Photodetectors based on junctions of two-dimensional transition metal dichalcogenides
Wei Xia1, Yan Fa-Guang1, Shen Chao1, Lv Quan-Shan1, Wang Kai-You1, 2, †
State Key Laboratory of Superlattices and Microstructures, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China
College of Materials Science and Opto-Electronic Technology, University of Chinese Academy of Science, Beijing 100049, China

 

† Corresponding author. E-mail: kywang@semi.ac.cn

Abstract

Transition metal dichalcogenides (TMDCs) have gained considerable attention because of their novel properties and great potential applications. The flakes of TMDCs not only have great light absorption from visible to near infrared, but also can be stacked together regardless of lattice mismatch like other two-dimensional (2D) materials. Along with the studies on intrinsic properties of TMDCs, the junctions based on TMDCs become more and more important in applications of photodetection. The junctions have shown many exciting possibilities to fully combine the advantages of TMDCs, other 2D materials, conventional and organic semiconductors together. Early studies have greatly enriched the application of TMDCs in photodetection. In this review, we investigate the efforts in photodetectors based on the junctions of TMDCs and analyze the properties of those photodetectors. Homojunctions based on TMDCs can be made by surface chemical doping, elemental doping and electrostatic gating. Heterojunction formed between TMDCs/2D materials, TMDCs/conventional semiconductors and TMDCs/organic semiconductor also deserve more attentions. We also compare the advantages and disadvantages of different junctions, and then give the prospects for the development of junctions based on TMDCs.

1. Introduction

Since the discovery of graphene by Geim and Novoselov,[1,2] many unique properties[37] have emerged, resulting in considerable attention on layered materials.[811] Many efforts have been put into the fundamental physical researches on properties of layered materials for next generation of electronics and optoelectronics. Until now there have already been hundreds of layered materials that can be thinned down to monolayers and retain their stability.[3] Each layered material has its own merits and demerits. For example, the carrier mobility in black phosphorus (BP) is higher than that in transition metal dichalcogenides (TMDCs), but TMDCs have larger light absorption compared with BP in visible range.

TMDCs have the general formula MX2 where M is a transition metal (such as Mo, W) and X is a chalcogen atom (such as S, Se, Te).[1215] The atoms forming bounds are arranged into planes (layers). The layers are combined by van der Waals interactions to form bulk crystals.[16] The weak out-of-plane interactions allow the isolation of single layers by micromechanical exfoliation.

Crystals of TMDCs have been investigated for more than five decades.[16,17] However, only in past decade, the thin layers of TMDCs were set off an upsurge, where many interesting properties were observed, such as charge density wave, super-conductivity, variation of band structure, and so on.[12,18,19] Among them, the variation of band structure is prominent in optical properties. For example, the bandgaps of TMDCs change from indirect to direct when the TMDCs are thinned down to monolayers. The direct bandgap leads to increased probability for photon to generate electron-hole pairs which results in better light absorptions. Moreover the bandgaps of thin layered TMDCs rely on the number of layers,[12,13,2024] which allows the absorption of light at different wavelength. Due to the van der Waals interactions between layers, we can also stack different types of TMDC flakes regardless of lattice mismatching to have different wavelength light absorption.[25]

Earlier, photodetectors based on field-effect transistor (FET) structure of TMDCs have been studied[2631] and have shown excellent performances, such as the highest photoresponsivity of 880 A·W−1 based on monolayer MoS2.[31] There have already been many reviews.[24,25,3236] Because the built-in electric field can effectively separate photogenerated electron–hole pairs, the junctions of TMDCs have gained more attentions for optoelectronics in these years. With the rapid development of the photodetectors based on the junctions of TMDCs, it is time to review the recent achievements. In this review, we mainly present photodetectors based on junctions of TMDCs flakes utilizing photoconductive effect and photovoltaic effect. Firstly the underlying physics of photodetection based on P-N junctions are briefly introduced. We then will discuss and compare the different kinds of junctions based on TMDCs (including techniques and performances). After that, we conclude the advantages and disadvantages of different junctions. Finally we give the prospects for the future developments of junctions based on TMDCs.

2. Two main mechanisms of photodetection: photoconductive effect and photovoltaic effect

Different materials normally have different work functions and Fermi levels. When junctions are formed between p type and n type materials, the band structures will change due to the alignment of the Fermi levels. The conduction and valance band edge of p type materials will bend down. On the contrary, the conduction and valance band edge of n type materials will bend up. Then the energy barrier will be generated between the two materials.

In the photoconductive effect, when the photon energy is higher than the bandgap, photon absorptions generate extra electron–hole pairs. The photogenerated electrons and holes can be drifted in opposite directions by built-in electric field and external source-drain voltage, which greatly reduces recombination of electron–hole pairs and generates photocurrent (Fig. 1(a)). The photocurrent adds to the dark current, resulting in the reduction of the resistance. Under negative external voltage, the internal electric field will be enhanced and the energy barrier will increase by qVR (Fig. 1(a)). Photogenerated electrons are easier to be drifted to low energy side (the side of n type material) and photogenerated holes are easier to be drifted to low energy side (the side of p type material). Thus, the photogenerated electrons and holes are separated more effectively.

Fig. 1. (a) Band alignment of PN junction under photoconductive mode. (b) Band alignment of PN junction under photovoltaic mode.

In the photovoltaic effect, photocarriers are also generated in the same way as the photoconductive effect. However, the difference lies in that photogenerated electron–hole pairs are separated only by built-in electric filed rather than external field Vds. The built-in electric field generally exists in PN junctions or Schottky junctions. The built-in electric field can effectively separate electrons and holes, which leads to a sizeable short-current Isc at Vds = 0. Moreover, when the circuit is open, photocarriers will be accumulated on both sides, and open circuit voltage Voc can be obtained (Fig. 1(b)).

There are two operating modes for the PN junction photodetectors: photoconductive mode (the photodetectors work under external bias, Fig. 1(a)) and photovoltaic mode (the photodetectors work without external electric field, Fig. 1(b)). The advantage of photovoltaic mode is the lowest dark current which improves the specific detectivity. The advantages of the photoconductive mode are as follows: i) the reduction of the junction capacitance increases the response speed; ii) under reverse bias, the recombination probability of the photogenerated electrons and holes are smaller, making it suitable for weak light detection; iii) under reverse bias, the width of depletion layer will be extended, which will increasing photon collection efficiency.

To facilitate comparison of the photodetectors, we mainly focus on the following figures of merits of photodetectors based on junctions: external quantum efficiency (EQE), responsivity (R), specific detectivity (D*), and response time (τ).

External quantum efficiency is a ratio between the number of the charge carriers generated and the number of incident photons on photodetectors. It depends on both the absorption of light and collection of the charges. Responsivity is the ratio between the photocurrent and the incident light power density on photodetectors, which implies the achievable electrical signal under certain illumination. Specific detectivity is equal to the reciprocal of noise-equivalent power, normalized per square root of the sensor’s area and frequency bandwidth, which indicates the measure of detector sensitivity. The time response of detectors is measured between 10% and 90% of the final output signal, either on the rising or falling edge.

