Monolayer MoS₂ field effect transistor with low Schottky barrier height with ferromagnetic metal contacts

Abstract

Two-dimensional MoS₂ has emerged as promising material for nanoelectronics and spintronics due to its exotic properties. However, high contact resistance at metal semiconductor MoS₂ interface still remains an open issue. Here, we report electronic properties of field effect transistor devices using monolayer MoS₂ channels and permalloy (Py) as ferromagnetic (FM) metal contacts. Monolayer MoS₂ channels were directly grown on SiO₂/Si substrate via chemical vapor deposition technique. The increase in current with back gate voltage (V_g) shows the tunability of FET characteristics. The Schottky barrier height (SBH) estimated for Py/MoS₂ contacts is found to be +28.8 meV (at V_g = 0V), which is the smallest value reported so-far for any direct metal (magnetic or non-magnetic)/monolayer MoS₂ contact. With the application of positive gate voltage, SBH shows a reduction, which reveals ohmic behavior of Py/MoS₂ contacts. Low SBH with controlled ohmic nature of FM contacts is a primary requirement for MoS₂ based spintronics and therefore using directly grown MoS₂ channels in the present study can pave a path towards high performance devices for large scale applications.

Introduction

Two-dimensional (2D) materials with their layered structures have attracted much attention as next generation device materials due to their extraordinary properties such as mechanical flexibility, large surface to volume ratio, and their easy integration in heterostructure junction devices^1–3. Graphene is well known example among these materials, which shows very rich physics resulting from its linear dispersion relation and massless Dirac Fermion⁴. Owing to remarkable properties such as very high mobility, large electrical and thermal conductivity, high Young’s modulus and small spin-orbit coupling (SOC), graphene became a promising candidate for wide range of applications, including high speed electronics, sensors, energy generation and storage devices as well as spintronics^5–7. However, semi-metallic nature (gapless band structure) of pristine graphene limits its application in semiconductor electronics as zero band-gap leads a low on/off ratio in graphene-based field effect transistors (FETs)^4,8. In addition to this, small SOC in graphene does not allow this material to have better control on generation and electrical manipulation of spins in spintronic devices.

Unlike graphene, molybdenum disulfide (MoS₂), which belongs to the family of transition metal dichalcogenides (TMDs) shows semiconducting nature with a sizable band-gap³. The type and value of band-gap in MoS₂ can be changed by varying the number of layers—MoS₂ shows an indirect bandgap (~1.2 eV) in the bulk form (multilayers) and a direct band-gap (~1.8 eV) when reduced to monolayer³. In addition to non-zero band-gap, MoS₂ also possesses considerable SOC along with unique spin-valley coupling to manipulate the spins, which makes the material very attractive for the next generation spintronic and other technological applications. The interest in mono-layer MoS₂ has further increased after the demonstration of a high on/off ratio (~10⁸) and high carrier mobility at room temperature FETs^9,10. However, the large electrical potential drop due to high contact resistance between MoS₂ and metal contacts may strongly limit the performance of MoS₂-based devices. Previous reports show large Schottky barrier heights (SBHs) for various metal/MoS₂ contacts¹¹, which can be reduced by different approaches such as insertion of insulating layers (h-BN¹², MgO¹³, TiO₂^14,15, Al₂O₃¹⁶) between metal and MoS₂, chemical doping^17,18 of MoS₂ and electrical gating^13,16. To study spin injection from ferromagnetic (FM) metal and spin transport in MoS₂, it is very important to investigate the contact behavior between FM metal and MoS₂ and to suppress the SBH that hinders efficient spin injection/detection. There are only few reports in the literature, which discussed behavior of FM/monolayer MoS₂ contacts and estimated SBH^12,13,16. From our knowledge, the MoS₂ in previous reports is either exfoliated from the bulk single crystal of MoS₂ or transferred from MoS₂ sample grown via chemical vapor deposition (CVD) technique. However, it is notable that these methods are not suitable for mass production of electronic devices and exfoliation and/or transfer methods can induce unwanted changes in physical and electronic properties of MoS₂. In addition, the FET device characteristics strongly depend on the growth methods of MoS₂ and FM electrodes. Hence, it will be interesting to study the behavior of FM/MoS₂ contacts, where MoS₂ is directly grown on substrate by CVD technique.

