1 Introduction
Two-dimensional (2D) materials have inspired a new research wave attempting to extend Moore’s Law and realize future electronic device miniaturization in the past decade since the discovery and rapid development of grapheme [
1-
4]. Numerous researches have revealed that graphene’s special zero-bandgap bring about low on/off ratio, which is unsatisfactory for applications in logic electronic and optoelectronic devices [
2,
5]. Since then, a variety of graphene analogs with the same atomically-thin layered structure have been extensively investigated for their abundant fantastic properties [
5-
8]. Transition metal dichalcogenides (TMDCs), which refer to a unique class of 2D materials with a sandwich structure where one layer of transition metal atoms is between two layers of chalcogen atoms [
9], have displayed adjustable band gaps [
10], high on/off ratio [
11] and extraordinary photoresponsivity [
12]. In addition, rich structural phases corresponding to different coordination and stacking order of metal and chalcogen atoms endow TMDCs a wealth of electronic and optoelectronic properties [
13-
17]. All of these outstanding features have put TMDCs in the position of promising candidates for next-generation nanoelectronics [
18,
19]. However, the actual performance of devices based on 2D TMDCs is far from the theoretical prediction by reason of the inevitable intrinsic defects in materials [
20-
22]. Heretofore, more significant efforts have been devoted to the defect engineering to improve the quality of TMDCs availably and remarkably.
In contrast to the bulk state, the nature of 2D TMDCs seems easier to be changed by different kinds of defects including vacancy, adatom, substitution and grain boundary due to the atomically thin structure [
5]. Specifically, the optical, electronic and optoelectronic properties of ultrathin TMDCs are all under the defect influence prominently [
23-
25]. For instance, the sulfur vacancies in molybdenum disulfide (MoS
2) could act as electron donors and increase the electron concentration to bring in n-type doping, so that the extra electrons tend to make trions negative and lower excitonic photoluminescence (PL) intensity [
26-
29]. On the other hand, the defects would have a great impact on the transport of conduction electron, which could serve as scattering centers and increase the probability of carrier scattering, sequentially leading to the low carrier mobility [
26]. To further study those intricate defects and precisely lessen unfavorable defects, it is necessary to utilize the advanced modern technologies of characterization to detect the defects existed in TMDCs at the atomic level. Multiple ways including scanning tunneling microscopy (STM) [
30], transmission electron microscopy (TEM) [
31] and scanning transmission electron microscope (STEM) [
32] could confirm the defect type and number, which can give a more accurate analysis on different kinds of defects. At present, plenty of approaches specific to the defect repairing have been demonstrated by experiments to improve the crystalline quality, such as regulating the growth parameters to increase the concentration of chalcogens in TMDCs [
33], choosing appropriate substrates [
34], sources [
35] and additives [
36], post annealing treatment in a certain temperature and flow rate [
37]. Moreover, extra protection layers (i.e., grapheme [
38] and h-BN [
39]), are also proved to prevent the formation of defects during the subsequent process of characterization and using. Although the investigations are extensive and the repair methods for defects are diverse, there seems to be a lack of systematic summary on the repairing techniques classified according to the typical defect types for the moment.
In this manuscript, we have reviewed the latest progress of defect engineering in 2D TMDCs triggered by the growth process and provide a coherent research idea for future studies on intrinsic defects introduced by growth. The scope of this review is shown in Fig.1. Firstly, we emphasize two important types of defects in TMDCs, point defect and grain boundary, and their influence on the electronic band structure in different conditions. There is no gainsaying that defects caused by intentional doping and etching could modulate the electrical and magnetic properties [
40-
42], and some significative defect designs are also beneficial to promote in-depth defect research [
43,
44]. For all that, defects are not ideal for TMDCs electronic device applications which have a strict requirement for the crystallinity to ensure high carrier mobility. Subsequently, we focus on some highly efficient methods suppressing defects during the process of synthesis. Despite various synthetic methods, chemical vapor deposition (CVD) is one of the most commonly used methods for low cost, simple operation, short growth cycle and large area preparation [
45-
47], while TMDCs grown by CVD usually produce chalcogen vacancies and grain boundaries [
48]. Hence, the assistance of halide, oxide and oxygen plasma aiming to control the vacancy generation and the selection of parameter, source, substrate as well as pretreatment and additive in allusion to grain boundary suppression are described respectively. Afterwards, we summarize the post-treatment strategies with dry and wet methods for representative chalcogen vacancies since defects cannot be eliminated completely during the growth. Interestingly, plasma with different gas and annealing in appropriate atmosphere could have a prominent effect on lowering the defect concentration by one or two order of magnitude, and thus the device properties are improved multiply. At last, we make a conclusion and put forward assumption and prospective for defect engineering in 2D TMDCs, which could provide a wider applied fields for TMDCs and inspire researches of other 2D materials.
