1.1 Introduction

Since graphene was first isolated from highly oriented pyrolytic graphite (HOPG) in 2004 [1], the excellent properties of graphene are motivating rapidly growing research enthusiasms in the layered materials, especially when they are thinned down to the atomic thickness. For layered transition metal dichalcogenides (TMDs), two-dimensional (2D, referring to few-layer and monolayer herein) ones exhibit distinct properties from the bulk counterparts owing to the quantum confinement effect and broken inversion symmetry.

Thus, increasing research efforts are devoting to this field, and the fantastic properties of 2D TMDs are continuously being revealed. Depending on the filling state for the d band of the metal elements, 2D TMDs can be semiconductors with varied band structures, semimetals, true metals, and superconductors, which determine their roles in the electronic devices. The already known layered TMDs are outlined in a periodic table with shadow background (Fig. 1.1), and a majority of their 2D counterparts have been theoretically predicted and experimentally synthesized. In this chapter, the atoms arrangement, band structure, and electronic applications of the most studied 2D TMDs are reviewed. Other aspects of the 2D TMDs such as synthesis methods, characterization technique, application in catalysts, and heterostructure assembly can be found in details in the following chapters. The readers can also refer to a comprehensive review for more information about 2D materials [2].

Fig. 1.1
figure 1

Known layered TMDs in the periodic table highlighted with shadow

Full size image

1.2 Group VIB TMDs (MoS2, WS2, MoSe2, WSe2, MoTe2, WTe2)

1.2.1 Atoms Arrangement and Band Structure

As one of the most widely studied group VIB TMDs in the twentieth century, MoS2 attracted a lot of research interest focusing on its tribological properties [3]. The renewed research interest in group VIB MX2 (M = Mo and W, X = S, Se, and Te) following the discovery of graphene concentrates upon the electronic and catalytic properties of 2D counterparts. The tunable bandgap, high carrier mobility, and large surface area of 2D MX2 render them potential application in electronics [4], optoelectronics [5], and catalysts [6]. Moreover, the reduced dimension along the vertical direction provides an ideal platform to study solid-state physics such as robust excitonic effects [7], spin-orbit coupling [8, 9], and many-body interactions [10], which are difficult to be observed in the bulk materials.

In similarity to graphite, MX2 monolayers are held together by interlayer van der Waals (vdW) force to construct the bulk materials. The vdW force is several orders of magnitude weaker than the in-plane covalent bonding, making it feasible to obtain the monolayers through top-down routes including mechanical and liquid exfoliation. As such, most of the early theory verification and demonstration of proof-of-concept electronic devices based on 2D MX2 were realized through this route [4, 5, 8, 11,12,13]. Different from the planar carbon sp2 hybridization within each graphene layer, monolayer MX2 is in fact composed of three sublayers of atoms, with one metal (M) sublayer sandwiched between two sublayers of chalcogenide (X) atoms, as shown in Fig. 1.2a [14]. Each M atom is sixfold coordinated with X atoms, forming either trigonal prismatic or octahedral geometry. In the trigonal prismatic atom arrangement, the above three X atoms are vertically above the below ones, while they stagger in the octahedral geometry.

Fig. 1.2
figure 2

a Ball-and-stick model of 1T, 2H, and 3R polymorphs of group VI TMDs [14]. Reproduced by permission of the Royal Society of Chemistry. b, c Atomic resolution STEM images of 1T (b) and 1H (c) phase of monolayer MoS2. Blue and yellow balls represent Mo and S atoms, respectively [15]. Copyright 2013. Reprinted with permission from Macmillan Publishers

Full size image

The bulk MoS2 shows polymorphs depending on the atoms arrangement in each monolayer and the relative position between monolayers. The three most commonly observed polymorphs of MoS2 are 1T, 2H, and 3R. The number (1, 2, and 3) represents the layer number in the unit cell, and the letter stands for symmetry (T-trigonal, H-hexagonal, and R-rhombohedral). Figure 1.2b and c shows the atomic resolution scanning transmission electron microscopy (STEM) images for monolayer 1T and 1H MoS2, respectively [15], consistent with the corresponding ball-and-stick model in Fig. 1.2a. For sulfide and selenide, the 2H phase is stable, and the 1T phase is metastable, while for telluride, the 1T phase is more energetically favorable. The 2H phase could be converted to 1T phase under certain conditions, such as Li-ion intercalation, strain, electron beam, laser beam irradiation, as well as plasma bombardment. Conversely, the 1T phase could also be converted to 2H phase [16,17,18,19,20]. The most prominent feature of the group VIB TMDs is the layer-dependent band structure tunability. In the bulk materials, they own an indirect bandgap, which changes to direct one when being thinned down to monolayer limit [11].

Figure 1.3a–c shows the MoS2 band structure of varied layer numbers calculated by density functional theory (DFT) [21]. In the bulk form, MoS2 owns an indirect bandgap of about 1.2 eV. The valence band maximum (VBM) locates at the Γ point, and the conduction band minimum (CBM) locates in the middle of Γ-K direction in the Brillouin zone (Fig. 1.3a). As the layer number reduces, the VBM and CBM shift due to quantum confinement effects (Fig. 1.3b). A direct bandgap of about 1.9 eV is finally achieved in monolayer MoS2 with both VBM and CBM sitting in the Brillouin zone K point (Fig. 1.3c). Other group VIB TMDs exhibit similar band structure evolution as a function of layer number [21,22,23,24]. Figure 1.3d shows the calculated band structure of monolayer WS2 [21]. A direct bandgap of ~2.1 eV is predicted, in contrast to the ~1.3 eV indirect bandgap in the bulk form [21].