In this paper, we study the device characteristics of FETs fabricated using monolayer MoS₂ channels, directly grown on SiO₂/Si substrate using salt-assisted CVD technique. 20 nm thick Ni₈₀Fe₂₀ (Py) electrodes were used as ferromagnetic contacts. The work function of Py is 4.83 eV¹⁹. To understand contact behavior of Py/MoS₂ contacts, I-V characteristics were studied with temperature and back gate voltages as controlling parameters. The SBH of Py/MoS₂ contacts was determined to be +28.8 meV at the zero-gate voltage. Such contacts with low SBH and ohmic nature can play a key role in future spin-based devices because the tuning of the SBH allows circumventing the conductance mismatch problem for injecting spins in semiconductors²⁰.

Experimental Details

The MoS₂ is grown on the thermally oxidized SiO₂/n⁺-Si substrate with SiO₂ thickness of ~285 nm via salt-assisted CVD technique (please see ref.^21,22 for the growth procedure). The monolayer crystal grown by this method were found to be almost free from defects/vacancies and any unintentional doping with alkali and/or halogen atoms²². Moreover, the sharp photoluminescence spectrum and the largest mobilities observed for these samples confirm its excellent crystal quality over exfoliated and CVD grown samples^22,23. Figure 1(a) shows an optical microscope image of as-grown monolayer MoS₂, which was processed in different channels shown in Fig. 1(b). Afterwards, Py electrodes of thickness 20 nm were deposited on the top of MoS₂ channels, capped by Ti(3 nm)/Au(50 nm) as shown in Fig. 1(c). The monolayer MoS₂ was confirmed by the Raman spectroscopy with a laser light of wavelength of 488 nm. Only monolayer MoS₂ was processed for the fabrication of FET devices. Figure 1(d) shows Raman spectra for MoS₂ samples. The two prominent peaks in the Raman spectra appear due to an in-plane (E_2g) mode located around 382.2 cm⁻¹ and an out-of-plane (A_1g) mode located around 398.2 cm⁻¹. The difference between these two modes is found to be ~18 cm⁻¹, which confirms the monolayer MoS₂²⁴. The MoS₂ channels [Fig. 1(b)] of length (l) 6–8 µm and width (w) 2–4 µm have been prepared by electron-beam (EB) lithography using TGMR resist (negative tone). In the first step, the MoS₂ channels were covered by TGMR resist and unwanted MoS₂ is etched by O₂ plasma etching for 20 seconds. In the next step, TGMR resist was lifted by wet etching using N-Methyl-2-pyrrolidone and then cleaned by acetone and isopropyl alcohol (IPA). Py(20 nm)/Ti(3 nm)/Au(50 nm) electrodes were deposited by EB deposition technique after patterning them by EB lithography using Poly(methyl methacrylate) (PMMA) resist. The center-to-center distance between two electrodes was 0.65 µm. The electrical measurements have been performed using helium-free cryostat in the temperature range 10–300 K. The gate voltage was applied from the backside of the Si substrate.

Figure 1.

Optical microscope images of monolayer MoS₂—(a) As grown MoS₂ (triangular shaped), (b) MoS₂ channels of different dimensions after processing (rectangular shaped) and (c) MoS₂ device with permalloy (Py) electrodes. (d) Raman spectra of monolayer MoS₂ performed with laser light of wavelength of 488 nm. (e) The schematic of a FET device and measurement configuration for two-probe source-drain current-voltage (I_DS-V_DS) curves.

