The Recent Progress of Two-Dimensional Transition Metal Dichalcogenides and Their Phase Transition

: Graphene is attracting much attention in condensed matter physics and material science in the two-dimensional(2D) system due to its special structure, and mechanical and electronic properties. However, the lack of electronic bandgap and uncontrollable phase structure greatly limit its application in semiconductors, such as power conversion devices, optoelectronic devices, transistors, etc. During the past few decades, 2D transition metal dichalcogenides (TMDs) with much more phase structures have attracted intensive research interest in fundamental studies and practical applications for energy storage, as catalysts, and in piezoelectricity, energy harvesting, electronics, optoelectronic, and spintronics. The controllable phase transition also provides another degree of freedom to pave the way for more novel devices. In this review, we introduce the abundant phase structures of 2D-TMDs, including 2H, 1T, 1T’ and charge density waves, and highlight the corresponding attractive properties and applications of each phase. In addition, all the possible methods to trigger the phase transition in TMDs are systematically introduced, including strain engineering, electron doping, alloying, thermal, electric field, and chemical absorption. Finally, the outlook of future opportunities in TMD phase transitions and the corresponding challenges, including both the synthesis and applications, are also addressed.


Introduction
Graphene, as a typical 2D material, did not widely concern researchers because many theoretical calculations suggested the instability of the monolayer structure. In 2004, physicists from the University of Manchester, A. K. Geim and K. S. Novoselov, successfully separated the stable monolayer graphene through micromechanical exfoliation for the first time, leading to rising in graphene [1]. The special bonding mechanism between carbon atoms also provides graphene with toughness [2,3]. In terms of electronic structure, graphene is semi-metal and has ultra-high conductivity, which is much higher than traditional precious metal conductive materials, such as gold, silver, and copper. As a 2D material, graphene has a 2D honeycomb-like structure formed by sp 2 -hybridization between six carbon atoms in a plane and has a large specific surface area, and, therefore, it is widely used as electrode material for lithium-ion batteries and supercapacitors [4].
In addition, graphene also shows excellent performance in electronic, optical, and energy storage devices, FET, sensors, nanogenerators, biomedicine, and many other fields [5][6][7]. However, graphene's zero band gap greatly limits its further applications [8]. Therefore, the search for a 2D material similar to graphene and able to avoid the shortcomings of graphene has become an urgent matter in physical, chemical, material, and device research.
In recent years transition metal dichalcogenide (TMD) has become a much more attractive novel 2D material due to its unique physical and chemical properties, leading to Mo, Ta, and W. Thermodynamic stable phase structure is always associated with the 2H or 1T phases, while other structures are always metastable phases. For example, among the TMDs of group VI B, there are six combinations of TMDs, of which five (except WTe 2 ) are the 2H phase as the most stable phase and the 1T phase as the metastable phase. common structure is the 1T' phase. This structure can be regarded as the periodic distorted 1T phase with two neighboring metals close to each other and forming a metal zig-zag chain. The metal-metal chemical bond is a "dimer bond" [41]. The 1T' is often observed in group V B TMDs. The stabilities of the 2H, 1T, and 1T' phases are very different depending on the combinations of transition metal elements (IV B to X B) and chalcogens (S, Se, and Te). At present, the known 2H and 1T phase TMD materials are mainly the dihalides of Nb, Mo, Ta, and W. Thermodynamic stable phase structure is always associated with the 2H or 1T phases, while other structures are always metastable phases. For example, among the TMDs of group VI B, there are six combinations of TMDs, of which five (except WTe2) are the 2H phase as the most stable phase and the 1T phase as the metastable phase.

Charge Density Waves
In addition to the above three structures, a series of more complex structures have also been observed in TMDs. The characteristics of these structures are that the crystal structure is distorted, based on 2H and 1T, which leads to a certain degree of modulation distribution of the charge density in TMDs, showing the characteristics of the periodic change of the charge density with the lattice [42][43][44]. Figure 2 shows the phase transition in multiple physical domains [45]. It can be seen that the CDW of the material varies under different temperature and pressure conditions. CDW was first discovered in the periodic distortion of one-dimensional metal chains at low temperatures. The main physical mechanism for its generation is coupling between electrons and phonons. When electrons in the same quantum state repel each other and open a certain band gap at the Fermi surface (metal-insulation transition), the crystal structure changes. Periodic distortion occurs, which leads to modulation of the spatial distribution of charge density within the material system [46]. Chan and Heines proved that CDWs do not appear in systems where electrons and phonons are not coupled [47].

Charge Density Waves
In addition to the above three structures, a series of more complex structures have also been observed in TMDs. The characteristics of these structures are that the crystal structure is distorted, based on 2H and 1T, which leads to a certain degree of modulation distribution of the charge density in TMDs, showing the characteristics of the periodic change of the charge density with the lattice [42][43][44]. Figure 2 shows the phase transition in multiple physical domains [45]. It can be seen that the CDW of the material varies under different temperature and pressure conditions. CDW was first discovered in the periodic distortion of one-dimensional metal chains at low temperatures. The main physical mechanism for its generation is coupling between electrons and phonons. When electrons in the same quantum state repel each other and open a certain band gap at the Fermi surface (metal-insulation transition), the crystal structure changes. Periodic distortion occurs, which leads to modulation of the spatial distribution of charge density within the material system [46]. Chan and Heines proved that CDWs do not appear in systems where electrons and phonons are not coupled [47].
The CDWs of TMDs have received extensive attention, mainly because of their unconventional superconducting phenomena and strongly correlated electronic systems [48]. In addition, due to the diversity of CDWs, some new phenomena and peculiar properties can be achieved by modulating CDWs, such as the implementation of superconductivity under chemical doping, and the catalytic performance of surfaces controlled by the absorptivity of reactants and active sites in CDWs [49][50][51][52]. Quasi-one-dimensional Sn metal atomic chains can be grown on the surface of superlattice NbTe 2 modulated by CDWs, and the transition between metalloid and metallicity induces the electronic properties of the surface [53]. This provides a great possibility for using CDWs to control the surface properties of materials for special hetero catalysis, epitaxial growth of low-dimensional materials, and other nano-electronic devices. controlled by the absorptivity of reactants and active sites in CDWs [49][50][51][52]. Quasi-onedimensional Sn metal atomic chains can be grown on the surface of superlattice NbTe2 modulated by CDWs, and the transition between metalloid and metallicity induces the electronic properties of the surface [53]. This provides a great possibility for using CDWs to control the surface properties of materials for special hetero catalysis, epitaxial growth of low-dimensional materials, and other nano-electronic devices. Phase and electronic structures of TMDs. General phase diagram of 1T-TaS2 in different physical domains (temperature, doping x(Fe), or pressure P). Where C refers to commensurate CDW, T refers to triclinic incommensurate CDW, NC refers to near commensurate and IC refers to incommensurate CDW. The phase evolution of a CDW can be characterized by the variation of the hexagonal CDW diffraction peak at the inverse vector Q (upper right corner). The upper right corner shows the SD 13-cluster, representing the C-CDW cells in real space. Lattice distortion within each star is coupled with strong charge density redistribution. Reproduced from [45], with the permission of the AAAS.

