Tunable electronic structure of two-dimensional transition metal chalcogenides for optoelectronic applications

2.1 Crystal structure

Transition metal dichalcogenides can be generalized with the formula MX₂, where M denotes a transition metal from group IIIB to group VIIIB of subgroup elements in the element periodic table (Mo, Ti, Zr, Nb, Ta, W, Pt, and so on), whereas X denotes a chalcogen (S, Se, or Te). A large number of TMDs exhibit graphite-like layered structure, which can be exfoliated into 2D layers. The layered structure is composed of 2D hexagonally packed X-M-X sandwich layers – metal atoms in the middle hexagonal plane covalently bonded to chalcogen atoms located in the top and bottom hexagonal planes, stacked layer-by-layer, as shown in Figure 1A. The adjacent X-M-X sandwich layers are held together to form bulk crystal by van der Waals interactions, which is much weaker than intralayer covalent bonds. The strong intralayer and weak interlayer interactions lead to the strong anisotropy in mechanical properties, responsible for a remarkably easy mechanical exfoliation to get thin layers or monolayer (a single X-M-X sandwich layer).

Figure 1:

Crystal and electronic structures.

(A) Three-dimensional schematic illustration of a typical TMD (MX₂) structure with 2H phase. The pink ball represents the chalcogen atom X, and the gray ball represents the metal atom M, which are also the same in (B) and (C). (B) The schematic illustration of the two basic octahedral and trigonal prismatic coordination in TMDs. (C) Schematics of the three most commonly involved structural phases: 1T (tetragonal symmetry, one layer per repeat unit, octahedral coordination), 2H (hexagonal symmetry, two layers per repeat unit, trigonal prismatic coordination), and 3R (rhombohedral symmetry, three layers per repeat unit, trigonal prismatic coordination). The dot-dashed vertical lines indicate the alignments of interlayer M and X atoms. (D) Simplified band structure of bulk MoS₂, showing the lowest conduction band c1 and the highest split valence bands v1 and v2. A and B are the direct-gap transitions, and I is the indirect-gap transition. E′g is the indirect gap for the bulk, and E_g is the direct gap for the monolayer. Reproduced with permission from Mak et al. [12]. Copyright 2010 the American Physical Society.

The bulk TMDs exhibit a variety of polymorphs, which determine the unique structure and electronic properties, depending on stacking orders and metal atom coordination by the chalcogens. As illustrated in Figure 1B, within each X-M-X sandwich layer, the transition metal atom is coordinated by six chalcogens in two basic geometries: octahedral coordination or trigonal prismatic coordination, leading to different symmetries. The octahedral coordination with AbC (the letters denote different atom-arrange positions, whereas the uppercase and lowercase letters denote chalcogen and transition metal atoms, respectively) stacking sequence holds inversion symmetry for monolayer, whereas the trigonal prismatic coordination with AbA stacking sequence results in broken inversion symmetry in monolayer. There are three most commonly involved structural phases [1]: 1T (one sandwich layer per unit cell with AbC stacking order, octahedral coordination), 2H (two sandwich layers per unit cell with AbABaB stacking order, trigonal prismatic coordination), and 3R (three sandwich layers per unit cell with AbACaCBcB stacking order, trigonal prismatic coordination), as illustrated in Figure 1C. As an example, the most widely studied group VIB bulk TMDs (e.g. MoS₂, WSe₂, MoTe₂) exhibit 2H or 3R phase [2], [13], [14], [15], [16], the group VB TMDs (e.g. TaSe₂, TaS₂) exhibit 2H or 1T phase [17], [18], [19], [20], [21], and all the group IVB TMDs (e.g. TiS₂, ZrS₂) are 1T phase [22], [23], [24], [25], [26], [27]. It has been found that phase changes can be induced by intercalation [28], [29], [30], [31], [32], [33], [34]. For example, 2H-MoS₂ can transform to 1T phase by lithium intercalation [2], [31], [33], [34], [35]. These structural phases featured with different symmetries and stacking orders, relating to the d-electron count, primarily play important roles in affecting the electronic properties of TMDs.

2.2 Electronic properties

Most TMDs have common features in their band structures originating from d orbital on metal atoms, as demonstrated by theoretical calculations and spectroscopic experiments [1]. Different d-electron counts of transition metal elements from group IIIB to group VIII fill the nonbonding d bands to different degrees, resulting in the wide variety of electronic properties including insulator, semiconductor, semimetal, metal, and superconductor [2]. In general, the partially filled group VIB TMDs, such as MoX₂ and WX₂, are semiconducting [36], [37], whereas the fully filled group VB TMDs, such as NbX₂ and TaX₂, are metallic [38], [39].

