CRYSTALLINE AND NANOSTRUCTURED MATERIALS BASED ON TRANSITION METAL DICHALCOGENIDES: SYNTHESIS AND ELECTRONIC PROPERTIES

Abstract

The article reviews the current state of research of the synthesis, structural, electrophysical and spectroscopic properties of layered transition metal dichalcogenides (TMDs) of group 4-7 elements (M = Zr, Hf, Nb, Ta, Mo, W, Tc, Re; Q = S, Se, Te) depending on their morphology and modification. The changes in the properties of these compounds are considered upon the transition from crystalline materials to nanostructured and monolayer samples. We discuss “fine tuning” of electronic properties by introducing heteroatoms into the TMD structure: intercalation, when a heteroatom is introduced between layers, and isovalent and/or non-isovalent substitution in cationic and anionic sublattices. Practical applications of fine tuning of the properties of TMD based materials are discussed.

INTRODUCTION

The origination and further development of modern branches of technology such as solid-state microelectronics, optoelectronics, nanoelectronics stimulate the search and study of new promising materials exhibiting a wider range of properties than those of already employed materials. The discovery of graphene drew attention to other related layered materials, e.g. transition metal dichalcogenides MQ₂ (TMDs). Even though many of these compounds have been already widely studied, they nowadays attract attention, along with other similar layered systems, due to the general interest to nanomaterials and nanotechnologies. Exceptional electronic, mechanical, thermal, and chemical properties of layered materials are now being explored on nanostructured samples as possible photo- and electro-catalysts, including in hydrogen evolution reactions [1-3], gas sensors [4], optoelectronic sensors [5], electrode materials for chemical current sources (batteries and accumulators [6, 7], including solar energy converters), solid lubricants, etc. [8, 9].

Modification of known materials is one way to construct new materials with improved properties finely tuned for each specific task. The uniqueness of TMD materials is related to their structure and provides wide possibilities for chemical modification. The first possibility is due to their layered structure: preparation of intercalated systems where a heteroatom or a molecule is introduced between the layers (e.g., intercalation with lithium and other alkali metals [10], copper [11], organic molecules [12]), as well as splitting and manufacturing of few-layer and monolayer samples [13-17]. The second “source of possibilities” is the substitution of metal atoms in the cationic sublattice [8] and chalcogen atoms in the anionic TMD sublattice [18].

This review is devoted to modified TMDs MQ₂ with M atoms from groups 4-7 and periods 2 and 3 (M = Zr, Hf, Nb, Ta, Mo, W, Tc, Re; Q = S, Se, Te), their preparation, structure, properties. Since MoS₂ is currently a popular “model TMD” that has been discussed in more than 26 thousand publications for the past five years (according to Web of Science), we consider it only briefly in this review and focus mainly on other TMDs.

SYNTHESIS OF TMDs AND MODIFIED TMDs

The development of methods for the synthesis of crystalline and polycrystalline TMD samples also affected the preparation of their crystalline intercalates or compounds doped with metals or chalcogens. The growing interest in micro- and nano-sized TMD materials (and the development of available methods for such studies) expanded the synthesis of such compounds up to solution methods and modifications of chemical vapor deposition (CVD). The development of synthesis methods for the fabrication of nanometer thick (“few-layer”) TMD crystallites is also interesting in terms of obtaining new physical characteristics upon the transition from bulk crystals to nanocrystals composed of a small number of molecular layers as well as to mono- and oligolayers.

Synthesis of crystalline and polycrystalline TMDs

Crystals and crystalline TMD powders can be prepared from elements in close to stoichiometric ratios using the oxidation of metals by chalcogens. The decrease of oxidation potentials from sulfur to tellurium requires increased synthesis temperatures in this series. The methods for the TMD synthesis for all metals considered in the present article was summarized in [19]. To some extent, chalcogen atoms in these reactions act as transport agents, and relatively high synthesis temperatures (> 700 °C) exclude the formation of trichalcogenides MQ₃. In the case of group 4 and 5 metals, the pressure of chalcogen (primarily, sulfur and selenium) in the TMD synthesis is important to exclude the formation of non-stoichiometric dichalcogenides. It is known that low sulfur pressure leads to the formation of so-called self-intercalates M_1+xS₂: 3R–Nb_1+xS₂, x > 0.07 [20], Ti_1+xS₂, 0.005 < x <0.014 [21] or 0.20 < x < 0.33 [22]. To prepare stoichiometric compounds MQ₂, pressure is controlled using chalcogen excess (up to 10%). Other transport agents are also used: iodine, bromine [23], Ar/H₂ mixture [24], phosphorus [20], phosphorus chalcogenides P₂S₅ and P₂Se₅ [25], and halides of chalcogens (SeCl₄) and metals (TaCl₅ [26]). The preparation of well crystallized ReS₂ and ReSe₂ from elements requires high temperatures (1000-1150 °C) and long-term syntheses in ampoules using transport agents [25]. The use of rhenium oxides or ammonium perrhenate in the synthesis of ReS₂ lead to the formation of a poorly crystallized product with an admixture of oxides [27]. It was observed that semiconductor properties of ReS₂ and ReSe₂ depend on the nature of the transport agent: n-type and p-type semiconductor crystals ReS₂ and ReSe₂ were obtained in case of bromine and iodine, respectively.

The use of selenium and tellurium melts to prepare TMD crystals with a size of ~1 mm was described in [28]. Gradual evaporation of chalcogen into the cold end of the ampoule lead to the formation of crystals NbSe₂, TaTe₂, ReSe₂ and chalcogen-mixed crystals Nb(S_1–xSe_x)₂.

Synthesis of TMD based micro- and nanomaterials

The methods of preparing TMD based micro- and nanomaterials can be represented by two groups: 1) “bottom-up”, i.e. using molecular or atomic precursors, 2) “top-down”, including the stage of exfoliation of crystalline objects.

“Bottom-up” synthesis of TMD nanomaterials. TMD materials of group 4-7 elements are prepared by various modifications of the CVD method since chemical vapor deposition provides reproducible synthesis of well-crystallized nanomaterials with controlled thickness and lateral dimensions. The sources of transition metals are usually their oxides or halides; the source of chalcogens are chalcogens as such or their organic compounds. Nanocrystals MS₂ (M = Zr, Hf) are successfully prepared using chlorides MCl₄ due to their low melting points (compared to those of oxides) and the absence of oxygen in the resulting products (ZrS₂ [29], HfS₂ [30]). Few-layer 3R–NbS₂ particles [31] (starting with monolayer ones) were prepared using Nb₂O₅ and sulfur in the presence of sodium chloride in the H₂/Ar atmosphere. Nanocrystals of NbS₂ and TaS₂ can also be obtained from corresponding metal pentachlorides MCl₅ and sulfur in a flow-type reactor in the presence of microcrystalline sodium chloride or h-BN particles used as the substrate (3R–NbS₂ modification is formed) [24]. When using metal oxides Zr, Hf, Nb, Ta, alkali metal halides are added; apparently, formation of metal oxohalides increases mass transfer during the synthesis [32] to increase the formation rate of TMD nanocrystals. Thin-layer crystals of disulfides of group 6 metals are commonly prepared using metal oxides (MO₃: M = Mo [33], W [34]) and sulfur (or hydrogen sulfide). Nanocrystals prepared from MO₃ oxides have impurities in the form of oxides, which is unacceptable for some applications, e.g. in optoelectronics. According to XPS data, using carbonyls M(CO)₆ (M = Mo, W) in TMD synthesis also leads to the formation of metal oxide impurities (e.g., in the synthesis of WSe₂ from W(CO)₆ and (CH₃)₂Se [35]). In some works, molybdenum dichalcogenides were successfully prepared from molybdenum halides: for example, MoS₂ nanoparticles (NPs) were fabricated from MoCl₅ and S [36], MoSe₂ NPs were fabricated from MoCl₅ and ((CH₃)₃Si)₂Se [37]. WS₂ was prepared from hexafluoride WF₆ in an atmosphere of H₂S/H₂ plasma [38]. Few-layer samples ReS₂ can be obtained by CVD either from ammonium perrhenate and sulfur [39] (or hydrogen sulfide [40]) or from metallic rhenium. In the first case, the synthesis was carried out at a rather low temperature (460 °C), but the resulting ReS₂ nanocrystals had an admixture of rhenium oxide (the material was used in the catalysis of crude oil hydrosulfonation). The use of metallic rhenium in the synthesis requires high temperatures (> 1000 °C) as well as transport agent (bromine or iodine). The authors of [41] fabricated ReS₂ NPs at 700 °C by lowering the temperature in the course of the synthesis from an eutectic rhenium–tellurium mixture (1:6) with the melting point 430 °C. Nanomaterials of transition metal ditellurides are also prepared by CVD. Review [42] summarized the synthesis parameters of thin-layer samples of group 4-6 transition metal ditellurides MTe₂: such particles were mainly prepared from tellurium and metal oxides in the presence of NaCl, KCl or KI at 650-800 °C in an Ar/H₂ flow.