2 Defect types
2.1 Point defect
As the simplest kind of crystal defects, point defects are also considered as defects in zero dimension owing to their small size at arbitrary direction in three dimensions [
49]. Those defects are deviating from the original arrangement of the crystal structure at the specific position or in the adjacent microscopic region. Familiar types of point defects [
50] are shown in Fig.2(a), including vacancies, adatoms, substitution and interstitial impurities. Notably, vacancy [
51] is the most important sort of point defects induced by thermal motion of atoms in the crystal, which is usually arisen during the synthesis process for high temperature. Generally, the vibration energy of atoms rises with the increasing temperature, and will be out of the balanced site when the energy is enough to resist bound energy, leaving isolated vacancies at the lattice site. Just as the Schottky vacancy [
52] in Fig.2(a), a metal vacancy and a chalcogen vacancy have been displayed, while dislocated atoms commonly enter into other vacancies or gradually transfer to the grain boundary or surface. When the atom from the crystal squeezes into the gap between the lattice, another defect called self-interstitial atom will be formed, simultaneously generating a corresponding vacancy. This pair of point defects containing a vacancy and an interstitial atom is Frenkel defect [
50] as known. Normally, the concentration of vacancy will increase with the rising temperature on the basis of statistical thermodynamics. In addition, foreign atoms can also be regarded as point defects for changing the normal lattice due to the different atom size and chemical electronegativity. Fig.2(a) also shows the smaller foreign atom could fill an interstice among the host atoms, while the foreign atom with suitable size replaces a host atom, called interstitial atom and substitutional atom respectively. What’s more, compared with substitutional and interstitial atoms, the adatom lying on the crystal surface is energetically more favorable to form [
53], which is also shown in Fig.2(a).
Furthermore, when defective points emerging continuously along the edge, a kind of one-dimensional defect called edge is formed [
54], which is shown in Fig.2(b). Those edge defects are generally created to provide active sites for catalysis and can be increased through constructing molecular clusters, nano-particles and other nanostructures. Many researches in catalysis have investigated how to expose more edge sites, so that the catalytic performance will be improved [
55].
2.2 Grain boundary
Grain boundary is a typical kind of interfacial defects with two dimensions, which is always existed in polycrystalline materials and separates two small grains of different directions [
49]. Due to different grain orientations, grain boundaries are ubiquitous at the interface when films driven by different domains overlap each other during the growth process of large-area TMDCs [
56]. According to the angle of misorientation between two neighboring grains, grain boundaries [
57] can be divided into low-angle grain boundary and high-angle grain boundary as shown in Fig.2(c). Apparently, the orientation of grains changes along the boundary, and thus irregular alignments endow the atoms at grain boundary with higher energy than those inside the grains, which could further affect the quality of crystals. Additionally, the energy will enhance following the increasing mismatch angle, since the structure of interface becomes more complicated, and finally lead to higher density and smaller size of domains [
58].
It is obvious that defects like vacancies and grain boundaries in the crystal will have a crucial influence on various properties of materials as a result of breaking the regulation of atomic arrangement. Especially, the energy level around the defects differs from that of regular lattice, inducing local energy levels in the forbidden band. In general, the type and concentration of defects could determine the shift of Fermi level [
59-
61]. Fig.2(d) shows four situations of the energy level caused by different defects in 2D system. When defects play the role of donor and bring higher electron concentration, the Fermi level will move towards the conduction band. For example, the sulfur vacancies in MoS
2 make the electrons of molybdenum unpaired, thereby turning into free electrons to participate in the conduction process. Therefore, the MoS
2 with sulfur vacancies would transform to an n-type semiconductor and the Fermi level representing the intensity of electron filling the energy level will improve [
59]. At a certain temperature, the Fermi level is closer to the bottom of the conduction band with the increasement of donor defects. On the contrary, the accepter-like defects will trigger hole conduction and gradually shift the Fermi level near to the valence band in the wake of increasing defect concentration. Ji
et al. [
62] have demonstrated that intrinsic W vacancies in WSe
2 monolayer via STM and
ab initio calculations are the critical cause inducing p doping, which is completely different from common intrinsic chalcogen vacancies in most TMDCs. With regard to fabricating lateral heterostructure, this p-type doping is always expected but rare to realize [
63].