Semiconducting TMDs such as MoX₂ and WX₂ possess bandgaps in the near-infrared to the visible region, promising interesting photonics and optoelectronics applications [3], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52]. Their bulk materials are indirect-bandgap semiconductors with the bottom of conduction band and the top of valence band maximum located at different high symmetry points of the Brillouin zone, respectively, whereas the monolayers change to larger direct-bandgap semiconductors [53], [54], [55], [56], [57], [58]. Compared to bulk TMDs, the electronic properties of 2D TMDs are considerably tuned as a result of quantum confined effect in the out-of-plane direction and the resulting reduced symmetries and alteration in hybridization between p_z orbitals on chalcogen atoms and d orbitals on metal atoms near Γ point when thinning down thickness [55], [57], [59]. Except the indirect-to-direct bandgap transition and bandgap size enlargement, few-layer or monolayer 2D TMDs exhibit strong excitonic effects due to the weak charge carrier screening and subsequent strong Coulomb interactions caused by reducing thickness [57], [60], [61], [62], [63], [64], [65], [66], [67].

A simplified band diagram of MoS₂, as a representative semiconducting TMD, is depicted pictorially in Figure 1D, showing the lowest conduction band c1 and the highest split valence bands v1 and v2. The indirect bandgap E′g for the bulk features conduction band minimum locates between Γ and K points and valence band maximum located at the Γ point, whereas the direct bandgap E_g for the monolayer locates at K point. The valence band around K point is split as a result of spin-orbit coupling [12], [68], [69], [70]. The direct-gap transitions between the maximum of the split valence bands and the minimum of conduction band, generally called the A and B excitons, will lead to two distinct low-energy resonance peaks in optical spectrum, while the indirect-gap transition I will give rise to an even lower-energy resonance peak. The indirect-to-direct bandgap transition and strong excitonic effect in monolayer TMDs have been widely demonstrated by absorption spectra and photoluminescence [12], [59], [70], [71], [72].

3.1 Quantum confinement by thickness thinning down

The thickness-dependent band structures of TMDs, due to the resulting quantum confinement effect of thinning thickness, greatly affect the optical properties and optoelectronic performance. Benefiting from the easily mechanical cleavage [7] and bottom-up growth by CVD [103], [104], [105], [106], [107] or metal–organic chemical vapor deposition [15], [108], [109], atomic thin layers can be achieved to provide opportunities for studying thickness-dependent physical properties in experiments. As first reported by Splendiani et al. [59] in 2010, a strong photoluminescence, which is absent in bulk, emerges in the monolayer MoS₂, indicating an indirect-to-direct bandgap transition in this d-electron material. Through characterization by absorption, photoluminescence, and photoconductivity spectroscopy, Mak et al. [12] traced the thickness-dependent electronic properties and the resulting optical properties of MoS₂ from one to six layers (Figure 2C). It shows that the indirect bandgap increases with decreasing thickness and changes into direct bandgap at the monolayer, whereas the direct excitonic transition energy almost keeps unchanged. In addition, the monolayer MoS₂ exhibits a considerable increase in luminescence quantum efficiency compared with bulk, as a result of the strong excitonic effect due to reducing thickness. The observation of thickness-dependent band structure and resulting optical properties evolution shows possibility of tuning electronic properties for optoelectronic applications by thinning thickness [3], [40], [42], [48], [110], [111].

Figure 2:

Quantum confinement with tuning thickness.

(A) Band structures of bulk MoS₂, its monolayer, and few layers calculated at the DFT/PBE level. The horizontal dashed lines indicate the Fermi level. The arrows indicate the fundamental bandgap (direct or indirect) for a given system. The top of valence band (blue/dark gray) and bottom of conduction band (green/light gray) are highlighted. Reproduced with permission from Kuc et al. [55]. Copyright 2011 American Physical Society. (B) ARPES spectra of monolayer, bilayer, trilayer, and 8 ML MoSe2 thin films along the G–K direction. White and green dotted lines indicate the energy positions of the apices of valence bands at the G and K points, respectively, with energy values written in the same colors. Reproduced with permission from Zhang et al. [70]. Copyright 2014 Springer Nature Limited. (C) Left: PL spectra for monolayer and bilayer MoS₂ samples in the photon energy range from 1.3 to 2.2 eV. Inset of left: PL QY of thin layers for N=1–6. Middle: Normalized PL spectra by the intensity of peak A of thin layers of MoS₂ for N=1–6. Feature I for N=4–6 is magnified, and the spectra are displaced for clarity. Right: Bandgap energy of thin layers of MoS₂, inferred from the energy of the PL feature I for N=2–6 and from the energy of the PL peak A for N=1. The dashed line represents the (indirect) bandgap energy of bulk MoS₂. Reproduced with permission from Mak et al. [12]. Copyright 2010 the American Physical Society. (D) Left: Schematic 3D view of single-layer transistor with hexagonal structure MoS₂ nanosheet, 50-nm-thick Al₂O₃ dielectric and ITO top-gate under monochromatic light. Middle: The schematic band diagrams of ITO (gate)/Al₂O₃ (dielectric)/single-layer (1 L), double-layer (2 L), and triple-layer (3 L) MoS₂ (n-channel) under the light (E_light=hν) illustrate the photoelectric effects for bandgap measurements. Right: Respective photocurrent dynamics of single-layer MoS₂-based top-gate transistors under monochromatic red and green lights. Reproduced with permission from Lee et al. [40]. Copyright 2012 American Chemical Society.

