Electrical and photo-electrical properties of MoS2 nanosheets with and without an Al2O3 capping layer under various environmental conditions

Abstract The electrical and photo-electrical properties of exfoliated MoS2 were investigated in the dark and in the presence of deep ultraviolet (DUV) light under various environmental conditions (vacuum, N2 gas, air, and O2 gas). We examined the effects of environmental gases on MoS2 flakes in the dark and after DUV illumination through Raman spectroscopy and found that DUV light induced red and blue shifts of peaks (E1 2 g and A1 g) position in the presence of N2 and O2 gases, respectively. In the dark, the threshold voltage in the transfer characteristics of few-layer (FL) MoS2 field-effect transistors (FETs) remained almost the same in vacuum and N2 gas but shifted toward positive gate voltages in air or O2 gas because of the adsorption of oxygen atoms/molecules on the MoS2 surface. We analyzed light detection parameters such as responsivity, detectivity, external quantum efficiency, linear dynamic range, and relaxation time to characterize the photoresponse behavior of FL-MoS2 FETs under various environmental conditions. All parameters were improved in their performances in N2 gas, but deteriorated in O2 gas environment. The photocurrent decayed with a large time constant in N2 gas, but decayed with a small time constant in O2 gas. We also investigated the characteristics of the devices after passivating by Al2O3 film on the MoS2 surface. The devices became almost hysteresis-free in the transfer characteristics and stable with improved mobility. Given its outstanding performance under DUV light, the passivated device may be potentially used for applications in MoS2-based integrated optoelectronic circuits, light sensing devices, and solar cells.

The electrical and photo-electrical properties of exfoliated MoS 2 were investigated in the dark and in the presence of deep ultraviolet (DUV) light under various environmental conditions (vacuum, N 2 gas, air, and O 2 gas). We examined the effects of environmental gases on MoS 2 flakes in the dark and after DUV illumination through Raman spectroscopy and found that DUV light induced red and blue shifts of peaks (E 1 2 g and A 1 g ) position in the presence of N 2 and O 2 gases, respectively. In the dark, the threshold voltage in the transfer characteristics of few-layer (FL) MoS 2 field-effect transistors (FETs) remained almost the same in vacuum and N 2 gas but shifted toward positive gate voltages in air or O 2 gas because of the adsorption of oxygen atoms/molecules on the MoS 2 surface. We analyzed light detection parameters such as responsivity, detectivity, external quantum efficiency, linear dynamic range, and relaxation time to characterize the photoresponse behavior of FL-MoS 2 FETs under various environmental conditions. All parameters were improved in their performances in N 2 gas, but deteriorated in O 2 gas environment. The photocurrent decayed with a large time constant in N 2 gas, but decayed with a small time constant in O 2 gas. We also investigated the characteristics of the devices after passivating by Al 2 O 3 film on the MoS 2 surface. The devices became almost hysteresis-free in the transfer characteristics and stable with improved mobility. Given its outstanding performance under DUV light, the passivated device may be potentially used for applications in MoS 2 -based integrated optoelectronic circuits, light sensing devices, and solar cells. transistors (FETs) cannot be efficiently switched off; that is, off-current is comparable with on-current. However, 2D transition metal dichalcogenides (TMDCs) possess a suitable band gap of ~1-2 eV, making these materials potentially applicable in nanoelectronics, sensing, and photonics. [6][7][8][9] Several studies have analyzed 2D TMDC-based FETs, photodetectors, and gas sensors. The earliest TMDC, namely WSe 2 crystal, was used in FETs and exhibits a high mobility (>500 cm 2 V −1 s −1 ) and ambipolar behavior with a ~10 4 on/off ratio at 60 K. [10] TMDCs are potential materials for molecular sensing

Introduction
Two-dimensional (2D) nanomaterials have sustained attention because of their unique properties and facile fabrication process despite the complexity of their structures. Graphene as a 2D material has become popular in recent years because of its outstanding linear dispersion relation, high charge carrier mobility, feasibility of chemical doping, and other unique physical properties conferred by its low dimensionality. [1][2][3][4][5] However, the zero band gap of graphene hinders its application in logic and optoelectronic devices. Graphene-based field-effect OPEN ACCESS spectra were recorded with a Renishaw microspectrometer equipped with a 514 nm laser at room temperature. DUV light (λ = 220 nm and average intensity of 11 mW cm −2 ) was used for illumination.