Fig. 1.3
figure 3

a–d DFT calculated band structure of bulk MoS2 (a), bilayer MoS2 (b), monolayer MoS2 (c), and monolayer WS2 (d). The arrows indicate the fundamental bandgap, and the horizontal dashed lines indicate the Fermi level [21]. Copyright 2011 by the American Physical Society. Reproduced with permission (e) PL spectra for mono- and bilayer MoS2 samples. The inset shows the PL quantum yield as a function of layer number [11]. Copyright 2010. Reproduced with permission from American Physical Society

Full size image

As a result of the band structure evolution, the photoluminescence (PL) emission dramatically increases when layer reduces (Fig. 1.3e). The PL quantum yield in monolayer MoS2 is 100-fold higher than bilayer MoS2 and four orders of magnitude higher than bulk MoS2 (inset in Fig. 1.3e). Similarly, the electroluminescence (EL) efficiency in monolayers is also higher than the bulk materials [12], due to more efficient exciton generation and photon emission. The monolayer bandgap in the range of 1–2 eV, corresponding to the visible and near infrared spectrum range, and the direct bandgap nature render monolayer group VIB TMDs for constructing high-efficiency photodetectors, photovoltaics, and light-emitting diodes.

1.2.2 Electronics

2D group VIB TMDs are promising candidates for next-generation high-performance transistor channel materials, ascribing to the high conductivity, atomic thickness, and ultra-flat surface without dangling bonds. The first monolayer MoS2 field-effect transistor (FET) is reported by Kis et al. [4]. The cross-sectional view of the device structure and the electrical connection (Fig. 1.4a), and the transfer and output curves of the device are displayed in Fig. 1.4b.

Fig. 1.4
figure 4

a Cross-sectional view of the monolayer MoS2 FET and the electrical connection for measurement. b Room temperature transfer and output (inset) curves for the monolayer MoS2 FET. The transfer curve is obtained at Vds = 10 mV, and the output curves are obtained at Vbg = 0, 1 and 5 V [4]. Reprinted by permission from Macmillan Publishers Ltd., copyright (2011) c 3D schematic (left) and false-colored SEM image (right) of MoS2 FET with SWCNT gate. d ID-VGS curves of a bilayer MoS2 FET at VBS = 5 V and VDS = 50 mV and 1 V. e ID-VDS curves for the device at VBS = 5 V and varying VGS. f ID-VGS curves at VDS = 1 V and varying VBS [36]. Copyright 2010. Reproduced with permission from AAAS

Full size image

From the curves, field-effect electron mobility of ~200 cm2/Vs and current on/off ratio exceeding 1 × 108 were extracted. The mobility is significantly boosted compared with its mostly reported 1–10 cm2/Vs on SiO2, ascribing to the suppression of Coulomb scattering by the top deposited HfO2 layer. Fuhrer and Hone [25] pointed out that the capacitive coupling between the top and bottom gates could result in an overestimation of the mobility and the true mobility is likely in the range of 2–7 cm2/Vs. And thus Hall effect measurements were later performed, from which the mobility, contact resistance, and capacitive coupling can be obtained simultaneously [25]. One device exhibited mobility of 168 cm2/Vs at 4 K and 60 cm2/Vs at 250 K [25]. Other factors besides Coulomb scattering such as defects, charged impurities, and metal–semiconductor contact are also liable for the measured relatively low mobility in a FET device compared with the theoretical phonon-limited value. For example, through a thiol chemistry route to repair the sulfur vacancies and interface modification, Wang et al. achieved >80 cm2/Vs electron mobility in monolayer MoS2 bottom-gated FET [26].

To reduce the Schottky barrier height at the interface between metal electrode and MoS2 channel, Chhowalla et al. [16] converted part of the MoS2 flake underneath the metal electrodes from 2H to 1T phase, decreasing the contact resistance from 0.7–10 kΩ µm to 200–300 Ω µm.

As a consequence, excellent device performance including high mobility, high on/off current ratio, current saturation, and low subthreshold swing (SS) was realized. Similarly, graphene was tested as an electrode material for MoS2 FET to realize an ohmic contact [27]. Duan et al. devised an electrode transfer method to address the problem of Fermi-level pinning at the metal–semiconductor interface [28]. Instead of the conventionally evaporated metal electrodes, the transferred metal pads which are vdW interacted with the pre-deposition MoS2 channel exhibited prominent advantages in maintaining the integrity of the underlying channel material and eliminating metal–semiconductor chemical bonding. The Fermi-level pinning in MoS2 channel is eliminated and the metal–semiconductor Schottky barrier height is tuned and predicted by Schottky–Mott model. High electron mobility of 260 cm2/Vs and hole mobility of 175 cm2/Vs were realized by choosing metals electrodes with appropriate work functions [28]. The electrical performance of MoS2 FETs is also closely correlated with the layer number and the substrate [29,30,31]. Compared with MoS2, the electrical performances of other 2D group VIB TMDs are less investigated. However, they were proposed to exhibit superior electrical performance dictated by the theoretical calculation and existing experimental studies [32,33,34,35].