As for 2D TMDCs, their electronic and optoelectronic properties are closely related to the change of band structure which affects the absorption and emission of light as well as the capture and release of carriers [
64]. Undeniably, the defect energy levels introduced in the forbidden band to some extent broaden the photoelectric applications on account of tuning the width of the band gap. However, these defects would inevitably lead to the formation of trap states, which usually reduce the mobility [
65]. In consequence, it is necessary to repair the defects through the defect engineering, so that the electrons will turn to extended states and the performance comprising mobility, on/off current ratio, photoresponsivity, etc. will be improved. Furthermore, repairing defects is an effective way to eliminate the boundaries and realize the large-area growth of 2D TMDCs.
3 Improve the quality during the growth
3.1 Vacancy
To reduce the vacancy, one strategy is optimizing the growth condition. Recently, Wang
et al. [
66] proposed a novel method to grow MoS
2 by introducing sodium halide NaX (X = Cl, Br and I) during the growth as shown in Fig.3(a). The results show that the sodium halide would not only facilitate the growth of the MoS
2 domain, but also suppress the excitons bound originating from the sulfur vacancy. By theoretically calculation, it was demonstrated that the X atoms would absorb at sulfur vacancy, and then passivated it and eliminate the trapping states. Moreover, introduced other metallic oxide powder as the blended precursor is also an effective strategy to decrease nonmetallic vacancy [
67-
69]. Li
et al. [
70] introduced few WO
3 as mixed precursor with MoO
3 during the preparation of CVD-grown MoSe
2, achieving the isoelectronic doping of W atoms in monolayer MoSe
2. Eventually, the Mo
0.98W
0.02Se
2 and Mo
0.82W
0.18Se
2 were obtained at the different weight ratio of WO
3/MoO
3. According to Fig.3(b), the HAADF-STEM measurement, the relative content of Se vacancy decreased from 4% to 2.5%, and 4% to 2% for Mo
0.98W
0.02Se
2 and Mo
0.82W
0.18Se
2 respectively, compared to MoSe
2, which can be ascribed to the doping W atoms and interpreted by a stronger hybridization between the outermost p orbitals of Se and the d orbitals of W. The PL intensity of Mo
0.82W
0.18Se
2 augmented more than 10 times as compared to pristine MoSe
2 because the defect-mediated nonradiative recombination reduced as the reduction of Se vacancy. For obtaining MoS
2 with large area and high crystal quality, Chen
et al. [
71] adopted oxygen-assisted CVD to prepare monolayer MoS
2. As shown in Fig.3(c−e), the domain size of MoS
2 can reach up to 350 μm and rarely S vacancies was observed. Significantly, the FET property performed excellent with mobility up to 90 cm
2·V
−1·s
−1. They speculated the oxygen can not only prevent the poisoning of precursors, but also eliminate defects during the growth.
3.2 Grain boundary
In order to reduce the density of grain boundary during the growth of TMDCs, many studies based on the optimization of the growth strategy have conducted to obtain larger domain size. Wang
et al. [
72] achieved a novel two-stage space confinement effect during CVD-growth of monolayer WS
2 as shown in Fig.4(a), where the stacked substrates formed the micro-reactors for the first stage and substrates placed in another one-side sealed tube for the second stage. As demonstrated in Fig.4(b), small size and overlap manner of monolayer WS
2 crystals were obtained without these two space confinement effects, while the crystals size was around 100 μm and even up to 450 μm with the first stage and these two stages respectively. Similarly, Tu
et al. [
73] achieved larger grain size, uniform MoS
2 monolayers via CVD method using a double-tube system with one-side sealed inner tube. Other than above strategies, Chen
et al. [
74] developed a two-stage CVD method to prepare MoS
2 monolayers where the induction stage was separated from the growth stage and obtained larger domain size of 305 μm compared to the one-stage CVD method. Both of them reduced the nucleation density effectively. Moreover, adjustment of the growth parameter, such as variety, flux and scale of carrier gas [
71,
75,
76], the distance between substate and source [
77], and the orientation of substrate [
78], can achieve the improvement of the domain size.