Performance and Application of 2D TMDs
Some bulk TMDs are similar to graphene and layered materials with van der Waals interaction between two neighboring layers. Due to the weak van der Waals interaction, 2D TMDs can be obtained using a stripping method similar to graphene. In addition, chemical vapor deposition (CVD) can also be used to prepare 2D TMDs [54]. The emergence of 2D TMDs has greatly enriched catalysis applications, energy storage, ferrous materials, electronics, and spintronics.

Catalyst of Hydrogen Evolution Reaction
As bulk TMD is reduced to a monolayer, van der Waals interactions are no longer present at the surface, while chalcogen atoms and transition metal atoms are exposed at the surface. Therefore, the surface of 2D TMDs often show high chemical activity serving as a catalyst in HER. Experimental studies reported that monolayer MoS2 has excellent catalysis [55,56]. However, the excellent catalytic performance of 2H-MoS2 should mainly be attributed to the high chemical activity from the metallic edge [57]. The activity of semiconducting 2H-MoS2 phase nanosheets is mainly limited by their high resistance, which hinders charge transfer. The charge transfer kinetics is a key parameter to further improve its performance as a HER catalyst [58]. Based on these reasons, the metallic 1T phase should present excellent catalytic performance, compared to the 2H phase. In 2013, Voirv et al. directly compared the catalytic performance of MoS2 in the 2H phase and 1T phase ( Figure 3) [59]. In their experiment, the edge of both 1T and 2H were gradually Phase and electronic structures of TMDs. General phase diagram of 1T-TaS 2 in different physical domains (temperature, doping x(Fe), or pressure P). Where C refers to commensurate CDW, T refers to triclinic incommensurate CDW, NC refers to near commensurate and IC refers to incommensurate CDW. The phase evolution of a CDW can be characterized by the variation of the hexagonal CDW diffraction peak at the inverse vector Q (upper right corner). The upper right corner shows the SD 13-cluster, representing the C-CDW cells in real space. Lattice distortion within each star is coupled with strong charge density redistribution. Reproduced from [45], with the permission of the AAAS.

Performance and Application of 2D TMDs
Some bulk TMDs are similar to graphene and layered materials with van der Waals interaction between two neighboring layers. Due to the weak van der Waals interaction, 2D TMDs can be obtained using a stripping method similar to graphene. In addition, chemical vapor deposition (CVD) can also be used to prepare 2D TMDs [54]. The emergence of 2D TMDs has greatly enriched catalysis applications, energy storage, ferrous materials, electronics, and spintronics.

Catalyst of Hydrogen Evolution Reaction
As bulk TMD is reduced to a monolayer, van der Waals interactions are no longer present at the surface, while chalcogen atoms and transition metal atoms are exposed at the surface. Therefore, the surface of 2D TMDs often show high chemical activity serving as a catalyst in HER. Experimental studies reported that monolayer MoS 2 has excellent catalysis [55,56]. However, the excellent catalytic performance of 2H-MoS 2 should mainly be attributed to the high chemical activity from the metallic edge [57]. The activity of semiconducting 2H-MoS 2 phase nanosheets is mainly limited by their high resistance, which hinders charge transfer. The charge transfer kinetics is a key parameter to further improve its performance as a HER catalyst [58]. Based on these reasons, the metallic 1T phase should present excellent catalytic performance, compared to the 2H phase. In 2013, Voirv et al. directly compared the catalytic performance of MoS 2 in the 2H phase and 1T phase ( Figure 3) [59]. In their experiment, the edge of both 1T and 2H were gradually passivated through oxygen. The HER performance decreased with the edge being oxidized in 2H, while it was not degraded during the oxidization in 1T, suggesting that the catalytic effect of 1T mainly derives from the inside of the nanosheet.
The above discoveries promoted research on the catalysis of the 1T phase, and researchers tried many methods to introduce the metallic 1T phase in many other materials. Voirv et al. synthesized MoS 2 nanosheets with 2H and 1T coexisting on the SiO 2 substrate. Compared to the pure 2H nanosheet, the catalytic efficiency of the 1T and 2H mixed nanosheet was greatly improved [57]. The in-depth analysis suggested that the low contact barrier between the 2H phase and 1T phase led to high electron transfer efficiency in the material in addition to the high activity of the 1T phase. Both factors increased the HER performance. Besides this, the catalytic performance of the 1T phase could be further improved by electron doping. For example, electrons can be injected into MoS 2 to weaken the H atom's adsorption, when single-walled carbon nanotubes (SWCNTs) were deposited on the 1T-MoS 2 Crystals 2022, 12, 1381 5 of 20 monolayer, leading to the acceleration of the HER process [60]. In addition to MoS 2 , Chen and Tsai et al. also found excellent HER performance in other TMDs, such as VS 2 , NbS 2 , and so on [61,62].
substrate. Compared to the pure 2H nanosheet, the catalytic efficiency of the 1T and 2H mixed nanosheet was greatly improved [57]. The in-depth analysis suggested that the low contact barrier between the 2H phase and 1T phase led to high electron transfer efficiency in the material in addition to the high activity of the 1T phase. Both factors increased the HER performance. Besides this, the catalytic performance of the 1T phase could be further improved by electron doping. For example, electrons can be injected into MoS2 to weaken the H atom's adsorption, when single-walled carbon nanotubes (SWCNTs) were deposited on the 1T-MoS2 monolayer, leading to the acceleration of the HER process [60]. In addition to MoS2, Chen and Tsai et al. also found excellent HER performance in other TMDs, such as VS2, NbS2, and so on [61,62].

Energy Storage
The lithium-ion battery (LIB) is a typical secondary battery with high energy density, long life, large charge and discharge capacity, and low self-discharge, and has, thus, attracted extensive attention. To meet the growing need for applications, such as portable electronic products, electric vehicles, aerospace and other fields, the development of highperformance battery materials is an important subject of current research. Anode materials are very important for the performance of LIBs. Ideal anode materials should possess large capacity, high conductivity, high ion mobility, low cost, chemical stability, and environmental friendliness [63]. TMDs are a kind of potential anode material due to their rapid lithium-ion diffusion interface and more inert surface than those of other materials. A recent study revealed that the typical TMD VS2 shows excellent cycling performance and capacity when used as anode material of LIBs [64]. The capacity of the VS2 nanosheet could reach 532.2 mA g −1 after 30 cycles at 50 mA g −1 charge-discharge current. After 100 cycles, the stable discharge capacity was 436.8 and 270.4 mAh g −1 at 100 and 500 mAg −1 , respectively. The excellent rate performance and cyclic stability of the material are attributed to higher electronic conductivity and its layered structure. Based on TMDs, there also exist other methods to further improve the performance of LIBs. Zhou et al. composited MoS2 and graphene to increase the specific surface area, and the stability and capacity of the MoS2/graphene were greatly increased (1060 mAh g −1 ) as the anode of LIBs [65]. In addition, the conductivity of 2H-MoS2 as an anode in LIBs could be improved by introducing the 1T' phase [66]. The resulting capacity was nearly twice that of the pure