As the traditional silicon-based metal-oxide-semiconductor FET (MOSFET) scales down to the limit, the short channel effects become adversely prominent. Benefiting from the ultrathin and surface bonding-free nature, the 2D group VIB TMDs are promising candidates to construct high-performance FETs at a reduced channel length. To implement such a conception, a single-walled carbon nanotube (SWCNT) was used as the bottom gate to tune the bilayer MoS2 FET (Fig. 1.4c) [36]. The effective channel length was about 1 nm in the on state and 3.9 nm in the off state inferred from the simulation. Figure 1.4d–f shows the electrical performance of the MoS2 FETs gated by SWCNT at a short channel length. A nearly ideal SS of ~65 mV per decade at room temperature and on/off current ratio of ~106 were realized. More complicated device components in logical circuits were also devised and preliminarily tested based on group VIB TMDs FET [37,38,39,40].

1.2.3 Photo-, Chemical, and Bio- Sensors

The high performance of 2D group VIB TMDs FETs, as well as the large exposed surface, provides a good platform for their applications in sensing light, chemicals, and biomaterials, through a stimulus to electrical signal switch process. For most of the sensing devices based on 2D materials FETs, the change of the carrier concentration in the channel material upon applying a stimulus is responsible for the electrical response, similar to the function of a top gate. The first single-layer MoS2 phototransistor was reported by Zhang et al. [5] and a photoresponsivity of 7.5 mA/W and a prompt response time of ~50 ms were achieved.

Later, high photoresponsivities of 880 and 2200 A/W were reported by Kis et al. [41] and Li et al. [42], respectively. Figure 1.5a shows a schematic of the single-layer MoS2 phototransistor and Fig. 1.5b shows the source-drain current versus source-drain voltage (Ids-Vds) curves of the phototransistor under dark and different illumination intensities. The increased current with the illumination intensity was ascribed to the photoelectric and photo-thermoelectric effects [43, 44]. The monolayer MoS2 FET is also suitable for detection of chemical vapors, especially those of electron donors (Fig. 1.5c). Figure 1.5d shows the conductivity change of monolayer MoS2 channel upon exposing to a sequence of 0.002% P0 triethylamine pulses (black line). The amplitude of conductivity change was in proportion to triethylamine concentration and the control experiment of exposure to nitrogen and water vapor confirmed that the observed response was due to interaction of the MoS2 with triethylamine. The MoS2 sensor is less sensitive to electron acceptors, owing to the weak n-type nature of the MoS2 flakes. The MoS2 FET for label-free bio-detection was also demonstrated [46]. In this device architecture (Fig. 1.5e), the MoS2 channel was covered with a dielectric layer, which was functionalized with receptors for explicitly capturing the target biomolecules.

Fig. 1.5
figure 5

a Schematic of the single-layer MoS2 phototransistor. b Ids-Vds curves of the phototransistor under dark and different illumination intensities. The inset image shows the spatial photocurrent map by raster-scanning of a focused laser beam over the device. Scale bar, 5 μm [41]. Copyright (2013). Reprinted with permission from Macmillan Publishers c Schematic of the device structure of monolayer MoS2 chemical sensor. d Conductivity variation of monolayer MoS2 sensor channel upon exposing to a sequence of 0.002% P0 triethylamine pulses (black line). The dashed blue line shows the pulse timing (15 s on/30 s off) and concentration. The solid red and green lines show the response to nitrogen and water vapor (0.025% P0), respectively. The purple line shows the response of graphene sensor to water vapor pulses (0.025% P0) [45]. Copyright 2013 American Chemical Society. Published with permission e Schematic of the device structure of MoS2 biosensor. f Ids-Vgs curves of the MoS2 biosensor at three PH values of the solution [46]. Copyright 2014. Reproduced with permission from American Chemical Society

Full size image

When the target biomolecule was captured by the receptor biomolecule, it would induce a gating effect and modulate the current between the source and drain electrodes. Figure 1.5f shows the Ids-Vgs (gate voltage) curves of the MoS2 biosensor at three PH values of the solution. At low PH value (PH = 3), the surface OH groups of the dielectric layer tended to be protonated, causing a positive surface charge and reduced threshold voltage. The source-drain current (Ids) increased with the decrease in pH value. The sensitivity as high as 713 for a pH change by 1 unit in the subthreshold region and a wide operable pH range from 3 to 9 were demonstrated. This kind of MoS2 biosensor was also used for specific protein detection at an ultralow concentration [46].

1.2.4 Tunneling FET

Tunneling FETs (TFET) have been realized in graphene-based van der Waals heterostructures [47, 48]. In those work, 2D MoS2 or WS2 layers were inserted between the graphene layers, acting as atomically thin barrier. High-performance TFET characteristics were obtained in such a device structure. Banerjee et al. [49] devised a novel TFET utilizing band-to-band tunneling (BTBT). The device structure is shown in Fig. 1.6a. Under a bias, electrons flow from degenerately doped p-type Ge to the MoS2 channel in a BTBT way and diffuse to the drain electrode.