A 10-nm-thick Al 2 O 3 passivation layer was deposited on the FL-MoS 2 at 200 °C through atomic layer deposition (ALD; Lucida D100 ALD by NCD Co., Ltd, Daejeon, Korea). During the growth process, 100 cycles with a growth rate of ~1 Å per cycle were completed to reach the desired thickness. Trimethylaluminum and deionized water were used as precursors. Ultra-pure N 2 (99.999%) was used as a carrier and purging gas. The pressure of the growth chamber was maintained at ~5.5 × 10 −3 torr during deposition.

Results and discussion
Raman spectroscopy is a fast and reliable tool that provides information about structural and local perturbation of 2D materials. The Raman spectrum of MoS 2 is dominated by two vibrational modes: E 1 2 g , which is attributed to in-plane vibrations of two S atoms with respect to the Mo atom, and A 1 g , which corresponds to the out-of-plane vibrations of S atoms in opposite directions. The E 1 2 g and A 1 g modes are sensitive to the number of layers forming the structure of MoS 2 . [17,18] The Raman spectra of FL-MoS 2 layers were recorded at room temperature with a laser excitation of ~514 nm and a small laser power of ~1.0 mW to avoid the heating effect. Figure 1(c) shows the Raman spectra of pristine and gas-treated (in the dark and under DUV light) FL-MoS 2 (five layers; see Figure S1). The FL-MoS 2 exhibited strong bands at ~383.2 and ~407.7 cm −1 , which is ascribed to in-plane vibrational (E 1 2 g ) and out-of-plane vibrational (A 1 g ) modes, respectively. The vibrational modes of FL-MoS 2 did not significantly change after treating the samples with O 2 and N 2 gases in the dark. However, we observed a shift toward a lower wave number after DUV treatment of FL-MoS 2 in the presence of N 2 gas. Subsequently we observed a shift toward a higher wave number in the presence of O 2 gas under DUV light (Figure 1(c)). After the DUV treatment in N 2 gas, the red shift of A 1 g and E 1 2 g was ~2.2 and ~1.6 cm −1 , respectively. The subsequent DUV treatment in O 2 gas environment made the blue shift of A 1 g and E 1 2 g by ~2.1 and ~1.5 cm −1 , respectively. The red and blue shifts in peak (E 1 2 g and A 1 g ) position are due to doping of electrons and holes on FL-MoS 2 by DUV treatment in N 2 and O 2 gas, respectively. The mechanism of Raman shifts by adsorption and desorption of these gas molecules is attributed to electron-phonon coupling of modes. The A 1 g mode couples more strongly with electrons than the E 1 2 g mode. The electron doping in MoS 2 causes softening specifically of its Raman-active A 1 g phonon, accompanied by increase in line-width of its Raman peaks. [19] In contrast, the other Raman mode with E 1 2 g symmetry is less sensitive to electron doping. The Raman spectrum application because of their high surface-to-volume ratio. MoS 2 sheets are sensitive detectors of NO, NO 2 , NH 3 , N 2 , and triethylamine gases. [11][12][13][14][15] However, the large variation in the electrical transport properties of MoS 2 is generally attributed to extrinsic or environmental effects. These effects may significantly limit the exploration of the intrinsic properties of MoS 2 . Investigating the environmental factors that influence the reliability and stability of MoS 2 FETs is important. Detailed and comparative investigations on environmental gases effects are needed to understand the overall performance of MoS 2 FETs and undesirable effects should be minimized or removed to enhance device reliability and stability.