Band structure calculations through the first principles show qualitative agreement with experimental researches. The thickness-dependent band structure of various TMDs has been studied by lots of theoretical works including ab initio calculations [36], [55], [58], [59], [112]. As an example shown in Figure 2A, the valence band maximum of MoS₂ shifts from Γ to K point when the thickness is thinned to monolayer, resulting in indirect-to-direct bandgap transition [55]. The unusual thickness-dependent band structure of TMDs is attribute to quantum confinement effect and the characters of d-electron orbitals that comprise the conduction and valence bands [59], [70], [71]. Theoretical calculations for MoS₂ [59] show that conduction band states at the K point are mainly composed of localized d orbitals on Mo atoms, located in the middle of the S-Mo-S sandwich layers and having relatively weak interlayer coupling. However, states near the Γ point originate from combination of d orbitals on Mo atoms and antibonding p_z orbitals on S atoms, having strong interlayer coupling and being more sensitive to thickness. Therefore, as thickness decreases, the conduction band states near K point are almost unchanged, while the conduction bands and valence bands near the Γ point are changed considerably [113]. The direct observation of the thickness-dependent electronic band structure evolution is achieved by angle-resolved photoemission spectroscopy (ARPES) measurements on mechanically exfoliated and CVD-grown MoS₂ and molecular beam epitaxy–grown MoSe₂ thin films [70], [71]. The significant step-by-step evolution of the valence band (Figure 2B) provides direct evidence of the valence band maximum shifts from Γ to K point (indirect-to-direct bandgap transition), when the thickness is thinned to monolayer [70]. A lot of TMDs including MoX₂ and WX₂ are expected to possess a similar indirect-to-direct bandgap transition with decreasing thickness to monolayer, covering the bandgap energy range of ~1 to 2 eV [12], [40], [55], [70], [71], [114], [115], [116].

The controllability of band structure with tuning thickness paves the way for new optoelectronics. The electronic band structures of TMDs described earlier directly influence their optical properties. A variety of photodetectors based on thickness-modulated TMDs have been demonstrated. Lee et al. [40] have realized photodetection of different wavelengths using MoS₂ layers with different thickness. As shown in Figure 2D, devices with single- and double-layer MoS₂, exhibiting significant energy bandgaps of 1.82 and 1.65 eV, respectively, are demonstrated effective for green light detection, whereas devices made of triple-layer MoS₂ with a bandgap of 1.35 eV are effective for red light detection. Yin et al. [110] have fabricated a single-layer MoS₂ phototransistor showing good photoresponsibility.

3.2 Defect engineering

Previous studies have shown 2D material–fabricated electronic devices could emit photons out of the materials, detect incident photons, or control the properties of incident light [117]. Photoluminescence transition of bulk to monolayer MoS₂ is directly related to their different electronic structures [43], [59]. Briefly speaking, photoluminescence consists of photoexcitation (i.e. photons that excite electrons to a higher energy level), relaxation processes, and emitting electromagnetic radiation, whose emitting intensity is proportional to the decrease in the concentration of defects. The defects of 2D materials are experimentally revealed: (1) point defects including foreigner dopants, vacancies, and ad adsorbed atoms; (2) line defects such as grain boundaries, edges, and so on. Hong et al. [118] have experimentally revealed monolayer MoS₂ of physical vapor deposition often showed antisite defects with one Mo atom replacing one or two S atoms (Mo_S or Mo_S2). However, mechanical-exfoliated and chemical vapor–deposited MoS₂ samples are dominantly shown with S vacancies with one (V_S) or two (V_S2) S atoms absent [118], [119]. The researchers exceptionally found that antisite defects (Mo_S or Mo_S2) reduced the mobility of monolayer MoS₂ by three times than by the presence of vacancies (V_S or V_S2), whereas the phonon-limited mobility of holes carried by the intrinsic valence band is much more sensitive. The measured carrier motilities of monolayer MoS₂ are mostly controlled by the primary type of defects, regardless of carrier density and contact resistance.

Influence of defects could not only change electronic structure and charge-carrier mobility of TMDs but also create midgap states in the bandgap and introduce the in-gap defect levels, which can significantly modify and improve the performance of optoelectronic properties [120]. As shown in Figure 3G, defect engineering and strong oxygen bonding on the defect sites of monolayer MoS₂ could tune the photoluminescence intensity up to at least thousands of times. The physical-adsorbed O₂/H₂O molecules at defect sites induced heavy p-type doping and switched the trions recombination to exciton recombination process. The reduced nonradioactive recombination from excitons at defect sites could guarantee the enhancement of photoluminescence intensity under the vacuum (Figure 3A). The strong bonding energy of 2.395 eV for an oxygen molecule adsorbed on an S vacancy of MoS₂ was used to controllably introduce defects and oxygen bonding in MoS₂ by oxygen plasma, resulting in a controllable manipulation of photoluminescence [120].

Figure 3:

Defect engineering.