Few-layer TMD samples are prepared from various metal and chalcogen (usually, sulfur) sources, some of which have complex compositions. Thus, the authors of [43] obtained a series of microcrystalline NbS₂ films using atmospheric pressure chemical vapor deposition (APCVD) from NbCl₅ and several sources of sulfur ((SiMe₃)₂S, t-Bu₂S₂, t-BuSH, HSCH₂CH₂SH) at 250-600 °C. Depending on the sulfur-containing initial substance and the synthesis temperature, the obtained NbS₂ samples had two polymorphic modifications: 1T (S(SiMe₃)₂ or t-Bu₂S₂ at 500-600 °C) and 3R (t-BuSH or HSCH₂CH₂SH at 500-600 °C). X-ray amorphous films NbS₂ and TaS₂ were prepared by thermolysis of aerosols of complexes [M(St-Bu)_5–x(NMe₂)_x] (M = Nb, Ta) [44]. Films 2H–NbSe₂ were fabricated by APCVD at 600 °C using t-Bu₂Se and NbCl₅ in an atmosphere of nitrogen [45].

High vacuum conditions can be used to obtain metal and chalcogen vapors which form TMD NPs after being deposited on a substrate. Ultrathin epitaxial NbSe₂ layers were prepared on a Se-terminated GaAs(111) in a high-vacuum chamber in the temperature range 425-530 °C, and, depending on the synthesis temperature, the samples contained crystal modifications of NbSe₂ 2Hb+3R (420 °C and 500 °C) or 2Ha+2Hb (530 °C) [46]. Oriented monolayers TaSe₂ [43] were obtained on a gold surface upon simultaneous evaporation of selenium and tantalum at 600 °C; the growth of NbSe₂ layers was conducted at 300-600 °C [47].

Since CVD allows controlling the thickness of nanocrystals, and chemical compositions of their surface are very close to nominal formulas MQ₂, these thin-layer semiconductor crystals are considered promising in digital electronics and optoelectronics. Heterolayer crystals can be prepared by combining the multitude of studied TMDs. Since each material has unique electronic structures and properties, the fabrication of heterostructures could exhibit improved properties or even new properties compared to individual TMDs. The authors of review [48] discuss TMD heterostructures, their CVD synthesis and growth features, structure, and prospects for the use in nanoelectronics.

TMDs were also prepared in the form of inorganic fullerenes (IFs) and nanotubes. A series of multi-layer IF–MS₂ (NbS₂ [49], TaS₂ [50], MoS₂ [51], WS₂ [52], ReS₂ [53]) were synthesized in reactions of metal chloride vapors (MCl₄ or MCl₅) with hydrogen sulfide at 550-750 °C and subsequent annealing in H₂S. Solvothermal synthesis of IF–MoSe₂ from ammonium molybdate and sodium selenite [54] and IF–WSe₂ from WO₃ and selenium vapor [55] was also reported. Solid solutions of NPs in the form of fullerenes can be prepared from initial substances containing another metal: Mo_1–xRe_xS₂ and W_1–xRe_xS₂ [23] and W_1–xNb_xS₂ [49]. Besides, binary and doped TMDs were prepared in the form of nanotubes: e.g., MoS₂ and WS₂ nanotubes were obtained upon the decomposition of ammonium tetrathiomolybdate (or tetrathiotungstate) in hydrogen at 1000-1200 °C [56]; NbSe₂ nanotubes were prepared by irradiating NbSe₂ powder by a beam of electrons [57].

Solution methods for the preparation of TMD NPs include colloidal and sol-gel methods. A characteristic feature of such TMD syntheses is high temperature, for example, when annealing the resulting gel or when carrying out colloidal synthesis. In some cases, the size of TMD particles can be controlled using the sol-gel method. Thus, NbS₂ particles can be obtained from a mixture of Nb(OEt)₅ with benzenesulfonic acid (precipitate of niobium ethoxysulfide is formed) [58] or from the thiolate complex Nb(2,6-Me₂C₆H₃S)₅ whose solution is bubbled by H₂S (sol-gel stage) [59] and subsequent annealing at 700-800 °C in H₂S. When preparing NbS₂ microparticles, the synthesis temperature can be also diminished by using the colloidal method: 10-60 nm thick nanoparticles were obtained by the reaction of NbCl₅ with CS₂ in oleylamine at 280-320 °C. The authors of [60] described colloidal synthesis of a series of NbS₂ nanoparticles with various shapes (Fig. 1) from NbCl₅ solutions in mixtures of oleic acid and oleylamine with carbon disulfide at 300 °C. The shapes of the particles (mono- or oligolayers, plates, hexagonal nanocrystals, nanorods, and nanospheres) depended on the ratio of oleic acid, oleylamine, and CS₂.

TMD nanomaterials of group 6 elements can also be obtained by the sol-gel method. For the study as memristors, films of 2H–MoS₂ nanoparticles were prepared from ammonium heptamolybdate and acetamide as a source of sulfur in water in the presence of a stabilizer; the resulting gel was annealed in an atmosphere of argon at 300-400 °C [61]. Composites MoS₂/C synthesized by the sol-gel method are used as electrodes in lithium batteries [62]. MoSe₂ nanoparticles for the study as catalysts in the reaction of hydrogen evolution were prepared from MoCl₅ and diphenyl diselenide Ph₂Se₂ in an ethanol solution. The gel was mixed with carbon particles and annealed at 700 °C to obtain 200 nm large particles with a “core–shell” structure [63].

Since the sol–gel method includes a solution stage with the participation of molecular starting compounds, the salts containing various metals can be mixed before preparing the sol to obtain a dichalcogenide based solid solution. Work [64] is a logical development of the sol–gel synthesis with respect to TMDs, namely, preparation of doped dichalcogenides. The authors synthesized doped compounds with several compositions: WS₂ doped with iron or molybdenum and MoS₂ doped with tungsten or rhenium. It was shown in [65] that MoS₂ nanoparticles with a thickness of 3-8 MoS₂ layers with a partial substitution of sulfur by nitrogen (the nitrogen content in different samples was 5.8-7.6 at.%) can be obtained from MoCl₅ and thiourea CS(NH₂)₂ and subsequent annealing at 650 °C.

Fig. 1

Various forms of NbS₂ particles prepared from NbCl₅ and CS₂. The figure is cited from [60], ©ACS Applied Nano Materials, American Chemical Society, 2018.

“Top-down” synthesis of TMD nanomaterials. Exfoliation of TMDs intercalated with alkali metals (lithium, sodium, potassium) allows obtaining a solution with TMD monolayers [66] or with particles with some other number of layers (2, 3), depending on the order of the intercalate. Intercalate A_xMQ₂ (A is an alkali metal, Q is most often S) is placed in a liquid medium (for example, water) which interacts with alkali metal atoms to form hydrogen that helps to separate the MQ₂ layers from each other. There are many TMDs for which lithium and sodium (less often potassium or magnesium) intercalates have been studied: TiQ₂ (Q = S [67], Se, Te [68]), ZrS₂, ZrSe₂ [69], NbS₂, NbSe₂ [70], MoTe₂ [71], MoSe₂, WS₂ and WSe₂ [72], ReS₂ [73]. Most often, these TMDs are used as electrode materials in chemical current sources. Being exfoliated in the presence of a stabilizer, such intercalates can give high yields of solutions with monolayer samples. TMD samples obtained upon deintercalation in solution can serve as matrices for the assembly of new intercalates. Thus, MoS₂ monolayers stabilized in solution are used to prepare intercalates [74] with organic cations [75] and organometallic complexes [76]; some of them have been studied as electrocatalysts in the reaction of water splitting associated with hydrogen formation.

There is a progress in solution methods intended for the exfoliation of TMDs not subjected to chemical modification. Upon such exfoliation, the crystalline powder or TMD crystals are usually located in a liquid medium which promotes the formation of few-layer TMD particles. Similarly to other layered structures (graphite oxide, layered metal hydroxides, clays, etc. [77]), TMDs can exfoliate due to cavitation: emerging bubbles of the gas dissolved in the dispersion medium burst and create high-pressure and high-temperature regions in the space around them to destabilize the crystals of the substance. Dynamic simulations show that the collapse of a 4-5 nm bubble in an isopropanol–water mixture increases the surface temperature of the dichalcogenide (MoS₂) up to 3000 K and the pressure up to 20 GPa [78]. This effect creates a 10 GPa shear stress inside MoS₂ to cause crushing and exfoliation of the crystals. The most common ways to cause cavitation is sonication (with a frequency up to 3 MHz) or, to a lesser extent, boiling TMDs in a liquid dispersion medium. The nanoparticles resulting from such treatment show significant dispersion of lateral sizes (ten to hundreds of nanometers); the thickness of the particles can range from 2, 3 MQ₂ layers to several micrometers. Dispersion media for the preparation of TMD colloidal solutions can stabilize TMD NPs in solutions. The Hansen solubility parameter theory, which can be used to estimate the minimal dispersion energy, is used when choosing the dispersion medium with respect to TMDs [14]. Thus, for disulfides MS₂, efficient dispersion media are isopropanol, N-methylpyrrolidone, N,N′-dimethylformamide; for diselenides MSe₂ such dispersion media are ethanol and acetonitrile. For other media, stabilizers are used: pyrene for the stabilization of MoS₂ in ethanol [79], urea for the stabilization of a wide range of MS₂, MSe₂ and MTe₂ in water [80].