In this study, we fabricated few-layer (FL) MoS 2 FETs and investigated electrical transport properties under various environmental conditions. We analyzed the change in on-state current under different environmental conditions. We observed the hysteresis behavior in transfer characteristics, where oxygen on the MoS 2 surface plays a vital role in hysteresis. Gas environment significantly affects the characteristics of MoS 2 FETs under deep ultraviolet (DUV) light. Thus, we investigated the time-dependent photoresponse of MoS 2 FETs in the presence of various environmental gases. We also studied the maximum photocurrent saturation, responsivity, detectivity, external quantum efficiency (EQE), linear dynamic range (LDR), and relaxation time after switching light ON and OFF under various environmental conditions. Finally, we investigated the characteristics of devices after passivating by Al 2 O 3 film on the MoS 2 surface. The devices became almost hysteresis-free and stable FETs with improved mobility. The effect of Al 2 O 3 on MoS 2 flakes was discussed to explore the basic and important parameters regarding light detection.

Experimental section
We fabricated FL-MoS 2 FETs using a conventional approach. FL-MoS 2 flakes were transferred to clean 300-nm-thick SiO 2 layer on p++ Si substrate with Scotch tape via mechanical cleavage method. [16] FL flakes were initially identified under an optical microscope ( Figure  1(a)) and further confirmed through Raman spectroscopy. The thickness of the flakes was estimated as ~3.6 nm by atomic force microscopy, which indicated the presence of five layers (Supporting Information Figure S1). Large patterns formed through photo-lithography for each device, and fine electrodes for source and drain were completed after electron beam-lithography. The source-drain contacts of Cr/Au (10/80 nm) were deposited via thermal evaporation. Electrical transport measurements were performed using a Keithley 2400 source meter (Beaverton, OR, USA) and a Keithley 6485 picoammeter (Beaverton, OR, USA). The final optical device image is illustrated in Figure 1(b). All of the measurements were carried out at room temperature. Raman for each case was obtained from five different points on the MoS 2 surface, and the results were consistent.
The electrical transport measurements of the FL-MoS 2 FETs (sample-1; L = 8.1 μm and W = 7.6 μm) were carried out under various environmental conditions. Transfer characteristics (drain-current I D as a function of back-gate voltage V g ) were measured at room temperature with fixed source-drain voltage (V ds ) = 1 V. Figure 2(a) presents the transfer characteristics of the FL-MoS 2 FETs in the dark under various environmental conditions. The measurements were performed in vacuum, followed by N 2 , air, and O 2 gas environment. The current level of the device was almost similar in the N 2 and vacuum environments but was substantially reduced in the air and O 2 environments. The I on /I off of our device was ~10 5 and remained almost same in vacuum, N 2 , air, and O 2 gas flow (Supporting Information Figure S2a). Figure 2(b) illustrates the hysteresis in the transfer characteristics of FL-MoS 2 in the dark under different environmental conditions. We swept the V g from -40 to 80 V repeatedly and observed a hysteresis in vacuum. Later on, the same devices were exposed in N 2 gas and found a similar hysteresis trend to that in vacuum. However, further exposure of devices to air and oxygen significantly changed the hysteresis because of adsorption of oxygen atoms/molecules on top of MoS 2 surface. [20] This kind of oxygen adsorption resulted in formation of charge-trapping sites and made the hysteresis prominent in FL-MoS 2 FETs. In transfer characteristics in Figure S2a, we found that the threshold voltages (V th ) of FL-MoS 2 FETs was near -35, -37, -34, and -28 V in vacuum, N 2 , air, and O 2 , respectively. Furthermore, we observed that the V th remained similar in vacuum and N 2 gas but shifted toward positive gate voltages when the devices were exposed to air and O 2 gas. The shift of V th to positive back gate voltages indicates an electron deficiency in FL-MoS 2 due to exposure to air and oxygen. Such behavior was previously reported for MoS 2 -related FET devices in O 2 environments, which was also attributed to the absorption of oxygen molecules into sulfur or defect states on the MoS 2 surface, which traps the charge carriers. [21,22]  for each device, L is the channel length, and W is the channel width. In the dark, we estimated the μ vacuum , μ N2 , μ air , and μ O2 of the FL-MoS 2 FETs to be ~15.6, 16.5, 9.1, and 8.1 cm 2 V −1 s −1 , respectively. However, under DUV illumination, the μ vacuum , μ N2 , μ air , and μ O2 of the FL-MoS 2 FETs were ~24.8, 50.6, 11.9, and 0.8 cm 2 V −1 s −1 , respectively. The FL-MoS 2 FETs exhibited a high I D and large mobility in N 2 but showed a small I D in air and O 2 environment under DUV illumination. [25] Figure 3 presents the output characteristics of the FL-MoS 2 FETs in vacuum, N 2 , air, and O 2 environments in the dark. The output characteristics (I D versus V ds ) were measured at a fixed V g ranging from −20 V to + 60 V with a step of 20 V. The nonlinear I-V characteristics supports that the Cr/Au metal electrodes make Schottky barrier due to the work function difference between metal and MoS 2 .