3. Photodetectors based on homojunctions of TMDCs

In conventional semiconductors, homojunctions are the semiconductor interfaces that occur at the similar semiconductor materials with equal bandgaps and different doping. Similarly, homojunctions exist at the interface of 2D materials such as graphene,[3739] WSe2,[4043] and black phosphorus.[44] In conventional semiconductors, elemental doping is the most common way to form the homojunctions. However, there are three ways to modulate the carrier type of TMDCs to create homojunctions. The first one is to change the surface properties via chemical treatment.[4548] The second one is to change the properties of the single-crystalline TMDCs by elemental doping.[49,50] The homojunctions can also be created by electrostatic gating except for doping.[4043] In the following, we will review the recent development concerning the homojunctions created by these three methods respectively.

3.1. Homojunction fabricated by surface chemical doping

In 2012, Hui et al. obtained p-doped WSe2 using physisorbed NO2. However, the doping decays almost within an hour because of the weak physisorption process.[51] Next year they realized n-doping of WSe2 and MoS2 by Pstassium, which could decrease the contact resistance.[45] Surface chemical doping is demonstrated to lead to the conduction transition of thin TMDC layers. Lately, lateral homojunction devices based on surface chemical doping were presented by Choi et al. in Fig. 2(a).[46] The stacks of MoS2 and boron nitride (BN) flakes in the devices rely on the mechanical exfoliation and transfer. BN was partially covered on MoS2 by polydimethysiloxane stamp. The AuCl3 chemical dopant was spin-coated onto the partially stacked h-BN/MoS2 heterostructure.

Fig. 2. (color online) (a) Lateral device schematic diagram of PN junction by AuCl3 chemical dopant.[46] (b) IdsVds characteristics measured in different gate configurations.[46] (c) EQE and D* with respect to the wavelength of light at Vds = +1.5 V and Vg = 0, 40 V and −40 V.[46] (d) Vertical PN junction schematic with CHF3 plasma-treated MoS2 layers.[52] (e) JdsVds characteristics of CHF3 plasma-treated PN junctions.[52] (f) EQE under different wavelength of light.[52]

The Au nanoaggregates can form due to the reduction of ions in the AuCl3 solution. Because of the reaction, the ions which are in direct contact with MoS2 can receive electrons from MoS2 layers. Thus, the p-type MoS2 will be formed by surface charge transfer, and then the homojunctions of P–N junctions based on MoS2.

The log-plotted IdsVds curves under different gate configurations show the reverse current can be decreased with increasing Vg, which is very useful for photodetectors. The large tunable conductance can be attributed to the gating tuning Fermi surface towards or backwards the valance band, which changes the hole transport. Meanwhile, the curves in Fig. 2(b) imply the rectifications resulting from the chemical doping.

Figure 2(c) plots the EQE (left axis) and D* (right axis) for different wavelength of light. The maximum EQE obtained is ∼ 7000% at Vds = 1.5 V, Vg = −40 V, and 350 nm. It is worth noting that the EQE under different conditions is around 100 times higher than that of the intrinsic MoS2 phototransistor.[53] Meanwhile the obtained D* is similar to the reported multilayer intrinsic MoS2 phototransistors. Therefore the AuCl3 doping and stacking processes have not influenced the sensitivity of the devices.

Another example presented here is the MoS2 P–N junctions fabricated by CHF3 plasma doping.[52] As shown in Fig. 2(e), the junction exhibits more prominent diode-like behavior than the Au/untreated MoS2/ITO heterostructure. The junction formed at CHF3 treated and untreated MoS2 surfaces shows the built-in potential of 0.7 eV. Under photovoltaic mode, the EQE of CHF3 treated MoS2 PN junction can reach a high value of 78%.

In 2016, Lei et al. fabricated planer p–n junctions based on 2D InSe by making use of the long pair electrons found in most of 2D metal chalcogenides rather than relying on lattice defects and physisorption methods.[48] Via a Lewis acid-base reaction, the long pair electrons of n type InSe can react with Ti4+ to form p type coordination complexes. Moreover the conduction type conversion in other 2D materials can also be realized via Lewis acids.

To conclude, surface chemical doping is an effective way to prepare photodetectors based on the homojunctions of TMDCs without influencing the sensitivity of the photodetectors. Not only under photoconductive mode but also under photovoltaic mode, the homojunctions have shown higher EQE than the intrinsic TMDCs. Moreover chemical surface doping can also be used to decrease the contact resistance between TMDCs and metal. However, the properties of the homojunctions of TMDCs made by surface chemical doping can easily be affected by external environment. Thus the stability of these devices is rather poor.

3.2. Homojunctions fabricated by elemental doping

To have more stable and controllable PN junction, elemental doping has been used to create the homojunctions. Practical applications require substitution of host atoms with dopants, where the doping is secured and stabilized by covalent bonding inside the lattice. Suh et al.[49] realized stable P-type doping in bulk MoS2 by substitutional niobium (Nb) and achieved degenerate hole density of ∼ 3×1019 cm−3. In order to form p-type MoSe2, Jin et al. also used Nb with five valance electrons to replace Mo which has six valance electrons.[50] Via Nb doping, n-type MoSe2 single crystal can be successfully converted to p-type (Fig. 3(a)). The thin n-MoSe2 and p-MoSe2 layers were stacked by mechanical exfoliation and transfer. Then the vertical homojunctions based on layered MoSe2 have been realized on SiO2/Si substrates (see insets in Fig. 3(b)).

Fig. 3. (color online) (a) Schematic diagram of Nb-doping for converting n-MoSe2 to p-MoSe2. (b) IdsVds characteristics measured in different gate configurations. The insets: optical image and schematic of the vertical homojunctions based on MoSe2. (c) JdVds curves under different laser powers. The inset: short circuit current and open circuit voltage as a function of PLaser. (d) EQE and D* with respect to the laser power under different Vds.[50]

The fabricated p–n junction exhibits typical rectification properties, as shown in Fig. 3(b). Not only that, the ideality factor was calculated to be ∼1, which implies the device is an ideal diode. With increasing illuminated light intensity, short-circuit current (Isc), open-circuit voltage (Voc) and reverse current are all increased, as shown in Fig. 3(c). This is consistent with the photovoltaic mechanism where the generated photocurrent can add to the reverse current. In Fig. 3(d), both the EQE and the responsivity are enhanced at Vds = 0 and Vds = −1 V as the laser power increased. Photocarriers can be more easily drifted to electrode across the depletion region under reverse bias. For the vertical MoSe2 homojunction by elemental doping, the maximum EQE and R value can reach ∼ 1049% and be more than 1 A·W−1 at Vds = 1 V, PLaser = 890 W·m−2, respectively.

The elemental doping is superior in terms of versatility and stability over surface chemical treatment based on absorption or intercalation of volatile species. However, the elemental doping is more suitable for TMDC bulk materials rather than very few layered materials. In order to fabricate PN junctions based on the few-layer TMDCs, mechanical exfoliation is often essential, which can induce uncertainty factors such as the thicknesses and the resistance between the flakes. Thus, the current method is not suitable for practical application. Recently, Kim et al. realized phosphorus doping in monolayer and bilayer MoS2 by laser.[54] This approach can be controlled effectively by varying the laser irradiation power and time. However, more solutions of elemental doping in thin layers are still needed. We expect to realize PN homojunctions based on layered TMDCs through the elemental doping during the CVD growth, which is important for the application of TMDCs.