Selection of the substrate and source also greatly influences the domain size. Yang
et al. [
79] adopted a Mo foil placed above the soda-lime glass growth substrate as the source and obtained highly uniform MoS
2 monolayers with domain size of more than 400 μm, much larger than the power source with 40 μm, which is shown in Fig.4(c). Pam
et al. [
80] treated WO
3 precursor with different method, including thermal annealing (AWO
3), air plasma (PTWO
3), both (PTAWO
3) and commercially available WO
3 powder (FWO
3). As shown in Fig.4(d), the AWO
3 obtained the largest domain size. In addition to the source, Zhang
et al. [
81] implemented large-size high quality MoS
2 monolayer with domain size up to 563 μm on molten glass, which was 3 times larger than the sapphire substrate. The result is shown in Fig.4(e) clearly. Significantly, corresponding mobility and on/off ratio of MoS
2-based FET were 20 times and two orders of magnitude higher than that transferred from sapphire, respectively. Even 2.5-millimeter single-crystal MoSe
2 monolayer was achieved on molten glass [
82]. Besides, large domain size of TMDCs films up to several hundred micrometers can be obtained on Au substrate [
83-
86]. Godin
et al. [
87] adopted substrates pretreated by oxygen plasma. Compared to nontreatment substrate, the average grain size increased by 78% from 56 μm
2 to 101 μm
2, which can be ascribed to lowered activation energy for growth and increased lateral diffusion on the substrate.
Catalysts and additives are increasingly used to produce TMDCs films with higher quality, such as alkali metal compound [
88-
94], PTAS [
95,
96] and other additives [
97-
99]. Huan
et al. [
93] developed a NaCl-assisted CVD method. Consequently, they obtained larger MoS
2 monolayers with domain size of 100 μm, much larger than 100 nm without the assist of NaCl. Moreover, the triangular MoS
2 domain can up to 300 μm through the optimization of growth parameter. Significantly, fast growth of large-domain MoS
2 monolayers was achieved with a growth rate ~5 μm·s
–1. Furthermore, studies showed that the growth system using NaCl as catalyst can obtain TMDCs thin films with crystal domain size of 632 μm [
89] or even 1.7 mm [
88]. Yang
et al. [
95] adopted different seeding promoters (PTAS, CuPc and CV) to explore their different influence on the CVD-grown MoS
2 monolayers. As shown in Fig.4(f), PTAS seeding performed best with the largest domain size of 70 μm while the domain size of MoS
2 with CuPc and CV was between 10 and 20 μm. Besides, the mobility of PTAS-assisted MoS
2 was 23.2 cm
2·V
−1·s
−1 and the current on/off ratio was 10
6−10
7. It can be explained by the presence of PTAS possibly increased the surface adhesive force of MoS
2, resulted in the layered growth of MoS
2. Interestingly, Zhang
et al. [
98] introduced hydroxide mixed with ammonium molybdate to form aqueous solution as Mo source to prepare MoS
2 monolayer. Eventually, as seen in Fig.4(g), MoS
2 monolayers grown with −OH possessed larger domain size of 200 μm than system without −OH of 500 nm, where the −OH groups were attached to the (001) surface of MoS
2, forming a MoS
2−OH bilayer structure. The surface −OH layer not only suspended the growth along the [001] axis and led to preferential growth of MoS
2 monolayers, but also protected the MoS
2 surface from oxidation in air. Significantly, the field-effect carrier mobility and on/off current ratio were 30 cm
2·V
−1·s
−1 and 10
7, respectively.