To measure the photodetector response, the MoS 2 nanosheet device was irradiated with DUV light (sample-2; L = 1.3 μm and W = 3.15 μm). In Figure 4 where dI ds /dV g is slope of the transfer curve in the linear region, C g is the gate capacitance (~115 aF μm −2 ) of Si/ SiO 2 substrates (300 nm), V ds is the source-drain current Two key figures of merit were measured to quantify detector performance. Current responsivity (R λ ) and detectivity (D * ) were defined as follows [26]: where ΔI ph is the photo-excited current, P is the light power intensity, and A is the effective area of the photodetector. Responsivity is defined as the photocurrent generated per unit power intensity of the incident light on the effective area of a photoconductor. The power intensity of our DUV incident light and the area of device were 11 mW cm −2 and 4.09 μm 2 , respectively. In Equation (3), e is the absolute value of the electron charge and I dark is the current density in dark. Figure 4(d) shows the responsivity and detectivity of our devices, where R λ was estimated to be ~65.9, 27.9, 14.7, and 154 2eI dark 60 s with alternating ON and OFF of the light source (V ds = 1 V and V g = 0 V) under various environmental conditions. Figure 4(a) illustrates that the photocurrent response depended on various environments. The photocurrent response was weak in O 2 gas but strong in N 2 gas. The MoS 2 nanosheet exhibited a repeatable and reasonably stable response to incident DUV light under various environmental conditions. We also found that the photocurrent response increased with V ds . The photocurrent (ΔI ph = I ph -I dark ) after 60 s DUV illumination reached ~14.7, 39.7, 83.3, and 88.5 μA at V ds = 1, 4, 6, 8 V, respectively (Supporting Information Figure S2b and c). In Figure 4(b), the photocurrent saturation was examined under various environmental conditions. We illuminated the FL-MoS 2 FETs with DUV light until the photocurrent saturation was obtained and then turned off the DUV light to observe photocurrent decay behavior. The maximum photocurrent values (ΔI max = I saturation -I dark ) of the FL-MoS 2 FETs were ~29, 12.3, 6.5, and 67.7 μA (V ds = 1 V and V g = 0 V) under various environmental conditions (Figure 4(c)). The largest and smallest photocurrent saturations were observed in N 2 and O 2 gas environments, respectively. important factor which is the denominator in the formula (EQE = hcR λ /eλ). We enhanced our EQE to 8600% using smaller wavelength together with N 2 gas environments. The N 2 gas provides supportive environment for photocurrent generation. The LRD (= 20 log (I ph /I dark ), where I ph is the photocurrent measured at a light intensity of 11 mW cm −2 ) was measured under DUV illumination. The measured LDR values of the MoS 2 nanosheet device were ~29.5, 23.16, 20, and 34.4 dB in vacuum, air, O 2 , and N 2 environments, respectively. The results demonstrate that MoS 2 nanosheet devices in N 2 gas can be potential photodetectors in the future. Relaxation response time or decaying behavior is another key factor in photodetector performance. Decaying behavior is observed after photocurrent saturation under various environmental conditions. Relaxation data were extracted from Figure 4(b), when light was turned off after photocurrent saturation. The response time obtained from some 1D nanostructures or graphene oxide-based photodetectors ranges from seconds to several tens of minutes. [33][34][35] The wide range of response time is attributed to the difference in the materials or device structures. The dynamic response to kAW −1 in vacuum, air, O 2 , and N 2 gas environments, respectively. D * , which is measured in Jones units (1 Jones = 1 cm Hz 1/2 W −1 ), ranged from ~10 13 -10 14 . These data indicated the best response in N 2 gas, which is more than ~10 4 times higher than that of the previously reported 2D, layered material devices on Si/SiO 2 substrates. [26][27][28][29][30] To investigate photodetector performance further, two other critical parameters, namely EQE and LDR (typically presented in the unit of dB), were measured as seen in Figure 5(a) and (b), respectively. The EQE (= hcR λ /eλ, where h is a plank constant, c is the speed of light, R λ is the responsivity at a DUV wavelength of 220 nm, and e is the electron charge) is defined as the number of electron-hole pairs excited by one absorbed photon. The EQE was estimated to be ~3700, 1500, 830, and 8600% in vacuum, air, O 2 , and N 2 environments, respectively. Compared with previously reported devices, our MoS 2 devices on the Si/SiO 2 substrate exhibited higher EQE in N 2 gas. [24,28,31,32] There are three factors which improve external quantum efficiency (EQE) of our devices; higher responsivity, light of lower wavelength and N 2 gas environment. The wavelength, λ, is an  films. [37,38] In the present study, electron-capturing impurity states were reduced under DUV light in N 2 gas environment but were enhanced under DUV light in O 2 gas environment. Thus, the decay time was significantly longer in N 2 environment and substantially shorter in O 2 environment. Evidently, decaying behavior was robustly dependent on the surrounding environment. We measured the electrical transport of the FL-MoS 2 FETs (sample-3; L = 1.5 μm and W = 1.2 μm) in vacuum at a bias voltage V ds = 1 V to study the passivation effect by high-k dielectric materials to circumvent the environmental effects. The transfer characteristic of the pristine device at room temperature in the dark is shown in Figure 7(a), where a pronounced hysteresis can be observed. A 10-nm-thick passivation layer of Al 2 O 3 was deposited on the same device through atomic layer deposition (ALD). After depositing the passivation layer, the transfer characteristics of the device were examined in the dark under various environmental conditions (Figure 7(b)). The MoS 2 device was almost hysteresis-free with increased drain current and exhibited n-type. The Al 2 O 3 layer on the MoS 2 surface served as a DUV illumination for relaxation time can be expressed as [28,36]: where A is a scaling constant, τ decay is a time constant for decaying, and t is the time after DUV light is switched on or off. The time constant (τ) can be calculated by fitting the experimental data. Figure 6(a-c) describes the decaying photocurrent behavior in N 2 , vacuum, air, and O 2 environments, respectively, and the red lines indicate the fitting data. The relaxation time was calculated to be ~215.9, 79.4, 11.1, and 3.7 s in N 2 , vacuum, air, and O 2 , respectively ( Figure 6(d)). These results demonstrated that the decay process was very slow in N 2 and significantly fast in O 2 gas. Thus, the adsorbates and defect states originating from ambient air, water, and oxygen atoms/molecules play an important role in photocurrent relaxation and electron-hole pair recombination dynamics. In general, the surface-adsorbed oxygen significantly affects the photoresponse of MoS 2 and ZnO of the device are shown in Figure S2d (Supporting Information). The time-dependent response under DUV light is described in Figure 8(a) with switching light ON and OFF for a 60 s cycle at (V ds = 1 V and V g = 0 V) in vacuum. As shown in Figure 8(a), the passivated device exhibited a slow response because of the suppression of impurities and defect states by the Al 2 O 3 layer. However, the photocurrent of the passivated device enhanced compared with that of the pristine device. In Figure 8(b), the maximum photocurrent saturation values (ΔI max = I saturation -I dark ) of the pristine and passivated FL-MoS 2 FETs were found to be ~20.6 and 24.2 μA, respectively. R λ and D * were also investigated as shown in Figure  8(c). The maximum responsivity and detectivity of the pristine device were ~11 × 10 3 AW −1 and ~6.3 × 10 12 Jones, respectively. However, those after deposition of the Al 2 O 3 layer were ~14 × 10 3 AW −1 and ~8.2 × 10 12 Jones, respectively. To investigate further the effect of passivation, the relaxation time (τ decay ) of the pristine and passivated devices was examined (Figure 8(d)).