Photoluminescence intensity mapping images of exfoliated monolayer MoS₂: (A) as-exfoliated, (B) annealed in vacuum for 1 h at 350°C, (C) pumping down to 0.1 Pa. The maps have the same color bar; photoluminescence intensity images of another monolayer MoS2: (D) as-prepared, (E) annealed in vacuum for 1 h at 500°C, (F) pumping down to 0.1 Pa. The images have the same color bar; (G, H) PL spectra taken from locations A to D in the images and change of normalized PL intensities (as compared to the original values) of locations B–D throughout the annealing and pumping process. Reproduced with permission from Nan et al. [120]. Copyright 2014 American Chemical Society.

3.3 Chemical composition modification

Ternary TMD alloys composed of two MX₂ compounds with controllable concentrations show composition-dependent band structure, providing an effective strategy for tuning optical properties [66], [74], [121], [122], [123], [124], [125], [126], [127]. As mentioned earlier, the electronic properties of TMDs range from insulating, semiconducting, to metallic, due to the d-electron counts related to chemical compositions of TMDs. A necessary bandgap for optoelectronic applications requires at least one of the alloyed material is semiconducting. Thus, alloys are investigated either between metals and semiconductors, or semiconductors to semiconductors. For successfully alloying different materials, lattice match and bond distance are crucial [124]. Figure 4A shows how these TMDs meet the requirements by plotting theoretical bandgap differences with lattice mismatch [124]. The shaded area in the upper left corner corresponds to the largest difference in bandgaps (metal–semiconductor alloys), while the smallest lattice mismatch, potentially indicating the most suitable compound pairs for bandgap-tunable alloying. In this metal–semiconductor–type alloys, the represented metal is VX₂, whereas the represented semiconductor is MoX₂ or WX₂. The shaded area in the lower left corner corresponds to the semiconductor–semiconductor alloys with small lattice mismatch and moderate bandgap variation with concentration. The represented semiconductor–semiconductor–type alloys are Mo–W dichalcogenides. In general, alloying TMDs within the same transition metal group or different chalcogenides are expected because they have the same lattice symmetry, as well as only small lattice mismatches. Figure 4B shows the calculated bandgaps for two representative types of TMD alloys (Mo_1−xW_xS₂ represents alloys of the same transition metal group, and MoSe_2(1−x)S_2x represents alloys of different chalcogenides) at different concentrations [124]. It shows significant modulation of bandgaps by composition concentrations, expected to be observed in optical absorption or emission experiments.

Figure 4:

Chemical composition tuning.

(A) Lattice constant matching for alloying pairs of 2H TMDs based on LSDA lattice constants. Triangles designate oxides; squares, sulfides; circles, selenides; and diamonds, tellurides. (B) Calculated bandgaps in Mo_1−xW_xS₂ (downward triangles, red) and MoSe_2(1−x) 2_x (upward triangles, blue) alloys as a function of concentration x. The bandgaps in the thermodynamic ground states are shown with larger symbols. (A) and (B) reproduced with permission from Kutana et al. [124]. Copyright 2014 The Royal Society of Chemistry. (C) PL spectrum of the complete composition MoSe_2(1−x)2_x nanosheets and a typical PL mapping of a single ternary nanosheet (the inset, scale bar, 7 um) excited with a 488-argon ion laser. Reproduced with permission from Li et al. [127]. Copyright 2014 American Chemical Society. (D) Left: Current–voltage characteristics at different laser powers (λ=633 nm) for five devices ranging in composition from MoS₂ (E_g=1.85 eV) to MoSe₂ (E_g=1.55 eV). The plots are enlarged to the same scale. Right: Photocurrent measured at a source-drain bias of 2 V as a function of laser power for single-layer MoS_1.2Se_0.4 (E_g=1.74 eV). The superlinear behavior is observed across compositions. Reproduced with permission from Klee et al. [128]. Copyright 2015 American Chemical Society.

Two-dimensional TMD alloys can be obtained by mechanical exfoliation of bulk alloys [122] and synthesis through physical vapor deposition [129], [130] or CVD [74], [131], [132], stimulating experimental exploration of bandgap engineering by chemical composition control. It is characterized by the combination of Raman and photoluminescence spectrum that 2D TMD alloys show tunable optical properties modified by the composition of alloys [66], [74], [122], [126], [127], [129], [131]. Figure 4C shows the photoluminescence spectra of the MoSe_2(1−x)S_2x nanosheet synthesized by CVD, with the spectral peaks continuously shifted from 668 nm (for pure MoS₂) to 795 nm (for pure MoSe₂) [127]. It indicates the bandgap of 2D TMD alloys can be continuously tuned through fully control of compositions. The 2D TMD alloys with continuously tunable bandgaps can have application in photodetection application, such as modulating cutoff detection wavelengths. An experimental study on photodetector based on MoSe_2(1−x)S_2x shows significant decrease of the photocurrent at fixed wavelength for Se-rich alloys compared to S-rich ones along with decreased diffusion length of photogenerated carriers (Figure 4D) [128]. An unusual superliner dependent of photocurrent on light intensity has also been observed. Both theoretical and experimental investigations show 2D TMD alloys possess the potential for tunable optoelectronics [133].

