbar array applications., 2011, 32(10): 1427-1429.

[2] FLOCKE A, NOLL T G. Proceedings of the 33rd European Solid-StateCircuits Conference (ESSCIRC, 2007), 2007: 328.

[3] LIU Z J. ZnO-based one diode-one resistor device structure for crossbar memory applications., 2012, 100(15): 041301.

[4] LEE W, PARK J, KIM S,. High current density and nonlinearity combination of selection device based on TaO/TiO2/TaO, structure for one selector-one resistor arrays., 2012, 6(9): 8166-8172.

高均一性二维碲化钼忆阻器阵列及其神经形态计算应用

何慧凯1, 杨蕊2,3, 夏剑3,4, 王廷泽3,4, 董德泉3,4, 缪向水2,3

(1. 中国电子科技南湖研究院, 嘉兴 314002; 2. 湖北江城实验室, 武汉 430205; 3. 华中科技大学 光学与电子信息学院, 武汉光电国家实验室, 武汉 430074; 4. 华中科技大学 材料科学与工程学院, 材料加工与模具技术国家重点实验室, 武汉 430074)

二维过渡金属硫化合物是构建纳米电子器件的理想材料, 基于该材料体系开发用于信息存储和神经形态计算的忆阻器, 受到了学术界的广泛关注。受制于低成品率和低均一性问题, 二维过渡金属硫化合物忆阻器阵列鲜见报道。本研究采用化学气相沉积得到厘米级二维碲化钼薄膜, 并通过湿法转移和剥离工艺制备得到碲化钼忆阻器件。该碲化钼器件表现出优异的保持性(保持时间>500 s)、快速的阻变(SET时间～60 ns, RESET时间～280 ns)和较好的循环寿命(阻变2000圈后仍可正常工作)。该器件具有高成品率(96%)、低阻变循环间差异性(SET过程为6.6%, RESET过程为5.2%)和低器件间差异性(SET过程为19.9%, RESET过程为15.6%)。本工作成功制备出基于MoTe2的3×3忆阻器阵列。在此基础上, 将研制的MoTe2器件用于手写体识别, 实现了91.3%的识别率。最后, 通过对MoTe2器件高低阻态的电子输运机制进行拟合分析, 揭示了该器件阻变源于类金属导电细丝的通断过程。本项工作表明大尺寸二维过渡金属硫化合物在未来神经形态计算中具有巨大的应用潜力。

二维材料; 碲化钼; 忆阻器阵列; 神经形态计算

TQ125

A

1000-324X(2022)07-0795-07

10.15541/jim20210658

2021-10-25;

2021-12-14;

2022-01-06

National Natural Science Foundation of China (U1832116, 51772112)

HE Huikai (1992–), PhD. E-mail: hehk@hust.edu.cn

何慧凯(1992–), 博士. E-mail: hehk@hust.edu.cn

YANG Rui, professor. E-mail: yangrui@hust.edu.cn

杨蕊, 教授. E-mail: yangrui@hust.edu.cn