Fig. 1.6
figure 6

a Schematic of the band-to-band tunneling FET. The red arrows show the electron transport path. The electrons band-to-band tunnel (BTBT) from Ge source to MoS2 channel, and laterally diffuse to the drain electrode. b Ids-Vgs curves for three different drain voltages of 0.1, 0.5, and 1 V, from which the SS were obtained [49]. Copyright (2015). Reproduced with permission from Macmillan Publishers

Full size image

In the off state, only the electrons above the conduction band of Ge could move to MoS2 according to the band alignment between Ge and MoS2. The current was low due to the lack of available electrons in the conduction band. As the gate voltage was increased, the MoS2 conduction band was lowered below the Ge valence band. The electrons in the Ge valence band could flow to the empty states in the MoS2 channel, leading to an abrupt increase in current. The drain current as a function of gate voltage at three different drain voltages is shown in Fig. 1.6b. Ultralow SS all below 60 mV per decade over about four decades of current were deduced from the three curves, breaking the fundamental limitations on SS in MOSFETs.

1.2.5 Light Emitters and Photovoltaics

Similar to the PL, EL in 2D group VIB TMDs originates from the excitonic recombination. They are supposed to exhibit more efficient light emission being direct bandgap semiconductors in the monolayer limit. In a pioneer work by Steiner et al. [12], a pair of electrodes were deposited and contacted with the single-layer MoS2. Electrons were injected from one electrode to the conduction band of MoS2 and combined with holes, forming excitons under a high source-drain voltage. This is known as a hot carrier process, and the light emission is localized in the vicinity of the contacts. 2D p-n junction is more efficient in exciton generation and requires a reduced source-drain bias. Three independent groups reported the construction of 2D p-n junctions within monolayer WSe2 through local electrostatic doping [50,51,52]. WSe2 is selected owing to the ambipolar transport behaviors.

Figure 1.7a shows the device structure of monolayer WSe2 with two split gates [51]. The two gates could locally dope the channel. When the same polar voltage was applied to the two gates, the devices operated as a resistor (blue and green dash lines in Fig. 1.7b). Whereas when the two gates were applied voltage of opposite polarity, the current rectification was observed (blue and green solid lines in Fig. 1.7b). Thus, the device could operate as a light emitter when a forward current is driven through the device and opposite voltages are applied to the two gates.

Fig. 1.7
figure 7

a Schematic of the monolayer WSe2 device with split-gate electrodes. b I–V curves of the device in the dark under different biasing conditions: p-n (solid green line; VG1 = −40 V, VG2 = 40 V), n–p (solid blue line; VG1 = 40 V, VG2 = −40 V), n–n (dashed green line; VG1 = VG2 = 40 V), p–p (dashed blue line; VG1 = VG2 = −40 V). c Electroluminescence emission spectra recorded at constant currents of 50, 100, and 200 nA, respectively. The green curve shows that no light emission is obtained under unipolar conduction. The left inset shows the two gate voltages and the right inset shows that the amplitude of the emission intensity on a double-logarithmic scale is linearly dependent on the current. d I–V curves of the device under illumination. The biasing conditions are the same as (b). Inset: electrical power (Pel, operating as a diode) versus voltage [51]. Copyright (2014). Reproduced with permission from Macmillan Publishers

Full size image

Figure 1.7c shows the EL emission spectra at different driving currents. The EL emission is found at the same energy as the PL and the emission intensity increases linearly with the driving current (inset in Fig. 1.7c), indicating that the EL arises from excitonic transition. Vertically standing van der Waals heterostructures are also good platforms for investigating the light emission behaviors of 2D semiconductors. In this kind of structure, the carriers flow perpendicular to the 2D plane. Compared with the planar structure where the carriers flow within the 2D plane, the vertical structure offers the advantages such as reduced contact resistance, higher current density, and whole device light emission.

For example, a sandwich structure of graphene/BN/MoS2/BN/graphene was devised [53]. The Fermi level of the two graphene layers could be tuned to align with the conduction and valence band edges of MoS2, resulting in electrons and holes injection through the BN barrier. EL emission with external quantum efficiency of ~10% was realized. The 2D semiconductors could also be attempted as a light absorption layer in a photovoltaic device. For instance, asymmetric electrodes of Pd and Au were applied in a MoS2 FET [54]. When the channel was illuminated by light, the generated electron-hole pairs could be separated under the influence of the built-in potential from the space charge at the contacts, giving rise to a photovoltaic effect. The 2D p-n heterostructure could be another viable way for electron-hole pair generation and separation. When the device in Fig. 1.7 was illuminated with light and the two split gates were applied by opposite voltages, a photovoltaic effect was observed (green and blue solid lines in Fig. 1.7d). The power conversion efficiency (PCE) of the device was 0.5% with a filling factor (FF) of 50%. The comparatively low PCE and FF are ascribed to the low absorption of the atomically thin photoactive layer. The strategy of combining 2D TMDs with plasmonic materials or tandem device structure for sufficient light absorption would be used to improve photo-to-electric conversion performance. Other n-type and p-type 2D TMDs would also be combined to construct the p-n junction to fulfill the photovoltaic functionalities [55].