3.3. Homojunction fabricated by electrostatic gating

Unlike other TMDCs such as MoS2, MoSe2, and WS2, WSe2 has been demonstrated to be ambipolar transport.[4042,55] The carrier type of WSe2 can be controlled through electrostatic gating. The homojunction based on WSe2 has been created successfully by this method.[40]

The schematic diagram of the locally-gated junction is presented in Fig. 4(a). The single-layer WSe2 flake was transferred on two local gates covered with BN as a dielectric layer. Except BN,[40,42,56] conventional dielectrics such as HfO2,[41] Si3N4,[43] Al2O3[39] have also been used in similar type devices. Compared with HfO2 and Si3N4, BN flake has atomically flat surface and charge traps-free interface, it can improve the interface between BN and WSe2, thus enhance the mobility of devices.

Fig. 4. (color online) (a) Device schematic diagram. (b) The log plotted IdsVg curve was measured under Vds = 500 mV. The color gradient implies the transition between hole- and electron-doping. Inset shows the data on a linear scale. (c) Color maps of Ids under different Vlg, Vrg and Vds which suggest the types of the junctions. (d) IdsVds characteristics under different illumination wavelength. The inset shows the responsivity (R) under Vds = −1 V and Vds = 0.[40]

In Fig. 4(b), the ambipolar Ids can be controlled by Vg, implying both hole and electron-doping can be accessed. This phenomenon can be attributed to that the Femi level EF can be effectively moved up and down across the bandgap by Vg in thin layers of WSe2. The moving of EF to the conductive band or valance band will induce different carrier types and concentrations. Similar behavior has also been reported by others.[42,43] So the types of the junctions (PP, PN and NN) can be adjusted by controlling Vlg and Vrg, as shown in Fig. 4(c).

Then we focus on the photoresponse of the PN junction formed by electrical gating. The maximum responsivity is 0.7 mA·W−1 at Vds = 1 V and 532 nm illuminations. Meanwhile the EQE reaches a maximum value of 0.1%. In addition, the response time is in the order of 10 ms. Though the EQE here is lower compared with the two doping methods mentioned above, locally electrostatic gating is still a novel way to tune the junctions and control their electrical properties.

To conclude, the homojunction photodetectors based on TMDCs have been introduced, which can be realized by three different ways: surface chemical doping, elemental doping and electrostatic gating. Compared with intrinsic TMDCs, photodetectors based on homojunctions can largely improve the EQE and R. Meanwhile homojunctions are fabricated by only one kind of TMDCs, which makes them promising as detectors for special wavelength applications from visible to near infrared.

4. Photodetectors based on heterojunctions of TMDCs

In contrast to homojunctions, heterojunction is the semiconductor interface that occurs between two dissimilar materials with unequal bandgaps. Flakes of TMDCs can be placed on conventional semiconductor materials, organic semiconductor or on other 2D materials by ways such as CVD, mechanical exfoliation and transfer and so on. Due to the unique layer thickness dependence of the bandgaps, the heterojunctions can also be made of only single TMDCs with two different layer thickness.

4.1. Heterojunctions based on TMDCs, conventional bulk semiconductors, and organic semiconductors
4.1.1. Heterojunctions based on TMDCs and conventional bulk semiconductors

Compared with heterojunctions based on two different 2D materials, it is easier and more feasible to combine a 2D layered material with conventional semiconductor materials such as MoS2/Si,[5762] WSe2/InAs,[63] MoS2/GaAs,[64] and MoS2/GaN.[65]

The MoS2/Si heterojunctions have been widely investigated.[5762] Most of such junctions were fabricated by mechanical exfoliation and transfer.[5760] Esmaeili-Rad et al. fabricated heterojunction of MoS2/a-Si and they obtained the photoresponsivity of 210 mA/W at green light.[57] Vertical MoS2/p-Si heterojunction was fabricated by transferring and the high photoresponsivity of 7.2 A/W was obtained.[60] In order to have large scale such heterojunctions, Wang et al.[61] and Hao et al.[62] fabricated the MoS2/Si heterojunctions by magnetron sputtering.

Figure 5(a) shows the device geometry fabricated by Wang et al.[61] The MoS2 film was deposited on the p-Si substrate with a predefined SiO2 window by magnetron sputtering. The deposited molecular layers of MoS2[61] are perpendicular to the substrate as shown in Fig. 5(b), which is different from the MoS2 prepared by mechanical exfoliation[57,59,60] and chemical vapour deposition.[58]

Fig. 5. (color online) (a) Device schematic of the MoS2/Si heterojunction photodetector.[61] (b) Schematic illustration shows the separation of the photocarriers under illumination.[61] (c) Photoresponse characteristics measured under different light intensities of 808 nm laser.[59] (d) Responsibility and specific detectivity with respect to the light intensity.[61] (e) Schematic structure of MoS2/GaAs heterojunction.[64] (f) Current density-voltage curves of MoS2/GaAs heterojunction under dark and illumination.[64]

The Isc and Voc increase with the increased light intensity in Fig. 5(c), which strongly suggests that photocarriers are separated by the electric field at the interface of MoS2 and Si. Responsivity can reach ∼ 300 mA·W−1 under the light intensity of 1 mW·cm−2 which is comparable to most of the devices based on MoS2.[26,50,57,66] The maximum value of specific detectivity is ∼ 1013 Jones at 1 mW·cm−2 which is three orders of magnitude higher than MoS2/Si heterojunction photodetectors by transferring.[60]

The rising and decaying time of the MoS2/Si heterojunctions were calculated to be 3 µs and 40 µs, which is by far the reported fastest response in TMDCs heterojunctions. The fast response is directly related with the structure and can be attributed to the following reasons: the strong build-in electric field which can efficiently separate photocarriers; the vertically standing layered structure of MoS2 layers which offers high-speed paths; thin SiO2 between Si and MoS2 which serves as a passivation layer to reduce the interface defects.

Lately, Xu et al. developed MoS2/GaAs heterojunctions which have wide response band width from ultraviolet to visible light.[64] In order to decrease the dark current, h-BN was inserted into the interface between MoS2 and GaAs. Owing to h-BN, the detectivity can reach 1.9 × 1014 Jones which is even higher than that of MoS2/Si heterojunction.[61] MoS2/GaN heterojunctions have also been prepared by Ruzmetov et al. via power vaporization technique and show very high-quality van der Waals interface.[65]

The MoS2/Si heterojunctions show not only high specific detectivity but also short response time.[61] The photodetectors based on MoS2/GaAs heterojunctions have wider photoresponse band width. The efforts above confirm that combining 2D TMDCs with conventional semiconductor technology is a promising way to make high performances PN heterostructural photodetectors.

4.1.2. Heterojunctions based on TMDCs and organic semiconductor

Compared with conventional bulk semiconductor, much cheaper organic semiconductor materials including pentacene,[67,68] C8-BTBT,[6971] CuPc,[72] and PTCDA[73] have also enriched the design of heterojunction based on TMDCs.

A lot of efforts have been made to epitaxially grow organic semiconductor thin film on 2D materials such as graphene, BN, and MoS2. He et al. grew high-quality few layer C8-BTBT on graphene and BN via van der Waals epitaxy. The thickness of C8-BTBT can be well controlled down to monolayer.