Energy Storage
The lithium-ion battery (LIB) is a typical secondary battery with high energy density, long life, large charge and discharge capacity, and low self-discharge, and has, thus, attracted extensive attention. To meet the growing need for applications, such as portable electronic products, electric vehicles, aerospace and other fields, the development of highperformance battery materials is an important subject of current research. Anode materials are very important for the performance of LIBs. Ideal anode materials should possess large capacity, high conductivity, high ion mobility, low cost, chemical stability, and environmental friendliness [63]. TMDs are a kind of potential anode material due to their rapid lithium-ion diffusion interface and more inert surface than those of other materials. A recent study revealed that the typical TMD VS 2 shows excellent cycling performance and capacity when used as anode material of LIBs [64]. The capacity of the VS 2 nanosheet could reach 532.2 mA g −1 after 30 cycles at 50 mA g −1 charge-discharge current. After 100 cycles, the stable discharge capacity was 436.8 and 270.4 mAh g −1 at 100 and 500 mAg −1 , respectively. The excellent rate performance and cyclic stability of the material are attributed to higher electronic conductivity and its layered structure. Based on TMDs, there also exist other methods to further improve the performance of LIBs. Zhou et al. composited MoS 2 and graphene to increase the specific surface area, and the stability and capacity of the MoS 2 /graphene were greatly increased (1060 mAh g −1 ) as the anode of LIBs [65]. In addition, the conductivity of 2H-MoS 2 as an anode in LIBs could be improved by introducing the 1T' phase [66]. The resulting capacity was nearly twice that of the pure 2H phase. The capacity could still maintain 636 mAh g −1 at charge and discharge current of 1A g −1 even after 2000 cycles.
Instead of LIBs, TMDs could also serve as the electrodes of supercapacitors due to their high electrical conductivity, small size, and large specific surface area. The supercapacitor shows a much higher capacity than the normal capacitor. More importantly, the lower voltage limitation fills the gap between electrolytic capacitor and rechargeable battery. Supercapacitors are applied in many electronics devices, including regenerative braking, short-term energy storage or burst mode power transfer, and as power backups for static random-access memory (SRAM), which requires fast charging-discharging cycling [67]. In 2007, Soon et al. applied MoS 2 film grown by CVD to the electrode of a supercapacitor. The experimental results showed that the specific surface area of the electrodes increased as the number of layers decreased, resulting in a substantially increased AC frequency and charge accumulation [68]. With the discovery of the 1T phase, Acerce et al. successfully used 1T-MoS 2 as the electrode material for supercapacitors [69]. Compared with previous studies, the volumetric capacitance of 1T-MoS 2 ranged from 400 to 700 Fcm −3 . Its chem-ical stability was also improved, and the Coulomb Efficiency remained above 95% after 5000 cycles. VS 2 also presented excellent performance when it was used as the electrode of supercapacitors. Its specific capacitance was up to 4760 µF/cm 2 ( Figure 4) [70].
power backups for static random-access memory (SRAM), which requires fast chargingdischarging cycling [67]. In 2007, Soon et al. applied MoS2 film grown by CVD to the electrode of a supercapacitor. The experimental results showed that the specific surface area of the electrodes increased as the number of layers decreased, resulting in a substantially increased AC frequency and charge accumulation [68]. With the discovery of the 1T phase, Acerce et al. successfully used 1T-MoS2 as the electrode material for supercapacitors [69]. Compared with previous studies, the volumetric capacitance of 1T-MoS2 ranged from 400 to 700 Fcm −3 . Its chemical stability was also improved, and the Coulomb Efficiency remained above 95% after 5000 cycles. VS2 also presented excellent performance when it was used as the electrode of supercapacitors. Its specific capacitance was up to 4760 μF/cm 2 ( Figure 4) [70].

Ferroelasticity, Ferroelectricity, and Ferromagnetism
Ferroelasticity, ferroelectricity, and ferromagnetism all describe a hysteresis phenomenon in material under external action induced by a state of low symmetry. The difference is that ferroelasticity is the hysteresis of stress under strain, ferroelectricity is the hysteresis of polarization under an external electric field, and ferromagnetism is the hysteresis of magnetization under an external magnetic field. Multiferroic material is the coupling among the above three ferroics, and, thus, it shows great potential in sensors, transducers, and other devices. The discovery of the above properties in 2D materials has promoted the miniaturization of many devices.

Ferroelasticity
Ferroelasticity is a phenomenon in which a material exhibits spontaneous strain. When stress is applied to a ferroelastic material, it transits among its different stable variants. TMDs show a variety of phase structures, including 2H, 1T, and distorted 1T' phases. Both the 2H phase and the 1T phase are high symmetry structures. However, due to the existence of lattice distortion in the 1T' phase, the symmetry of the crystal structure is reduced leading to ferroelasticity. In 2016, Li et al. reported the mechanical straininduced orientation transition of the 1T' phase and the thermodynamic stability through first-principles calculation [71]. Figure 5c depicts the three different variants of the 1T' phase in TMDs, which can be regarded as different ferroelastic domains. The strain can be used to regulate variants, and the transition barrier among them could also provide the Reproduced from [70], with the permission of the American Chemical Society.

Ferroelasticity, Ferroelectricity, and Ferromagnetism
Ferroelasticity, ferroelectricity, and ferromagnetism all describe a hysteresis phenomenon in material under external action induced by a state of low symmetry. The difference is that ferroelasticity is the hysteresis of stress under strain, ferroelectricity is the hysteresis of polarization under an external electric field, and ferromagnetism is the hysteresis of magnetization under an external magnetic field. Multiferroic material is the coupling among the above three ferroics, and, thus, it shows great potential in sensors, transducers, and other devices. The discovery of the above properties in 2D materials has promoted the miniaturization of many devices.

Ferroelasticity
Ferroelasticity is a phenomenon in which a material exhibits spontaneous strain. When stress is applied to a ferroelastic material, it transits among its different stable variants. TMDs show a variety of phase structures, including 2H, 1T, and distorted 1T' phases. Both the 2H phase and the 1T phase are high symmetry structures. However, due to the existence of lattice distortion in the 1T' phase, the symmetry of the crystal structure is reduced leading to ferroelasticity. In 2016, Li et al. reported the mechanical straininduced orientation transition of the 1T' phase and the thermodynamic stability through first-principles calculation [71]. Figure 5c depicts the three different variants of the 1T' phase in TMDs, which can be regarded as different ferroelastic domains. The strain can be used to regulate variants, and the transition barrier among them could also provide the possibility to trigger phase or variant transition in experiments. For example, the domain transition barrier was about 0.23 eV/f.u. in WTe 2 , while it was 0.21 eV/f.u. in MoS 2 . These results indicated that the ferroelastic domain transition could be achieved under conventional experimental conditions. Regrettably, there is no experimental verification yet, because the calculation results are obtained at 0 K. In addition, the shape memory effect and superelasticity in ferroelasticity also need to be further studied.

