Direct Contacting of 2D Nanosheets by Metallic Nanoprobes ^†

Abstract

We present a simple and fast methodology to realize metal contacts on two-dimensional nanosheets. In particular, we perform a complete characterization of the transport properties of MoS₂ monolayer flakes on SiO₂/Si substrates by using nano-manipulated metallic tips as metallic electrodes directly approached on the flake surface. We report detailed experimental investigation of transport properties and contact resistance in back-gated field effect transistor in which the Si substrate is used as the gate electrode. Moreover, profiting of the n-type conduction, as well as the high aspect ratio at the edge of the MoS₂ flakes, we also explored the possibility of exploiting the material as a field emitter. Indeed, by retracting one of the metallic probes (the anode) from the sample surface, it has been possible to switch on a field-emitted current by applying a relatively low external electric field of few-tens of Volts for a cathode-anode separation distance below 1 µm. Experimental data are then analyzed in the framework of Fowler-Nordheim theory and its extension to the two-dimensional limit.

1. Introduction

Molybdenum disulfide (MoS₂) is one of the most investigated transition-metal dichalcogenides (TMDs) for exploitation in next-generation two-dimensional (2D) devices, including field-effect transistors [1,2,3,4], solar cells [5], photodetectors [6,7], field emission devices [8,9,10,11], chemical or biological sensors [12,13], etc.

MoS₂ has a crystal structure characterized by a hexagonal layer of Mo atoms between two layers of S atoms. Layers are bonded together by van der Waals forces. MoS₂ flakes can be fabricated either by mechanical exfoliation or chemical vapor deposition [14]. Bulk MoS₂ has 1.2 eV indirect bandgap, while mono-layer (1 L) and bilayer (2 L) MoS₂ have 1.8 eV and 1.6 eV indirect bandgap, respectively [15]. Consequently, both 1 L and 2 L MoS₂ can be used to realize field-effect transistors with high On/Off ratio and photoresponse [16]. On the other hand, carrier mobility is typically limited to few-tens cm² V⁻¹ s⁻¹. Moreover, ohmic contacts (with low resistance) are crucial to improving device performance [17].

In this paper, we demonstrate a simple method of realizing electrical contacts on MoS₂ flakes by using nanomanipulated metallic probes inside a scanning electron microscope (SEM). We show that this technique allows complete characterization of the back-gated field-effect transistor (FET), as well as checking the field emission properties of the MoS₂ flake.

2. Materials and Methods

The MoS₂ flakes studied in this work have been grown on Si/SiO₂ substrates by means of a chemical vapour deposition technique, in which S powder and a saturated ammonium heptamolybdate solution have been used as precursors. Few-layer MoS₂ flakes have been characterized by micro-Raman spectroscopy ($\lambda$ = 532 nm). The experimental setup for electrical characterization is realized inside a SEM chamber (see Figure 1a) provided with two piezo-driven nano-manipulators for precise positioning (step resolution $~$5 nm) of metallic probes (tungsten tips). A semiconductor parameter analyzer (Keithley 4200-SCS) is then used as a source-measurement unit, to apply bias up to ±100 V and to measure current with resolution better than 0.1 pA. Electrical measurements are performed at room temperature and in high vacuum (10⁻⁶ mbar) after gently approaching the tungsten tips on the MoS₂ flake (a real image taken inside the SEM chamber is shown in Figure 1b) and using the Si substrate as a back gate.

Micro-Raman analysis of the MoS₂ flake has shown a spectrum (see Figure 1c) with two peaks corresponding to the $E_{2g}^1$ and $A_{1g}$ modes, separated by about $20{ \mathrm{cm}}^{-1}$, indicating that the sample under investigation is a monolayer.

3. Results and Discussion

In Figure 2a, we report the output characteristics ($I_{ds}-V_{ds}$) measured in the range of ±0.5 V for different values of the gate voltage ($V_{gs}$). We notice a slight rectification that can be explained as the result of asymmetric Schottky barriers forming at the tungsten/MoS₂ interfaces [3]. By varying the distance between the two tungsten tips, we can modulate the channel length of the FET, thus realizing an experiment based on the Transfer Length Method (TLM) [18,19] to evaluate the contact resistance at the tungsten/MoS₂ interface. In Figure 2b, we show the measured total resistance $R_{tot}$ versus $d$, with $R_{tot}=2R_c+\frac{R_s}{W}d,$ where $R_c$ is the contact resistance, $R_s$ is the MoS2 sheet resistance, W is the channel width (assumed to be equal to the tip diameter, 200 nm), and $d$ is the channel length, i.e., the separation between the two tips. Experimental data have linear behavior, from which specific area contact resistivity and sheet resistance can be evaluated as ${\rho }_c\approx 4\times {10}^{-2}{ \mathsf{\Omega }\mathrm{cm}}^2$ and $R_s\approx {10}^8 \mathsf{\Omega }/\square$ from the intercept and the slope of the linear fit, respectively.