4 Improve the quality by post-treatment
4.1 Sulfur vacancy
4.1.1 Post-treatment by dry method
To solve the limitation of TMDCs application in the field of electronic devices, which is caused by intrinsic defects, mostly the sulfur vacancies, besides repair methods during growth, many post-treating methods have been conducted. Jiang
et al. [
100] developed a method as shown in Fig.5(a), treating the multilayered WS
2 gained by mechanical exfoliation with nitrogen plasma, for repairing the sulfur vacancies. The high-resolution spherical aberration correction STEM images in Fig.5(b) and the calculation indicated that an average density of sulfur vacancy (Vs) was smaller than pristine samples from 0.181 nm
−2 to 0.049 nm
−2, which was induced by the N atoms adsorption at sulfur vacancies and the formation of W−N bonds. Furthermore, because of the nitrogen plasma treatment, the performance of WS
2 FETs enhanced obviously, with the field-effect mobility increased from 29.7 to 184.2 cm
2·V
−1·s
−1 and the on/off current ratio increased by three orders of magnitude. There are many other approaches with similar repair mechanisms where these vacancies were boned with other atoms. Nan
et al. [
101] adopted post-treatment to exfoliated MoS
2 films with mild oxygen plasma and achieved controllable defect concentration particularly S vacancy in MoS
2. Eventually, the PL intensity increased by nearly 100 times, by which O atoms can chemisorb on S vacancy and achieve the p-type doping. Pierucci
et al. [
102] treated monolayer MoS
2 grown by CVD with hydrogen, as represented in Fig.5(c), and the XPS measurement displayed gradually decrease of the substoichiometric signature peak owe to Mo atoms around S vacancy, from 18% to 13%, even nearly disappeared as the increased dose of hydrogen compared to pristine MoS
2. The decrease of S vacancy can be explained by the partial saturation of these vacancies with hydrogen atoms forming Mo−H bonds, as shown in Fig.5(d). Significantly, the conversion of electronic properties of monolayer MoS
2 was achieved from the intrinsic electron (n) to hole (p) doping after hydrogen treatment.
Besides the introduction of foreign atoms to form chemical bonds, annealing in the same atmosphere can also achieve the purpose of vacancy repair [
103-
105]. Liu
et al. [
105] annealed CVD-grown MoS
2 monolayers under ultrahigh vacuum conditions at different temperature, which is shown in Fig.5(e), and obtained nearly defect-free films with lower defect concentration of 1.2 × 10
11 cm
–2 after annealing at 400 K compared to the usual value of 10
13 cm
–2. Moreover, laser irradiation also can repair nonmetallic vacancy effectively and enhance PL intensity eventually [
106-
108].
4.1.2 Post-treatment by wet method
There are many kinds of solution treatment which can also achieve defect repair through the mechanism that vacancies were filled by foreign atoms to form chemical bonds. Recently, bis-(trifluoromethane) sulfonimide (TFSI) treatment shows excellent vacancy repair potential [
109-
116]. Dai
et al. [
110] reported the chemical method treated by bis-(trifluoromethane) sulfonimide (TFSI) to repair the S vacancies in monolayer CVD-grown MoS
2 and WS
2. After the treatment, the XPS spectra of MoS
2 shows the increase of S/Mo ratio from 1.532 to 1.832, so as to indicate successful defect repair of Vs. The PL intensity enhanced nearly 50 times. Amani
et al. [
114] obtained similar results as shown in Fig.6(a), where the S/Mo ratio increased from 1.84 to 1.95 compared to as-exfoliated MoS
2 and PL intensity enhanced over two orders of magnitude and photoluminescence quantum yield increased from 0.6% to 95% after the treatment of TFSI. Roy
et al. [
111] conducted the same study and the ADF-STEM images in Fig.6(b) showed that there was an obvious reduction of the Vs and V
S2’s relative content in MoS
2 from 2.65% to 0.32% and 0.27% to 0.1% after TFSI treatment respectively, confirming the success of the defect repair. Significantly, they revealed the mechanism of the repair is that S vacancy sites were directly filled by the excess S atoms supplied by the TFSI molecules. Moreover, Dhakal
et al. [
112] confirmed the vacancy repair effect of TFSI through different enhancement of photoluminescence intensity in different regions of MoS
2. The results displayed that Raman intensity ratio of the 2LA(M) and A
1g peaks, where a smaller value of ratio indicates more defects, enhanced from 3.8 to 5.4 at the edge, while 3.8 to 4.7 in the center after the TFSI treatment. According to previous research, there were more S vacancies at the edge of 1L-MoS
2 than inner region, which indicated why more effective improvement occurred at the edge. Namely, it was further proved the chemical treatment with TFSI can repair S vacancies in 1L-MoS
2 effectively.
Besides the TFSI, poly(4-styrenesulfonate) (PSS) has also been employed to repair the TMDCs. It was found in Fig.6(c) that the PSS treatment would decrease the sulfur vacancy. It was reported the S/Mo ratio of MoS
2 had a remarkable increase from 1.67 to 1.86 after PSS treatment. The electron concentration would decrease by 643 times. Furthermore, the PL intensity and photoresponsivity which reaches up to 339.2 A·W
−1, increased about 2-fold and 5.6-fold respectively after the treatment [
117,
118]. Zhang
et al. [
119] found that the PSS treatment would improve the stability of the MoS
2 in air, where after exposure 12 months, the property has no obvious decay.