Ferroelectricity
Ferroelectricity refers to spontaneous polarization and polarization switching under an external electric field. Ferroelectric polarization switching is often accompanied by the lattice constant changing. Therefore, ferroelectric materials can be used to fabricate electric field-controlled smart driving materials. On the contrary, the electrical polarization could also be changed through strain, and this property is the so-called piezoelectricity. Piezoelectric materials, such as MoS 2 , WS 2 , and WTe 2 show potential in energy conversion devices and sensors.
Duerloo et al. studied ferroelectricity in a series of TMDs through first-principal calculations, in which semiconductive 2H MoS 2 , MoSe 2 , MoT 2 , WS 2 , WSe 2, and WTe 2 were proved to be ferroelectric, while metallic 2H NbSe 2 and TaSe 2 were not ferroelectric [72]. Uniaxial strain was also applied on MoS 2 and WS 2 to investigate piezoelectricity, as shown in Figure 5b. The polarization of MoS 2 and WS 2 along the x-direction varied linearly with strain and was stronger compared with other 2D materials, such as BN. In addition to the natural ferroelectricity of 2H TMDs, ferroelectricity was also observed in some organic functional group absorbed TMDs, such as -CH 3 , CH 2 OCH 3 , CH 2 F, and so on. The electrical polar orientation of these functional groups could be changed under an external electric field. The above ferroelectric materials present great potential for multifunctional electronic devices, such as ferroelectric field-effect transistors with ultra-high on-off ratios, Dirac fermion topology transistors for electron/hole conversion, ultra-high resistance ferroelectric heterojunctions, and multiferroic spin valves controlled by electric fields [73].
In addition to the 2H phase, the 1T' phase is also a potential ferroelectric material due to its low symmetry crystal structure. In 2019, Pankaj et al. discovered ferroelectricity in 1T' phase WTe 2 , and the piezoelectric response of this material could reach the nanoscale [74]. As shown in Figure 5a, the ferroelectric domain structure and domain wall of WTe 2 can be clearly observed. Meanwhile, the hysteresis loop and the transition process of the ferroelectric domain could be observed simultaneously by the applied electric field. The saturated iron electrodes could be up to~0.19 µC/cm 2 . Since the above tests were all tested at room temperature, the results proved that WTe 2 had ferroelectricity at room temperature, which was helpful for its practical application. In addition, the theoretical calculations indicated that the potential barrier was about 0.29 eV/f.u. during the transition of the ferroelectric domain. In the typical ferroelectric material BaTiO 3 , the ferroelectric domain transition barrier was about 0.43 eV/f.u. Obviously, ferroelectric domain transition in 1T'-WTe 2 was much easier under the external electric field, suggesting a faster external field response.
The ferroelectricity of all the above-mentioned TMDs originates from the in-plane spontaneous polarization. Besides in-plane polarization, Qing Yang et al. proposed the out-of-plane spontaneous polarization in double-layer 1T-WTe 2 through calculations [75]. Meanwhile, the reversal of the ferroelectric polarization direction can be realized by interlayer sliding. Since the metal ion W is positively charged and the non-metal ion Te is negatively charged, adjusting the stacking of the bilayer WTe 2 by interlayer slip could generate vertical upward or downward electrical polarization. Calculations showed that the sliding barrier of the bilayer structure in the mutually perpendicular electric polarization states (State I and II) was 0.6 MeV.

Ferromagnetism
With the rise of spintronics, spin, another intrinsic property, is treated as the carrier, instead of the electron, expanding the degree of freedom and performance of electronic devices. The unique properties of TMDs meet the relevant performance requirements and have become a novel candidate for spintronic devices. The spin injection is very important for spintronics. The direct method of injection spin to nonmagnetic TMDs is doping magnetic atoms leading to the so-called dilute semiconductor. Ashwin et al. studied the magnetic and electronic structure of Mn-doped 2H-MoS 2 by first-principles and Monte Carlo calculations [76]. When the content of Mn reached~12.5%, strong ferromagnetic coupling among Mn atoms could be observed, which derived from the double exchange interaction between Mn and the neighboring S. The MoS 2 was still semiconducting after Mn doping. In addition to Mn, it was found that Ti, V, Cr, Fe, Co, Ni, Cu, and Zn doping could also introduce magnetism into MoS 2 [77].
Magnetism can also be induced from the adsorbing atoms on the surface of TMDs such as H, B, C, N, O, and F. At the same time, theoretical calculations show that the adsorption of O and F on MoS 2 has a large spatial spin density and a long-range antiferromagnetic coupling. For H and N atom adsorption, n-type and p-type doped magnetic semiconductors can be achieved [78,79], respectively. However, the Curie temperature of magnetic semiconductors (diluted magnetic semiconductors) obtained by this kind of micro-doping is too low (only~5 K Curie temperature was observed in Mn-doped MoS 2 ), which greatly limits the application of this material in spintronics devices [76].
To overcome the disadvantages of dilute magnetic semiconductors, Liang Cai et al. proposed inducing the second phase in non-magnetic semiconductors (Figure 5d) [80].
In their experiments, MoS 2 contained both 2H and 1T phases and a certain amount of S vacancies. The exchange between Mo and S was strengthened and then led to 395 K Curie temperature. In addition, recent studies on natural magnetic transition metals V, Cr, and Mn TMDs also showed a high Curie temperature. The Curie temperature could even reach~1000 K in strained 1T CrSe 2 and 1T CrTe 2 [81]. During the straining process, the transition from antiferromagnetic to ferromagnetic could also be realized.
Magnetism can also be induced from the adsorbing atoms on the surface of TMDs such as H, B, C, N, O, and F. At the same time, theoretical calculations show that the adsorption of O and F on MoS2 has a large spatial spin density and a long-range antiferromagnetic coupling. For H and N atom adsorption, n-type and p-type doped magnetic semiconductors can be achieved [78,79], respectively. However, the Curie temperature of magnetic semiconductors (diluted magnetic semiconductors) obtained by this kind of micro-doping is too low (only ~5 K Curie temperature was observed in Mndoped MoS2.), which greatly limits the application of this material in spintronics devices [76].
To overcome the disadvantages of dilute magnetic semiconductors, Liang Cai et al. proposed inducing the second phase in non-magnetic semiconductors (Figure 5d) [80]. In their experiments, MoS2 contained both 2H and 1T phases and a certain amount of S vacancies. The exchange between Mo and S was strengthened and then led to 395 K Curie temperature. In addition, recent studies on natural magnetic transition metals V, Cr, and Mn TMDs also showed a high Curie temperature. The Curie temperature could even reach ~1000 K in strained 1T CrSe2 and 1T CrTe2 [81]. During the straining process, the transition from antiferromagnetic to ferromagnetic could also be realized.

