MoS2-Based Photodetectors Powered by Asymmetric Contact Structure with Large Work Function Difference

2D Mo2C was produced by a modified chemical vapor deposition method, and 2D Mo2C-Au was formed as an asymmetric contact structure with a large work function difference. Mo2C/MoS2/Au photodetectors powered by asymmetric contact structure can work under self-powered condition with a responsivity of 10−1 mA W−1. The detection performance of the photodetectors can be stable for at least 110 days. 2D Mo2C was produced by a modified chemical vapor deposition method, and 2D Mo2C-Au was formed as an asymmetric contact structure with a large work function difference. Mo2C/MoS2/Au photodetectors powered by asymmetric contact structure can work under self-powered condition with a responsivity of 10−1 mA W−1. The detection performance of the photodetectors can be stable for at least 110 days. Self-powered devices are widely used in the detection and sensing fields. Asymmetric metal contacts provide an effective way to obtain self-powered devices. Finding two stable metallic electrode materials with large work function differences is the key to obtain highly efficient asymmetric metal contacts structures. However, common metal electrode materials have similar and high work functions, making it difficult to form an asymmetric contacts structure with a large work function difference. Herein, Mo2C crystals with low work function (3.8 eV) was obtained by chemical vapor deposition (CVD) method. The large work function difference between Mo2C and Au allowed us to synthesize an efficient Mo2C/MoS2/Au photodetector with asymmetric metal contact structure, which enables light detection without external electric power. We believe that this novel device provides a new direction for the design of miniature self-powered photodetectors. These results also highlight the great potential of ultrathin Mo2C prepared by CVD in heterojunction device applications.


Introduction
Self-powered and high-sensitivity miniature photodetectors are very attractive for a broad range of applications, spanning the imaging, environmental monitoring, and biomedical diagnostics fields [1][2][3][4][5][6][7]. Two-dimensional (2D) materials offer the opportunity to produce the performance of self-powered and high-sensitivity miniature photodetectors [8][9][10][11][12][13][14]. 2D material-based photodetectors with various structures have been widely reported [15][16][17][18][19][20][21][22][23]. The fabrication of asymmetric metal contact structures based on 2D materials is an effective method to obtain self-powered photodetectors [24,25]. One advantage of self-powered devices driven by an asymmetric contact is that they are suitable for long operation times [26]. The large difference in work function between asymmetric source and drain contact metal is beneficial for the fabrication of efficient detection devices [27,28]. However, common metal electrodes have a high work function, which makes it difficult to create an asymmetric structure with a large work function difference between the contacts. Obtaining ultrathin and stable electrode materials with a low work function is still a major challenge. Some ultrathin transition metal carbides (TMCs) with low work function may be potential candidates for the fabrication of asymmetric contact structures [29][30][31][32]. Nevertheless, the materials prepared by chemically etching methods (known as MXenes) always exhibit surface terminations, such as hydroxyl, oxygen, or fluorine, which affect their properties and structures [33]. The small lateral size and thermal instability of MXene also limits its application in planar optoelectronic devices [34]. The chemical vapor deposition (CVD) method offers a new way to synthesize large-area 2D TMCs (such as Mo 2 C, TaC, and WC) and their derivatives, without surface terminations [35][36][37][38][39][40]. Ultrathin Mo 2 C crystals with a work function of 3.8 eV, prepared by the CVD method, are both thermally (they remain stable in air at 200 °C) and environmentally stable under H 2 plasma treatment [41][42][43]. It has been reported that CVD-grown ultrathin Mo 2 C (CVD-Mo 2 C) has significant superconducting properties, with a high critical temperature [35] and is suitable for the preparation of asymmetric metal structures. However, dense nucleation sites are often formed during the Mo 2 C preparation by the CVD method, which limits the lateral growth of Mo 2 C.
In this work, we first optimized the CVD process for preparing Mo 2 C. The use of graphene-intercalated copper can reduce the number of Mo 2 C nucleation sites in the initial growth stage, thereby facilitating the synthesis of large-area Mo 2 C. Then, a self-powered Mo 2 C/MoS 2 /Au device with an asymmetric metal contact structure was fabricated. The large work function difference (1.3 eV) between Mo 2 C and Au favors the fabrication of an efficient MoS 2 -based photodetector. The Mo 2 C/MoS 2 /Au-structured device could enable the detection of light without external electric power. We believe that this novel device provides a new direction for the design of self-powered miniature detectors. The present work also highlights the great potential of ultrathin CVD-Mo 2 C in electrode applications.