3.4 Foreigner species intercalation

Intercalation of metal atoms, molecules, and ions into 2D-layered TMDs has been widely studied and demonstrated to be an effective method for tuning electronic and optical properties of the host materials [21], [29], [30], [31], [32], [33], [34], [134], [135], [136]. Intercalation is a chemical process to reversibly insert various foreign species, such as zero-valent metal atoms, ions, and organic molecules, into the interlayer gaps of a host material (Figure 5A). Various intercalation methods, such as ion exchange intercalation, electrochemical intercalation, and zero-valent intercalation, have been developed for fundamental researches and technical applications [32]. The weak interlayer van der Waals interaction in TMDs leads to relatively large interlayer gap, showing that the intercalation into TMDs is easily achieved and affords a convenient way to modify and tune the electronic and optical properties for future optoelectronics applications.

Figure 5:

Foreigner species intercalation.

(A) Schematic for intercalation in the layered TMD materials, where guest species can be inserted into the interlayer gap by diverse methods. Reproduced with permission from Jung et al. [32]. Copyright 2016 the Partner Organisations. (B) Left: Schematic of the band structure of 2H-MoS₂ at the K-point in the Brillouin zone without Li intercalation. Optical transitions due to the A and B excitons are depicted. These transitions possibly cause absorption peaks in 2H-MoS₂. Right: Schematic of the band structure of 1T-Li_x=1MoS₂ at the K-point in the Brillouin zone after Li intercalations. 1T-Li_x=1MoS₂ shows a metallic behavior as the conduction band, and the valence band becomes overlapped, and the Fermi level lies in an incompletely filled band. This metallic band structure with no bandgap substantially reduces the light absorption. (C) Left: Experimental setup for electrochemical tuning of Li concentration in MoS₂. Middle: Optical transmission spectra of MoS₂ flakes with thicknesses of 9 nm before (red) and after (blue) Li intercalations. Right: Thickness dependence of MoS₂ optical transmissions at 500 nm. The results showed enhancement in transmission after Li intercalation. (B) and (C) reproduced with permission from Xiong et al. [31]. Copyright 2015 American Chemical Society. (D) Top: Active control of PL in the 2D MoS₂ nanoflakes. PL images of the nanostructured MoS₂ film at the intercalating voltages of 0 V, −2 V, −4 V, −6 V, −8V, −10 V, and −12 V. Bottom: In situ PL intensity change of a selected area “1” in the MoS₂ film under different intercalating/deintercalating voltages of ±2, ±4, ±6, ±8, ±10, and ±12 V in the 0.1 M LiClO₄ solution (blue line) at the step duration of 60 s, whereas a selected uncoated area “2” within the substrate area is chosen as the reference (red line). Reproduced with permission from Wang et al. [134]. Copyright 2013 American Chemical Society. (E) Left: PL images of films made of drop-casted 2D MoS₂ in initial and at the intercalating voltages of −4, −6, −8, and −10 V (at a broadband excitation light source covering wavelengths ranging from 400 to 500 nm). The changes in PL of 2D nanoflakes at different applied voltages are presented. Yellow light is emitted from the 2D MoS₂ nanoflakes at the initial stage. When a voltage of −4 V is applied to the sample, the PL level is only slightly reduced. When the applied voltage is decreased from −6 to −10 V, PL intensity drops dramatically, and there is a near-total photoquenching at the voltage of −10 V. Right: UV-vis absorbance spectra of the 2D nanoflakes under different electrochemical forces, in which the pristine 2D MoS₂ is used as the differential reference. Main absorption peaks in the experimental are observed at 332, 380, 430, 498, and 714 nm. The wavelengths for theoretical estimates of plasmonic peaks as a function of intercalated Li⁺ ions are listed along dotted lines. An intercalated fluoride-doped tin oxide (FTO) peak is observed at 325 nm, showing a minimal interference. Reproduced with permission from Wang et al. [135]. Copyright 2015 American Chemical Society.

It has been reported that intercalation of alkali metal ions (Li⁺, Na⁺, K⁺) into 2D-layered TMDs induces structural phase change and alters electronic properties, which result in modified optical properties [29], [30], [31], [33], [134], [135]. A commonly studied example is the 2H-to-1T phase transition in Li-intercalated 2D-layered MoS₂, which consequently results in alteration of electronic band structure. It is predicted by calculations that the 2H phase of the original 2D MoS₂ is semiconducting with a sizeable bandgap, whereas the 1T phase of Li-intercalated MoS₂ is metallic with overlapped conduction and valence bands (see schematic in Figure 5B) [31]. This consequently affects many optical properties and opens up new opportunities for optoelectronics. Xiong et al. [31] observe that without bandgap in 1T phase the light transmission in ultrathin MoS₂ layers after Li intercalation is substantially enhanced due to the light absorption reduction (Figure 5C). Wang et al. [134] report large modulation of photoluminescence of liquid phase–exfoliated MoS₂ nanoflakes by electrochemical controlled ion intercalation, showing significant application prospect in bio-optical sensors, as well as optical modulators or switches (Figure 5D). The observed modulation of photoluminescence is attribute to the lattice expansion, as well as the transition from originally semiconducting 2H into metallic 1T crystal phase after ion intercalation. In addition, they successfully achieved plasmon resonances of 2D MoS₂ nanoflakes in the visible and near-UV regimes by the electrochemical intercalating Li⁺ ions into 2D MoS₂ nanoflakes, providing a great application potential for future plasmonic biosensing and optical systems (Figure 5E) [135]. The emergence of plasmon resonances is ascribed to the formation of semimetallic states controlled by doping level of intercalated Li⁺. The phase change induced by alkali metal intercalation may be attributed to the d-electron count alteration as a result of charge transfer from s orbital of the alkali metal to the d orbital of the transition metal.