Synthesis of M′MQ ₂ intercalates

The synthesis of intercalates and doped TMDs allows altering their electronic structure and, consequently, their physicochemical properties by stabilizing polymorphic modifications where TMDs are active in electrocatalysis (1T–MoS₂, WS₂), improving thermoelectric properties of TMDs, increasing the superconducting transition temperature, and other properties. When atoms M′ are introduced into the space between the MQ₂ layers, the infinite layers acquire a negative charge. Table 1 summarizes the data on the synthesis and study of properties of group 4-6 TMD intercalates.

We first consider the self-intercalation phenomenon, i.e. formation of superstoichiometric compositions M_1+xQ₂ where metal atoms (group 4 or 5) constituting infinite MQ₂ layers also enter inside the interlayer space. A small number of interlayer metal atoms are usually arranged randomly but tend to get ordered for stoichiometric M_1.25Q₂. Thus, self-intercalate 2H–Nb_1.25S₂ with orderly arrangement interlayer niobium atoms ($\sqrt{3}a\times \sqrt{3}a$ lattice) was obtained by the reaction of niobium with hydrogen sulfide at 1290 °C [81]. The packing of sulfur layers is described as BBCC, and those of sulfur and niobium as …BABA*CACA**BA… The authors of [82] synthesized a series of self-intercalated selenides 2H–Nb_1+xSe₂ (0.04 < x < 0.29) in evacuated quartz ampoules at prolonged heating (800-1000 °C) followed by a long-term cooling or quenching at 0 °C. All obtained self-intercalates had larger parameters with than those of NbSe₂ (12.55 Å): from 12.56 Å (x = 0.01) to 13.02 Å (x = 0.29).

Table 1 Summary of the Synthesis, Structure, and Properties of Synthesized Intercalates ${{{M}'}_{x}}M{{Q}_{\text{2}}}$

Among TMD intercalates, lithium intercalates (less often, intercalates of other alkali metals) seem to be the most popular ones thanks to the study of lithium batteries [7]. There is also an impressive list of intercalates TiS₂, ZrS₂, NbS₂, TaS₂ with organic cations: amines, amides, phosphines [12], amino acids and polypeptides [83]; N containing organic rigid molecules MoS₂ [75, 84, 85]. We here consider ${{{M}'}_{x}}M{{Q}_{\text{2}}}$, where M′ is a transition or post-transition metal. The most common approaches to the synthesis of intercalates ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ are preparation from elements and interaction of TMDs with the intercalant metal or with its compound; the use of a transport agent e is also possible. Electrochemical intercalation is less common. The list of intercalant elements is quite wide and includes transition and nontransition metals of groups 4-10. For example, a large series of intercalates ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ (x = 1/3, 2/3, 1) with four TMD matrices MQ₂ (M = Nb, Ta; Q = S, Se) with ordered positions In, Tl, Sn, Pb, Bi were prepared in two ways: by heating mixtures of elements at 700-900 °C and by keeping initial TMDs in the vapors of intercalant metals [86]. Atoms M′ are in an octahedral environment, atoms M are in a trigonal prismatic environment, like in the initial 2H–MQ₂ matrices. Series M_xNbSe₂ (M = Bi, Sb, Pb, W, Ge) [87] was prepared by the reduction of salts containing intercalant cations, in organic solvents, and in the presence of crystalline NbSe₂. 2H–TaS₂ and 2H–TaSe₂ were intercalated by indium upon the diffusion of indium atoms from its bulky sample [88]. Finally, there is a “layer-by-layer deposition” of intercalates for the MoS₂ matrix when intercalate LiMoS₂ was exfoliated in a liquid medium, and intercalates ${{{M}'}_{x}}\text{Mo}{{\text{S}}_{\text{2}}}$ were prepared by adding salts of M′ = Fe [89], Cu, Ag, Cd, Hg, Pd, Pb [74].

Synthesis of doped TMDs

In doped compounds ${{{M}'}_{x}}{{M}_{1-x}}{{Q}_{\text{2}}}$ or $M{{{Q}'}_{y}}{{Q}_{\text{2}-y}}$, heteroatoms replace some atoms in the cationic or anionic subsystem of initial MQ₂. The methods for preparing doped TMDs depend on the form of the desired product (crystals, NPs, few-layer samples) and vary similarly to the methods of preparing binary TMDs. Crystals of doped TMDs are usually obtained by the formation of the TMD in the presence of from elements or compounds serving as the sources of dopant atoms; NPs are prepared using solution methods (sol-gel, solvothermal synthesis) (Table 2). In some works, doped few-layer TMDs are prepared by introducing atoms into defected sites of TMD samples. Substitution (doping) of atoms in both subsystems (cationic or anionic) can be isovalent or non-isovalent.

Isovalent substitution in the anionic sublattice. Chalcogen-mixed polycrystalline TMDs $M{{{Q}'}_{x}}{{Q}_{\text{2}-x}}$ (Q = S, Se, Te) are prepared from stoichiometric mixtures of from elements or corresponding dichalcogenides at the synthesis temperatures of unsubstituted TMDs. The first syntheses of such TMDs date back to 1970. Sulfides and selenides usually form a continuous series of solid solutions for 0 ≤ x ≤ 2: WS_(2–x)Se_x [122], ZrS_2–xSe_x [123], ReS_xSe_(2–x) [124]; the substitution by tellurium leads to the formation of inhomogeneity zones [122].

Non-isovalent substitution in the anionic sublattice. It was shown that MoS₂ NPs with a thickness of 3-8 layers and partial replacement of sulfur by nitrogen (the nitrogen content in different samples ranged from 5.8 at.% to 7.6 at.%) can be obtained by the sol-gel method from MoCl₅ and thiourea CS(NH₂)₂ and subsequent annealing at 650 °C [65]. There is also an interesting method of preparing phosphorus-, oxygen-, and fluorine-substituted few-layer MoS₂ crystals from a plasma source of these atoms. The fabricated few-layer MoS₂ particles were doped with phosphorus [125] upon the interaction with a PH₃/He plasma; as a result, phosphorus atoms occupied defects in the sulfur sublattice on the surface of the MoS₂ particles. The measured current-voltage characteristics of obtained particles testified that the hole-type of doping. The authors of [126] reported that P doped 3.3 at.% MoS₂ (obtained by the treatment of MoS₂/C NPs by phosphorous acid at 400 °C) had increased electrical conductivity and somewhat improved catalytic properties in the hydrogen evolution reaction. In [127], the interaction of few-layer MoS₂ samples with O and F containing plasmas led to the introduction of oxygen (fluorine) ions into the samples so that the ions took some part of electron density from the MoS₂ layers, resulting in a hole-type doping of the latter. In a similar work [128], few-layer MoS₂ samples were etched by oxygen plasma to prepare monolayers, and it was shown that predominant formation of sulfur vacancies upon such treatment leads to electron doping of MoS₂ layers.

Isovalent substitution in the cationic sublattice. Isovalently substituted TMDs are known for 4-6 group metals. They are usually prepared from elements, but also by other approaches. For example, MoQ₂ (Q = S, Se, Te) was doped with tungsten in a series of works: solid solutions Mo_xW_1–xQ₂ were obtained by direct vapor transport from elements [129], a MoQ₂ + WQ₂ mixture [130], and from halides, e.g., by the interaction of a stoichiometric mixture MoCl₅ and WCl₆ with Na₂S [131] or MoO₃ + WO₃ oxides with a chalcogen Q [132]; few-layer samples on a substrate can be prepared by CVD [133] and low-temperature solution-phase [134] methods.

Non-isovalent substitution in the cationic sublattice. Non-isovalent substitution in zirconium and hafnium dichalcogenides are mainly studied by theoretical calculations. One example of experimental non-isovalent doping is TaS₂, for which products $1\text{T}-{{{M}'}_{x}}\text{T}{{\text{a}}_{1-x}}{{\text{S}}_{\text{2}}}$ were obtained from elements using doping by the following metals: Ti, Zr, Hf, [135], Cr, Mn, Fe [136], Co, Cu [137]. For M′ = Co, Cu, quenching is needed to stabilize the metastable 1T form.

