Two-dimensional (2D) materials are composed of atomically thin crystals with an enormous surface-to-volume ratio, and their physical properties can be easily subjected to the change of the chemical environment. Encapsulation with other layered materials, such as hexagonal boron nitride, is a common practice; however, this approach often requires inextricable fabrication processes. Alternatively, it is intriguing to explore methods to control transport properties in the circumstance of no encapsulated layer. This is very challenging because of the ubiquitous presence of adsorbents, which can lead to charged-impurity scattering sites, charge traps, and recombination centers. Here, we show that the short-circuit photocurrent originated from the built-in electric field at the MoS2 junction is surprisingly insensitive to the gaseous environment over the range from a vacuum of 1 × 10−6 Torr to ambient condition. The environmental insensitivity of the short-circuit photocurrent is attributed to the characteristic of the diffusion current that is associated with the gradient of carrier density. Conversely, the photocurrent with bias exhibits typical persistent photoconductivity and greatly depends on the gaseous environment. The observation of environment-insensitive short-circuit photocurrent demonstrates an alternative method to design device structure for 2D-material-based optoelectronic applications.
Generally, two-dimensional (2D) materials, composed of atomically thin crystals, exhibit an enormous surface-to-volume ratio, and the physical properties of 2D materials, including electrical, optical, and mechanical properties, are easily subjected to the change of the chemical environment1,2,3,4,5,6. Regarding the electrical property, the carrier transport in 2D materials is very sensitive to the presence of extrinsic adsorbents, which typically cause charged-impurity scattering, charge trapping, and recombination centers7,8,9,10,11, leading to degradation of the transport characteristics12,13,14. Although various encapsulation methods have been developed15,16,17,18,19, it is intriguing to explore methods to control the transport properties in the circumstance of no encapsulated layer. Here, we demonstrated that the short-circuit photocurrent enabled by the built-in electric field at the MoS2 junction is surprisingly insensitive to the gaseous environment, which is very uncommon in the photoresponse of thin transition-metal dichalcogenides (TMDCs). We exploit the unique property of 2D TMDCs, in which the electronic band structures are associated with the number of the layered materials20,21, to create a junction structure. For molybdenum disulfide (MoS2), the transition changes from direct bandgap (1.9 eV) in monolayer MoS2 to indirect bandgap (1.3 eV) in bulk MoS222,23,24. This layer-dependent electronic structure therefore offers a distinct approach for designing the MoS2 junction based on homogeneous material of the TMDCs25,26,27,28.
In this work, we fabricated atomic thin MoS2 junction phototransistors with different MoS2 layers and studied their photoresponse behavior at different source-drain bias and gaseous environment. The difference of band gap in different thickness of few layer MoS2 was utilized to create a built-in electric field. Interestingly, we observed that the short-circuit photocurrent due to the photovoltaic (PV) effect29 is insensitive to the gaseous environment over the range from a vacuum of 1 × 10−6 Torr to ambient condition. This environmental insensitivity can be well ascribed to the unique characteristic of diffusion current that is associated with the carrier density gradient. Conversely, the photocurrent under bias exhibits persistent photoconductivity (PPC) and highly depends on the gaseous environment. The scanning tunneling microscopy and spectroscopy (STM and STS) measurements reveal the energy profile at the MoS2 junction, confirming the presence of the band offset. Moreover, we show that the MoS2 junction phototransistors exhibit gate-voltage tunable open-circuit voltage and short-circuit photocurrent, demonstrating the capability for regulating the current-voltage characteristics of the MoS2 junction via electric field effect.
The details of sample fabrication processes are described in the Supporting Information S1. In brief, MoS2 junctions with different layers were prepared by mechanically exfoliating MoS2 crystals onto SiO2(300 nm)/Si substrates. We then identify MoS2 flakes with different numbers of layers using optical microscopy. The thicknesses of the flakes of MoS2 were characterized using atomic force microscopy and Raman spectroscopy (Supporting Information S2). We then employ a resist-free approach to define the electrical contact to the MoS2 junction to avoid the resist residue which may cause extra carrier scattering. We then deposited Au (50 nm) for the electrical contacts by e-beam evaporation at a base pressure of 1 × 10−7 Torr.