Phase change from 1T to 2H has been observed in TaS₂ after Li intercalation [137]. Except alkali metals, molecular intercalation in 2D MoS₂ also induces phase transition from semiconducting 2H to metallic 1T phase [34]. Local phase transition has been proposed to achieve 2H–1T hybrid structures [138]. The influence of intercalation on optical properties is not limited to 2D TMDs. Many nontransition metal chalcogenides are demonstrated to show evident modulated optical properties after intercalation. Bi₂Se₃ nanoplates intercalated by zero-valent Cu atoms have been studied to show dramatic enhancement of optical transmission, due to the increase in the effective bandgap caused by the increased free electron density introduced via intercalation of Cu atoms [139]. Organic molecules–intercalated Bi₂Se₃ nanoribbon has been reported to show a wide tunability of the photonic and plasmonic properties of the host material [140], [141]. In short, the ability to modify the electronic structures and optical properties of 2D TMDs by intercalation provides a new effective control manner for future tunable optoelectronics and attracts a new round of attention.

3.5 Strain engineering

Strain engineering has both theoretically and experimentally proved to be an effective approach to continuously tune the bandgap of TMDs, which subsequently modulates the electronic, optical, and photonic properties of TMDs [84], [85], [86], [87], [88], [89], [90], [91], [142], [143], [144]. The ability to continuously tune the band structure of a TMD is most desirable for optic and photoelectronic applications, motivating a vast array of theoretical works that study the strain effect on band structure. Applying strain on a crystal will cause lattice stretching or compressing in different directions, resulting in the change of lattice constant. This causes the change of overlapping and hybridization of electron orbital and can be responsible for the modified electronic and optical properties. In order to study the influence of mechanical strains on the monolayer group VIB TMDs MoX₂ and WX₂, Johari and Shenoy [84] performed first-principles density functional theory based calculations and revealed that both tensile and shear strain continuously reduce the bandgap, whereas tensile strain decreases the bandgap more rapidly (Figure 6A). Specially, they found that application of biaxial tensile strain causes a direct-to-indirect bandgap and semiconductor-to-metal transition of the monolayer TMDs. Another theoretical work by Yun et al. [85] found, for the monolayer of MoX₂ and WX₂, the tensile strain reduces the bandgap, whereas the compressive strain enhances bandgap. For those beyond group VIB TMDs, density functional theory calculations suggest different strain effects. For example, the bandgap of group IVB TMDs (TiX₂, ZrX₂, HfX₂) is predicted to increase with the tensile strain. All these calculations suggest a reversible and flexible manner to engineer the electronic and optic properties of TMDs.

Figure 6:

Strain engineering on tuning electronic structures.

(A) Bandgap of monolayer TMDs with respect to strain, ε, which varies from 0 to 10%. Strain is applied to the optimized structures (ε=0) through various approaches, such as uniaxial expansion in x-direction (xx), y-direction (yy), homogeneous expansion in both x- and y-directions (xx+yy), expansion in x-direction and compression in y-direction (xx−yy), and compression in x-direction and expansion in y-direction (yy−xx) with the same magnitude of strain. The first three strain profiles correspond to tensile strain, whereas the latter two represent pure shear strain. Left: MoS₂. Right: WS₂. Reproduced with permission from Johari and Shenoy [84]. Copyright 2012 American Chemical Society. (B) Left: Schematic diagram of a cantilever used in the experiment. It is made of a square-shaped PMMA substrate of length L=9.5 cm and thickness t=0.245 cm. MoS₂ samples were deposited near a corner of the substrate. Strain was applied on the samples by clamping one edge (gray) and bending the opposite edge of the substrate. Strain on the sample was calibrated by the sample location x and the cantilever deflection δ as described in the text. Insets: Atomic structure of monolayer MoS₂ and optical image of one sample with both monolayer (1L) and bilayer (2L) regions. The scale bar is 5 μm. Right: Absorption (left panel) and PL (right panel) spectrum of a monolayer MoS₂ sample under tensile strains up to 0.52% along the zigzag direction. The dashed blue and red lines are guide to the eye of the red shift of the peaks. Reproduced with permission from He et al. [87]. Copyright 2013 American Chemical Society. (C) Top left: Schematic of an inhomogeneously strained membrane ribbon with varying width λ (dashed line). The two electrodes mechanically impose a vertical displacement on the central region of the membrane. Without the two electrodes, the equilibrium geometry of the membrane is flat. Top right: Schematic device setup for elastic strain–engineered artificial atom (not drawn to scale). Notice that although the bulk of the electrodes is metallic (e.g. transparent conducting oxide), they are coated with semiconducting buffers to facilitate selective quasi-particle collection. Bottom left–right: Three broad-spectrum solar energy funneling mechanisms arising from a different band bending and exciton binding profile in the strain-engineered semiconducting membrane. Reproduced with permission from Feng et al. [145]. Copyright 2012 Macmillan Publishers Limited.