The transfer characteristics ($I_{ds}-V_{gs}$) reported in Figure 2c have been measured for different gate voltage ranges up to ±60 V, with $V_{ds}=-5 \mathrm{V},$ and by positioning the tungsten tips at separation of $13 \mathsf{\mu }\mathrm{m}$. The device has n-type behavior, with a threshold voltage of about $-10 \mathrm{V}$, and it can be explained in terms of chemisorption of oxygen on MoS₂ or sulphur vacancies [20,21,22].

From the transfer characteristic measured in the range ±50 V, we have estimated the on/off ratio as ~${10}^5$, a subthreshold swing of $SS\approx 4 \frac{V}{decade}$, and a mobility of $\mu =1{ \mathrm{cm}}^{2 }{\mathrm{V}}^{-1 }{\mathrm{s}}^{-1}$, a value within the typical range ($0.02-100 {\mathrm{cm}}^{2 }{\mathrm{V}}^{-1} {\mathrm{s}}^{-1}$) reported for MoS₂-based FETs on SiO₂ [23,24]. The low mobility can be attributed to the high contact resistance and to high defects or traps density [25].

In Figure 3c, we show the transfer characteristics measured by sweeping the gate voltage different ranges, from ±20 V up to ±60 V. The curves have a clear hysteresis that we explain as being caused by negative charge trapping [20]. We observe that the hysteresis width ($H_W$), estimated at $I_{ds}=0.1 \mathrm{nA}$, has linear dependence on the $V_{gs}$ sweeping range (See Figure 3d). This behavior can be ascribed to the trapping process driven by the gate voltage and the effects on the MoS₂/Si-substrate capacitor.

Finally, we also investigated the field emission (FE) properties of the MoS₂ flake, profiting of the n-type conduction and the high aspect ratio of the flake side. By retracting the tip-anode at a distance h = 900 nm from the MoS₂ edge, we can measure the current emitted from the flake under the application of an external electric field (Figure 3a). More precisely, we applied a voltage bias of up to 120 V on the anode, and we measured the current emitted from the flake (cathode) with a resolution better that 0.1 pA. The current−voltage ($I_{ds}-V_{ds}$) curves have been measured at fixed cathode-anode separation h and for two different values of gate voltage.

The FE characteristics have been measured by applying a bias voltage on the anode up to +120 V, by keeping a fixed gate voltage of 10 V and 40 V, respectively. The measured curves are reported on a linear scale (Figure 3b) and on a logarithmic scale (Figure 3c). Interestingly, we observe that the FE current is larger for $V_{gs}=40 \mathrm{V},$ suggesting that the gate voltage increases the n-doping of the MoS₂ flake [26].

We analyzed the FE curves in the framework of the Fowler-Nordheim (FN) theory [27], for which the FE current is expressed as

$$I_{ds}=A\frac{{\left(\beta V_{ds}/h\right)}^2}{\Phi }S·exp\left(-B\frac{{\Phi }^{\frac{3}{2}}}{\left(\beta V_{ds}/h\right)}\right),$$

where $A=1.54\times {10}^{-6} {\mathrm{A} \mathrm{V}}^{-2} \mathrm{eV}$ and $B=6.83\times {10}^7 {\mathrm{V} \mathrm{cm}}^{-1} {\mathrm{eV}}^{-3/2}$, $\Phi$ is the work function of the emitter, $S$ is the emitting surface area, and $\beta$ is the field enhancement factor. Accordingly, for FE curves, it is expected that $ln\left(I_{ds}/V{ }_{ds}^2\right)$ versus $1/V$ is linear (FN plot), and $\beta$ can be evaluated from its slope.

In Figure 3d, we report the FN plots that demonstrate the FE nature of the measured current. For $V_{gs}=40 \mathrm{V}$, we found a turn-on field $E_{on}=40 {\mathrm{V} \mathsf{\mu }\mathrm{m}}^{-1}$ (defined as the field to obtain a FE current of $1 \mathrm{pA}$) and $\beta \approx 200$.

4. Conclusions

We demonstrate a simple and fast methodology to realize metal contacts on two-dimensional nanosheets by gently approaching nanomanipulated tungsten tips inside a scanning electron microscope. We contacted a MoS₂ monolayer to form a back-gated FET, and we performed complete electrical characterization, reporting specific area contact resistivity of $4\times {10}^{-2}{ \mathsf{\Omega } \mathrm{cm}}^2$, sheet resistance of ${10}^8 \mathsf{\Omega }/\square$, on/off ratio of 10⁵, subthreshold swing of 4 V/decade, and mobility of $1 {\mathrm{cm}}^2 {\mathrm{V}}^{-1 }{\mathrm{s}}^{-1}$. Finally, by retracting the tip-anode, we performed field emission characterization of the MoS₂ flake, reporting that the FE current can be modulated by the gate bias.