Preparation of MoS 2
Monolayer triangular MoS 2 was synthesized through an atmospheric pressure CVD system with a two-temperature-zone tube furnace. Thoroughly cleaned n-type SiO 2 (300 nm) on Si was used as the substrate. Molybdenum trioxide powder (Mo source) and sulfur powder (S source) were vaporized at temperatures of 850 and 180 °C, respectively, for the synthesis of MoS 2 on the SiO 2 substrate. Finally, the furnace was cooled down to room temperature naturally.

Preparation of Mo 2 C
Three-layer, 25-μm-thick Cu foil (Alfa Aesar, 99.95% purity) was cut into pieces and placed on a 50-μm-thick Mo foil (Alfa Aesar, 99.95% purity). The Cu/Mo substrate was placed in a single-temperature-zone CVD system and heated to above 1096 °C under 200 sccm Ar (type I method, see below). The other Mo foil was placed near the Cu/Mo substrate (type II method). CH 4 (5 sccm) and H 2 (300 sccm) were introduced into the chamber to grow the Mo 2 C crystals. Finally, the tube furnace was naturally cooled to room temperature before collecting the Mo 2 C sample.

Device Fabrication
After cleaning the Mo 2 C on Cu/Mo with H 2 plasma to remove graphene from the surface of the sample [43], the 1 3 Mo 2 C on the Cu/Mo substrate was placed face down on the silicon oxide wafer on which MoS 2 was grown. A 1 M (NH 4 ) 2 S 2 O 8 aqueous solution was then used for etching the Cu layer, following which the Mo 2 C on the substrate surface dropped onto the silicon oxide wafer. After the sample was dried, a photoresist was spin-coated on the Mo 2 C-MoS 2 sample at 4000 rpm for 60 s. After drying at 90 °C for 1.5 min, the target site for the vapor deposition of a gold electrode on the Mo 2 C-MoS 2 was exposed with a 200 mW cm −2 laser. Then, the sample was placed in the developing solution and allowed to stand for 1 min. After rinsing the developing solution with deionized water, gold was evaporated onto the sample. Finally, we obtained the multifunctional and horizontally structured Mo 2 C/MoS 2 / Au sample by washing away the remaining photoresist with acetone.

Theoretical Calculations
Structural relaxations and electronic calculations were performed by first-principles simulations based on density functional theory, as implemented in the CASTEP package [44]. The exchange-correlation interaction was treated within the generalized gradient approximation (GGA), using the Perdew-Burke-Ernzerhof (PBE) functional and a plane-wave basis with a kinetic energy cutoff of 500 eV [45,46]. The long-range van der Waals interactions were considered using the DFT-D2 dispersion correction proposed by Grimme [47]. The atomic positions and cell vectors were relaxed until the maximum force and maximum stress tolerance were less than 0.01 eV Å −1 and 0.02 GPa, respectively. Vacuum gaps of at least 15 Å were used to minimize the interactions between adjacent images of the single-layer structure. The reciprocal space was sampled with dense grids of 16 × 16 × 1 (for structural optimizations) or 20 × 20 × 1 (for accurate band structure calculations) k-points in the Brillouin zone. The lattice constant, Mo-S bond length, and S-Mo-S bond angle of MoS 2 after the structural optimization were 3.218 Å, 2.437 Å, and 80.63°, respectively. The calculated bond length of 3.218 Å is slightly different from that of the optimized lattice structure reported by Saha et al. [48] (a = b ≈ 3.19 Å).

Characterization
Raman spectroscopy (LabRAM HR800, He-Ne laser excitation at 532 nm) was used for structural characterization. Optical images were acquired by a Leica DM4000 M microscope. Field-emission scanning electron microscopy (FE-SEM, FEI Nova Nano-SEM 450) and transmission electron microscopy (TEM, Tecnai G220 U-TWIN) were used for investigating the morphology and structure of the samples. The current-time characteristics of the photodetector were measured by a low-temperature cryogenic probe station (Lake Shore CRX-6.5 K), a semiconductor parameter analyzer (Keithley 4200-SCS), and a light source (Energetiq EQ-1500).