In experiments, mechanical strain can be applied through bending of samples on flexible substrates or via lattice mismatch between substrates and epitaxy layers. Using a cantilever device to apply a uniaxial tensile strain, it has been observed that the A and B resonance peaks in both absorption and photoluminescence spectrum, corresponding to direct bandgap transitions, show a similar red-shift rate with increasing the tensile strain in both monolayer and bilayer MoS₂ (Figure 6B) [87], whereas the I peak related to the indirect bandgap transition in bilayer MoS₂ has a larger red-shift rate. This experimental result indicates applying a uniaxial tensile strain can continuously reduce the bandgap of atomic thin MoS₂ layer, which agrees well with the first-principles calculations. On the contrary, through a piezoelectric substrate, the compressive strain was applied to trilayer MoS₂, and it has been demonstrated to increase the bandgap, manifested as both the photoluminescence peak and Raman modes are blue-shifted [88], [146].

The effective strain tunability of the electronic structure of TMDs promises a wide range of applications in optoelectronics. Applying local tensile strain, TMDs can exhibit spatially varying bandgap [86], and subsequent confinement potentials for photoinduced excitons can be generated to trap excitons for quantum optics, photodetection, and photovoltaics because of the exciton funnel effect that excitons will move to lower bandgap regions induced by locally applied tensile strain before recombining. This funnel effect has been utilized to design a photovoltaic device by introducing inhomogeneous elastic strain [145]. An “artificial atom” made of a nanoindented MoS₂ monolayer (Figure 6C) has been proposed and predicted to be able to absorb a broad range of the solar spectrum along the elastic strain gradient and concentrate photoexcited charge carriers. Introducing compressive strain, photoresponsibility is controllably modulated by the piezo-phototronic effect in a flexible MoS₂/WSe₂ van der Waals photodiode, leading to excellent strain-tunable photodetection performance [52]. Benefiting from the various manners of applying mechanical strains and the richness of TMDs, the development of novel optoelectronics is on the way.

3.6 Heterostructure constructing

Forming heterostructures of materials is a common strategy to modify the electronic band structures of the materials [147]. The heterostructures composed of different 2D TMDs stacked along the out-of-plane direction (vertical heterostructure) or in-plane direction (lateral structure) [148] can exhibit significant different electronic and optical properties compared with each component material, having great potential for atomically thin optoelectronic and photovoltaic applications [75], [149], [150], [151], [152]. For a heterostructure, many aspects, such as the stacking order, lattice mismatch, component layers, and so on, can affect its electronic structure. For example, it is theoretically found that heterobilayers composed of particularly stacked different monolayer TMDs exhibit a direct bandgap, where electrons and holes are physically separated and localized in different layers (Figure 7A) [153]. The predicted direct bandgaps of the heterobilayer TMDs ranging from 0.79 to 1.15 eV are much smaller than the direct bandgaps of monolayer WS₂ (2.1 eV) and MoS₂ (1.8 eV), which extends the optical applications in the infrared range. Additionally, the band offset calculations show both the conduction band minimum and valence band maximum of WX₂ are higher than those of MoX₂ due to the higher energy of the 5d orbital of W than that of the 4d orbital of Mo (Figure 7B) [76]. As a result, MoX₂/WX₂ heterostructures will form a type II band alignment with the conduction band minimum and valence band maximum located in MoX₂ and WX₂, separately. A type II band–aligned heterostructure, in which free electrons and holes prefer to spontaneously stay at separate positions (Figure 7B), is suitable for optoelectronics and light harvesting. Accordingly, TMD heterostructures enable bandgap engineering, which could further offer more flexible choices for the construction of novel high-performance optoelectronic devices.

Figure 7:

Optoelectronics of 2D TMD heterostructures.