The MoS₂ nanoparticles doped with manganese [138] or iron [139] can be obtained in solvothermal syntheses from ammonium molybdates and sources of M′ and sulfur (M′ = Mn: manganese(II) acetate and CS₂; M′ = Fe: iron(II) sulfate and cysteine). Doping MoS₂ with manganese is also possible using CVD [140]. MoS₂ doped with cobalt or simultaneously with cobalt and niobium was also synthesized by bombarding MoS₂ crystals by Co⁺ and Nb⁺ ions [141]. Cobalt doped crystalline MoS₂ was prepared also from elements by chemical vapor transport (CVT) [142]. A Co_xW_1–xS₂ layer [143] was grown by CVT on tungsten oxide microtubes to fabricate catalyst particles.

Doping MoS₂ with niobium can be carried out by the synthesis from elements: crystals Nb_0.005Mo_0.995S₂ [144] were prepared by CVT at the temperature range 1050-935 °C. The fraction of niobium atoms in the MoS₂ structure increases up to 15% (our work, JSC, 2018); metallic properties increase in all cases: electrical resistance decreases with increasing Nb content.

Table 2 Summary of the Synthesis, Structure, and Properties of Synthesized Doped TMDs

The layers of Re doped MoS₂ (0.1 at.%, 0.20 at.%, 0.4 at.%) [145] were prepared by CVD from oxides ReO₃ and MoO₃ in an S/Ar flow at 700 °C. The photoluminescence spectra demonstrate that the band position depends on the doping density. MoS₂ with high rhenium content (up to 20%) [146] was prepared by CVT from elements. Contrary to the expectation that electrical conductivity would be increase by adding electron-rich rhenium (d³ vs d² for Mo⁴⁺), the crystalline samples showed stronger electrical resistance and diamagnetism due to due formation of rhenium clusters (HRTEM and magnetic measurement data), similarly to the clusters in the ReS₂ structure.

CRYSTAL STRUCTURE AND POLYMORPHS OF TMDs, MODIFIED (INTERCALATED AND DOPED) TMDs

Crystal structure and polymorphs of TMDs

The crystal structures of TMDs are packings of infinite [MQ₂]_2∞ layers interacting with each other via van der Waals contacts Q…Q. Inside the infinite layers, metal atoms are connected with six neighboring ones via bridging chalcogenide ions Q^2– located above and below the layer of metal atoms. TMDs can form several types of polymorphs depending on which coordination polyhedron of six chalcogen atoms contains the metal atom (octahedral or trigonal prismatic coordination) and on the arrangement of infinite [MQ₂]_2∞ layers relative to each other. The main polymorphic modifications and structural types are shown in Fig. 2. Metal atoms occur in an octahedral environment of chalcogen atoms in the 1T modification (tetrahedral symmetry) and in a trigonal prismatic environment in 2H and 3R modifications. The structure of Re, Tc dichalcogenides slightly differs from other TMDs. It is known that disulfides and diselenides of rhenium and technetium belong to the triclinic crystal system [124], while TcTe₂ is monoclinic. Rhenium does not form layered ditelluride; rhenium telluride Re₆Te₁₅ is a framework compound [157]. Dichalcogenides ReS₂ and ReSe₂ are isostructural, their rhenium atoms are surrounded by octahedra of six sulfur atoms and form metal clusters Re₄ linked in chains along axis b to form layers parallel to plane ab. Dichalcogenides TcS₂ and TcSe₂ are isostructural with each other. The structural motif of the layers is similar to the distorted one of ReS₂, but the packing of the layers is different. Technetium ditelluride TcTe₂ is monoclinic; the layers are parallel to the ab plane, and the packing resembles that of the high-temperature MoTe₂ form [158], but axes a and b in TcTe₂ are doubled.

The fact that the TMD crystal structure has polymorphs with different coordination environments of metal atoms and relative arrangements of infinite layers MQ₂ as well as the electronic structure of TMDs explain the appearance of a large number of works devoted to the modification of the electronic structure of TMDs to study them as possible functional materials.

Fig. 2

Main TMD polymorphs.

Structure of ${{{M}'}_{x}}M{{Q}_{2}}$ intercalates

TMDs, being layered compounds, are characterized by weaker Q…Q interactions between the layers than covalent bonds within each layer. Therefore, there have been many successful attempts to intercalate different species into the interlayer space: from lithium atoms to organometallic complexes [159] and long organic molecules [12]. The structure of intercalates ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ (M′ is a transition or post-transition metal) considered in the present work shows a number of interesting features: relative arrangement of MQ₂ layers, ordering or the degree of ordering of intercalant atoms, and consequently, the intercalate order. These characteristics are related to the changes in the unit cell parameters of these compounds compared to initial MQ₂ which can be sometimes used to determine positions of atoms M′; more reliable data are obtained by XRD. As a rule, M′ atoms in ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ intercalates are located in ordered positions between MQ₂ layers after x reaches some value. Such situation is referred to as the unit cell increase (transformation) relative to initial MQ₂ parameters.

Series of intercalates with a variable content of the intercalated metal are of the greatest interest as far as studying the crystal structures and properties of such compounds. The authors of [105] prepared a series of intercalates Cr_xNbSe₂ (0 < x < 0.5). Chromium atoms occupied octahedral voids in the interlayer space; at 0 < x ≤ 0.25, the structure remained the same as that of initial 2H–NbSe₂. At x = 0.25, the ordering of Cr atoms and the 2a₀×2a₀ superlattice (space group P6₃/mmc) were observed. At increasing chromium contents (0.25 < x < 0.5), another superstructure was found ($\sqrt{3}{{a}_{0}}\times \sqrt{3}{{a}_{0}}$, space group P6₃22). At x > 0.33, magnetic ordering occurs. According to the ⁵³Cr NMR and magnetic susceptibility data for Cr_0.33NbSe₂, the oxidation state of Cr is 4+, and there is 0.7 unpaired electrons per one Nb atom. Ordering of intercalant atoms was observed also in 2H–Fe_0.25NbSe₂ [160] and 2H–Fe_0.25TaS₂ [161].

Parameters a and c were varied in a series of intercalates ${{{M}'}_{x}}\text{Zr}{{\text{S}}_{\text{2}}}$ (0 < x < 1) M′ = Fe [92] (also M′ = Co, Ni, Cu [162]) containing iron atoms inside tetrahedral voids and exhibiting a relative arrangement of ZrS₂ layers similar to that in initial ZrS₂ (1T modification): while parameter a decreases slightly, parameter c increases at x = 0.5 (by 1%) and then gradually diminishes with increasing x. Band structure calculations and comparison of ZrS₂ absorption spectra with those of intercalates Fe_xZrS₂ suggest that intercalation increases the energy of bands p_z and ${{d}_{{{z}^{2}}}}$that are sensitive to the increase of the distance between ZrS₂ layers. According to the magnetic susceptibility data for Fe_xZrS₂ intercalates, their iron ions have a 2+ charge and, presumably, form clusters $\text{Fe}_{2}^{2+}(\text{Z}{{\text{r}}^{\text{3}+}}\text{S}_{2}^{2-})$. This system is unusual because iron atoms occupy tetrahedral voids in the interlayer space while a more common situation is octahedral or trigonal prismatic voids, which would be indicative of the 1T to 3R polymorph transition.

In a more difficult case, two types of atoms can be intercalated into the dichalcogenide matrix. In compound 6R–Cu_xTa_1+yS₂ (x ≈ 0.23, y = 0, 0.06) [163], intercalated atoms of tantalum and copper are located in different interlayer spaces: tantalum and copper atoms occur in octahedral and tetrahedral voids, respectively. Intercalated atoms are partially ordered; 3D superlattice 2a₀×2b₀×2c₀ was found. Interestingly, the following dependence of the crystallographic type of matrix on the composition and cooling rate during the intercalate synthesis was determined for Cu_xTa_1+yS₂ [164]: quenching the ampoule during the synthesis led to the formation of 2H–Cu_xTa_1+yS₂ (0.29 < x < 0.34, 0.28 < y < 0.30), while slow cooling lead to the formation of 6R–Cu_xTa_1+yS₂ (0.30 < x < 0.31, 0.03 < y < 0.04). Moreover, transformation of 2H–Cu_xTa_1+yS₂ into 6R–Cu_xTa_1+yS₂ was observed upon slow cooling. It was shown for Cu-intercalated ZrTe₂ that different crystallographic positions of copper atoms in polycrystalline compounds Cu_xZrTe₂ (0 < x < 0.4) are possible depending on temperature. In the temperature range 250-750 °C, Cu atoms occupy mainly tetrahedral positions in the interlayer space and mainly octahedral positions above 900 °C [165].

Most of the works devoted to the intercalation of TMDs reported charge transfer from the intercalant to the host layers while M′ atoms acquire a formal positive charge and the negative charge is distributed over the MQ₂ layers. Depending on the nature of the host (i.e., on the electronic configuration of the metal in MQ₂ and the chalcogen nature), the acquired charge (electron density) can provoke changes in the structure of the host layers. Thus, in the intercalate Cu_0.69NbTe₂ [117], the structure type of the host is CdI₂, but the NbTe₂ sublattice is distorted with respect to the trigonal structure so that zigzag chains of niobium atoms are formed in these layers in the same direction within one layer; in this case, the formal electron configuration of Nb atoms is d². The authors of [11] prepared copper intercalates for ditellurides of group 5 metals: Cu_0.85VTe₂, CuNbTe₂, and Cu_0.89TaTe₂. Zigzag chains of V, Nb, Ta atoms were found in their crystal structures, which was confirmed by the calculations of the electron density localization function. A similar behavior would be expected for group 6 TMDs (MoQ₂, WQ₂) with d² configuration of metal atoms. In fact, the formation of zigzag chains of molybdenum atoms was observed in MoS₂ as it passed into the 1T modification (for example, upon the intercalation by organic cations [75].