1.2.6 Valleytronics

2D group VIB TMDs also provide a good platform for the study of valley physics. In monolayer group VIB TMDs with broken inversion symmetry, the valence bands are split by spin-orbit coupling. The spin splitting must be opposite at the two in-equivalent valleys, labeled K and K’ in the first Brillouin zone (Fig. 1.8a). Right (left)-handed circularly polarized light only couples to the K (K′) valleys according to the optical selection rules [56, 57], as has been experimentally demonstrated by valley polarization using polarization resolved PL (Fig. 1.8b) [8, 9, 58, 59]. The circularly polarized EL emission, which is electrically switchable, was also demonstrated in a WSe2 p-i-n junction [60]. The p- and n-regions are defined by the electric double layer at the electrolyte-solid interface, which could induce a large gate field and break the inversion symmetry of few-layer WSe2. Figure 1.8c shows the circularly polarized EL emission for two opposite current flow directions. The circular polarization was reversed when the source-drain bias was exchanged. The result was explained by the electron-hole overlap controlled by in-plane electric field [60].

Fig. 1.8
figure 8

a Valley contrasting optical selection rules in a 2D hexagonal lattice with broken inversion symmetry [57]. Reprinted by permission from Macmillan Publishers Ltd., copyright (2014). b Polarization resolved PL spectra under circularly polarized excitation [8]. Reprinted by permission from Macmillan Publishers Ltd., copyright (2012). c Circularly polarized EL spectra for two opposite current directions. The contribution to EL from two valleys is shown in the bottom [60]. Copyright 2014. Reprinted with permission from Science

Full size image

1.3 Group VIIB TMDs (ReS2 and ReSe2)

1.3.1 Atoms Arrangement and Band Structure

Rhenium dichalcogenide (ReX2, X = S and Se) is a representative-layered group VIIB TMD, although the element Re is discovered late [61]. Similar to group VIB TMDs, the ReX2 layers are held together by weak vdW force and the Re and X atoms within each layer are covalently bonded.

The difference is that the extra electron in each Re atom brings in a chemical bond between Re atoms. Each four Re atoms arrange in a parallelogram shape and the parallelograms connect to form atomic chains along the principle b-axis, which is the shortest axis in the basal plane (Fig. 1.9a) [62]. The a-axis is 118.97° apart from the b-axis, corresponding to the second-shortest axis. This kind of atoms arrangement of both ReS2 and ReSe2 is experimentally verified by aberration-corrected STEM, as shown in Fig. 1.9b and c [63]. For ReS2, the distance between two neighboring parallelograms in the direction of b and a is measured as 0.34 and 0.31 nm, and that for ReSe2 is 0.39 and 0.35 nm, respectively, in agreement with the theoretical calculations [64].

Fig. 1.9
figure 9

a Side view (top panel) and top view (bottom panel) of the atoms arrangement in monolayer ReX2 (X = S and Se) [62]. Reproduced with permission from Wiley-VCH Verlag GmbH (b, c) ADF STEM images of monolayer ReS2 (b) and ReSe2 (c) [63]. Copyright 2015. Reprinted with permission from American Chemical Society

Full size image

Unlike the stable H phase observed in MX2 (M = Mo and W, X = S and Se), ReX2 shows a distorted 1T structure originating from the Peierls distortion [65], which prevents interlayer-ordered stacking and minimizes the wavefunctions overlap. As a result, the interlayer interaction in ReX2 is over an order of magnitude lower than that in MoS2 (18 vs. 460 mV per unit cell). The bulk ReX2 behaves like vibrationally and electronically decoupled while stacked monolayer counterparts [65]. Thus the bulk ReX2 can be directly used for study of 2D physics, circumventing the necessity of preparing monolayer samples. The bandgap of layered TMDs is strongly influenced by the layer number. For group VIB TMDs, an indirect to direct bandgap transition occurs when the bulk materials are thinned down to monolayer limit, caused by the exciton confinement effects [11]. As a consequence, both the PL quantum yield and PL emission position vary. However, for ReX2, the electronic band structure is barely influenced by the layer number, due to the weak interlayer interaction.

Figure 1.10a shows the band structure for monolayer, trilayer, and five-layer ReS2 by ab initio calculation [66]. The CBM and VBM share the identical position in the Brillouin zone, respectively, giving rise to direct bandgap for monolayer, trilayer, and five-layer ReS2. The bandgap only varies slightly, from 1.44 eV for monolayer ReS2, 1.40 eV for trilayer ReS2, to 1.35 eV for five-layer ReS2. In fact, the bulk ReS2 is also a direct bandgap semiconductor with a bandgap of 1.35 eV, predicted by generalized gradient approximation (GGA) [67]. As such, the PL emission for ReS2 with different layer numbers is a superposition of the emission from individual monolayers. The PL intensity increases as the layer number increases and no significant emission position change is observed due to the similar bandgap (Fig. 1.10b) [65].

Fig. 1.10
figure 10

a Ab initio calculated band structures of monolayer, trilayer, and five-layer ReS2 [66]. Reprinted by permission from Macmillan Publishers Ltd., copyright (2015). b PL signal of ReS2 with different layer numbers [65]. Copyright 2014. Reprinted with permission from Macmillan Publishers

Full size image

1.3.2 Anisotropic Optical and Electrical Properties

Deriving from the atoms arrangement, ReX2 crystals are supposed to exhibit anisotropic optical and electrical response behaviors. The anisotropic properties of the bulk materials have been documented [68, 69] and that of the few-layer and monolayer counterparts are the current research focus. The unique anisotropy is not only important for the understanding of physical properties but also fundamental for designing novel kinds of electronic devices.