Lately the heterojunctions based on monolayer MoS2 and p-type C8-BTBT in Fig. 6(a) were also presented by He et al.[70] MoS2 was mechanically exfoliated on SiO2/Si as an epitaxy substrate and then few-layer C8-BTBT on MoS2 was grown by epitaxial growth. The sharp and clean interface can be formed between the MoS2 and the organic C8-BTBT, which is important for the performance of the heterojunction. Under illumination, obvious Isc and Voc response are observed in the heterojunction as shown in Fig. 6(b) and the responsibility can reach 22 mA/W under photovoltaic mode. At the same time, Jariwala et al. fabricated heterojunctions based on MoS2 and p-type pentacene.[67] As shown in Fig. 6(d), the devices show apparent diode-like behavior and large photoresponse. This kind of hybrid heterojunctions has attracted a lot of attention and can show good performances by selecting appropriate device structures and materials. Compared with conventional semiconductors, organic semiconductors can be made in inexpensive methods. Moreover the low annealing temperature of organics allows for flexibility of substrate choice. It is promising to fabricate cheaper photodetectors by combining TMDCs and organic semiconductors.

Fig. 6. (color online) (a) Microscope image and device schematic of vertical C8-BTBT/MoS2 heterojunction.[70] (b) IdsVds under white light illumination with different power.[70] (c) Structure of pentacene/MoS2 heterojunction.[67] (d) IdsVds under dark and illumination.[67]
4.2. Photodetectors based on heterojunctions of only TMDCs
4.2.1. Fabrication of heterojunctions based on only TMDCs

Because of the weak van der Waals interactions, we can stack any two kinds of 2D materials together in order to form heterojunctions by mechanical exfoliation and transfer regardless of the lattice mismatch, which will give us wide variety selections of layered materials. A lot of efforts have been made to stack vertical heterojunctions of 2D TMDCs by mechanical exfoliation and transfer as shown in heterojunctions such as MoS2/WS2,[74,75] MoS2/WSe2,[7682] MoS2/MoSe2,[83] MoS2/MoTe2,[84] MoSe2/WSe2,[85] and so on. Although such vertical heterojunctions show very good performance in photodetection, the yield and efficiency of making such devices is very low.

In order to reduce the steps of mechanical exfoliation and transfer, Cheng et al. combine physical vapor deposition process, mechanical exfoliation, and transfer together.[86] However, the efficiency and the quality of junctions still have not been improved much.

Since MoS2 layers were grown by chemical vapor deposition (CVD),[8789] CVD has been regarded as a feasible and efficient way to directly fabricate heterojunctions on a large scale. Moreover, CVD-grown devices showed stronger interlayer interaction compared with their transferred counterparts.

In 2014, Gong et al.[90] grew both vertical and lateral MoS2/WS2 heterojunctions by CVD. As the FET, the vertical MoS2/WS2 heterojunctions have very high ON/OFF ratio of 106 and the estimated mobility is in the range of 15 to 34 cm2·V−1·s−1, which is faster than MoS2/WS2 made by mechanical transfer (0.51 cm2·V−1·s−1). The high mobility can be attributed to the clean interface between WS2 and MoS2, which reduces the unwanted defects between layers to improve the charge transfer. Meanwhile, the lateral MoS2/WS2 heterojunctions have been demonstrated to be intrinsic monolayer PN junctions without external electrostatic gating.

Afterwards scalable production of few-layer MoS2/WS2 vertical heterojunction array was reported by Xue et al.[91]

The few-layer MoS2/WS2 vertical heterojunctions (Fig. 7(a)) were fabricated by sulfurization of patterned WO3 and Mo sheets with an ambient pressure thermal reduction process. There are two Raman-active modes (1E2g, A1g) that can be generally observed due to the selection rules for scattering radiation or limited rejection of the Rayleigh scattering radiation.[93] In the Raman spectra of MoS2/WS2, the position of 1E2g and A1g peaks do not change, implying that MoS2 and WS2 are individual sheets. So the two-step chemical vapor deposition approach has been demonstrated to effectively prevent the mixture of MoS2 and WS2, resulting in a well-defined interface between MoS2 and WS2. Moreover, the thicknesses of MoS2 and WS2 in the heterojunctions can be adjusted by controlling the thicknesses of the predeposited WO3 and Mo sheets during the fabrication.

Fig. 7. (color online) (a) Optical micrograph of the MoS2/WS2 vertical heterojunction device.[91] (b) Raman spectra of MoS2, WS2 and MoS2/WS2.[91] (c) Optical image of vertical MoS2/WS2 heterojunctions grown by chemical vapor deposition.[92] (d) IV measurement across a MoS2/WS2 heterointerface.[92]

In the same year, Li et al. realized the controlled epitaxial growth of lateral WSe2/MoS2 heterojunction.[94] WSe2 was grown on c-plane sapphire substrates through van der Waals epitaxy and then MoS2 was epitaxially grown at the edge of WSe2. The atomically sharp transition in compositions at junctions can be achieved through precisely controlling the two-step epitaxial growth. Very recently, Choudhary et al.[92] reported large-area (> 2 cm2) vertical MoS2/WS2 heterojunctions grown by chemical vapor deposition. The heterojunctions show diode-like current rectification, which means them to be high-quality van der Waals PN junctions (Fig. 7(d)).

To conclude, although mechanical exfoliation and transfer are still widely used to stack two kinds of 2D materials, CVD techniques have been demonstrated to be a reliable way to fabricate large scale vertical and lateral heterojunctions of TMDCs which are more promising for application. Meanwhile, some new methods also appear. Tan et al. achieved 2D semiconductor hetero-nanostructures by the epitaxial growth in liquid phase at room temperature.[95] Mahjouri–Samani et al. fabricated array of lateral MoSe2/MoS2 heterojunction by lithographic patterning and controllable conversion of exposed MoSe2 to MoS2, which also provides some new ideas for device fabrication.[96]

4.2.2. Photoelectric properties of vertical heterojunctions based on only TMDCs

Furchi et al.[77] fabricated vertical stacks of WSe2/MoS2 exactly by mechanical exfoliation and transfer. The EQE is a low value of about 1.5%. A much higher performance in similar vertical WSe2/MoS2 heterojunctions was reported by Cheng et al. (Fig. 8(a)).[86]

Fig. 8. (color online) (a) Schematic of vertical WSe2/MoS2 heterojunction; (b) IdsVds curves measured in the dark and under the illumination whose wavelength is 514 nm and power is 5 µW. The inset shows response of the current with respect to periodic on/off of the illumination; (c) EQE under different power of 514 nm and 633 nm laser at Vds = 0 and Vg = 0.[86]

The vertical WSe2/MoS2 heterojunctions have obvious open-circuit voltage and short-circuit current under laser illumination (514 nm, 5 μW) (Fig. 8(b)). Also a fast response time of 100 μs has been demonstrated. The maximum value of EQE can reach 12% under laser illumination (514 nm, 0.5 μW) (Fig. 8(c)) which is higher than homojunctions based on WSe2.[41,43] The high performance can be resulted from the efficient separation of photocarriers at the interface of WSe2 and MoS2.