Results and Discussion
We studied the photoresponse behavior of the MoS2 junction devices to investigate the PV effect due to the built-in electric field. Figure 1a,b show a schematic of the device structure of MoS2 1L-3L junction with Au electrodes and an optical microscopy image of a typical MoS2 junction device (sample A), respectively. For sample A, the total channel length is 5 μm, and the thin and thick regions of the MoS2 junctions are 0.7 and 1.9 nm, respectively. We performed the photocurrent measurement with a photoexcitation (532 nm, 20 kW/cm2) focused at the MoS2 junction interface at VG = 60 V. Figure 1c compares the photocurrent at VSD = 0 mV and VSD = 50 mV, in which we readily found drastically different photoresponse between zero and small VSD. For VSD = 50 mV, the photocurrent exhibits a PPC which is the photocurrent that persists after the photoexcitation is terminated. For a disordered system, it is common to use the stretched exponential decay to describe the PPC relaxation30,31,32. Here, we can analyze the photocurrent relaxation by a single stretched exponential decay33
where τ is the decay time, and β is the exponent (0 < β < 1), yielding τ = 90 s for VSD = 50 mV. The decay time is comparable to the previous study30, suggesting that the observed PPC effect in the MoS2 sample is originated from random localized potential fluctuation30,31,32. In contrast, the short-circuit photocurrent (ISC) at zero bias is simply due to the PV effect driven by the built-in electric field, which exhibits a fast switching behavior and returns to the dark current level rapidly after illumination is terminated. The photocurrent relaxation can be fitted by a normal exponential decay (β = 1 in Equation 1)34,35,36. The value of τ is extracted to be approximately 60 ms, which are smaller than that under bias (VSD = 50 mV) by 3 orders of magnitude (Supporting information 3).
The drastic difference between photoresponse with and without applying VSD is intriguing and can be attributed to the difference in the carrier transport mechanism37. Here, we mainly focus on the slow-changing photocurrent that distinguishes the two scenarios. The current in the channel can be expressed as Jn = Jn(drift) + Jn(diffusion) = eμnnE − eDndn/dx. The carrier density is varied temporally due to the trapping and de-trapping process in the illumination and dark condition, respectively. Because the photoconductivity (PC) is determined by the drift current, which is proportional to the carrier density, the PC is greatly affected by the presence of the charge traps. In contrast, when no bias voltage is applied, ISC is induced by the built-in electric field at the junction and then driven by the diffusion current in the channel, which is proportional to the gradient of the carrier density dn/dx. Because dn/dx is related to the gradient of quasi-Fermi level dFn/dx but not the carrier density, it is much less sensitive to the trapping/de-trapping processes. Therefore, ISC exhibits a much faster photoresponse and the PPC effect can be greatly suppressed when the photocurrent is dominated by the short-circuit current.
Interestingly, the short-circuit photocurrent is virtually insensitive to the variation of the gaseous conditions over a range from a vacuum of 1 × 10−6 Torr to ambient condition. Here, we present this unique characteristic by showing theVSD dependent photocurrent (Iph) in different gaseous environment. Figure 2a compares Iph of a 1L-3L MoS2 junction (sample B) atVSD = 0 mV among vacuum, N2 (1 atm), and ambient condition, revealing that the magnitude of the ISC (ΔISC) is insensitive to the environment. We note that in all 5 1L-3L samples that we measured, the photoresponse behavior is comparable and therefore the data shown here is representative. Under very different pressure and chemical substances, ΔISC exhibits essentially the same value of approximately 140 pA. Moreover, the decay time (τ) values for these different conditions are also similar, as described in detail in Supporting information S3. In contrast, the photocurrent of sample B under bias critically depends on the environmental conditions. Figure 2b shows Iph atVSD = 5 mV under vacuum, N2, and ambient conditions, indicating the changes in photocurrent (ΔIPC) for these three conditions are 263, 133, and 41 pA, respectively. Here we note that while ΔISC is only governed by the built-in electric field, ΔIPC is referred to PC that is determined by both built-in field and external bias. Moreover, the sample exhibits largerτ under vacuum (90 sec) as compared to the τ under N2 (16 sec) and ambient (8 sec) conditions. This dependence of Iph on the chemical environment is typical for the 2D-material devices. The extracted parameters of mobility, threshold voltage, ΔISC, and ΔIPC are listed in Table 1.