Superstructures are often formed in group 5 and 6 TMD intercalates when at least some intercalant atoms in the interlayer spaces are ordered and superstructures $\sqrt{3}a\times \sqrt{3}a$ or 2a×2a (a is the initial host lattice parameter) are formed (Table 2). Superstructure $\sqrt{13}a\times \sqrt{13}a$ is typical for the case of commensurate charge density wave. Such electronic structure rearrangements were observed both for niobium and tantalum dichalcogenides and for their intercalates (see more details below).

Structure of substituted dichalcogenides

Chalcogen-substituted TMDs form a series of solid solutions, and the structural type in the homogeneity zone is preserved. Sulfides and selenides usually form continuous series of solid solutions at 0 ≤ x ≤ 2: ZrS_2–xSe_x [123], ReS_xSe_2–x [124] WS_2–xSe_x [122]. Lattice parameters a and c increase with increasing selenium content. Telluride-containing TMDs have discontinuities in the homogeneity region and crystallize in different structural types depending on x: WSe_2–xTe_x crystallizes in the structural type 2H at 0 ≤ x ≤ 0.62 and in the structure type WTe₂ (Pnm2₁) at 1.48 ≤ x ≤ 2.0; WS_2–xTe_x are similarly monophasic at 0 ≤ x ≤ 0.3 and 1.8 ≤ x ≤ 2.0 [122].

We now consider isovalent substitution in the cationic sublattice on the example of Mo–WQ₂. It is known that a continuous series of solid solutions 2H–Mo_xW_1–xQ₂ can be prepared for Q = S, Se [129, 166]. In the case of Q = Te, there are both pure 2H, 1T′, T_d(γ) phases and the regions of their coexistence, depending on x, and the reported data vary greatly. Substitution of molybdenum atoms in MoTe₂ by tungsten atoms leads to the destabilization of phases 2H and 1T, and tungsten contents exceeding 34-37% lead to the formation of the orthorhombic T_d(γ) phase. At the tungsten content 15-34%, both phases 1T and T_d coexist simultaneously; at the tungsten content < 15%, phase 1T is stable; and at the tungsten content < 4%, the 2H phase may exist [167-169]. In this case, according to [170], a low critical tungsten concentration of ~8% stabilizes the orthorhombic T_d(γ) phase at room temperature, and the Fermi surface of this phase is similar to that of WTe₂.

Introducing the heteroatom while preserving the structural polytype affects lattice parameters according to Vegard′s law: Mo_1–xNb_xS₂, W_1–xNb_xS_2–ySe_y (our works, JSC 2017 and 2018), M_xTc_1–xS₂ [171], Au_xRe_1–xSe₂ [172]. Introducing Au atoms in the main ReSe₂ lattice slightly increases its lattice parameters and reduces the stability of the triclinic structure under pressure.

Sometimes the introduction of heteroatoms causes more significant changes. For example, phase 2H is destabilized and phase 3R appears after introducing 1% of Re atoms into the cationic sublattice 2H–WS₂ [173]. It was shown for Re_xMo_1–xS₂ that high concentration of rhenium atoms leads to the clustering of the latter [147].

PROPERTIES OF TMDs

Electronic structure of TMDs

Electronic structure of binary TMDs. The electronic structure of stoichiometric TMDs MQ₂ depends on the nature of the transition metal and the chalcogen and changes with the number of layers in the crystal so that these compounds can vary from wide-bandgap semiconductors to metals. The band structure of TMDs was qualitatively systematized as far as in 1970-80-x [177]. The evolution of the TMD band structure depending on metal can be schematically considered on the example of disulfides ZrS₂, NbS₂, and MoS₂ (Fig. 3) [174].

Group 4 TMDs (Ti, Zr, Hf) usually crystallize as 1T polymorphs. In ZrS₂ and HfS₂, chemical bonds have a significant ionic component, all d electrons of the transition metal are involved in the bonding with sulfur; therefore, it is reasonable to assume that the metal ions have the d⁰ configuration, and the charge distribution corresponds to the ${{\text{M}}^{\text{4}+}}\text{S}_{2}^{2-}$ model (Fig. 3) [174]. Metal ions in these compounds occur in an octahedral environment of sulfur, and the crystal-field splitting of d orbitals gives a qualitative structure of the unfilled conduction band (Fig. 3) with predominantly ${{d}_{{{z}^{2}}}}$ states forming the lower part of the conduction band. In turn, the valence band, which is formed mainly by p orbitals of sulfur, is located much lower on the energy scale; thus, both ZrS₂ and HfS₂ with a more pronounced ionic nature of bonding are relatively wide-bandgap semiconductors (~ 1.16 eV and ~ 1.44 eV, respectively [175]). Replacing sulfur with selenium and tellurium leads to a decrease in the ionic component of M–Q bonds decrease. The semiconductor band gap between the valence p band and the conduction d band is still large enough in ZrSe₂ and HfSe₂ (≈ 0.6-0.7 eV), but ditellurides already show an overlap between p and d bands equal to several tenths of electronvolts and, respectively, exhibit semimetallic properties [176].

Group 5 TMDs (V, Nb, Ta) form polytypes with trigonal prismatic and octahedral types of metal ion coordination. For 1T – structures , the presence of an additional d electron compared with group 4 TMDs, would have resulted in the d¹ metal configuration, i.e. partially filled d conduction band and, consequently, good metallic properties. However, special features of TMDs such as pronounced layered structure and, consequently, a quasi-2D electronic subsystem make possible the development of the Peierls instability whereby a superstructural modulation (charge density waves, CDW) is formed with such a period that an additional band gap appears at the Fermi level due to the increased unit cell size (see more details below). In contrast to quasi-1D chained compounds where the Peierls instability can make a system totally dielectric, structural modulation in quasi-2D TMDs affects only a part of charge carriers, and the very possibility of its formation strongly depends on the geometry of the Fermi surface. Among group 5 TMDs, 1T–TaS₂ and 1T–TaSe₂ have been found most strongly susceptible to superstructural distortions, and even at room temperature they exhibit only weak metallic properties.

Group 5 TMDs can be also prepared with trigonal prismatic coordination, usually with the 2H polytype. Even though this coordination increases the energy due to the Coulomb repulsion between negatively charged chalcogen atoms located directly above each other in the structure, the splitting of d levels of the metal ion in this configuration is such that the energy of the ${{d}_{{{z}^{2}}}}$ band additionally decreases and, therefore, the total energy balance favors the stabilization of the 2H structure (Fig. 3). Close energies of these two structural types make possible the existence of both 1T and 2H polytypes, for example, for TaS₂ and TaSe₂. A more complex topology of the Fermi surface of compounds with the 2H modification reduces the effect of superstructural modulations thus increasing their metallic properties. Some compounds, e.g. NbSe₂, are superconductors ([174] and references therein).

Fig. 3

Band structure of TMDs with octahedral and trigonal prismatic coordinations of metal ions; filled states are dashed; the dashed line shows the Fermi level. The figure is adapted from [174], ©Advances in Physics, Taylor & Francis, 1987.

Group 6 TMDs usually form trigonal prismatic structures. Since metal ions have the d² configuration where the ${{d}_{{{z}^{2}}}}$ band is separated from the bands formed by the rest of d orbitals (Fig. 3), MoS₂, MoSe₂, WS₂, and WSe₂ are semiconductors whose top of the valence band is formed by p and d electron levels. In contrast to diselenides and disulfides of Mo and W, ditelluride MoTe₂ can crystallize also in the polymorphic structure with distorted octahedral (1T′) coordination, and such strongly distorted 1T′ structure is the main one for WTe₂ [177-180]. Owing to the changes in the crystal field splitting of d bands in 1T′ and increased p–d hybridization, the band gap between valence and conduction bands in Mo and W 1T′ ditellurides disappears and these compounds become semimetallic [174, 177-181].

As the d shell of metal ions fills up to the d³ level in group 7 TMDs such as Re dichalcogenides, electronic instability of the partially filled d band leads to structural distortions, a fourfold increase of the unit cell size, and formation of Re ion chains. As a result of the unit cell size increase, six completely filled electron subbands are formed to create semiconducting properties in ReQ₂ with a band gap up to ~1.4 eV [182-184]. Even though the reasons underlying the superstructure and dielectricity in 1T–TaS₂ and ReS₂ are similar (decreasing energy of d electrons), the superstructure in ReS₂ does not correlate with the shape of the Fermi surface and more resembles the Jahn–Teller distortion [182].

