Polarity independent resistive switching in MoS2 nanosheets and PEO-based nanocomposite films

Here, we report the exfoliation of bulk MoS2 (molybdenum disulfide) into few-layer nanosheets and then prepared nanocomposite films (MoS2-PEO) with poly(ethylene oxide) as the host. We observed nonpolar or polarity independent bistable resistive switching memory in two-terminal devices with indium tin oxide and aluminum (Al) as bottom and top electrodes, respectively. In both bipolar and unipolar operations, it is observed that the biasing direction controls the current conduction mechanism. When the positive bias is applied at the top Al electrode, the low resistance state (LRS) conduction is ohmic type. But in the opposite biasing condition, LRS conduction is space charge controlled. The current–voltage characteristics of bipolar and unipolar switching are distinctly different in terms of their RESET process. In bipolar, the RESET process is very sharp, whereas in unipolar operation it is staggered and step-wise.


Introduction
Recently, MoS 2 in its two-dimensional (2D) form as a monolayer or as few-layer nanoplatelets has drawn considerable interest for many nanoelectronic applications, including areas like resistive switching memory, transient electronics, bioelectronics, etc. 1-3) Specifically, its resistive switching characteristics are quite diverse owing to its unique properties such as tunable bandgap and charge confinement effects. There are reports suggesting that MoS 2 nanoflakes show both bipolar and unipolar resistive switching (URS) memory characteristics in the same device. 1,4) Moreover, MoS 2 with different device architectures can show typical memristive switching and neuromorphic computing operations. [5][6][7][8] To further advance the application potential in the field of flexible and wearable electronics, two-dimensional MoS 2 nanosheets are lately investigated in nanocomposite films with different polymer hosts. MoS 2 nanoflakes in polymer hosts like PVA, PMMA, PVP and PVDF-HFP mostly show bipolar resistive switching (BRS) and WORM type memory operations. [9][10][11][12] The switching mechanism could be redox or charge-trapping type. However, unlike the case of bare MoS 2 , there are no reports of observing both unipolar and BRS in the polymeric nanocomposites of MoS 2 nanosheets. Resistive switching devices showing both unipolar and BRS properties are known as polarity independent or nonpolar resistive switching devices. The nonpolar resistive switching devices offer the advantage of flexibility in peripheral circuit design and high storage density in comparison to the conventional BRS memories. 13,14) Here, in this work, we are reporting the observation of nonpolar resistive switching memory in MoS 2 nanosheets based poly(ethylene oxide) (PEO) composites.

Experimental methods
The bulk MoS 2 powders and PEO (M w ∼100 000) were purchased from Alfa Aesar and Sigma-Aldrich. Dimethylformamide (DMF) was used as solvent for liquid phase exfoliation of MoS 2 due to good wettability and miscibility of MoS 2 in DMF 15) and was bought from Merck. Indium tin oxide (ITO) coated polyethylene terephthalate (PET) substrate was purchased from Sigma-Aldrich. According to the reports given for good yield of MoS 2 , we followed grinding assisted liquid-phase exfoliation. 16) Following this method, MoS 2 nanosheets were prepared from the bulk MoS 2 in DMF medium. The detailed experimental procedure was already reported in our previous work. 17) Exfoliation of MoS 2 into few-layer nanosheets was confirmed from atomic force microscopy (AFM) and the average layer thickness of MoS 2 nanosheets is 5 nm. The composite solution of MoS 2 nanosheets and PEO are deposited onto ITO coated PET substrate by spincoating method. We studied the current-voltage characteristics in MoS 2 -PEO nanocomposite films with device structure as ITO/MoS 2 -PEO/Al where Al works as the top electrode.

Preparation of MoS 2 -PEO nanocomposites
To prepare nanocomposite films with PEO, initially, 600 mg of PEO was added to 10 ml of DMF and stirred at 60°C for 2 h in a magnetic stirrer. Then one wt% of exfoliated MoS 2 was added to the solution of PEO and stirred for 12 h at 300 rpm to get a homogeneous mixture. The nanocomposite films were prepared by the solution cast method. They were dried at 50°C on a hot plate followed by evacuation at 40°C for 6 h. To observe current-voltage characteristics in the composite films, two-terminal devices were prepared on ITO coated PET substrate. The solutions of MoS 2 nanosheets and PEO in DMF were deposited by spin-coating on ITO surface at 4000 rpm for one minute. The prepared films were then dried at 50°C for 6 h in a vacuum furnace, followed by deposition of the top aluminum electrode having a diameter of ∼100 μm by using the thermal evaporation technique. Figure 1 presents the X-ray diffraction (XRD) pattern of prepared nanocomposite film made up of MoS 2 nanosheets dispersed in the PEO host. Bragg peaks corresponding to both 2H-MoS 2 and PEO are observed in the XRD pattern. The diffraction peaks at 14.8°, 32.65°, 36.1°, 39.8°and 44.6°a re assigned to (002), (100), (012), (103) and (006) planes of MoS 2 respectively. 11,18,19) The (120), (112) and (222) peaks of PEO are observed at 19.05°, 23.2°and 26.4°respectively indicated by (*) mark. 20,21) Further, we separately investigated the (002) peak of MoS 2 . The average crystallite size was calculated to be ≈8 nm by using Debye-Scherrer formula.