More recently, in the MoS2/WSe2 heterojunction, the ultrafast electron transfer from WSe2 to MoS2 sheet was demonstrated to be 470 fs and a 99% efficiency upon optical excitation. This implies MoS2/WSe2 heterojunction could have big potential for high-speed photodetectors.[82]

4.2.3. Photoelectric properties of lateral heterojunctions based on only TMDCs

In addition to vertical heterojunctions, lateral 2D heterojunctions have also emerged. Lateral heterojunctions based on TMDCs are commonly fabricated such as MoSe2/MoS2,[96,97] MoSe2/WSe2,[98] WS2/MoS2,[90] WS2/WSe2,[97] WSe2/MoS2.[94]

Duan et al. realized the lateral epitaxial growth of WS2/WSe2 heterojunctions (Fig. 9(a)).[97] Afterwards, a 50-nm-thick Al2O3 was deposited on part of WSe2 to insulate it from the contact for WS2. Compared with the IdsVds curve in the dark, the heterojunctions exhibited obvious photovoltaic with an open-circuit voltage of 0.47 V and short-circuit current of 1.2 nA (Fig. 9(b)). Moreover, the response time is less than 100 μs which is two orders of magnitudes less than the response time of the lateral WSe2 homojunction.[40] The faster response speed is originated from the carriers transport mainly in surface of lateral heterojunctions. The value of EQE is about 9.9% for these devices. Li et al. investigated the WSe2/MoS2 lateral heterojunctions.[94] They found that the depletion width of the heterojunction is 320 nm and device shows clear photovoltaic effect.[94]

Fig. 9. (color online) (a) Optical image of lateral WS2/WSe2 heterojunction.[97] (b) IdsVds characteristics in the dark and under laser illumination (514 nm, 30 nW). The inset shows the photocurrent response under periodic on/off laser illumination.[97] (c) Color-coded rendering of the real space imaging of band profile plotted in terms of log(dI/dV) in monolayer-bilayer WSe2.[99] (d) Band alignment in monolayer-bilayer WSe2 heterojunctions.[99] (e) Selective subset of log(dI/dV) spectra. The green and red arrows represent valance and conduction band of the interface states.[99]

Moreover, due to the thickness dependence of bandgap and band edges of TMDCs, a novel heterojunctions have appeared, which consist of different layers of the same TMDCs. Interlayer states can exist in these heterojunctions such as monolayer-bilayer WSe2 heterojunctions[99] and monolayer-multilayer MoS2 heterojunction.[100,126] Zhang et al. studied interlayer states between monolayer and bilayer WSe2 by low-temperature scanning tunneling microcopy and spectroscopy (STM/S). The interlayer states have a gap value of 0.8 eV as shown in spectrum #13 and spectrum #14 of Fig. 9(e). Because of the interlayer states, the band bending and build-in electric field can be induced between interfaces (Fig. 9(d)), which implies that they are a novel class of heterojunctions for photodetectors. The advantage of this kind of heterojunctions is that gaps can be adjusted by thickness. Thus through the combination of the same TMDCs of different thickness, detecting wavelength can be tuned and extended.

To conclude, along with development of different techniques, large scale vertical and lateral heterojunctions have been demonstrated and they have exhibited some excellent properties such as the fast response time which is less than 100 μs. Heterojunctions which consist of different layers of the same TMDCs are also greatly promising for new design of photodetectors. Different structures provide a more solid foundation for more complex structures such as PIN junctions and quantum well structures. However, it is still not good enough for applications. For example, the area of heterojunctions with higher quality is not larger enough and EQE is still very low. Hence, the fabrication and performances of TMDC/TMDC heterojunctions all need to be improved.

4.3. Photodetectors based on heterojunctions of TMDCs and other 2D materials

Though TMDCs have the advantage of strong light absorption, applications based on TMDCs are limited by the natures such as low carrier mobility compared with graphene[2,5,101] and few-layer black phosphorus.[102104] Thus, in order to make devices with better performance, a native way is to combine good properties of TMDCs and the advantages of other 2D materials by stacking their layers together.

The approach has already been demonstrated by Deng et al.[105] The MoS2/BP heterojunctions combine high carrier mobility of BP with strong light absorption of MoS2. The heterojunctions show high responsivity of 418 mA·W−1, but low EQE of 0.3%. Due to built-in electric field, the photocarriers can be effectively separated and then the photoelectrons are injected into BP layer, rather than trapped in MoS2.

More recently, the similar study of WSe2/BP heterojunctions (Fig. 10(a)) were reported by Chen et al.[106] Due to the weak screening effect and the ambipolar transport of WSe2[4042,55] and BP,[103,107] different types of heterojunctions can be well controlled by electrostatic gating. As shown in Figs. 10(b) and 10(c), the rectification properties almost disappear when Vg > −33 V and Vg < 25 V. The values of EQE are shown in Fig. 8(d) and the maximum value is 3.1% at Vg = 31 V, which is larger than 0.3% in the MoS2/BP heterojunctions.[105] By stacking TMDCs and other 2D materials together, their merits can be combined in the same device such as the bipolar transport of WSe2 and BP, light absorption of MoS2 and high mobility of graphene, and so on.[108111]

Fig. 10. (color online) (a) Device schematic diagram and optical image of WSe2/BP heterojunction. (b) and (c) IdsVds characteristics under different Vg. The insets show the rectification ratio with respect to Vg. (d) Relations between EQE and power of laser under Vg = −40 V and Vg = 31 V.[106]

Although the junctions have shown excellent properties, there are still many methods to improve performances by using graphene as electrodes to reduce the influence of barrier,[112,113] facilitating the carrier collection by designing a vertical sandwiched structure[76,113,114] and using BN as a dielectric layer to improve interface and increase the reflection between the layers.[56,115]

Early Luo et al. has proven the response time of photodetectors based on InSe can be decreased to 120 μs by using graphene as electrodes, which is about 40 times faster than the detectors using metal electrodes.[112] Lee et al. fabricated vertical MoS2/WSe2 heterojunction by using top and bottom graphene electrodes (Fig. 11(a)).[76] The device had photoresponsivity of ∼ 10 mA·W−2, which is five times larger than that of MoS2/WSe2 heterojunction with metal contacts. There are several advantages to use graphene as electrodes. Firstly ultrafast carrier mobility of graphene is very useful in the separation of photogenerated carriers, which can enhance the photoresponsivity and further improve detectivity. Secondly the graphene can absorb few light which has little effects on the light absorption of the heterostructure. Finally the overlapping of the top graphene and bottom graphene facilitates the carrier collection by reducing the vertical transport channel, which can also improve the response speed. So using graphene as electrodes to improve the performances of photodetectors based on TMDCs and other 2D materials was demonstrated to be an effective way.

Fig. 11. (color online) (a) Schematic diagram of MoS2/WSe2 junction sandwiched between top and bottom graphene electrodes.[76] (b) The relationship between EQE and excitation energy.[76] (c) Possible existence of the interfaces between materials. (d) Photocurrent with respect to laser power.[115]

Except using graphene as electrodes, in order to improve the performances of TMDCs, Wasala et al. investigated the effects of BN on the optical properties of MoS2.[115] As shown in Fig. 11(d), the photocurrent increased with increasing the thickness of the underlying BN layer. Because the BN layers can increase the reflections between layers, which can result in the enhancement of photocurrent. It implies BN also can be used to enhance the photocurrent response properties of TMDCs such as photoresponsivity and detectivity. Moreover because the h-BN has atomically flat and smooth surface with lack of dangling bonds, it has been widely used as an ideal subatrate and dielectric layer for 2D materials such as graphene and TMDCs.[56,116,117] The h-BN can also be used as intercalation which could decrease the dark current and further improve detectivity.[61]

In conclusion, completely vertical structure can reduce the transport channel length, which can greatly enhance the photoresponse speed of photodetectors. Moreover, using graphene electrodes and BN reflection layer have been demonstrated to be promising methods to improve photocurrent response. Combing those methods, high-speed and highly sensitive photodetector could be realized by stacking of TMDCs and other two-dimensional materials.