As previously discussed, ISC is driven by the diffusion current that is related to the gradient of carrier density. This gradient is associated with the condition of illumination, but is negligibly affected by the trapping process and the presence of adsorbents. Consequently, both the magnitude and the response time of ISC are very insensitive to the gaseous environment, despite the pressure and the chemical composition being very different. We note that this insensitivity of the photocurrent in MoS2 junctions is very uncommon in 2D-material-based devices, considering the ultrahigh surface-to-volume ratio that leads to a large area being exposed to absorbents. Conversely, when bias voltage is applied, the carriers could be trapped or released during illumination and dark conditions, leading to the charging and the PPC effect, as seen in Fig. 2b.
To further investigate the observed PV effect due to the built-in electric field, we present the VSD dependence of the photocurrent. Figure 3a shows a schematic diagram of energy band alignment and photoinduced carrier dynamics in the MoS2 heterojunction. Type-I band alignment is implied based on the STS results shown in Fig. 4d,e. At zero bias, the photocurrent is mainly driven by the built-in field under illumination (Supporting Information S4), whereas both the built-in and the external field co-exist when bias voltage is applied. It is noted that only the electron conduction is considered here because all the MoS2 junction samples exhibit n-type semiconducting behavior (Supporting Information S1). We considered the photocurrent under two polarity of bias (the detailed calculation of ΔISC and ΔIPC are described in Supporting Information S5). The fast photoresponse of the photocurrent is negative, regardless of the polarity of bias, indicating that ISC is induced by the built-in electric field. We then compared the photocurrent of sample B corresponding to the PV and the charging/discharging effect as a function of VSD in small bias regime, as shown Fig. 3b. ΔISC is found to be relatively independent of VSD; this response is ascribed to the cancellation of the external field. In contrast, ΔIPC, and thus the charging/discharging effect, increases linearly with VSD; this response may be due to the phenomenon that the carriers are driven to the local potential minimum more efficiently with increasing bias.
Figure 4a shows a schematic of the experimental setup for the STM measurement on the MoS2 junction transistors. Before MoS2 exfoliation, we deposited a thin TiOx film of 5 nm38, which served as electron transfer dopant on MoS2 flakes39, resulting in the reduction of MoS2 sheet resistance. The deposition of a TiOx layer is therefore crucial for preventing of the STM tip from crashing during the measurement. Figure 4b shows a topography image of the MoS2 junction device (sample C) at the junction, revealing the step with two flat terraces. The apparent step height is 2.6 nm, corresponding to 4 layers of MoS2. Because the thinner MoS2 is characterized by AFM as 4 layers, the studied MoS2 sample is therefore a 4- to 8-layer junction. Figure 4c shows a high magnification topography image of MoS2 that reveals crystalline structure, indicating a pristine MoS2 surface.
To investigate the band alignment at the MoS2 junction, the STS technique was utilized to measure the normalized dI/dV as a function of bias, which corresponds to the local density of state (LDOS). Figure 4d shows the bias dependence of dI/dV curve for the two MoS2 terraces with different thicknesses. The onsets of the normalized dI/dV curves at positive and negative bias correspond to the conduction band edge (EC) and the valance band edge (EV), respectively (Supporting information S6). From the STS measurement, we deduce the following: the EC and EV in 8-layer MoS2 are 0.29 eV and −1.12 eV, respectively; the EC and EV in 4-layer MoS2 are 0.44 eV and −1.22 eV, respectively. By spatially mapping the normalized dI/dV curves, we plot the band alignment across the 4- to 8-layer junction, as shown in Fig. 4e. The deduced energy profile suggests that the band alignment of the MoS2 junction is type I, as depicted in Fig. 3a.
Finally, we present the current-voltage characteristics at different VG to analyze the field effect of the photocurrent in the MoS2 junction devices. Figure 5a compares the ISD − VSD curves of sample B in dark and under illumination with a 532 nm laser and 20 kW/cm2 at VG = 60 V. The ISD − VSD curves exhibit a linear behavior, indicating that resistor behavior, rather than the rectifying effect, dominates the transport property of the junctions. The slope of the ISD − VSD curves therefore approximates the conductance (G) of the device. Under illumination, G is enhanced by 40 times compared with G in dark; this enhancement is attributed to the generation of the photoinduced carriers. The open-circuit voltage (VOC) and ISC can be extracted as 60 μV and 0.12 nA, respectively. We note that the ISC can be further enhanced by reducing the Schottky barrier height via contact engineering.