Field Effect Transistor
The field effect transistor (FET) is an electronic device that controls the current through an electric field. The carrier conductivity can be adjusted through the shape of the conduction band under an electric field. FET is widely used in digital integrated circuits, information storage, and other logical operation devices. During the past decades, graphene has attracted considerable attention as the main material of FETs due to its abundant physical properties and high mobility [8]. The special electronic properties and moderate mobility of TMDs provide the possibility to construct excellent FETs. In theoretical calculation, the mobility of single-layer MoS 2 reached 200 cm 2 V −1 s −1 , the mobility of ZrS2 reached 1237 cm 2 V −1 s −1 , and the mobility of ZrSe 2 reached 2316 cm 2 V −1 s −1 [82,83]. Yonatan Vaknin fabricated FET based on MoS 2 as shown in Figure 6a [84]. It was also found that the barrier height at the gold/molybdenum disulfide (MoS 2 ) interface decreased with the increase of drain voltage, reaching a reduction of 0.5-1 V. An increase in the grid voltage resulted in a decrease in the additional barrier. Radisavljevic fabricated a MoS 2 FET on HfO 2 [83]. The mobility could reach 200 cm 2 V −1 s −1 at room temperature. In addition, the current on/off ratio could reach 108 at room temperature, with ultra-low standby power consumption. In subsequent further studies, the team revealed a temperature-dependent dependence of electron mobility, suggesting that the top dielectric had an inhibitory effect on charge impurity scattering. At the same time, the gate voltage applied to MoS 2 induced metal-insulator transition, which enhanced the interaction between electrons [85]. In a FET, the materials used for the gate, drain, and source are semi-conductive and metallic, respectively, and, therefore, the metal-semiconductor contact interface of different materials leads to a higher Schottky barrier [85], and higher contact resistance (0.7 kΩ µm-10 kΩ µm).
To solve this problem, Zhang and his teammates grew 2H/1T'-MoTe 2 lateral heterojunctions through CVD and constructed them as FETs, in which 1T' and 2H phases were used as source and drain, respectively. The Schottky barrier between 1T' and 2H was onlỹ 0.25 eV [86]. This method was then applied to construct a 2H/1T' heterojunction FET in MoS 2 . The contact resistance of the device was only 200-300 Ω µm. In addition, IV B TMDs also showed great potential in FET applications. For example, the on/off ratio of FET composed of ZrS 2 -/hBN could reach 10 6 . The theoretical mobility of HfSe 2 was up to 3500cm 2 V −1 s −1 , while its on/off ratio was 7.5 × 10 6 [38].

Photodetector
A photodetector is used to detect light or other electromagnetic radiation. The core component of the photodetector is the p-n junction, which can convert photons into current or electron-hole pairs in the depletion zone. The photodetector has a wide range of applications, including light-emitting diode, ion detector, photomultiplier tube, photovoltaic power generation unit, displays, etc. In the application of this device, the essential mechanism is the transition of electrons excited by light radiation. Therefore, the electronic structure is very important to the performance of this device, which mainly

Photodetector
A photodetector is used to detect light or other electromagnetic radiation. The core component of the photodetector is the p-n junction, which can convert photons into current or electron-hole pairs in the depletion zone. The photodetector has a wide range of appli-cations, including light-emitting diode, ion detector, photomultiplier tube, photovoltaic power generation unit, displays, etc. In the application of this device, the essential mechanism is the transition of electrons excited by light radiation. Therefore, the electronic structure is very important to the performance of this device, which mainly affects the absorption and radiation of light. In semiconductors with a direct band gap, the bottom of the conduction band and the top of the valence band are in the same position in the reciprocal space. The electron transition is only related to the band gap and the light frequency, thus leading to very high photoelectric efficiency. In the indirect band-gap semiconductor, the bottom and top of the conduction band are located at different positions in the reciprocal space. In addition to the value of band gap and the frequency of light, the electrons need to shift spatially during the electron transition process, which results in lower absorption excitation efficiency in the semiconductors. As the number of layers decreases, many TMDs transit from indirect band gap semiconductors to direct band-gap semiconductors, such as MoS 2 , making TMDs one of the potential candidates for use in photodetectors [89]. In addition, TMDs have a very high response rate, quantum efficiency, and rapid light response as photodetector devices.
At present, many achievements have been made in the study of photoelectric detectors for TMDs. Monolayer and multilayer 2H-MoS 2 have semiconductor properties, and their band gaps are 1.8 eV and 1.6 eV, respectively. They can be effectively detected at exactly the frequency range of green light. When 2H-MoS 2 grows to three layers, the 1.3 eV band gap is suitable for red light detection. In addition to common TMDs, such as MoS 2 and WS 2 , other TMDs belonging to the IV B family also show excellent performance in optical and electronic applications. In 2015, He et al. fabricated ultra-sensitive phototransistors using HfS 2 as the main material, with photoelectric response efficiency exceeding 890 AW −1 at a gate voltage of 80 V [90]. In addition to their excellent photoelectric properties, TMDs can be further controlled by changing external conditions, such as stress and electric field, due to the obvious change of band structure under external conditions. For example, in a multilayer WS 2 /graphene van der Waals heterostructure, Francesco regulated the photoluminescence in WS 2 through a polymer micro actuator [91]. Then, Fatemeh prepared a MoSe 2 /WSe 2 heterojunction used as a solar cell [92]. The electron-hole interaction in this device was very strong. The performance of the device could be improved to 350% in the near-infrared frequency range. Luo prepared a high-response photovoltaic photodetector based on the MoTe 2 /MoSe 2 II heterojunction (Figure 6b) [87]. This self-powered photoelectric detector showed a response of 1.5 AW -1 . Due to the efficient separation of the electron-hole pair and the ultra-fast charge transfer, the light-induced on-off ratio of the current switch was greater than 10 4 at zero bias. In addition to the above photoelectric devices, another important application of photodetectors is solar cells. Researchers found that the light conversion efficiency of the solar cell composed of three layers of graphene/MoS 2 /n-Si could reach 11.1% [93].