Figure 12 and Table 1 summarize some experimental results from 2D material photodetectors in the review. The responsivity and response time for part of the reviewed devices are summarized in Fig. 12. In Table 1, we list the configuration, the fabrication technology, EQE, responsivity, specific detectivity and rise time. All those present the performances of the detectors.

Fig. 12. (color online) Responsivity against response time for some devices reviewed.
Table 1.

Summary of literature data for photodetectors based on two-dimensional TMDCs. The table contains the configuration, the fabrication technology, mode, and some benchmark parameters (external quantum efficiency (EQE), responsivity (R), specific detectivity (D*), rise time (τr)).

.
5. Conclusion and perspectives

This review provides an overview of the recent development in photodetectors based on junctions of TMDCs from structures, fabrications and performances. Owing to the built-in electric field in those junctions, the photogenerated electrons and holes can be more effectively separated. So compared with photodetectors based on FET type of intrinsic TMDCs, photodetectors based on junctions of TMDCs show many advantages (such as lager responsivity or faster response speed) as shown in Table 1 and Fig. 12. Moreover, selection of the materials and the device fabrication processing all have great influence on the performance of photodetectors.

Homojunctions can be created by the following three methods: surface chemical doping, elemental doping and electrostatic gating. Although surface chemical doping is suitable for thin layers of TMDCs, the homojunctions based on the surface chemical doping are easy to be affected by the external environment. Elemental doping is a fesiable way for vertical stable homojunctions with high performance. Although electrostatic gating can also be used to create homojunctions, the EQE of these junctions is still very low. Because of the limination of charge mobility, lateral homojunctions fabricated by surface chemical doping and electrostaic show slow photoresponse. However, homojunctions are based on the similiar materials, which makes them easier to realize the sensitive detection of special wavelength.

Heterojunctions can be fabricated by mechanical exfoliation and transfer, CVD and some other methods. Compared with mechanical exfoliation and transfer, CVD method has been demonstrated to be the most effective and promising method to obtain large scale heterojunctions based on TMDCs with atomically sharp interface. Moreover, CVD-grown junctions showed stronger interlayer interaction compared with transferred counterparts.

In vertical heterojunctions, using graphene as top and bottom electrodes can effectively shorten the vertical transport distance of photogenerated carriers, which reduces the response time by several orders of magnitude. The photoresponse can be stongly enhanced using BN as the reflection layer. Taking the advantages of graphene and BN, the heterojunciton photodetectors should have potential appliacations in high speed and weak signal photodetections.

Photodetectors with wide response spectrum are important in many applications. Heterojunctions can combine TMDCs and other semiconductor materials with different absorption wavelength together. MoS2/GaAs heterojunctions have been already demonstrated to have wide spectrum response from ultraviolet to visible light. Thus by selecting two appropriate materials in heterojunctions, it has great potential to realize from ultraviolet to far infrared wider response spectrum photodetectors or multicolor photodetectors.

Moreover, in TMDCs, spin and valley degrees of freedom are intimately coupled due to inversion symmetry breaking together with strong spin–orbit interaction, which has also attracted a great deal of interest.[118] By circularly polarized optical pumping, spin-valley coupled polarization can be generated,[119] which has been observed in optoeletronic devices based on MoS2[120] and WSe2.[121,122] It should be of great significance to introduce spin-polarization and valley-polarization in photodetectors based on junctions of TMDCs.

With the development of TMDCs, different types of junctions based on these materials have been demonstrated in photodetections. However, there is still a long way before they can be actually applied. In application, more controllable and effective way to fabricate large scale junctions is highly desired. Also the stability of these junctions still needs to be improved. Moreover, in theory, electronic transport in heterojunctions has been studied for many years by conventional modes, which cannot be applied well in few layer cases. A comprehensive understanding of electronic transport in heterojunctions based on 2D materials is urgently needed. Nevertheless, these junctions will greatly enrich studies in photodetectors based on the junctions of TMDCs.