Because the Fermi level in thin materials can be effectively tuned by the external electric field, it is intriguing to study the field effect of the current-voltage characteristics. Indeed, we observed a VG dependence of the ISD − VSD curves under illumination, as shown in Fig. 5b, in which G increases with increasing VG. To understand this field effect, we plot VOC and ISC as a function of VG, as shown in Fig. 5c. ISC is found to increase with increasing VG, similar to the ISD − VG curve (see Figure S1 of SI 1). This similarity is reasonable because ISC depends on the collection probability of the photoinduced carriers, which is correlated to the diffusion current and G. Moreover, as VG increases, the contact resistance may decrease due to Schottky barrier thinning40, leading to higher ISC. Conversely, the response of VOC decreasing with increasing VG can be attributed to the reduction of built-in electric field in the MoS2 junction. Because monolayer MoS2 is subjected to a stronger field effect compared with the few-layer MoS2 due to the different density of states, the rising of Fermi level in monolayer MoS2 is greater than that in few-layer MoS2 as VG increases, resulting in the reduction of the built-in electric field (see Fig. 3a) and thus VOC.
We further present the excitation power dependence of ISC at zero bias voltage (Fig. 5d) to examine the mechanism of the photocurrent generation41. We observe that ISC follows a power law ISC ∝ Pα, and the exponent α can be been extracted as 0.89 and 0.82 for VG = 0 and 60 V, respectively. For the PC and the PV process, the photoinduced carrier density is directly proportional to the rate of absorbed photons; therefore, α = 142. However, because the PC is excluded here (VSD = 0), the value of α therefore suggests that the PV is the dominant mechanism in the measured photocurrent. The deviation of the extracted α from unity may be attributed to electron-hole recombination at the MoS2 junction and/or nonradiative recombination centers43,44.
In conclusion, we demonstrated a unique environment-insensitive and gate-controllable short-circuit photocurrent in a MoS2 junction with differences in the number of layer. The environmental insensitivity of the short-circuit photocurrent can be attributed to the characteristic of the diffusion current. Conversely, the photocurrent with bias exhibits the typical PPC that greatly depends on the amount of the extrinsic adsorbents. The STM/STS measurement confirms the quality of the MoS2 junction samples and suggests the type-I band alignment of the junction. In addition to the effect of source-drain bias, the MoS2 junction devices exhibit strong back-gate voltage dependence, indicating the feasibility to control the photocurrent via field effect. The environment-insensitive photocurrent therefore shows an alternative method to design the device structure for TMDC-based electronic and optoelectronic applications.
The sample with the 1L-3L MoS2 junction was produced via mechanical exfoliation of MoS2 layers from the bulk MoS2 (SPI supplies) onto SiO2 (300 nm)/Si substrates. Next, a resist-free technique with a shadow mask (TEM grids) was utilized to deposit electrical contacts. The advantage of the resist-free technique is the lack of resist residue on the MoS2 surface resulting from the device fabrication process. We deposited Au (50 nm) as the electrical contacts using an electron-beam evaporator at a base pressure of 1.0 × 10−7 Torr. All of our MoS2 junction devices were measured in a cryostat (Janis Research Company, ST-500) under vacuum condition of 1.0 × 10−6 Torr. We performed DC electrical measurement using a Keithley 237 sourcemeter and applied the back-gate voltage using a Keithley 2400 sourcemeter. We employed solid-state CW laser (Nd:YAG, 532 nm) as the light source in the Raman spectroscopy and photoresponse measurements. The incident light beam was focused by an objective (100×, NA 0.6) with a spot size of ~0.9 μm.
How to cite this article: Shih, F.-Y. et al. Environment-insensitive and gate-controllable photocurrent enabled by bandgap engineering of MoS2 junctions. Sci. Rep. 7, 44768; doi: 10.1038/srep44768 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the Ministry of Science and Technology, Taiwan under contract numbers MOST 103–2112-M-001-020-MY3.
About this article
2D Materials (2018)