Giant Magnetoresistance and Spin Valve
Spin, another intrinsic property of electrons, was ignored in electronic devices. After the giant magnetoresistance (GMR) effect was first discovered in 1988, utilizing spin began to attract much more attention in electronic devices, which was named as Spintronics. Spintronics extends the degree of freedom of carriers to the intrinsic property of spin. Electronic devices with both electron spin and charge as carriers have shown great potential in modern information technology applications. With the development of 2D materials, breaking the degeneration of electron spin orbits, electrons exhibit new degrees of freedom (valley degrees of freedom) in different spins, providing new possibilities for the development of smaller electronic devices [94]. The simplest spin valve device based on the GMR effect has a three-layer structure, in which two ferromagnetic layers (common materials are magnetic metals, such as Fe, Co, and Ni) are arranged on both sides separated by a non-magnetic layer. One of the two ferromagnetic layers is fixed through the pinning effect from the antiferromagnetic layer, so that the magnetization orientation cannot be changed by the external magnetic field. The other layer is the free layer, the magnetization orientation of which can be changed by the external magnetic field. When the magnetization orientations of the two ferromagnetic layers are parallel to each other, the spin-up electrons have the lowest resistance, while the spin-polarized electrons, with different magnetization orientations, show higher resistance. When the magnetized orientation of the ferromagnet is antiparallel, the electrons, regardless of spin polarization, are in a high resistance state and are not conductive.
The tunneling magnetoresistance (TMR) effect is similar to GMR, except that the intermediate layer material is made of a non-magnetic insulative or semiconductive material [95]. In TMR, electrons are tunneled from one ferromagnetic layer to another through the tunneling effect. When the magnetization orientations of the two ferromagnets are parallel, the resistances of the polarized electrons are low and the chances of tunneling are high, which characterizes the conduction state. When the magnetization orientation of the two ferromagnets is inversely parallel, the polarized electrons have higher resistance and lower chances of tunneling, characterizing the non-conductive state. The spin valve structure of the TMR is also known as the magnetic tunnel junction (MTJ). Since the tunneling current density is generally small, the magnetoresistance of MTJ devices is usually higher than that of GMR spin valves.
It was observed that a higher magnetoresistance could be obtained by composing TMDs with other materials to form a multi-layer spin valve structure. For example, NiFe/MoS 2 /NiFe/Co spin valves have a magnetoresistance of 0.4% at room temperature (Figure 6c) [88]. On this basis, researchers further increased their magnetoresistance tõ 0.73% by adding an extra Au layer between the MoS 2 and NiFe layers to avoid oxidation of the NiFe layer. However, the high Schottky limits the spin polarization current and the effective injection of spin. Jen-Ru Chen et al. effectively reduced the Schottky barrier and increased the spin injection efficiency by adding an extra MgO layer between the ferromagnetic electrodes as a transition layer in MoS 2 [96]. In addition, researchers are attempting to reduce Schottky barriers using heterostructures composed of different phases of TMDs as the means of chemical synthesis are improved.

Thermal Conductivity
Like graphene, TMD material has high mobility of charge carriers at room temperature. However, the thermal conductivity of TMDs is much lower than that of graphene. The thermal conductivity of monolayer TMD material graphene is 2-3 orders of magnitude lower. Cai et al. calculated the room temperature thermal conductivity of MoS 2 as 23.2 MW·m −2 k −1 bythe nonequilibrium Green's function calculation [97]. Monolayer TMD materials are limited by their ultrathin thickness and highly localized Joule heating, resulting in hot spots [98]. This is extremely beneficial for the use of TMDs as thermoelectric materials. Meantime, this greatly affects the performance and reliability of electronic devices based on TMDs. In addition, phonon scattering at the interface plays a key role in heat transfer at the interface [99]. TMD material sheets and other material surface contacts can be divided into two types: side contact and edge contact [100]. In side contact, the metal surface interacts with the basal surface of MoS 2 through van der Waals. At the same time, MoS 2 and graphene have different phonon dispersions (Figure 7) [101], which results in additional electron scattering at the interface, so the interface thermal conductivity is usually lower at the side contact interface. The interfacial thermal conductivity of MoS 2 / graphene is 13.8 MW·m −2 k −1 [102]. In the edge contact, the interface between monolayer MoS 2 and other materials forms chemical bonds, resulting in high heat transport. The interfacial thermal conductivity of MoS 2 / graphene is 6420 MW·m −2 ·k −1 [103]. When TMDs are used as optoelectronic devices, cooling strategies are needed, which can effectively conduct heat through edge contact with other materials.
The thermal conductivity of TMDs is not only affected by material intrinsic properties and interfaces. It is also related to size, temperature and mechanical strain. Zhang et al. calculated the 1H and 1T'-MoS 2 two-phase thermal conductivities of monolayers by first principles [104]. The thermal conductivity of 1H and 1T'-MoS 2 monolayer increased with the length of the material and decreased with increase in temperature. Under the influence of mechanical strain, 1H-MoS 2 was almost unaffected, while 1T'-MoS 2 decreased with increase in tensile strain.
between monolayer MoS2 and other materials forms chemical bonds, resulting in high heat transport. The interfacial thermal conductivity of MoS2/ graphene is 6420 MW·m −2 ·k −1 [103]. When TMDs are used as optoelectronic devices, cooling strategies are needed, which can effectively conduct heat through edge contact with other materials.
The thermal conductivity of TMDs is not only affected by material intrinsic properties and interfaces. It is also related to size, temperature and mechanical strain. Zhang et al. calculated the 1H and 1T'-MoS2 two-phase thermal conductivities of monolayers by first principles [104]. The thermal conductivity of 1H and 1T'-MoS2 monolayer increased with the length of the material and decreased with increase in temperature. Under the influence of mechanical strain, 1H-MoS2 was almost unaffected, while 1T'-MoS2 decreased with increase in tensile strain.

TMDs' Phase Transition
The abundant phase structures in TMDs, and their corresponding different physical and chemical properties, suggest that phase transitions in 2D TMDs should be more attractive because of the possibility for flexible, low-power, and transfer-efficient electronics. In addition, efficient bulk regulation of phase transitions can also be applied in various fields, such as hydrogen evolution reaction and catalysis. At present, there are two main methods to realize the phase transition in TMDs: strain and electron modulation.