The anisotropic Raman response of ReX2 is revealed in several studies [70,71,72,73]. Compared with the group VIB TMDs, the Raman spectra of ReX2 exhibit more bands due to its low symmetry. Figure 1.11a shows an optical image of ReS2 samples with varied layer numbers obtained by mechanical exfoliation [70]. The Raman spectra for the 1L, 2L, 3L, and 4L ReS2 are shown in Fig. 1.11b. Note that the peak position of Raman mode I continuously changes with the layer number, from 133.1 ± 0.1 cm−1 in the 1L to 136.2 ± 0.2 cm−1 in the 4L, while the position of modes III, IV, and V barely varies. Thus the peak position difference between Raman mode I and other modes, such as mode III, can be used to identify ReS2 layer number in few-layered samples, similar to the work done on MoS2 [74]. Figure 1.11c exhibits the orientation-dependent polarization response of mode V in the 4L region with unpolarized (black), cross-polarized (blue), and parallel-polarized (red) collection. The cross- and parallel-polarized spectra yield 4-lobed and 2-lobed shapes, respectively, indicating the anisotropic Raman response behavior of thin ReS2 by angle-resolved polarized Raman spectroscopy technique.

Fig. 1.11
figure 11

a Optical microscopy image of a mechanical-exfoliated ReS2 sample with 1-4 layers identified. b Raman spectra taken in the 1L, 2L, 3L, and 4L ReS2 regions with an orientation of θ = 120°. c Raman intensity of mode V in the 4L region with unpolarized (black), cross-polarized (blue), and parallel-polarized (red) collection [70]. Copyright 2015. Reproduced with permission from American Chemical Society

Full size image

1.3.3 Anisotropic Electrical Properties

The unique atomic structure also renders an anisotropic electron transport behavior for 2D ReX2. The theoretical calculations have predicted that the conductivity along b-axis is higher than other directions [68, 69]. The FET devices were fabricated with the electrodes patterning perpendicular to the ReS2 a- and b-axes, respectively, and the transfer curves are shown in Fig. 1.12a [66].

Fig. 1.12
figure 12

a Transfer curves of a quadrilateral-shaped five-layer ReS2 flake in the A and B directions. The left inset shows the optical microscopy image of the FET device, scale bar, 10 mm, and the right inset shows the four-probe resistance of the same devices as a function of Vbg. b Experimentally measured field-effect mobilities of a six-layer ReS2 device along 12 evenly spaced directions in polar coordinate. The blue dots show the calculated mobilities in the same polar coordinate. The optical image of the device is shown inset [66]. Copyright 2015. Reprinted with permission from Macmillan Publishers

Full size image

Obviously, the current along the b-axis is higher than that along the a-axis, and the current ratio was also gate dependent. To eliminate the contact resistance, the four-probe resistance was measured (inset in Fig. 1.12a) and the result was consistent with the current trend.

The anisotropic transport behaviors were further evaluated by the device shown in the inset of Fig. 1.12b. The 12 electrodes were separated by 30° apart. The transfer curves of each pair of diagonally positioned electrodes were measured, and the normalized mobility was plotted in a polar coordinate. The field-effect mobility is highly angle dependent and the largest value is 60° from the direction with the lowest value. The anisotropic ratio of mobility was ~3.1, which is larger than 1.8 for anisotropic black phosphorus [75].

1.3.4 Electronics and Optoelectronic Devices

Since ReX2 sheets have exhibited a high room temperature mobility above 30 cm2/Vs and a high on/off current ratio of over 106 [76, 77], they are promising for application in electronics. As the synthesis technology for ReX2 is rapidly developing, especially the controlled sample preparation by CVD with predefined layer number and size [78,79,80,81], increasingly more electronic devices based on these materials are emerging. Figure 1.13 shows some elementary logical gates constructed with few-layer ReS2 channel [82]. Monolayer graphene was applied as the source, drain, and gate electrode, and a kind of ion gel with a high capacitance was used as the dielectric layer.

Fig. 1.13
figure 13

a Optical image and schematic band diagram of the NOT gate. b Voltage transfer characteristics and signal gain of the NOT gate. (c, d) Optical images and schematic band diagrams of the c NAND and d NOR gates. e The input and output voltages of the NAND and NOR gates as a function of time [82]. Copyright 2017. Reproduced with permission from American Chemical Society

Full size image

The graphene electrodes avoid the trap states and Fermi-level pinning in the ReS2 channel, and the ion gel can tune the electron concentration of the channel within a large range. Figure 1.13a shows an optical image of the logical NOT device and Fig. 1.13b exhibits the reasonable voltage transfer characteristics and the voltage gain exceeding 3.5. For the NAND gate, three transistors are connected in series (Fig. 1.13c). When either or both of the inputs are in the logic states “0,” the Vout is in the logic state “1,” and the Vout logic state “0” is only achieved when both the inputs are held at “1” (Fig. 1.13e). For the NOR gate, one transistor is connected in series with two transistors connecting in parallel (Fig. 1.13d). When either or both of the inputs are in the logic states “1,” the Vout is in the logic state “0,” and the Vout logic state “1” is only achieved when both the inputs are held at “0” (Fig. 1.13e).