Results and Discussion

Figure 1(e) shows the schematic for FET device and measurement configuration. Source-drain current-voltage (I_DS-V_DS) characteristics were performed by applying DC voltage and recording the current between two probes. We measured five devices (device #A1, #A2, #B1, #B2 and #C1) from three different samples (A, B and C) to check the consistency of our experimental results. In the main text, we mainly discuss FET characteristics of device #A1 while results of other devices are used as supporting information and are shown in Supplementary Material.

To understand the electrical behavior of the Py/MoS₂ contacts, we carried out systematic two-probe I_DS-V_DS curves measurements as a function of V_g and temperature for device #A1, and the results are shown in Fig. 2. Figure 2(a) shows I_DS-V_DS characteristics at 300 K as a function V_g. The I_DS-V_DS curves are linear and symmetric under a small range (±0.1V) of V_DS (inset figure), however show small deviation from linearity and asymmetric behavior under high range (±1V) of V_DS (main figure). I_DS-V_DS curves measured for device #A2 are almost linear and symmetric even under high range (±1V) of V_DS (see Supplementary Fig. S1), which reflects better device quality (we note that this device was unfortunately broken during the temperature dependent measurements). It can also be noted from Fig. 2(a) that I_DS increases with increasing V_g—for a fixed applied V_DS, at V_g = 20 V, obtained I_DS is at least 15 times higher than that of V_g = 0 V. It shows the tunability of FET characteristics with the application of V_g and suggests that the Schottky barrier is modified at the Py/MoS₂ interface, which can result in the reduction of the SBH. As temperature is lowered, I_DS-V_DS characteristics show strong deviation from linearity and significant reduction in I_DS, as can be seen in Fig. 2(b). This suggests that the device goes in the off state at low temperatures.

Figure 2.

Source-drain current-voltage (I_DS-V_DS) characteristics of a MoS₂-based FET (a) at T = 300 K as a function of back gate voltage (V_g), (b) at T = 10 K as a function of V_g, (c) at V_g = 0 V as a function of temperature, and (d) at V_g = 20 V as a function of temperature. The inset in (a) shows I_DS-V_DS characteristics under V_DS of range ±0.1V_. The arrow in (c,d) shows the direction of increase of I_DS with increasing temperature.

Figure 2(c,d) show I_DS-V_DS characteristics as a function of temperature at V_g of 0 V and 20 V, respectively. The arrows in these figures show the direction of increase of I_DS with increasing temperature. At fixed temperature and V_DS, the I_DS is higher at V_g = 20 V than V_g = 0 V. Indeed, the Schottky barrier is lowered at 20 V, therefore more carriers can be thermally activated and overcome the barrier.

SBH can be extracted using an activation energy method, the commonly used one. The advantage of the activation energy method is that we do not need information of electrically active area under the contacts to extract the SBH²⁵. The SBH for 2D materials can be extracted by employing the 2D thermionic emission equation^26,27,

$$I_{DS}=A{A}^{⁎}{T}^{\frac{3}{2}}\exp \left[-\frac{q}{k_{B}T}\left({\phi }_B-\frac{V_{DS}}{n}\right)\right],$$ 1

where A is the contact surface area, A* is the effective Richardson constant, q is the electronic charge, k_B is the Boltzmann constant, ϕ_B is the Schottky barrier height and V_DS is source-drain voltage, and n is the ideality factor. From above equation, the activation energy is given by $E_A=q\left({\phi }_B-\frac{V_{DS}}{n}\right)$. After rearranging few terms, the equation can be written as

$$ln\left(I_{DS}/T^{\frac{3}{2}}\right)=\ln A+ln{A}^{⁎}-\frac{E_A}{k_B}\left(\frac{1}{T}\right)$$ 2

From Eq. (2), it is clear that E_A can be estimated from the slope of $ln(I_{DS}/T^{\frac{3}{2}})$ vs. 1/T, called the Arrhenius plot. Once E_A is estimated, the SBH (ϕ_B) can be extracted by simply taking the intercept of E_A vs. V_DS plot, which takes into account an effect of band bending of MoS₂ by an application of the source-drain voltage.