Strain-Induced Phase Transition
Strain-induced phase transition is one of the most common control methods. The Gibbs free energy (G) is adopted to determine the lowest energy path of the strain-induced phase transition process. According to basic phase transition theory, the Gibbs free energy G of a material can be expressed as: where E coh is the cohesive energy of the material (the energy released by the combination of isolated atoms into the material); V is the volume; S is the entropy.
To compare the stability of two different structures, it is necessary to compare the Gibbs free energy changes of the two structures: The value of the first two items of Equation (2) is much larger than the entropy change, so the Gibbs free energy can be simplified as: All terms in Equation (3) can be easily obtained. The change of the cohesive energy is the difference in the total energy of the two different structures, and the pressure P is the first derivative of the energy to the volume change: It is worth noting that the pressure P in the MoS 2 to be calculated is the applied stress value. The two structures are considered to undergo phase transition when the Gibbs free energies are equal. Therefore, the lowest energy path for the phase transition is the common tangent of the Gibbs free energies of the two structures. As shown in Figure 8, V 1 /V o represents the change of the current volume relative to the original volume.
Since the lattice constants of the 2H phase and the 1T or 1T' phase are different, the phase transition of the 2H phase to the 1T or 1T' phase can be achieved by applying a certain degree of strain to the 2H phase in the ground state. Chen et al. provide a feasible solution for low energy consumption small spin electronic devices by adjusting strain to convert 1T'-CrS 2 into spin-up or spin-down semimetals [105]. Reed et al. systematically studied the phase transition process between 2H-1T' of Mo and W chalcogenides under uniaxial and biaxial strains through theoretical calculation [106]. It can be proved that strain is an effective method to adjust the relative energy of two phases. In addition to WTe 2 , the 2H phase can be transited into 1T' through uniaxial or biaxial stretching. When free energy is close to zero, its phase transition range is ε y = 0.5% to 2.4% (this range is the coexistence of 2H and 1T'). When the strain is greater than 2.4%, the 2H phase completely transits to the 1T' phase. At room temperature, the phase transition range is ε y = 0.3% to 3%. Subsequently, Song et al. experimentally fabricated monolayer MoTe 2 and applied tensile strain to it. Their experimental results demonstrated that the 2H phase gradually undergoes a phase transition from 2H to 1T' when the strain reaches 0.2% at room temperature, which is accompanied by a transition from semiconducting to metallic (Figure 9a) [107].

Electron-Controlled Phase Transition
In addition to directly changing the lattice constants of 2H phase and 1T/1T' phase to induce phase transition, controlling the number of electrons in the 2D TMD system can also achieve phase transition due to the different electronic structures and atomic bonds of 2H and 1T/1T' [108,109]. Depending on the specific electronic means of control, phase transition can be achieved by plasma thermoelectrons, beam radiation, electrostatic electron injection, alloying doping, and alkali metal atom adsorption.

Plasma Thermoelectron Induced 2H-1T Phase Transition
In 2014, Yimin Kang proposed a method to induce 2H-1T in MoS2 by plasma thermoelectrons [96]. Au nanoparticles were deposited on a single 2H-MoS2 nanosheet. Since the work function of Au atoms is about 5.1 eV, the electrons on the nanoparticles escape and transfer to the MoS2 nanosheet when an incident laser with a wavelength of 488 nm is applied to irradiate Au nanoparticles. The Schottky barrier between Au nanoparticles and 2H-MoS2 is less than 0.8 eV in theoretical calculation. Therefore, it provides the basis for electronic injection. In addition to electrons escaping from Au nanoparticles, Au nanoparticles are located above the top S atom. The interaction between Au and MoS2 weakens the top Mo-S bond, making the top S atom more slippery and the 2H phase unstable [110]. The 2H-MoS2 is formed because the 4d orbital of the Mo atom splits into three groups through the crystal field of the six nearest S atoms, and its lowest energy level is dz 2 , which results in the 2H phase being the most stable at room temperature. At this time, the symmetry centered on the Mo atom is D3h. The 4d orbits of the Mo atom in the 1T phase are split into two groups, one of which is the triple degenerate t2g orbits occupied by two electrons at the lowest energy level. The second is the double degenerate eg orbits. Considering that the dz 2 orbits are already fully occupied, the

Electron-Controlled Phase Transition
In addition to directly changing the lattice constants of 2H phase and 1T/1T' phase to induce phase transition, controlling the number of electrons in the 2D TMD system can also achieve phase transition due to the different electronic structures and atomic bonds of 2H and 1T/1T' [108,109]. Depending on the specific electronic means of control, phase transition can be achieved by plasma thermoelectrons, beam radiation, electrostatic electron injection, alloying doping, and alkali metal atom adsorption.

Plasma Thermoelectron Induced 2H-1T Phase Transition
In 2014, Yimin Kang proposed a method to induce 2H-1T in MoS 2 by plasma thermoelectrons [96]. Au nanoparticles were deposited on a single 2H-MoS 2 nanosheet. Since the work function of Au atoms is about 5.1 eV, the electrons on the nanoparticles escape and transfer to the MoS 2 nanosheet when an incident laser with a wavelength of 488 nm is applied to irradiate Au nanoparticles. The Schottky barrier between Au nanoparticles and 2H-MoS 2 is less than 0.8 eV in theoretical calculation. Therefore, it provides the basis for electronic injection. In addition to electrons escaping from Au nanoparticles, Au nanoparticles are located above the top S atom. The interaction between Au and MoS 2 weakens the top Mo-S bond, making the top S atom more slippery and the 2H phase unstable [110]. The 2H-MoS 2 is formed because the 4d orbital of the Mo atom splits into three groups through the crystal field of the six nearest S atoms, and its lowest energy level is dz 2 , which results in the 2H phase being the most stable at room temperature. At this time, the symmetry centered on the Mo atom is D 3h . The 4d orbits of the Mo atom in the 1T phase are split into two groups, one of which is the triple degenerate t 2g orbits occupied by two electrons at the lowest energy level. The second is the double degenerate e g orbits. Considering that the dz 2 orbits are already fully occupied, the electrons can only be distributed above higher energy levels when the electrons are injected into the 2H phase, while the extra electrons in the 1T phase can occupy the unoccupied t 2g orbits, making the 1T more stable.

Electron Beam Radiation
Although these methods can effectively control the 2H-1T phase transition in MoS 2 , they are not suitable for large-area phase transition or fine control due to the low efficiency of Au nanoparticle deposition and the uncontrollable direction of electron escape. Subsequently, by using in-situ scanning transmission electron microscopy (ISTEM), Yung-Chang Lin found that 2H-MoS 2 can gradually transit to 1T after a certain amount of time of beam radiation [111]. After the irradiation of 2H phase, the S atom in the top layer slides to form a sTable 1T phase. In addition to phase transition, the advantage of this method is that the phase transition region can be arbitrarily controlled to form coexisting heterojunctions of 2H and 1T phases. Metal-semiconductor heterojunctions constructed from different materials are usually difficult to achieve, so it is very important for applications in nanoelectronic devices. The patterning of phase structure in a single MoS 2 layer can be easily achieved through beam radiation, so electronic devices can be constructed according to specific requirements.

Electrostatic Doping
Phase transition in TMDs can also be achieved by applying a gate voltage (electrostatic doping). In 2016, Reed et al. proposed a method to realize phase transition in TMDs [112]. Their theoretical calculation showed that a certain voltage applied to MoTe 2 (grown on HfO 2 substrate) would promote electrons or holes being injected into MoTe 2 to achieve a phase change between 2H-1T'. The gate voltage in this process was about 2-4 V. The electron concentration change during phase transition in MoTe2 was −0.29 e/f.u. (cavity injection) or 0.35e/f.u. (electron injection). At the same time, the gate voltage induced phase transition could be effectively reduced in the W-doped MoTe 2 (Mo x W 1-x Te 2 ) (Figure 9c). Subsequently, Wang et al. experimentally demonstrated that electrostatic doping is an effective method to control phase transition in TMDs. To achieve electrostatic doping to regulate phase transition, they constructed a FET with a single layer of MoTe 2 as the main component, which was covered with an ionic liquid drop. As the external voltage increased, the electrons in the ionic droplets tended to enter MoTe 2 under the action of an electric field, thus enabling electron injection and phase transition [113]. Reversible phase transition could be achieved by adjusting the gate voltage. This greatly enriches its application prospects in many new applications, such as 2D memory. However, ionic liquid easily reacts with MoTe 2 at room temperature, which greatly limits the phase transition environment of MoTe 2 .