Since the direct bandgap nature of ReX2 is independent of the layer number, highly performance photodetectors are expected in the few-layer samples owing to improved light absorption compared with the monolayers. A similar device structure designed for group VIB TMDs is also utilized for ReX2 photodetectors, and high device performance was reported [77, 81, 83,84,85,86,87]. The polarization-sensitive photoresponsivity originating from the high in-plane optical anisotropy was also observed in a ReSe2 photodetector [84]. The EL was also preliminarily proposed and reported in ReS2 [88] and ReS2 p-n junction [89], and more efforts are necessary in this area to realize devices with higher performance.

1.4 Group VIII TMDs (PtS2 and PtSe2)

1.4.1 Atoms Arrangement and Band Structure

PtX2 (X = S and Se) crystal possesses a typical 1T-type hexagonal crystal structure, similar to the 1T phase of group VIB TMDs. The Pt atom is in a tilted octahedral site and coordinated by six X atoms. The lattice constants of the hexagonal structure are defined in Fig. 1.14a [90]. Figure 1.14b shows an atom-resolved STEM image of few-layer PtSe2 [91]. The atoms can be clearly identified from the image contrast and the intensity profile along the red line. The fast Fourier transform image is also in agreement with the hexagonal structure of PtSe2. Different from the weak interlayer interaction in group VIIB TMDs and the moderate interlayer interaction in group VIB TMDs as discussed above, the interaction between PtX2 layers is strong, resulting from the overlapping of the pz orbital of interlayer X atoms [90,91,92,93,94,95].

Fig. 1.14
figure 14

a 3D, top, and side view of the 1T PtX2 (X = S and Se). The blue and yellow balls represent Pt and X atoms, respectively [90]. Reproduced with permission from Wiley-VCH Verlag GmbH b Atoms resolved STEM image of few-layer PtSe2. The up inset is the fast Fourier-transformed image showing hexagonal structure and the bottom inset shows the intensity line proline along the red line [91]. Copyright 2017. Reproduced with permission from Wiley-VCH Verlag GmbH

Full size image

As a result, the band structure of PtX2 changes dramatically with the layer number. Monolayer PtSe2 is a semiconductor with indirect bandgap. The bandgap narrows in the bilayer and becomes zero in the trilayer, as predicted by the first-principle calculation (Fig. 1.15a–c) [96].

Fig. 1.15
figure 15

a–c First-principle calculated band structure of defect-free a monolayer, b bilayer, and c trilayer PtSe2 [96]. Reprinted by permission from Macmillan Publishers Ltd., copyright (2018), d Experimentally measured and DFT calculated bandgap as a function of PtS2 layer number [90]. Copyright 2016. Reproduced with permission from Wiley-VCH Verlag GmbH

Full size image

Also, the indirect bandgap could be transformed to direct bandgap under strain [93]. PtS2 shares a similar band structure evolution with PtSe2 as a function of layer number. The experimentally verified and DFT calculated bandgaps versus number of layers are plotted in Fig. 1.15d [90]. The layer-dependent semiconductor to metal transition is unique in the group VIII TMDs. Note that the value of the bandgap varies depending on the calculation method and the level of theory, as well as the defect concentration [90,91,92,93,94,95,96].

1.4.2 Electronic Devices

PtX2 is predicted to exhibit phonon-limited carrier mobility exceeding 1000 cm2/Vs at room temperature [97]. Thus they are suitable for high-speed electronic devices. The FETs based on different layer numbers of PtS2 and PtSe2 were constructed and tested [90, 91]. The output and transfer curves for an 11-nm-thick PtSe2 device are shown in Fig. 1.16a and b. The Ids increased with the positive back gate voltage, indicating the n-type carrier transport. The room temperature electron mobility derived from the linear region of the transfer curve was 210 cm2/Vs. This mobility is much higher than that of 2D group VIB and VIIB TMDs. Moreover, the device is rather stable, which exhibits negligible degradation after storing in air for one year [91]. A FET fully composed of PtSe2 was also reported [98]. A thinner PtSe2 sheet is used as a semiconductor channel material, and the thicker PtSe2 sheet is used as an electrode. The same kind of material within one device yields an Ohmic contact between electrode and channel, which is beneficial for carrier injection. P-type carrier transport was also reported in PtSe2 FET [99]. The different 2H phase resulted from the different synthesis route may be accounted for the observed p-type carrier transport. Photodetectors using the planar FET device configuration based on PtX2 channel materials have also been realized [91, 96, 98], in analogy to the photodetectors based on group VIB and VIIB TMDs.

Fig. 1.16
figure 16

a, b Output and transfer curves of a 11 nm PtSe2 transistor. The optical image of the device is shown in the inset in (a) [91]. Reproduced with permission from Wiley-VCH Verlag GmbH (c, d) Time-resolved photocurrent of c monolayer and d bilayer PtSe2 FET at a bias voltage of 0.1 V and zero gate voltage under laser illumination with a wavelength of 632 nm, 1.47 µm, and 10 µm, respectively [96]. Copyright (2018) Reproduced with permission from Macmillan Publishers

Full size image

The unique advantages for group VIII TMDs photodetectors are the tunable narrow bandgaps, which is an excellent choice for mid-infrared light detection. Figure 1.16c and d shows the time-resolved photocurrent when the device was illuminated with lasers of different wavelengths. For the monolayer PtSe2 FET, a photoresponsivity of 0.9 and 0.15 A W−1 was obtained for 632 nm and 1.47 µm illumination, respectively. However, the device showed negligible photoresponse to a mid-infrared laser illumination (λ = 10 µm) since the photon energy of the incident laser is far below the bandgap of monolayer PtSe2 (Eg ~ 1.2 eV).