The data for decaying time were obtained from Figure  8(b) and exponentially fitted by the single exponential Equation (4). The carrier lifetimes for the pristine and passivated devices were ~30.3 and 284.5 s, respectively.
protecting or capping layer. Hence, the effect of external environments was effectively disregarded. In addition, field-effect mobility increased after passivation by Al 2 O 3 film. The improvement of device performance after deposition of a high-k dielectric is associated with the suppression of Coulomb scattering, modification of phonon dispersion, the difference of Al 2 O 3 and SiO 2 dielectric constants (3.9 for SiO 2 and 9.0 for Al 2 O 3 ), and the removal of impurities during ALD growth at 200 °C. [39][40][41] Extensive theoretical work including the calculation of phonon dispersion relations in MoS 2 , the calculation of scattering rates on phonons and charge impurities, is necessary to describe these phenomena completely. After the deposition of a 10-nm-thick Al 2 O 3 layer, the electrical properties were also measured under DUV illumination in vacuum at V ds = 1 V. A huge increment in drain current was observed (Figure 7(c)). The field-effect mobility of the pristine and passivated devices in the dark and in the presence of DUV illumination was measured in vacuum, and the mobility was ~18, 32, and 104 cm 2 V −1 s −1 , respectively (Figure 7(d)).
We explored the photoresponse of the FL-MoS 2 FETs (sample-4; L = 2 μm and W = 8. As depicted in Figure 8(d), the relaxation time of the passivated device was significantly longer than that of the pristine device. This result can be attributed to the fact that the intermediate states of defects and oxygen constitutes are reduced after Al 2 O 3 deposition and those excited electrons take a long time to relax.

Conclusions
In conclusion, we fabricated FL-MoS 2 FETs and studied the effects of environmental gases in the dark and under DUV illumination. We determined that DUV light induced red and blue shifts of Raman peak (E 1 2 g and A 1 g ) positions in the presence of N 2 and O 2 gases, respectively. We investigated the electrical transport measurements of FL-MoS 2 FETs under various environmental conditions and found that device performance did not degrade in N 2 gas in the dark. However, O 2 gas significantly reduced the ON-current state. In the dark, the environmental gases did not modify the I on /I off of the device and remained consistent at ~10 5 . The threshold voltage and hysteresis were nearly similar in vacuum and N 2 gas but shifted toward positive gate voltage accompanied with a large hysteresis loop in air and O 2 gas. However, the environmental gases changed the characteristics of the FL-MoS 2 FETs under DUV illumination. The photocurrent response and saturated photocurrent of the FL-MoS 2 FETs under various environmental conditions were examined to investigate light detection parameters, such as responsivity, detectivity, external quantum efficiency, and linear dynamic range. The results demonstrated a strong photoresponse in N 2 gas but a poor photoresponse in O 2 gas. The photocurrent relaxation after turning off DUV light was also examined under various environmental conditions. The lifetime of a carrier was longer in N 2 gas but shorter in O 2 gas, indicating that N 2 gas helped the device recover from defects and impurities, while O 2 gas adversely affected device characteristics. Furthermore, we passivated the device by depositing the Al 2 O 3 layer to protect MoS 2 from environmental effects. We observed almost hysteresis-free transfer characteristics with improved electrical and photoelectrical responses and field-effect mobility after deposition of the passivation layer. We also found that τ decay increased after Al 2 O 3 layer deposition. This phenomenon may be attributed to the screening of impurities and defects from the MoS 2 surface. Thus, we developed improved, stable, reliable, and hysteresis-free FL-MoS 2 FETs for various nanotechnology applications.