Film morphology
The surface morphology of the nanocomposite film was investigated using optical microscopy, scanning electron microscopy (SEM) and AFM. Figure 2(a) shows that the PEO and MoS 2 nanosheets grew into spherulites with lots of interfacial amorphous regions. A magnified spherulite is presented in the inset showing radial lamellar structures. Further, the SEM image in Fig. 2 Fig. 2(d). The average roughness is ∼15 nm   with some big spike in heights due to some MoS 2 agglomerates present on the surface. = 203 J g −1 which is the melting enthalpy of 100% crystalline PEO. 22,23) It is observed that the percentage of crystallinity fraction (χ c %) has increased significantly up to 59.34% in PEO-MoS 2 films in comparison to 17.34% as that of calculated for pure films. This suggests that the MoS 2 nanosheets have an influence on the crystallization kinetics of the PEO. The exact nature of the influence depends upon two competing phenomena namely nucleation to spur crystallization and confinement effect to hinder it. There are reports that nanostructured materials like carbon nanotube and MXene at lower loading aids crystallization process. 24,25) Hence, we suppose one wt% of MoS 2 nanosheets in the PEO matrix improves the crystallinity fraction.

Electrical characterization
To investigate the electrical characterization of the fabricated two-terminal device, the most common technique used is measuring the current through the sample as a function of an externally applied voltage. Here we studied the currentvoltage (I-V ) behavior of the device by employing a Keithley 4200-Semiconductor Characterization System. The sample is placed on a probe station and contact is made to the metal-insulator-metal structure using tungsten tips as the probe. The schematic diagram of the two-terminal device is shown in Fig. 4(a). The I-V characteristics of the twoterminal devices prepared as the sandwich structure of  strikingly different. The RESET or ERASE process is very sharp in the bipolar switching, whereas it is staggered and step-wise in the case of unipolar.
To further understand the switching mechanism, the I-V characteristics for bipolar and unipolar operations are replotted in Figs. 5 and 6 on log-log scale. Figure 5(a) shows the SET process of positive bipolar operation. The slope of the I-V curve was observed to rise in three steps, with the slope rising from 2.24 to 3.77 and then the cell reached the LRS. The slope value of more than 2 confirms that the conduction mechanism during the SET process is due to trapcontrolled space charge limited current (TC-SCLC). 26) In the LRS, the slope is ∼1, indicating ohmic type conduction in LRS. The inset shows the linear relationship between ln(I) and V 0.5 in the region between 0.04 and 0.12 V. This suggests that the electrons are injected at the aluminium (MoS 2 -PEO) film interface by Schottky emission. 26) During the RESET process, the log-log plot of the I-V characteristic is depicted in Fig. 5(b). Starting from 0 V up to the RESET voltage (2.2 V), the slope remains one due to the ohmic conduction process. As the RESET occurs at 2.2 V, the maximum slope observed is 1.29 indicating the detrapping process of electrons from the trap sites of the nanocomposite films. For the bipolar switching operation where SET is observed in negative bias, as shown in Fig. 5(c), the HRS current is again TC-SCLC. But the LRS current is not ohmic like in the previous case. The LRS slope decreases gradually from 2.36 to 1.56 and hence, indicating space charge limited current (SCLC). 27) This is the major difference between these two bipolar operations. The log-log plot of the unipolar operation along the both the polarities are presented in Fig. 6. The operation done in the positive bias side as shown in Figs. 6(a) and 6(b) show HRS and LRS currents are governed by TC-SCLC and ohmic mechanisms, respectively. Whereas in the negative bias side presented in Figs. 6(c) and 6(d), the LRS is non-ohmic and instead governed by SCLC mechanism similar to the bipolar operation with SET process observed on the negative bias side. So, we infer that in both bipolar and unipolar operations, the LRS conduction mechanism is either ohmic or SCLC type, depending solely on the biasing direction. This may be due to the asymmetric electrode combinations used in the memory cells. The top aluminum electrode is oxidizable and can diffuse through as ions when a positive bias is applied during either of the bipolar or unipolar operations. Hence, when a SET process is completed by applying a positive bias to the top Al electrode, the conductive bridge may be formed assisted by electron trapping at MoS 2 sites as well as by electrochemical reduction of Al at the ITO electrode. But when the ON state is achieved by reversing the bias i.e. applying positive bias to the ITO side, may be exclusively an electronic effect governed by electron trapping at MoS 2 sites present in the PEO matrix. However, these inferences need further corroboration from other experimental observations. Nevertheless, the MoS 2 nanosheets and PEO composite films offer the tunability in resistive switching memory operations. A better understanding of its switching mechanism will pave the way for more control of switching operations and its parameters.

Conclusions
Exfoliated MoS 2 (molybdenum disulfide) nanosheets were prepared by the grinding assisted liquid-phase exfoliation method and a composite made with PEO. The nanocomposite thin films show versatile switching operations. Nonvolatile memory is observed with both bipolar and unipolar modes with different biasing directions. This suggests a polarity independent or nonpolar resistive switching memory operation achieved in Al/MoS 2 -PEO/ITO memory cells. During the switching cycles, the current in HRS is governed by TC-SCLC irrespective of the mode of operation (bipolar or unipolar) or biasing direction. But the current in LRS is either ohmic or space charge limited, controlled mainly by biasing directions. During bipolar operation, the RESET process is very sharp, but it is a step-wise staggered process in unipolar operation.