Reference
[1] Novoselov K S Geim A K Morozov S Jiang D Zhang Y Dubonos S A Grigorieva I Firsov A 2004 Science 306 666
[2] Novoselov K Geim A K Morozov S Jiang D Katsnelson M Grigorieva I Dubonos S Firsov A 2005 Nature 438 197
[3] Novoselov K Jiang D Schedin F Booth T Khotkevich V Morozov S Geim A 2005 Proc. Natl. Acad. Sci. USA 102 10451
[4] Lee C Wei X Kysar J W Hone J 2008 Science 321 385
[5] Xia F Mueller T Lin Y M Valdes-Garcia A Avouris P 2009 Nat. Nanotech. 4 839
[6] Mueller T Xia F Avouris P 2010 Nat. Photon. 4 297
[7] Wang L Meric I Huang P Gao Q Gao Y Tran H Taniguchi T Watanabe K Campos L Muller D 2013 Science 342 614
[8] Butler S Z Hollen S M Cao L Cui Y Gupta J A Gutiérrez H R Heinz T F Hong S S Huang J Ismach A F 2013 ACS Nano 7 2898
[9] Xu M Liang T Shi M Chen H 2013 Chem. Rev. 113 3766
[10] Koski K J Cui Y 2013 ACS Nano 7 3739
[11] Das S Robinson J A Dubey M Terrones H Terrones M 2015 Annu. Rev. Mater. Res. 45 1
[12] Mak K F Lee C Hone J Shan J Heinz T F 2010 Phys. Rev. Lett. 105 136805
[13] Radisavljevic B Radenovic A Brivio J Giacometti V Kis A 2011 Nat. Nanotech. 6 147
[14] Wang Q H Kalantar-Zadeh K Kis A Coleman J N Strano M S 2012 Nat. Nanotech. 7 699
[15] Chhowalla M Shin H S Eda G Li L J Loh K P Zhang H 2013 Nat. Chem. 5 263
[16] Wilson J Yoffe A 1969 Adv. Phys. 18 193
[17] Frindt R F Yoffe A D 1963 Pro. R. Soc. London A: Mathematical, Physical and Engineering Sciences 273 69
[18] Cao Y F Cai K M Li L J Lu W J Sun Y P Wang K Y 2014 Chin. Phys. Lett. 31 077203
[19] Costanzo D Jo S Berger H Morpurgo A F 2016 Nat. Nanotech. 11 339
[20] Ayari A Cobas E Ogundadegbe O Fuhrer M S 2007 J. Appl. Phys. 101 14507
[21] Splendiani A Sun L Zhang Y Li T Kim J Chim C Y Galli G Wang F 2010 Nano Lett. 10 1271
[22] Eda G Yamaguchi H Voiry D Fujita T Chen M Chhowalla M 2011 Nano Lett. 11 5111
[23] Yun W S Han S Hong S C Kim I G Lee J 2012 Phys. Rev. B 85 033305
[24] Liu Z Lau S P Yan F 2015 Chem. Soc. Rev. 44 5638
[25] Geim A K Grigorieva I V 2013 Nature 499 419
[26] Yin Z Li H Li H Jiang L Shi Y Sun Y Lu G Zhang Q Chen X Zhang H 2012 ACS Nano 6 74
[27] Larentis S Fallahazad B Tutuc E 2012 Appl. Phys. Lett. 101 223104
[28] Wu C C Jariwala D Sangwan V K Marks T J Hersam M C Lauhon L J 2013 J. Phys. Chem. Lett. 4 2508
[29] Das S Appenzeller J 2013 Appl. Phys. Lett. 103 103501
[30] Huo N Yang S Wei Z Li S S Xia J B Li J 2014 Sci. Rep. 4 5209
[31] Lopez-Sanchez O Lembke D Kayci M Radenovic A Kis A 2013 Nat. Nanotech. 8 497
[32] Wang H Liu F Fu W Fang Z Zhou W Liu Z 2014 Nanoscale 6 12250
[33] Koppens F H Mueller T Avouris P Ferrari A C Vitiello M S Polini M 2014 Nat. Nanotech. 9 780
[34] Zhang W Wang Q Chen Y Wang Z Wee A T 2016 2D Mater. 3 022001
[35] Bablich A Kataria S Lemme M 2016 Electronics 5 13
[36] Pospischil A Mueller T 2016 Appl. Sci. 6 78
[37] Farmer D B Golizadeh-Mojarad R Perebeinos V Lin Y-M Tulevski G S Tsang J C Avouris P 2009 Nano Lett. 9 388
[38] Peters E C Lee E J Burghard M Kern K 2010 Appl. Phys. Lett. 97 193102
[39] Lemme M C Koppens F H Falk A L Rudner M S Park H Levitov L S Marcus C M 2011 Nano Lett. 11 4134
[40] Groenendijk D J Buscema M Steele G A Michaelis de Vasconcellos S Bratschitsch R van der Zant H S Castellanos-Gomez A 2014 Nano Lett. 14 5846
[41] Baugher B W Churchill H O Yang Y Jarillo-Herrero P 2014 Nat. Nanotech. 9 262
[42] Ross J S Klement P Jones A M Ghimire N J Yan J Mandrus D Taniguchi T Watanabe K Kitamura K Yao W 2014 Nat. Nanotech. 9 268
[43] Pospischil A Furchi M M Mueller T 2014 Nat. Nanotech. 9 257
[44] Buscema M Groenendijk D J Steele G A van der Zant H S Castellanos-Gomez A 2014 Nat. Commun. 5 4651
[45] Fang H Tosun M Seol G Chang T C Takei K Guo J Javey A 2013 Nano Lett. 13 1991
[46] Choi M S Qu D Lee D Liu X Watanabe K Taniguchi T Yoo W J 2014 ACS Nano 8 9332
[47] Yang L Majumdar K Liu H Du Y Wu H Hatzistergos M Hung P Tieckelmann R Tsai W Hobbs C 2014 Nano Lett. 14 6275
[48] Lei S Wang X Li B Kang J He Y George A Ge L Gong Y Dong P Jin Z 2016 Nat. Nanotech. 11 465
[49] Suh J Park T E Lin D Y Fu D Park J Jung H J Chen Y Ko C Jang C Sun Y 2014 Nano Lett. 14 6976
[50] Jin Y Keum D H An S J Kim J Lee H S Lee Y H 2015 Adv. Mater. 27 5534
[51] Fang H Chuang S Chang T C Takei K Takahashi T Javey A 2012 Nano Lett. 12 3788
[52] Wi S Kim H Chen M Nam H Guo L J Meyhofer E Liang X 2014 ACS Nano 8 5270
[53] Choi W Cho M Y Konar A Lee J H Cha G B Hong S C Kim S Kim J Jena D Joo J 2012 Adv. Mater. 24 5832
[54] Kim E Ko C Kim K Chen Y Suh J Ryu S G Wu K Meng X Suslu A Tongay S 2016 Adv. Mater. 28 341
[55] Huo N Yang J Huang L Wei Z Li S S Wei S H Li J 2015 Small 11 5430
[56] Flöry N Jain A Bharadwaj P Parzefall M Taniguchi T Watanabe K Novotny L 2015 Appl. Phys. Lett. 107 123106
[57] Esmaeili-Rad M R Salahuddin S 2013 Sci. Rep. 3 2345
[58] Lopez-Sanchez O Dumcenco D Charbon E Kis A 2014 arXiv:1411.3232 [cond-mat.mes-hall]
[59] Lopez-Sanchez O Alarcon Llado E Koman V Fontcuberta i Morral A Radenovic A Kis A 2014 ACS Nano 8 3042
[60] Li Y Xu C Y Wang J Y Zhen L 2014 Sci. Rep. 4 7186
[61] Wang L Jie J Shao Z Zhang Q Zhang X Wang Y Sun Z Lee S T 2015 Adv. Funct. Mater. 25 2910
[62] Hao L Liu Y Gao W Han Z Xue Q Zeng H Wu Z Zhu J Zhang W 2015 J. Appl. Phys. 117 114502
[63] Chuang S Kapadia R Fang H Chang T C Yen W C Chueh Y L Javey A 2013 Appl. Phys. Lett. 102 242101
[64] Xu Z Lin S Li X Zhang S Wu Z Xu W Lu Y Xu S 2016 Nano Energy 23 89
[65] Ruzmetov D Zhang K Stan G Kalanyan B Bhimanapati G R Eichfeld S M Burke R A Shah P B O’Regan T P Crowne F J 2016 ACS Nano 10 3580
[66] Tsai D S Liu K K Lien D H Tsai M L Kang C F Lin C A Li L J He J H 2013 Acs Nano 7 3905
[67] Jariwala D Howell S L Chen K S Kang J Sangwan V K Filippone S A Turrisi R Marks T J Lauhon L J Hersam M C 2016 Nano Lett. 16 497