I_DS-V_DS curves recorded in high temperature range (290 −190 K) were employed to calculate the SBH. At low temperatures, the device remains in off state because the thermal energy supplied to carriers is not enough to overcome the barrier and therefore the thermionic emission theory cannot be applied successfully. Figure 3(a–c) show Arrhenius plots for various V_DS at different V_g. To determine the slope of the plot, the experimental data were fitted by Eq. (2) as shown in Fig. 3 as solid black lines. It is worth to note that the slope of the Arrhenius plot for V_g = 0 V is negative [Fig. 3(a)] and changes its sign from negative to positive with the application of V_g [Fig. 3(b,c)]. This indicates that the SBH decreases from a positive value (at V_g = 0 V) to negative values for V_g = 10 V and 20 V.

Figure 3.

Arrhenius plots, $ln(I_{DS}/T^{\frac{3}{2}})$ vs. 1000/T as a function of source-drain voltage (V_DS) for different back gate voltages, (a) V_g = 0 V, (b) V_g = 10 V, and (c) V_g = 20 V. The solid black line is a linear fit to Arrhenius plot to extract the slope.

The slope of the Arrhenius plot as a function of V_DS is depicted in Fig. 4 for different V_g and fitted by linear function. The intercept from the linear fit gives the value of SBH, which is found to be +34.5 meV for V_g = 0 V. The SBH estimated for V_g = 10 and 20 V from the intercept in Fig. 4(b,c) are found to be −6.8 and −3.2 meV, respectively. To confirm the reproducibility of our results, we also estimated SBH for other FET devices fabricated on different samples. The SBHs estimated for devices #B1, #B2 and #C1 were found to be +26.0, +24.6 and 30.1 meV, respectively (see Supplementary Material), which strengthens the central claim of this study. It is quite important to estimate the SBH in TMD-based devices for future spintronic applications, because formation of the Schottky barrier strongly hinders efficient spin injection and spin detection. This is the reason we have been focusing on the SBH of the Py/MoS₂ by modulating the source-drain and gate voltages, and indeed, it is significant that we have clarified appearance of the approximately zero barrier height. Meanwhile, in a fundamental point of view, it is also important to estimate SBH at flat band gate voltage. Hence, we fabricated device #C1 and measured both the zero- gate voltage and the flat-band SBH in the same device. As can be noted from Fig. S4 in the supplementary Material that the zero- gate voltage and flat band SBHs are 30.1 and 21.8 meV, respectively. The observed SBH value of +28.8 meV (average value) at V_g = 0 V is 64% lower than the previously reported SBH in Py/MoS₂ contacts¹⁶. The small zero gate bias SBH observed in our devices indicates the ohmic nature of Py/MoS₂ contacts. In the previous reports the SBH for ferromagnetic contacts such as Py/MoS₂ and Co/MoS₂ was reported to be 80.2 and 60.6 meV, where in the first case, the authors fabricated FET devices after transferring CVD grown MoS₂ to SiO₂/Si substrate via PMMA stamping method¹⁶ and in the latter case, the FET devices were fabricated using the MoS₂ channels exfoliated from bulk MoS₂ crystals¹³. The values of SBH estimated in previous reports for various FM and non-magnetic (NM) metal contacts on monolayer MoS₂ channel are compared in Table 1. From the Table 1, it is clear that the observed SBH in our study is the smallest value reported so-far in any direct FM (non-magnetic)/monolayer MoS₂ contact. The best value reported previously for direct metal/monolayer MoS₂ contact, patterned by EB lithography was ~38 meV^12,28. In fact, the SBH becomes small for transferred Ag electrodes with multilayer MoS₂ channels²⁹. However, multilayer MoS₂ is not a direct band-gap semiconductor and spin-valley locking effect is somewhat suppressed³⁰. Since one of the purposes of our study is exploring a potential of combination of ferromagnet and monolayer TMDs in spintronics viewpoints, realization of low SBH in Py/monolayer MoS₂ is crucial, although the low SBH formation to a multilayer TMD is also notable. It can be noted that MoS₂ channel used in previous reports were either exfoliated and/or transferred from CVD grown MoS₂. This implies that FET devices fabricated using MoS₂ directly grown on substrate via CVD technique decreases the chance of introducing distortion-induced defects during the exfoliation or the transferring methods and/or that of surface contamination, unlike exfoliated and transferred MoS₂ techniques. It is worth to recall that in our case MoS₂ is grown by salt-assisted CVD technique, which shows excellent crystal quality over exfoliated and CVD MoS₂ as demonstrated by the smallest SBH reported so-far in exfoliated and/or CVD MoS₂. High performance devices fabricated on directly grown CVD MoS₂ confirms the possibility of mass production of MoS₂-based devices. As aforementioned, circumventing the formation of the Schottky barrier is quite significant to realize efficient spin injection and detection in FM/semiconductor heterostructure, our results demonstrate the importance of an integration of directly grown MoS₂ channels.