For the bilayer PtSe2 FET, a much higher photoresponsivity of 6.25, 5.5, and 4.5 A W−1 was observed for 632 nm, 1.47 µm, and 10 µm illumination, respectively. The broadband mid-infrared photoresponse of bilayer PtSe2 FET was proposed to be originated from the defect-induced bandgap reduction [96]. The PtSe2 FET could also be used as a NO2 gas sensor [100], in which ultrahigh sensitivity and extremely fast response time are obtained.

A photovoltaic device was proposed and fabricated by transferring PtSe2 thin films onto pre-patterned Si substrates [100], forming PtSe2/n-Si Schottky barrier diodes (SBDs). The schematic of the device structure is shown in Fig. 1.17a and the rectifying characteristics are shown in Fig. 1.17b. A clear photoresponse was observed in the reverse bias region under illumination, as shown in Fig. 1.17c. The energy band diagram at the junction of the SBD under illumination is presented in Fig. 1.17d. J-V curves of the devices with different PtSe2 thicknesses under an incident light intensity (Pin) of 4.7 mW/cm2 are plotted in Fig. 1.17e. As the thickness of the PtSe2 layer increases, light absorption enhances and thus the power conversion efficiency increases.

Fig. 1.17
figure 17

a Schematic of the PtSe2/n-Si Schottky barrier diode. b J-V curve of the device with a PtSe2 film synthesized from 4-nm-thick Pt on a linear scale. c J-V curves of the device with a PtSe2 film synthesized from 4-nm-thick Pt on a semilogarithmic scale under dark and illumination. d Energy band diagram of the Schottky junction between PtSe2 and n-Si under reverse bias. e J–V characteristics of the devices with different PtSe2 thicknesses under an incident light intensity (Pin) of 4.7 mW/cm2 [100]. Copyright 2016. Reprinted with permission from American Chemical Society

Full size image

1.5 Other 2D TMDs

In addition to the TMDs discussed above, more 2D TMDs are emerging and attracting increasing attentions. These materials exhibit dramatically different properties mainly due to the varied d-orbit electron states of the transition metals and provide versatile platforms for studying fundamental physical conceptions such as charge density wave (CDW), phase transition, and superconductivity.

In the group VB TMDs, VX2 (X = S and Se) exhibits intrinsic ferromagnetic behaviors and the magnetic moments and strength of magnetic coupling could be tuned by isotropic strain, as predicted by the first-principles calculations [101]. The ferromagnetism has been experimentally verified in both VS2 and VSe2 [102, 103]. The CDW behavior was also discovered in VSe2 [103] and the CDW transition temperature is dependent on the thickness [104, 105]. For VS2, the stable phase (H or T) relies on the temperature and the thickness [106]. Few-layer VS2 is metallic [107, 108] and the metal-insulator transition is also observed and influenced by the vdW interaction [108].

The DFT calculation reveals that bulk and few-layers VSe2 in T and H phase, and monolayer VSe2 in T phase are metallic, while the H monolayer is a semiconductor [109]. 2D NbSe2 is also investigated and the coexistence of CDW and superconducting phase has been observed in monolayer limit [110]. The electronic phase transitions can be achieved by electrostatic gating [111]. A plenty of low-temperature electronic states also appear in 1T-TaS2, especially the first reported superconductivity in a pristine 1T-TaS2 [112]. A series of phase transitions in 1T-TaS2 thin flakes are reported by gate electric field induced Li-ion intercalation [113]. For 2H TaS2, the superconductivity persists down to 3.5 nm, and the critical temperature increases from 0.5 to 2.2 K as the layers are thinned down [114]. Similarly, the CDW and superconductivity phases were also reported in 2D group IVB TMDs, like TiSe2 [115,116,117]. The transition temperature could be tuned by the electric field [116]. 2D HfSe2 in group IVB TMDs is a kind of n-type semiconductor, with a relatively small bandgap of ~1.1 eV revealed by scanning tunneling spectroscopy (STS) on a molecular beam epitaxy grown sample [118] and high phonon-limited mobility over 3500 cm2/Vs predicted by calculation [97].

1.6 Conclusions

In this chapter, the atomic arrangement, band structure, and electronic applications of most of the widely studied 2D TMDs are summarized. Differentiating from graphene and bulk TMDs, 2D TMDs exhibit unique properties and application potential. As the synthesis technique is continuously developing, more 2D TMDs are obtained. For example, 47 kinds of binary, alloy, and heterostructured 2D TMDs have been synthesized by a salt-assisted CVD method in a recent report [119]. The availability of these materials paves the way for a deeper understanding of the physical origins and device applications. At the same time, apart from the optical and electrical applications, 2D magnetism is attracting increasing attentions, exhilarated by the particular magnetic properties discovered in transition metal Cr based 2D materials [120, 121]. As such, low-power and ultra-compact spintronic applications could be expected based on 2D TMDs.