[68] Kim K Santos E J Lee T H Nishi Y Bao Z 2015 Small 11 2037
[69] Wu B Zhao Y Nan H Yang Z Zhang Y Zhao H He D Jiang Z Liu X Li Y 2016 Nano Lett. 16 3754
[70] He D Pan Y Nan H Gu S Yang Z Wu B Luo X Xu B Zhang Y Li Y 2015 Appl. Phys. Lett. 107 183103
[71] He D Zhang Y Wu Q Xu R Nan H Liu J Yao J Wang Z Yuan S Li Y 2014 Nat. Commun. 5 5162
[72] Zhang L Yang Y Huang H Lyu L Zhang H Cao N Xie H Gao X Niu D Gao Y 2015 J. Phys. Chem. C 119 4217
[73] Zheng Y J Huang Y L Chen Y Zhao W Eda G Spataru C D Zhang W Chang Y H Li L J Chi D 2016 ACS Nano 10 2476
[74] Hong X Kim J Shi S F Zhang Y Jin C Sun Y Tongay S Wu J Zhang Y Wang F 2014 Nat. Nanotech. 9 682
[75] Tongay S Fan W Kang J Park J Koldemir U Suh J Narang D S Liu K Ji J Li J 2014 Nano Lett. 14 3185
[76] Lee C H Lee G H Van Der Zande A M Chen W Li Y Han M Cui X Arefe G Nuckolls C Heinz T F 2014 Nat. Nanotech. 9 676
[77] Furchi M M Pospischil A Libisch F Burgdoörfer J Mueller T 2014 Nano Lett. 14 4785
[78] Chiu M H Li M Y Zhang W Hsu W T Chang W H Terrones M Terrones H Li L J 2014 ACS Nano 8 9649
[79] Fang H Battaglia C Carraro C Nemsak S Ozdol B Kang J S Bechtel H A Desai S B Kronast F Unal A A 2014 Proc. Natl. Acad. Sci. 111 6198
[80] Chiu M H Zhang C Shiu H W Chuu C P Chen C H Chang C Y S Chen C H Chou M Y Shih C K Li L J 2015 Nat. Commun. 6 7666
[81] Nourbakhsh A Zubair A Dresselhaus M S Palacios T 2016 Nano Lett. 16 1359
[82] Peng B Yu G Liu X Liu B Liang X Bi L Deng L Sum T C Loh K P 2016 2D Mater. 3 025020
[83] Ceballos F Bellus M Z Chiu H Y Zhao H 2014 ACS Nano 8 12717
[84] Zhang K Zhang T Cheng G Li T Wang S Wei W Zhou X Yu W Sun Y Wang P 2016 ACS Nano 10 3852
[85] Rivera P Schaibley J R Jones A M Ross J S Wu S Aivazian G Klement P Seyler K Clark G Ghimire N J 2015 Nat. Commun. 6 6242
[86] Cheng R Li D Zhou H Wang C Yin A Jiang S Liu Y Chen Y Huang Y Duan X 2014 Nano Lett. 14 5590
[87] Lee Y H Zhang X Q Zhang W Chang M T Lin C T Chang K D Yu Y C Wang J T W Chang C S Li L J 2012 Adv. Mater. 24 2320
[88] Liu K K Zhang W Lee Y H Lin Y C Chang M T Su C Y Chang C S Li H Shi Y Zhang H 2012 Nano Lett. 12 1538
[89] van der Zande A M Huang P Y Chenet D A Berkelbach T C You Y Lee G H Heinz T F Reichman D R Muller D A Hone J C 2013 Nat. Mater. 12 554
[90] Gong Y Lin J Wang X Shi G Lei S Lin Z Zou X Ye G Vajtai R Yakobson B I 2014 Nat. Mater. 13 1135
[91] Xue Y Zhang Y Liu Y Liu H Song J Sophia J Liu J Xu Z Q Xu Q Wang Z 2016 ACS Nano 10 573
[92] Choudhary N Park J Hwang J Y Chung H S Dumas K H Khondaker S I Choi W Jung Y 2016 Sci. Rep. 6 25456
[93] Verble J Wietling T Reed P 1972 Solid State Commun. 11 941
[94] Li M Y Shi Y Cheng C C Lu L S Lin Y C Tang H L Tsai M L Chu C W Wei K H He J H 2015 Science 349 524
[95] Tan C Zeng Z Huang X Rui X Wu X J Li B Luo Z Chen J Chen B Yan Q 2015 Angew. Chem. 127 1861
[96] Mahjouri-Samani M Lin M W Wang K Lupini A R Lee J Basile L Boulesbaa A Rouleau C M Puretzky A A Ivanov I N 2015 Nat. Commun. 6 7749
[97] Duan X Wang C Shaw J C Cheng R Chen Y Li H Wu X Tang Y Zhang Q Pan A 2014 Nat. Nanotech. 9 1024
[98] Huang C Wu S Sanchez A M Peters J J Beanland R Ross J S Rivera P Yao W Cobden D H Xu X 2014 Nat. Mater. 13 1096
[99] Zhang C Chen Y Huang J K Wu X Li L J Yao W Tersoff J Shih C K 2016 Nat. Commun. 7 10349
[100] Howell S L Jariwala D Wu C C Chen K S Sangwan V K Kang J Marks T J Hersam M C Lauhon L J 2015 Nano Lett. 15 2278
[101] Avouris P Chen Z Perebeinos V 2007 Nat. Nanotech. 2 605
[102] Koenig S P Doganov R A Schmidt H Neto A C Oezyilmaz B 2014 Appl. Phys. Lett. 104 103106
[103] Li L Yu Y Ye G J Ge Q Ou X Wu H Feng D Chen X H Zhang Y 2014 Nat. Nanotech. 9 372
[104] Liu H Neal A T Zhu Z Luo Z Xu X Tománek D Ye P D 2014 ACS Nano 8 4033
[105] Deng Y Luo Z Conrad N J Liu H Gong Y Najmaei S Ajayan P M Lou J Xu X Ye P D 2014 ACS Nano 8 8292
[106] Chen P Zhang T Zhang J Xiang J Yu H Wu S Lu X Wang G Wen F Liu Z 2016 Nanoscale 8 3254
[107] Xia F Wang H Jia Y 2014 Nat. Commun. 5 4458
[108] Zhang W Chuu C P Huang J K Chen C H Tsai M L Chang Y H Liang C T Chen Y Z Chueh Y L He J H 2014 Sci. Rep. 4 3826
[109] Henck H Pierucci D Chaste J Naylor C H Avila J Balan A Silly M G Asensio M C Sirotti F Johnson A C 2016 Appl. Phys. Lett. 109 113103
[110] Li B Huang L Zhong M Li Y Wang Y Li J Wei Z 2016 Adv. Electron. Mater. 2 1600298
[111] Yang J Huo N Li Y Jiang X W Li T Li R Lu F Fan C Li B Nørgaard K 2015 Adv. Electron. Mater. 1 1500267
[112] Luo W Cao Y Hu P Cai K Feng Q Yan F Yan T Zhang X Wang K 2015 Adv. Opt. Mater. 3 1418
[113] Yan F Zhang D Luo W Hu P Feng Q Shen C Wei X Chang K Wang K 2016 rXiv:1606.05469
[114] Yu W J Liu Y Zhou H Yin A Li Z Huang Y Duan X 2013 Nat. Nanotech. 8 952
[115] Wasala M Zhang J Ghosh S Muchharla B Malecek R Mazumdar D Samassekou H Gaither-Ganim M Morrison A Lopez N P 2016 J. Mater. Res. 31 893
[116] Lee G H Yu Y J Cui X Petrone N Lee C H Choi M S Lee D Y Lee C Yoo W J Watanabe K 2013 ACS Nano 7 7931
[117] Cui X Lee G H Kim Y D Arefe G Huang P Y Lee C H Chenet D A Zhang X Wang L Ye F 2015 Nat. Nanotech. 10 534
[118] Xiao D Liu G B Feng W Xu X Yao W 2012 Phys. Rev. Lett. 108 196802
[119] Xu X Yao W Xiao D Heinz T F 2014 Nat. Phys. 10 343
[120] Eginligil M Cao B Wang Z Shen X Cong C Shang J Soci C Yu T 2015 Nat. Commun. 6 7636
[121] Yuan H Wang X Lian B Zhang H Fang X Shen B Xu G Xu Y Zhang S C Hwang H Y 2014 Nat. Nanotech. 9 851
[122] Zhang Y Oka T Suzuki R Ye J Iwasa Y 2014 Science 344 725
[123] Yang S Wang C Ataca C Li Y Chen H Cai H Suslu A Grossman J C Jiang C Liu Q 2016 ACS Appl. Mater. Inter. 8 2533
[124] Gao A Liu E Long M Zhou W Wang Y Xia T Hu W Wang B Miao F 2016 Appl. Phys. Lett. 108 223501
[125] Zhang W Chiu M H Chen C H Chen W Li L J Wee A T S 2014 ACS Nano 8 8653
[126] Kallatt S Umesh G Bhat N Majumdar K 2016 Nanoscale 8 15213