Although strong chemical bonds in the layered TMD structure are mainly formed within the layers, and the interlayer (van der Waals) interaction is relatively weak, the electronic structure of TMDs (except for group 7 TMDs) depends significantly on the crystal thickness and changes upon the transition from the bulk material to the monolayer (Fig. 4) [175, 185, 186].

For example, bulk MoS₂ is an indirect band gap semiconductor with a band gap of ~1.2-1.3 eV whereas a monolayer of this material is a direct band gap semiconductor whose band gap increases up to ~1.9 eV [185, 186]. Similar changes occur in the band structure of WSe₂ [190] thus providing possibilities of practical application of Mo and W dichalcogenides in semiconductor electronic devices to be used in electroluminescence and photoelectronics [187-189].

The fact that the band structure of Mo and W dichalcogenides changes from indirect band gap semiconductors and direct band gap semiconductor upon the transition to monolayers is more their specific feature rather than a universal property of TMDs. For example, photoemission data and theoretical calculations for Hf and Zr disulfides and diselenides show that they remain indirect band gap semiconductors when their band structures change upon the transition to monolayers [175, 190].

As the thickness of crystallites decreases on the way to monolayers, the band structures of conductive TMD compounds also change significantly. As an illustration, Fig. 5 shows calculated [191] Fermi surfaces for monolayers, bilayers, and bulk crystals of two TaSe₂ polytypes. It can be seen that the interlayer interaction in bilayers leads to the appearance of additional sheets of the Fermi surface, its complication, and eventually to the appearance of the “corrugated” along the c axis Fermi surface The corrugation, i.e. dependence of the Fermi surface on the wave vector along the axis c reflects the dispersion of electrons along the normal to the layers and the degree of electron transport anisotropy.

The band structure of TMDs is highly sensitive not only to interlayer interactions but also to distances between ions within each layer, which gives additional tool for modifying their properties. For example, it was shown for monolayers ZrQ₂ and HfSe₂ by theoretical calculations [192] that the band gap and related physical properties can be controlled using deformation. The band gap of the HfSe₂ monolayer is 0.6 eV, and calculations show that it should increase approximately linearly up to 1.34 eV with the tensile strain up to 8-9% [192]. In the case of ZrS₂, calculations show that the band gap increases upon the tensile strain up to 6% and starts decreasing after reaching the maximum value of 1.63 eV, while the compound should remain an indirect band gap semiconductor up to 10% deformation. In turn, the band gap decreases monotonically as the strain increases upon lattice compression, the semiconductor–metal transition is expected at the 8% compression. By tuning the deformation one should be able to obtain required band gaps and band structure parameters (e.g., band degeneration [193]) that are optimal for the required physical property. For example, a 6% tensile strain should significantly improve thermoelectric properties of the ZrS₂ monolayer: the thermoelectric efficiency increases up to ZT = 2.4 at 300 K, which is 4.3 times higher than that of the initial system [194]. According to the theoretical calculations, conductive TMDs should also exhibit high strain sensitivity; for example, biaxial tensile strain in the plane of the HfTe₂ monolayer should lead to the metal–semiconductor transition, while uniaxial tensile deformation should preserve the metallic nature [195]. High elasticity and stability of thin TMDs layers make structural deformation an attractive method for modifying TMD properties. In particular, structural deformation can be obtained using epitaxial growth of MQ₂ layers on substrates with different lattice constants.

Fig. 4

Band structure calculated for bulk MoS₂ (a) and the MoS₂ monolayer (b); the arrows show the minimum energy distance between the valence band and the conduction band. In the case a, the electron wave vector changes upon the minimal energy transition, i.e. it is an indirect band gap semiconductor; in the case b, the transition does not affect the wave vector, i.e. it is a direct band gap semiconductor. The figure is cited from [186], ©ACS Nano, American Chemical Society, 2013.

The sensitivity of TMD physical properties to weak deformations can be used to design various sensors and devices. For example, it was shown that deformation of a MoS₂ monolayer creates piezoelectric response which, in turn, changes the Schottky barrier on metal contacts to MoS₂ and, consequently, their electrical resistance. The electromechanical device based on MoS₂ monolayers with metal contacts demonstrated an unprecedented coefficient of sensitivity to deformations [196]. The change of conductive properties upon strain was also demonstrated on a polycrystalline sample of substituted molybdenum disulfide Mo_0.95Re_0.05S₂ [197].

Influence of intercalation. The most well studied class of intercalates are TMDs intercalated by alkali and alkaline earth metals ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ ([10, 174, 198] and references therein), which have been widely studied since 1960s. Upon the intercalation by alkaline earth metals and, particularly, alkali metals, the latter occupy positions in the interlayer space and transfer significant part of their outer electron density to the MQ₂ layers, which is the main effect of intercalation on the physical properties of these compounds [10, 69, 70, 174, 198-203]. (Below, the changes in the degree of electron band filling as a result of changes in the compound′s chemical composition or due to the effect of electric field will be referred to as “electron/hole doping”). In semiconductor compounds MQ₂ such as disulfides and diselenides of Hf, Zr, Mo, electron doping leads to the partial filling of the conduction band so that the compound acquires metallic and superconducting at low temperatures properties (Fig. 6) [10, 198, 199, 202]. In turn, intercalation by alkali metals and electron doping of conductive TMDs leads to the gradual conduction band filling and the metal–semiconductor transition [174].

Fig. 5

Evolution of the Fermi surface of 2H–TaSe₂ (a-c) and 1T–TaSe₂ (d-f) upon the transition from monolayers (a, d) to bilayers (b, e) and bulk materials (c, f). The figure is cited from [191], adapted. ©Scientific Reports, Springer Nature, 2015.

The properties of TMDs intercalated with alkali metals were compared with those of few-layer samples doped with electrons directly by the electrostatic method ( like in field effect transistors) and were found completely similar (Fig. 6) [198, 199, 204]. This is another confirmation that the main result of intercalation with alkali metals and alkaline earth metals is the appearance of a negative charge on MQ₂ layers, i.e. supplying the empty or partially filled conduction band with electrons. The fact that intercalated crystals of both group 4 (Hf, Zr) and Mo dichalcogenides (Fig. 6) exhibit superconductivity with close parameters to those of NbSe₂, NbS₂, TaS₂, TaSeS [174, 198] prove the similarity in properties of TMDs , that differ basically in the degree of d band filling.

The second most important consequence of intercalation of metals and, even more so of, large organic molecules into the TMD interlayer space is the decrease of interlayer interaction, which allows one to control the dimensionality of the system. For example, it was shown by angle-resolved photoemission spectroscopy (ARPES) that even the intercalation with potassium transforms the initial 3D electronic structure of bulk 1T–HfTe₂ into an almost purely 2D structure [205]. An even more obvious transition to a purely 2D structure was observed in TMDs doped with organic molecules where the interlayer distance could reach 60 Å excluding any significant interlayer interaction [177]. By analogy with the change of the band structure upon the transition from a bulk crystal to a monolayer, the intercalation affects the band structure by increasing, in particular, the band gap between p and d bands and decreasing the p-d hybridization and the width of d bands. One most interesting fundamental result obtained for intercalated TMDs and confirmed later for monolayers is the demonstration of the possibility of purely 2D superconductivity [174, 199, 204, 206].

Fig. 6

Superconducting transition temperature of MoS₂ subjected to electrostatic electron doping (up to ~2·10¹⁴ cm^–2) and chemical intercalation by alkali metals. The figure is cited from [199], © Science, The American Association for the Advancement of Science, 2012.

Structural distortions are another example of intercalation effects that are significant for physical properties. For example, deposition of sodium on the surface of HfSe₂ single crystals splits the valence band due to the deformations caused by intercalated sodium atoms, while the electrons from sodium atoms pass to the conduction band [207]. HfTe₂ and ZrTe₂ tend to form non-stoichiometric compounds whose nonstoichiometry is caused by Te vacancies [208, 209]. Upon the intercalation of HfTe₂ with lithium using gas-transport reaction, the effect of tellurium vacancies in Li_xHfTe₂ samples (whereby electrons of a defect remain localized near the latter) is weaker than the intercalation effect (whereby electrons from the lithium atom are transferred to the conduction band). Intercalation with Li slightly changes the lattice parameter a, while parameter c changes, at most, by 0.2 Å for x = 0.5 [210].

Intercalation of ReS₂ by alkali metal [211] and fluorine [212] atoms was studied by calculation methods. The calculations show that the changes in the electronic structure caused by alkali metal intercalation can be used to increase the catalytic activity of TMDs in hydrogen production reactions [211]. Fluorine intercalation also affects the band structure. According to the calculations, F atoms located above Re chains induce metallic states in the middle of the forbidden band, and the states are coupled ferromagnetically within Re chains and antiferromagnetically between the chains. In turn, F atoms between the Re chains should induce non-magnetic semiconductor states in the forbidden band [212].