Results and Discussion
A schematic illustration and the calculated electronic band structure of monolayer MoS 2 are shown in Fig. 1a. The calculated electronic band structure shows that the band gap of monolayer MoS 2 is 1.62 eV, which can efficiently absorb light with a wavelength below 765 nm. Figure 1b shows a schematic illustration and the calculated electronic band structure of monolayer Mo 2 C. Mo 2 C exhibits good metallic properties, similar to other TMCs without surface terminations. A schematic diagram of the Mo 2 C/MoS 2 /Au device and its energy band structure is shown in Fig. 1c; the device exhibits an asymmetric metal contact structure. The monolayer MoS 2 absorbs light and internally generates electron-hole pairs. The photogenerated electrons and holes move to molybdenum carbide and gold, respectively, under the potential difference induced by the asymmetric metal contacts.
A schematic diagram of the CVD growth process of Mo 2 C is shown in Fig. 2a. Methane is rapidly cracked at 1096 °C, providing the carbon source for the growth of Mo 2 C. There are two ways to grow molybdenum carbide on a copper/ molybdenum film substrate using the principle of metal immiscibility. In the first method (denoted as "type I"), the copper foil is placed on molybdenum foil as a growth substrate for Mo 2 C crystals. Mo atoms diffuse toward the surface of the copper foil at high temperatures and eventually react with carbon atoms to form molybdenum carbide. This method can produce highly crystalline 2D Mo 2 C on the in situ grown graphene [36]. However, under the growth conditions of the type I method, the rapid diffusion of Mo atoms in the copper foil at high temperature results in the formation of dense Mo 2 C nucleation sites on the surface of the copper foil. Figures 2b and S1 show optical images of Mo 2 C prepared by the type I method with growth times of 20 and 180 min, respectively. The optical images reveal the presence of high-density, small-area molybdenum carbide on the surface of the copper foil. As shown in Fig. 2c, in the type I growth mode, the diffusion process of Mo atoms from the molybdenum foil to the liquid copper surface is unhindered. We developed a new Mo 2 C growth method, denoted as "type II." In this method, copper foil is not placed on top of the molybdenum foil, but next to it. The distance between the Mo and the Cu foil is 2 mm. Copper vapor is deposited on the molybdenum foil under high-temperature and methane gas flow conditions. Graphene is continuously formed on the liquid copper surface during the copper vapor deposition process. Finally, Mo 2 C is formed on the surface of the copper film. Figure 2d shows an optical image of Mo 2 C prepared by the type II method with a growth time of 180 min. The comparison of Fig. 2b and S1 shows that the molybdenum carbide crystals prepared by the type II method exhibit significantly fewer nucleation sites as well as larger areas. Figure S2a, b shows the thickness of the Cu layer in the Cu/Mo substrate after 180 min of Mo 2 C growth by the type I and type II methods, respectively. The Cu layer thickness in Figure S2a is thicker than that in Figure S2b.
Generally, the thinner the copper layer in the substrate, the higher the amount of molybdenum carbide grown on its surface [49]. However, under the same growth conditions, the copper layer obtained by the type II method is thinner and less molybdenum carbide is formed. Therefore, we can infer that the growth process of molybdenum carbide in the type II method is similar to that shown in Fig. 2e, which shows the diffusion process of molybdenum atoms in copper. Graphene has a very fast growth rate. During the deposition of copper vapor onto the molybdenum foil, a large amount of graphene crystals is formed on the copper surface and further covered by the newly deposited copper, eventually forming a graphene-intercalated copper structure. The intercalation of graphene in copper hinders the diffusion of molybdenum atoms, resulting in a small number of Mo atoms diffusing into the copper surface. Graphene on the copper surface also has a passivation effect on the nucleation of molybdenum carbide. Finally, only a small number of Mo 2 C nucleation sites are formed on the copper surface, which leaves room for the formation of larger Mo 2 C crystals. Figure S3a shows an optical image of Mo 2 C with graphene, which transferred on the SiO 2 from the Cu/Mo substrate. Figure S3b shows the Raman spectra of graphene corresponding to the point marked "a" in Figure S3a.  Figure S4 displays optical images of Mo 2 C and MoS 2 on SiO 2 . The Raman spectrum of MoS 2 is shown in Fig. 3f, where the E 1 2g and A 1g peaks are found at 385.5 and 404.9 cm −1 , respectively. The difference between the positions of the two peaks (19.4 cm −1 ) is characteristic of monolayer MoS 2 [50]. To confirm the stoichiometry of Mo 2 C, the elemental distribution and electron energy-dispersive spectroscopy (EDS) data of Mo 2 C are shown in Figures S5  and S6, respectively.