Similarly, thiol molecules can adsorb on S vacancies of TMDCs and complete defect repair. It was studied that (3-mercaptopropyl)- trimethoxysilane(MPS), a kind of thiol molecules, can repair S vacancies through two steps. The detailed reaction is shown in Fig.6(d). Firstly, S atom in MPS chemisorbed onto a defect site of S vacancies, with the fracture of S−H bond. Secondly, the S−C bond broke and the S vacancy was repaired thoroughly with the formation of byproduct trimethoxy(propyl)silane. Significantly, as illustrated in Fig.6(e), the density of S vacancy has decreased from 6.5 × 10
13 cm
−2 to 1.6 × 10
13 cm
−2 and the S/W ratio increased obviously from 1.62 to 1.98 after the MPS treatment. Besides, the FET mobility enhanced from 40 cm
2·V
−1·s
−1 to 80 cm
2·V
−1·s
−1 [
120,
121]. Similarly, the S vacancy density decreased from 3% to 1.2% after the butanethiol (C
4H
10S) treatment, with the augment of the
ION/
IOFF ratio and the field-effect mobility from 2 × 10
6 to 4 × 10
7 and 1.4 to 8.0 cm
2·V
−1·s
−1 respectively [
122]. Makarova
et al. [
123] exploited thiol molecules to adsorb on S vacancies in MoS
2 and repair these vacancies by the auxiliary effect with the STM tip. The series of STM images in Fig.6(f) displayed that pristine MoS
2 had some darker points which are the S vacancies, however, it got to be bright after the adsorption of thiol molecules. Significantly, after removal of the alkyl by the brushing of the STM tip, the darker points disappeared. All the results indicated the effective repair of S vacancies, and the probable repairing processes are shown in Fig.6(g). Moreover, it was reported that PL intensity and electrical performance of MoS
2-based FET would be improved after the treatment of thiol molecules with the field-effect mobility enhanced by 3 times [
124-
126]. Furthermore, thiol-based organic compounds with different functional group could achieve different types of doping, n-type doping and p-type doping [
127].
There are many other chemical reagents available for vacancy repair, such as ethylene diamine tetraacetic acid (EDTA) [
128], titanyl phthalocyanine (TiOPc) [
129], CuCl
2 [
130], oleic acid (OA) [
131], Ag-NPs (Nanoparticles) [
132], ionic liquid [
133], Protoporphyrin (H2PP) [
134], N,N’-ditridecylperylene-3,4,9,10-tetracarboxylic diimide (PTCDI-C13) [
135], 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F4TCNQ) and 7,7,8,8-tetracyanoquinodimethane (TCNQ) [
136], 1, 2-dichloroethane (DCE) [
137], which can enhance the mobility [
128,
130,
131,
135], the on/off ratio [
129,
137], and the PL intensity [
131-
133].
4.2 Selenium vacancy repair
4.2.1 Post-treatment by dry method
As a kind of TMDCs,repairing selenium vacancy of selenide would heavily improve its photoelectrochemical properties. Many studies developed an oxygen passivization where the oxygen chemisorbed onto the Se vacancies in monolayer CVD-grown WSe
2. After this treatment, the height of WSe
2 flakes increases for the presence of oxides which is shown in Fig.7(a), and the XPS results display the peaks contributed to the W−O, namely confirmed the chemisorption of oxygen in WSe
2. Significantly, the W
4+/Se ratio increased from 1:1.81 to 1:1.94 and 1:2.06 after laser passivization in vacuum and ambient conditions respectively. Moreover, it was experimentally confirmed that the Se vacancies passivization by oxygen enhanced the conductivity and photoconductivity of monolayer WSe
2 respectively by 400 times and 150 times [
138]. Moreover, Wang
et al. [
139] developed one similar strategy shown in Fig.7(b) to conduct laser annealing on fewer layer MoSe
2. The XPS measurement displayed obvious increase of Se/Mo
4+ ratio from 1.69 to 1.84 after annealing. Meanwhile, the PL and Raman peak intensity both improved. Significantly, light-to-energy conversion efficiency of MoSe
2-based electrochemical cell increased by 110%−280%.