Figure 4.

Source-drain voltage (V_DS) dependence of slopes (−E_A/1000k_B) extracted from Fig. 3 for (a) back gate voltages, V_g = 0 V, (b) V_g = 10 V and (c) V_g = 20 V. The solid red line is a linear fit to the experimental data to extract the intercept and therefore Schottky barrier height (ϕ_B). (d) V_g dependence of Schottky barrier height, showing positive value for V_g = 0 V and negative values for V_g = 10 and 20 V.

Table 1.

Comparison of Schottky barrier heights (SBHs) for direct metal (magnetic and non-magnetic) contact on monolayer MoS₂ channel.

Metal contact	MoS2 nature	SBH (meV)	Reference
Ni80Fe20 (Py)	Monolayer, directly grown CVD	device #A1 = 34.5 device #B1 = 26.0 device #B2 = 24.6 device #C1 = 30.1 (flat band = 21.8) Average = 28.8	This work
Ni80Fe20 (Py)	Monolayer, transferred CVD	80.2	16
Co	Monolayer, Exfoliated	60.6	13
Co	Monolayer, Exfoliated	~38	12
Ti	Monolayer, Exfoliated	230	31
Cr	Monolayer, Exfoliated	130	31
Au	Monolayer, Exfoliated	320	31
Pd	Monolayer, Exfoliated	300	31

Conclusions

In conclusion, we fabricated FET devices using directly grown monolayer MoS₂ channels on SiO₂/Si substrate via salt-assisted CVD technique. The electrical properties of FET devices were studied by measuring two-probe I-V characteristic as a function of temperature and back gate voltage. The SBH estimated at V_g = 0 is found to be +28.8 meV, which is the smallest SBH reported so-far for any direct ferromagnetic as well as non-magnetic metal contact on monolayer MoS₂. The small zero- gate voltage SBH indicates ohmic Py/MoS₂ contacts. Ferromagnetic contacts with the smallest SBH and controllable ohmic contacts studied in the present study can open a route for practical realization of high performance MoS₂ based spintronic devices.

Author contributions

S.G. and M.S. planned and design the study. S.G. fabricated devices, collected data and analyzed results. F.R., R.O. and Y.A. helped in the experiments. T.E. and Y.M. supplied monolayer MoS₂. S.G. and M.S. wrote the manuscript. All authors discussed the results.

Data availability

Data measured or analyzed during this study are available from the corresponding author on request.

Competing interests

The authors declare no competing interests.

Supplementary information

is available for this paper at 10.1038/s41598-019-53367-z.