Intercalates ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ can be prepared also with 3d transition metals. In addition to the high-temperature synthesis, their specific feature is that ions intercalated in the interlayer space preserve electrons in their magnetic d shell when transferring outer electrons to the conduction band (Mn, Fe, Co, Ni acquire the 2+ state and V, Cr acquire the 3+ state [174]). As a result, the intercalates become ferro- or antiferromagnets with magnetic moments of intercalated ions interacting through superexchange interactions via M′–Q bonds or via conduction electrons of MQ₂ layers.

Changes in the electronic structure and properties of TMDs upon intercalation with organic molecules are determined by the same main factors as in the case of alkali metals: changed band fillings and decreased interlayer interactions, though with a different priority. Introducing organic molecules between the layers pushes MQ₂ layers apart thus turning them almost into the monolayer state with the corresponding changes in their electronic structure, while the changes in the degree of band filling are significantly weaker compared to alkali metals and depend on the ability of a particular intercalated molecule to transfer electron density or to withdraw it from the MQ₂ layer [174].

Effect of cationic and anionic substitutions on the electronic structure. Isovalent substitution of chalcogen atoms in the anion sublattice of dichalcogenides decreases the band gap in the series S → Se → Te [213, 214]. In the case of HfSe_2–xS_x, the band gap decreases with decreasing sulfur content, independently on chalcogen positions in the calculated structure [215].

The band gap also depends on the substitutions in the transition metal sublattice. For example, isovalent substitution of molybdenum for tungsten in Mo_xW_1–xSe₂ (0 ≤ x ≤1) leads to the formation of a solid solution where the optical band gap systematically changes depending on the composition [216]. In turn, non-isovalent cationic substitutions have a more complex effect that strongly depends on the nature of the dichalcogenide and the dopant. For instance, band structure calculations show that doping monolayer HfS₂ with different atoms can lead to a) hole doping and the Fermi level shift in the valence band, e.g. upon doping with La, Ce, Pr, Gd, Tb, Dy, Ho, Lu [217]; b) appearance of new states in the forbidden band upon doping with Nd, Pm, Sm, Eu, Er, Tm, Yb [217] as well as by transition metal atoms [218, 219] and P, As atoms [220].

The effect of substitution of rhenium atoms in rhenium disulfide and rhenium diselenide by metal atoms (V, Cr, Mn, Fe Co, Nb, Mo, Ta, W) on the band structure was studied by calculation methods in Refs. [221, 222]. In the case of dilute solutions, the least energy favorable dopant is Co both for rhenium disulfide and for rhenium diselenide; Ta is most energy favorable for ReS₂, and Nb for ReSe₂. It was shown that the band gap decreases significantly for all the studied dopants, with the exception of Mn. It is predicted that the electronic structure of ReQ₂ compounds doped with Cr, Fe, and Co is to be unstable with respect to spontaneous spin polarization; as a result, the ground state becomes ferromagnetic thus making these compounds 2D ferromagnetic semiconductors that can be possibly used in spintronics. Ferromagnetism of iron-doped ZrS₂ was demonstrated experimentally [223].

According to the calculations [223], doping a TcS₂ monolayer with transition metals Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn leads to the formation of additional states in the form of sharp peaks near the host′s valence band or conduction band due to the electronic levels of introduced atoms. In general, the band gap decreases: it reaches the smallest value (0.03 eV) when Tc is substituted by Fe, then by Cr (0.07 eV), Mn (0.14 eV), Co (0.15 eV), Sc (0.18 eV), Zn (0.22 eV), Cu (0.28 eV), Ni (0.35 eV), V (0.44 eV), and the largest value is reached for Ti (0.62 eV) [223].

Experimental studies of Mo_xRe_1–xS₂ single crystals (x ≤ 0.01) have shown that doping ReS₂ with a small amount of molybdenum leads to the formation of deep impurity levels, which has almost no effect on the magnitude of the indirect band gap between the valence band and the conduction band, but diminishes the mobility of charge carriers and, consequently, decreases the electrical conductivity [224]. Optical studies of single crystals Nb_0.01Re_0.99S₂ [225] also showed that the introduction of niobium into the ReS₂ lattice only slightly decreases the indirect band gap.

In the case of niobium and tantalum dichalcogenides, doping leads to additional changes due to the existence of CDW regions on the phase diagram (see below) and superconducting states. For example, in 1T–TaS₂ doped with copper (1T–Cu_xTa_1–xS₂, homogeneity region 0 ≤ x ≤ 0.25), copper in the Cu⁺¹ state, similarly to some other doping atoms of higher valencies, leads to the Anderson localization of charge carriers remaining after the CDW formation; as a result, the electrical resistance increases strongly and Curie paramagnetism emerges at low temperatures [137]. Substituting tantalum in 1T–TaS₂ with titanium (having one d electron less) changes the concentration of charge carriers and, respectively, the CDW period determined by the Fermi surface geometry. However, substituting Ta by Hf does not affect the CDW period [154]. Instead of changing the concentration of charge carriers, the Hf ion (which has significantly higher energies of d orbitals) creates a point defect with a hole localized near the ion. Thus, the effect of Hf ions on the CDW is that they cause spatial pinning of the charge density wave thus preventing the CDW to transform from the incommensurate state into the commensurate one, broadening the transition, and lowering its temperature [154].

CDW ordering

The instability of the metallic state of 1D conductors, which was predicted by Peierls in 1930s and experimentally confirmed by numerous studies of quasi-1D chain compounds, is necessarily associated with a structural distortion with wave vector q twice the Fermi wave vector (q = 2k_F) [226]. Such a structural distortion in 1D conductors leads to an increase in the unit cell size and the opening of the gap exactly at the Fermi level, with the decreased energy of occupied electronic states due to the band gap being the driving force of the structural transition. In 1D systems, the Fermi surface is two planes separated by a distance of 2k_F, which are perfectly superimposed by the CDW wave vector q = 2k_F; as a result, the emerged band gap covers the entire Fermi surface, and the compound becomes dielectric. More generally, the appearance of the Peierls instability and the CDW in compounds that are not 1D conductors is due to the presence of significant flat areas of the Fermi surface which can be superimposed using some translation vector q. Historically, TMDs were found to be most suitable objects for the studies of CDW and related phenomena in 2D systems [227]. The Peierls instability and CDW formation are clearly manifested in dichalcogenides MQ₂ with the 1T structure, whose Fermi surfaces have quite a simple structure with large flat areas (Fig. 7a).

Fig. 7a schematically shows the Fermi surface of 1T–TaS₂ (and 1T–TaSe₂) formed by six electronic “pockets” having large nearly flat areas. The appearance of flat Fermi surface areas in TMDs is is directly related to the weakness of the interlayer interaction, i.e. their quasi-2D nature which makes the electron dispersion along the c axis negligible, while the Fermi surface very weakly depends on the wave vector components along axis c (line LM in Fig. 7a). Flat areas of electronic pockets can be superimposed by the translation vector directed along lines ГM and ГK (Fig. 7b), but in the first case flat areas of four pockets out of six are superimposed [227]. Detailed electron diffraction experiments showed that 1T–TaS₂ and 1T–TaSe₂ always exhibit superposition of charge density waves with three vectors q_i rotated by 120° relative to each other; as a result, a periodic superstructure $\sqrt{13}a\times \sqrt{13}a$ is formed and the unit cell size in the plane increases by 13 times (Fig. 8a). In fact, such superstructure is a result of three CDWs with vectors q₁, q₂, q₃ formed in individual TaS₂ monolayers packed in the crystal as repeating stacks (the unit cell triples along axis c) [227, 228]. Such CDW superposition allows encompassing maximum of the initial Fermi surface and excluding the maximum fraction of charge carriers from the conduction so that the electrical resistance of 1T–TaS₂, for example, increases by dozens of times (Fig. 8b).

Being optimal with respect to the Fermi surface topology, the CDW periodicity does not necessarily imply commensurability of the resulting wave with the crystal lattice. However, in the case of an incommensurate CDW (IC–CDW), spatial positions of the maxima of CDW electron density do not coincide with positively charged ions, thus resulting in a loss in the Coulomb interaction energy. As a result, CDWs in most cases undergo significant evolution with temperature changes: (a) at high temperatures (above 500-600 K), the initial Fermi surface with no signs of CDW can be observed, (b) at lower temperatures, IC–CDW with a weak lattice response appears, (c) the tendency to minimize the separation between the charges of the electron density wave and the ion core changes the CDW period so that the CDW is adjusted to the crystal lattice, i.e. transition to the commensurate CDW state (C–CDW) at T_d takes place. In particular, such evolution was observed for 1T–TaSe₂ [227] where the CDW wavelength decreases by ~5% to adjust to the lattice. Sulfide 1T–TaS₂ shows a more complex CDW evolution. Instead of a direct transition to the C–CDW phase, 1T–TaS₂ first demonstrates a transition to nearly commensurate phase (NC–CDW) at ~355 K, and totally commensurate CDW is established only at ~200 K (Fig. 8b, c) [227, 229, 230]. In the NC–CDW phase, a quasi-periodic superstructure is formed from the domains of commensurate CDW (Fig. 8c): inside the domains, the CDW phase adjusts to the lattice to provide the maximum possible dielectric state, and the excess electron density is pushed off to the domain boundaries where the metallic state is preserved [231].