Alloying Doping and Alkali Metal Adsorption
Another way to adjust the number of electrons in TMDs is alloying or adsorbing alkali metal atoms on the surface of TMDs. For example, 2H and 1T coexistence could be observed in Mo 1−x W x S 2 [114]. Moreover, both theoretical and experimental studies show that the energies of the 2H and 1T phases gradually approach each other as the W content in MoS 2 or MoTe2 increases (no more than 0.5). That is, the transition barrier between the two phases gradually decreases. This indicates that the 1T and 2H phase structures easily undergo phase transition after alloying (Figure 9b) [115].
In theoretical calculation, alkali metal elements, such as Li and Na, easily transfer electrons to TMDs when they are adsorbed on the surface of TMDs, thus changing the number of electrons in TMDs. Theoretical calculations show that the Li absorption increases the stability of the 1T'phase in MoS 2 [116]. When the concentration of Li is beyond 20%, 1T' is more stable than 2H, providing the thermal dynamics conditions for 2H to 1T' transition. Reproduced from [112], with the permission of the Springer Nature.

Summary and Outlook
Since the emergence of TMD, due to its unique physical and chemical properties, it has received extensive attention. This paper reviewed the phase structure of 2D TMD materials, including 2H, 1T, 1T' and CDW. The singular physical and chemical properties of 2D TMDs were also introduced, such as catalytic properties, ferroelasticity, ferroelectricity, ferromagnetism and thermal conductivity properties. Meanwhile, the application of TMDs in electrocatalysis, lithium-ion batteries, supercapacitors, FETs, photodetectors, giant magnetoresistance and spin value are introduced in detail combined with TMDs phase and its characteristics. In addition, we reviewed current methods for modulating TMD phase transitions, including strain-induced and electron-regulated transitions. In particular, we analyzed the advantages and disadvantages of various methods of electron regulated phase transition, including plasma hot electrons, beam radiation, electrostatic electron injection, alloy doping and alkali metal atom adsorption. Further, we looked into the prospect of low symmetry TMDs phase structures for miniaturization applications and edge characteristics of electromechanical devices. Zhu et al. and Chen et al. proposed ultrathin transistors and single-atom layer-thick FinFET devices, respectively [117,118]. The size of electronic components can already reach the thickness of a single layer, and the miniaturized electromechanical devices of this single layer of atoms are also expected to be realized in the future and will have applications in many fields, from bionic robotics to the aerospace industry. However, conventional smart actuation materials exhibit poor performance at the nanoscale, due to detrimental forcelength ratio relationships, or performance degradation, due to defective surfaces. Recent studies have shown that 2D smart driver materials are promising candidates for drivers in such devices. These include graphene, graphene oxide, lithium-doped phosphene, and graphite diaynes. However, the limited number of high-performance, 2D smart driving materials severely limits the on-demand design of various applications. The 1T' phase in Figure 9. Electron controlled phase transition structure. (a) Temperature-force phase diagram for semiconducting 2H and metallic 1T' MoTe 2 . Reproduced from [107], with the permission of the American Chemical Society. (b) Phase structure of WTe 2 . Under the doping of Mo atom, WTe 2 changed from the H phase to T'phase. Reproduced from [115], with the permission of the American Chemical Society. (c) Phase boundary at a constant charge in monolayer MoTe 2 and MoS 2 . Reproduced from [112], with the permission of the Springer Nature.

Summary and Outlook
Since the emergence of TMD, due to its unique physical and chemical properties, it has received extensive attention. This paper reviewed the phase structure of 2D TMD materials, including 2H, 1T, 1T' and CDW. The singular physical and chemical properties of 2D TMDs were also introduced, such as catalytic properties, ferroelasticity, ferroelectricity, ferromagnetism and thermal conductivity properties. Meanwhile, the application of TMDs in electrocatalysis, lithium-ion batteries, supercapacitors, FETs, photodetectors, giant magnetoresistance and spin value are introduced in detail combined with TMDs phase and its characteristics. In addition, we reviewed current methods for modulating TMD phase transitions, including strain-induced and electron-regulated transitions. In particular, we analyzed the advantages and disadvantages of various methods of electron regulated phase transition, including plasma hot electrons, beam radiation, electrostatic electron injection, alloy doping and alkali metal atom adsorption. Further, we looked into the prospect of low symmetry TMDs phase structures for miniaturization applications and edge characteristics of electromechanical devices. Zhu et al. and Chen et al. proposed ultrathin transistors and single-atom layer-thick FinFET devices, respectively [117,118]. The size of electronic components can already reach the thickness of a single layer, and the miniaturized electromechanical devices of this single layer of atoms are also expected to be realized in the future and will have applications in many fields, from bionic robotics to the aerospace industry. However, conventional smart actuation materials exhibit poor performance at the nanoscale, due to detrimental force-length ratio relationships, or performance degradation, due to defective surfaces. Recent studies have shown that 2D smart driver materials are promising candidates for drivers in such devices. These include graphene, graphene oxide, lithium-doped phosphene, and graphite diaynes. However, the limited number of high-performance, 2D smart driving materials severely limits the on-demand design of various applications. The 1T' phase in TMDs has the properties of CDW, that is, coupling between electrons and phonons. At the same time, the 1T phase is very similar to the 1T' phase and has a large change in the character constant. Therefore, by tuning the CDW state of 1T' or the phase transition between 1T' and 1T, it is expected that smart driving materials with excellent performance could be developed.
It is worth mentioning that MoS 2 in practical applications is not an infinite ideal state, and there are always boundaries. Since the crystal structure and properties of the material at the edge are generally quite different from the internal properties of the material, the material has many novel properties. For example, the 2H phase is used to develop topological insulator materials and properties such as quantum spin Hall effect, edge ferromagnetism [119], and so on. Harnessing the physical properties of the edge is one of the great advantages of 2D materials. Due to the existence of such a low-symmetry phase as that of 1T', its edges may have more diverse edge structures and properties than other phase structures with higher symmetry. So far, there is little research on the edge structure and properties of the 1T' phase. Therefore, through research on the nanoribbons that compose the 1T' phase of MoS 2 , it is expected that novel physical and chemical properties will be discovered and will expand the application scope of this type of material.