(A) Left: Bilayer formed by overlapping WSe₂ on WS₂ with stacking type B denoted as bilayer (WS₂, WSe₂, B) and bilayer formed by placing WSe₂ on MoS₂ with stacking type A denoted as bilayer (MoS₂, WSe₂, A). In staking type B, the transition metal atoms are on top of the chalcogen atoms. In stacking type A, the chalcogen atoms are on top of each other. The dotted lines indicate the alignment of atoms between layers, and the arrows correspond to the distance dS-Se between chalcogen atoms of different layers. Right: Type 2 family band structure showing the direct bandgap of bilayer (WS₂, WSe₂, B), bilayer (WSe₂, MoS₂, A), and the infinite number of layers case: Crystal (WS₂, WSe₂, A). The red line indicates the top of the valence band. Reproduced with permission from Terrones et al. [153]. Copyright 2013 Springer Nature Limited. (B) Left: Calculated band alignment for MX₂ monolayers. Solid lines are obtained by PBE, and dashed lines are obtained by HSE06. The dotted lines indicate the water reduction (H⁺/H₂) and oxidation (H₂O/O₂) potentials. The vacuum level is taken as zero reference. Middle and right: Charge densities of VBM (middle) and CBM (right) states for monolayer WX₂-MoX₂ lateral heterostructures with common X. Reproduced with permission from Kang et al. [76]. Copyright 2013 American Institute of Physics. (C): (i) Bottom left: Schematic diagram of a van der Waals–stacked MoS₂/WSe₂ heterojunction device with lateral metal contacts. Top: Enlarged crystal structure, with purple, red, yellow, and green spheres representing Mo, S, W, and Se atoms, respectively. Bottom right: Optical image of the fabricated device, where D1 and D2 (S1 and S2) indicate the metal contacts for WSe₂ (MoS₂). Scale bar, 3 μm. (ii) Schematic illustrations of exciton dissociation (top) and interlayer recombination (bottom) processes. Top: Horizontal and vertical arrows represent charge transfer and intralayer recombination processes, respectively. Bottom: Red and blue arrows indicate Shockley–Read–Hall (SRH) and Langevin recombination processes, respectively. (iii) Photoresponse characteristics at various gate voltages under white-light illumination. Inset: Color plot of photocurrent as a function of voltages V_ds (x axis) and V_g (y axis). The dashed line represents the profile of short-circuit current density J_sc at V_ds=0 V. (iv) Top: Photocurrent map of the device presented in (i) for V_ds=0 V and 532-nm laser excitation. The junction area and metal electrodes are indicated by dashed and solid lines, respectively. Scale bar, 3 μm. Bottom: Photoluminescence spectra measured from the isolated monolayers (blue curve for MoS₂; red curve for WSe₂) and the stacked junction region (brown curve). (v) Photoluminescence spatial maps for emission at 1.66 eV (top) and 1.88 eV (bottom), corresponding to direct gap transitions of monolayer WSe₂ and MoS₂, respectively. The junction area is indicated by dashed lines. Scale bars, 3 μm. Reproduced with permission from Lee et al. [46]. Copyright 2014 Macmillan Publishers Limited.

Experimentally, TMD heterostructures are widely available through sequential transfer or various growth methods [149]. Stimulated by the graphene stacked on h-BN showing ultrahigh mobility, concerted efforts are expected to explore the modified physical properties of heterostructures formed by different 2D layers. In case of TMDs, high-quality heterostructures have been demonstrated not only to improve electrical transport properties, but also to enhance the optoelectronic performance [41], [44], [46], [50], [78], [80], [81], [154]. Researchers observed significant photoluminescence quenching and gate-tunable efficient photocurrent generation in a vertical heterostructure with the n-type monolayer MoS₂ transferred onto the top of a p-type monolayer WSe₂ (Figure 7C) [46]. The observed gate-tunable photovoltaic response originates from spontaneous dissociation of photogenerated excitons into free electrons and holes in different layers and subsequent tunneling-assisted interlayer recombination driven by large band offsets in a heterostructure with type II band alignment. The charge transfer processes in photoexcited MoS₂/WS₂ heterostructures have been proved to be ultrafast through photoluminescence mapping and femtosecond pump–probe spectroscopy [78]. It is found that hole transfers from the MoS₂ layer to the WS₂ layer within 50 fs after optical excitation.

Besides the heterostructures composed of two different TMDs, there are a lot of heterostructures formed by TMDs with other 2D materials, showing tunable optoelectronic properties [49]. The first to mention is the TMDs/graphene heterostructures [41], [44] because their distinct properties could be utilized to realize novel functionalities. The high mobility of graphene means fast response rate, whereas the large direct bandgap of TMDs results in strong light absorption as mentioned earlier. It is observed the strong light–matter interactions in TMDs/graphene heterostructures lead to enhanced photon absorption and photocurrent generation, allowing development of extremely efficient flexible photonic and optoelectronic devices [41]. Besides, ultrathin p-GaTe/n-MoS₂ heterostructure, which poses the type II band alignment, is constructed and proved to have high photovoltaic and photodetecting performance [50]. The light-emitting diodes based on n-type monolayer MoS₂ and p-type Si stacked vertically are fabricated to realize rectification and light emission from the entire surface of the heterojunction [155], [156].

It is worth to mention that geometrical alignments between different layers in heterostructures also play an important role in tuning the electronic structure. Heterostructures, including TMD bilayers, with well-defined interlayer twist angle can be obtained via transfer one monolayer onto the top of another. Experimental and theoretical study found that the interlayer twist angle has a strong influence on the indirect optical transition energy and second-harmonic generation [81]. As the interlayer twist affects the interlayer distance, which determines the interlayer interactions, band structures engineering by tuning twist angle is available for optoelectronic applications.