4.2.2 Post-treatment by wet method
Similar to the methods of sulfur vacancy repair, some chemicals can also be used to repair selenium vacancies. It was reported that the ratio of the Se to Mo atoms of layered MoSe
2 grown by chemical vapor deposition had a significant increase from 1.68 to 1.97 after the treatment of hydrohalic acid like HBr. Significantly, there was over 30 times improvement of PL intensity contributed to the repair of Se-related defects as shown in Fig.7(c), through the synergistic effect from Br atoms and the conversion of the bridging Se
22− to the Se
2− [
140]. Meng
et al. [
141] treated the monolayer MoSe
2 grown by CVD method with EDTA for healing the defect such as Se vacancy in MoSe
2. The FET based on single layer MoSe
2 device’s electronic mobility gains about 300 times from 0.1 cm
2·V
−1·s
−1 to 30 cm
2·V
−1·s
−1 and hole mobility over 10 cm
2·V
−1·s
−1. Besides, coating the CVD-grown MoSe
2 with ZnPc which is a kind of organic dopants named MPc was demonstrated a simple method to passivate the Se-vacancies, which has been demonstrated in Fig.7(d) and (e), where ZnPc bonded by van der Waals on the surface of TMDCs can reduce the defect density effectively for absorbing at the Se-vacancies in MoSe
2 [
142].
Tab.1 has summarized recent researches of defect repairing in CVD-grown TMDCs. Different strategies for undesirable vacancies and grain boundaries effectively improve the performance of TMDCs, including the carrier mobility, on/off current ratio, PL intensity, etc. In particular, large-area grown TMDCs are gradually realized by controlling the domain density during the growth to reduce grain boundaries, which is expected to meet the industry demand.
5 Conclusion and outlook
In recent years, two-dimensional layered TMDCs with unique atomic structure have attracted extensive attentions for breaking the zero-band gap limitation of graphene. Abundant band structures depending on thickness give them a wide range of electrical properties which have covered metals, semiconductors and even superconductors. Hence, TMDCs are expected to achieve the controllable property manipulation and the emerging application extension in the field of electronics, optoelectronics, biosensors, catalysis, and so forth. Up to now, a large number of experimental results and theoretical calculations have shown that the lattice defects play a key role in the optical and electrical properties of TMDCs [
143]. Accordingly, researchers have made significant efforts to study the defect repairing for the sake of performance modification in all kinds of TMDCs. In this review, we have emphasized key developments in defect engineering of 2D layered TMDCs, mainly focusing on (i) the repair methods of vacancies and grain boundaries during the CVD growth process, such as appropriate additive introduction, oxygen plasma assistance and growth procedure design. Other than that, (ii) the post treatment aiming at the non-metallic defects in synthesized films like S vacancy and Se vacancy and their influence are comprehensively discussed from both dry and wet methods perspectives. Annealing, plasma and special solution treatment, such as TFSI, PSS, and HBr, are all effective to reduce the undesirable defects in TMDCs and enhance their electrical performance. Nevertheless, there are still plenty of challenges to surmount in the future study on defect engineering as follow.
5.1 Revealing defect repairing mechanism
To date, few kinds of ultra-thin wafer-scale TMDCs are realized due to grain boundaries which could impede their epitaxial growth. On top of that, vacancy is also a crucial factor to consider in the defect engineering, which could hinder the high-performance electronic devices. Current researches have studied how to diminish those unwished defects generated during the synthesis through growth parameter alteration and how to repair the inherent defects of as-grown TMDCs by post-treatment. However, the repair mechanism is ambiguous in the present and the performance enhancement is limited and uncontrollable. On one hand, the exhibited properties of TMDCs are influenced by many parameters including the distance between the source and substrate, the flow rate, the ambient pressure, etc. On the other hand, considering the wide variety of TMDCs, it is complicated to precisely tailor the growth parameters corresponding to different types. Therefore, further researches should focus on exploring the specific defect repairing mechanism and the synthetic mechanism of 2D TMDCs, so that people could control their size and orientation more accurately in essence.
5.2 Optimizing repairing process design
Although repairing methods via plasma irradiation and chemical treatment are two promising avenues to solve the deficiency in TMDCs, there are still some problems existed like increasing energy consumption and tail gas in the dry methods, as well as unavoidable to introduce contamination and other defects in the wet methods. Thus, defect repairing process optimization ensuring the low defect density is required for the prospective defect engineering to improve the conductivity and photoconductivity. Meanwhile, both efficiency and cost of defect repairing are also needed to consider in the design. In this way, we can predict that the progressive development of defect engineering in the near future will make a major stride toward electrical, optoelectronic, magnetic and spin electronic applications.