In order to investigate the characteristics of Mo 2 C as an electrode material, we prepared a Mo 2 C/MoS 2 /Mo 2 C device and tested its I-V characteristics in the dark and under illumination. Figure 4a shows the optical image of the Mo 2 C/MoS 2 /Mo 2 C device. At zero bias, the device does not have photodetection capabilities. At small bias voltages, the dark current of the device does not increase significantly, while the light current shows a marked increase, as shown in Fig. 4b. This result indicates that Mo 2 C can form a good metal-semiconductor (MS) contact with MoS 2 , although the Mo 2 C/MoS 2 /Mo 2 C device In the type I method, copper foil is placed on molybdenum foil to directly grow molybdenum carbide. In the type II method, the molybdenum foil is placed near the copper foil. Copper evaporates at high temperatures and adsorbs on the molybdenum foil for the growth of Mo 2 C. b Optical image of Mo 2 C prepared by the type I method with a growth time of 20 min. c Schematic diagram of the type I Mo 2 C growth process. d Optical image of Mo 2 C prepared by the type II method. e Schematic diagram of the type II Mo 2 C growth process. The black line represents graphene with symmetrical electrode structure cannot provide selfpowered detection of light. Figure 5a shows the morphology and a schematic illustration of the Mo 2 C/MoS 2 /Au photodetector. The size of the channel between the gold electrode and the Mo 2 C material is approximately 5 μm. Au and Mo 2 C are connected to the source and drain electrodes, respectively. Since electrons in n-type MoS 2 are more likely to move from the valence to the conduction band under forward bias conditions, a larger source-drain current is generated when the device is illuminated under forward bias. Figure 6 shows the band energy diagram of Mo 2 C/MoS 2 / Au, without considering the surface states of MoS 2 . As MoS 2 is connected to Au and Mo 2 C, a potential difference ∆E eff is generated between Au and Mo 2 C. The value of ∆E eff corresponds to the difference between the work functions of Au (Φ Au = 5.1 eV) and Mo 2 C (Φ Mo2C = 3.8 eV): ∆E eff = (5.1-3.8) eV = 1.3 eV. The contact between n-type MoS 2 and Au forms an electron blocking layer at the interface. Because MoS 2 induces the accumulation of a large amount of holes in the electron blocking layer and of electrons on the Au surface, its energy band is bent upward at the interface and generates a built-in electric field (E 1 ) directed from MoS 2 to Au. The work function of MoS 2 on SiO 2 is 4.49 eV [51]. In the contact between MoS 2 and Au, the barrier height on the  Figure 7a shows that the response of the device to 0.5 mW cm −2 light at 600 nm is slightly higher than that at 400, 500, and 650 nm. The Mo 2 C/MoS 2 /Au photodetector exhibits almost no response to 700 nm light. Figure  S7 shows the current responses of the Mo 2 C/MoS 2 /Au photodetector to light of various wavelengths. When the wavelength of the incident light is increased to 700 nm, the photocurrent response of the device drops rapidly; therefore, the band gap of MoS 2 is about 1.77 eV. Figure 7b shows the photodetector response to 600 nm light of different intensities. The photocurrent increases with increasing light intensity. At an energy density of 1.78 mW cm −2 and a wavelength of 600 nm, the device has a switching ratio of approximately 10 and a responsivity of approximately 10 −1 mA W −1 . As shown in Fig. 7c, the response and recovery times of the photodetectors are 23 and 28 s, respectively. Moreover, Fig. 7d shows that the responsivity of the photodetector remains approximately constant over 110 days, indicating the high reliability and stability of self-powered Mo 2 C/MoS 2 /Au photodetectors.
The characteristics of the Mo 2 C/MoS 2 /Au device are compared to those of photodetectors with different structures in Table 1. Au/MoS 2 /Mo 2 C shows a larger difference between the work functions of the electrodes than that of other asymmetric structures. The self-powered Au/ CVD-MoS 2 /Mo 2 C device has a slightly faster response speed than Au/CVD-MoS 2 /Au which should work at bias voltage. Although the responsiveness of the Au/MoS 2 / Mo 2 C devices measured in this study is not outstanding, the responsivity of photodetectors driven by an asymmetric contact can be improved by reducing the spacing between the electrodes or decorating light-absorbing materials with quantum dots [27,52]. We thus believe that the Mo 2 C/semiconductor/Au configuration is an effective asymmetrical structure for self-powered photodetectors.

Conclusions
In conclusion, our results show that the use of grapheneintercalated copper to prepare Mo 2 C can affect the diffusion of Mo atoms in copper and reduce the number of Mo 2 C nucleation sites in the initial growth stage, thereby facilitating the synthesis of large-area Mo 2 C. The physical and chemically stable Mo 2 C has a low work function and is well suited for the preparation of asymmetric metal contact structures. The MoS 2 -based photodetectors powered by asymmetric contact structure with large work function difference can detect light of wavelength below 700 nm without external power. The responsivity of Mo 2 C/ MoS 2 /Au photodetectors is approximately 10 −1 mA W −1 under light irradiation at 600 nm and 1.78 mW cm −2 . The response and recover times are 23 and 28 s, respectively. This novel device may open new avenues for the design of self-powered multifunctional miniature devices. The present study also reveals the great potential of ultrathin CVD-Mo 2 C in electrode applications.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.