Fig. 7

Fermi surface of 1T–TaS₂ [230] formed by six electronic “pockets” with a close to cylindrical shape (a); displacement of the Fermi surface by vector q₁ directed along axis ГM resulting in the superimposition of flat areas of four out of six electronic “pockets” (b). The figures from [227] are adapted, ©Advances in Physics, Taylor & Francis, 1975.

Fig. 8

Structure of 1T–TaS₂ with indicated CDW induced local ionic displacements: 12 ions of a cluster shift to the central one to form a periodic $\sqrt{13}a\times \sqrt{13}a$ [229] superstructure (a); temperature dependence of the electrical resistance of TaS₂ single crystal showing two stages of commensurate CDW (CCDW) formation with the transition from the incommensurate CDW (ICCDW) to the nearly commensurate CDW (NCCDW) phase at ~355 K and the first-order phase transition at ~200 K to the CCDW phase [230] (b); crystal structure evolution with varying temperature: asterisks indicate the clusters where the CDW period is commensurate with the crystal lattice [229] (c). The figure is compiled from works [229], ©Chinese Physics B, IOP Publishing, 2019 and [230], ©Scientific Reports, Springer Nature, 2014.

In other polytypes such as 2H and 4Hb, the CDW formation is complicated by a more complex Fermi surface which decreases the transition temperature to the CDW phase and diminishes its influence on electron transport characteristics. However, these polytypes preserve the same characteristic features of CDW formation. For example, in 2H–TaSe₂ at T_ICCDW ~ 120 K, IC–CDW with a similar set of three wave vectors q_i appears [227, 232, 233] and passes to the commensurate state at ~90 K. Similarly to 1T–TaS₂, the CDW adjustment to the lattice proceeds through the formation of commensurate domains and the expulsion of “excessive” electron density (hindering the commensurability) to the domain boundaries. The only difference is the domain topology, which has a shape of stripes in 2H–TaSe₂ [232, 233].

Sensitivity of the band structure to polymorphism (packing of layers in the crystal) suggests a rich variety of electronic states in TMDs. An X-ray microdiffraction study of thin (7-36 nm) crystalline plates prepared by mechanical exfoliation of a bulk single crystal 2Ha–TaS₂ [234] revealed a number of polymorphs, including those not observed in bulk samples. All the polymorphs had significantly different properties with the CDW transition temperature ranging from 0 K to 300 K; depending on the polymorph, the electrical resistivity could increase or decrease with the CDW formation [234].One regularity exhibited by the TMD series is increasing temperature of transition to the commensurate CDW phase as the chalcogen atom is replaced by a heavier one: this is due to the changes in the phonon spectrum (CDW adjustment to the lattice is caused by the electron-phonon interaction) [227].

The main method of direct CDW observation is X-ray and electron diffraction allowing one to detect and determine parameters of lattice superstructures [227, 228]. Significant ionic displacements by CDW, e.g. Ta ions in the C–CDW phase in 1T–TaSe₂, makes it possible to directly observe the ionic superstructure in real space using high-resolution transmission electron microscopy [235].

After clarifying the basic principles of CDW formation in TMDs, extensive research has been undertaken to develop the methods for the control of CDW and transitions between its various states, since the possibility of obtaining and controlling various exotic electronic states provides possibility of their practical use, in particular, in nanoelectronics.

The most obvious way to influence a CDW state is to change the concentration of charge carriers in the sample using intercalation, cationic substitutions, electric field applied along or across the layers , and optical excitation [229-231, 234, 236, 237]. Changes in the concentration of charge carriers affects the topology of the Fermi surface and, consequently, affects the possibility of superimposing the flat regions of the Fermi surface to form the CDW and modify the superstructure vectors required for such a superimposition [174, 227]. The relation between the wave vector of the IC–CDW superstructure and the degree of conduction band filling was visually demonstrated for samples 1T–Ta_1–xTi_xSe₂ where the wave vector of the CDW superstructure perfectly fitted dependency q² ∝ (1–x) from q² ~ 1/13 at x = 0 to q² ~ 1/100 at x = 0.9 [227], i.e. a tenfold decrease of conduction band filling upon the substitution of 90% Ta by Ti (Ti does not supply electrons to the conduction band) caused almost as many times increase of the superstructure cell′s size.

Whereas the appearance and periodicity of IC–CDW are determined almost solely by the Fermi surface topology, the CDW transition to the nearly commensurate state and, even more so to the commensurate state is related to the interaction of the electron wave with the lattice. Even small changes of the IC–CDW wave vector q can significantly affect the ability of the CDW to adjust its period to the crystal lattice period. Consequently, the temperature of CDW transition to nearly commensurate state and commensurate states is extremely sensitive to any influences affecting the Fermi surface topology and the IC–CDW wave vector. For example, transition to the C–CDW state in 1T–TaS₂ is easily suppressed by the electric field applied in the field-effect transistor geometry and causing hole doping [230, 231]. Changes in the Fermi surface topology upon the transition to thin crystallites or under pressure enhancing the interlayer interaction also significantly affect the temperature of CDW transition to the commensurate state. In 1T–TaS₂ crystals, the NC–CDW to C–CDW transition disappears at a crystal thickness less than 30 nm, while transition IC–CDW to NC–CDW survives at least up to 7 nm [230]. Vice versa, in the diselenide crystal of 1T–TaSe₂ the temperature of CDW transition to the commensurate state increases with decreasing film thickness and reaches 570 K for a thickness of 3 nm (about 100 K higher than in bulk crystals) [238].

The impurity potential due to ionic substitutions or structural defects relatively weakly affects the appearance of IC–CDW but can easily suppress CDW transition to the commensurate state. When trying to adjust its phase to the chaotic Coulomb potential of impurities, CDW breaks down into fragments and can no more form a wave that would be commensurate with the lattice. In titanium doped crystals 1T–Ta_1–xTi_xS₂, the transition temperature to NC–CDW decreases with a rate of ~10 K/at.%, while the transition to C–CDW is totally suppressed already at x = 0.5% [230]. Doping with copper in 1T–Cu_xTa_1–xS₂ also suppresses CDW transition to the commensurate phase [137].

Most interesting is the possibility of rapid and reversible impact on the CDW state in TMDs. This possibility is provided by the above mentioned charge doping of thin crystallites or monolayers by a perpendicular electric field in the field-effect transistor configuration. In fact, hole doping of crystals with C–CDW removes electrons from the static commensurate wave, and the CDW breaks down into commensurate domains separated by domain boundaries [231]. The percolation network of domain walls accumulating the introduced holes provides an integral metallic state of the sample, similarly to the NC–CDW state (Fig. 8). It was found out that the same effect can be achieved by applying electric field along the layers [237]. It takes several picoseconds for a current pulse passing through a layer to transform the C–CDW into a set of nanodomains separated by conducting boundaries (Fig. 9), i.e. it switches the crystal from the dielectric state into the metallic state. After the passage of the current pulse, the conductive state may preserve for quite a long time, depending on temperature. Similar metastable conducting states can be obtained in thin (< 60 nm) 1T–TaS₂ crystals by rapid cooling of nearly commensurate CDW states to temperatures below 100 K [230].

An effect similar to charge doping can be achieved by irradiating a thin crystal with a femtosecond laser pulse [236, 239]. The irradiation-induced conductive state in 1T–TaS₂ turned out to be long-term stable below 60-80 K [236]. Super fast switching [239], several orders of magnitude change of electrical conductivity, and long-term stability (at low T) make effects related to the switching of CDW states highly attractive for applications. At higher temperatures, where the lifetime of the conducting state decreases to the nanosecond level, switching the CDW state in TMD nanolayers can be used to create highly sensitive bolometers, i.e. broadband radiation detectors [229, 240, 241].

The authors of [242] demonstrated the possibility to “draw” conductive nanogrids on a thin 1T–TaS₂ crystal existing in the insulating ground state by using voltage pulses from the tip of a scanning tunneling microscope to transform the CDW into a metastable incommensurate state.

From viewpoint of fundamental science, of considerable interest is the influence of CDW on the appearance of superconductivity in TMDs, in particular, in 2H–NbSe₂, 2H–TaS₂, and 2H–TaSe₂ where coexistence and competition of CDW and superconductivity were observed [229, 243]. For example, the temperatures of CDW and superconductivity occurrence in 2H–TaS₂ are T_CDW = 78 K and T_c = 0.8 K. In 3R–TaSe₂, superconductivity with T_c = 1.6 K was observed along with a weak CDW transition [244]. The maximal T_c = 7 K was observed in 2H–NbSe₂ [245]. The charge density wave and superconductivity compete for the areas of the Fermi surface where the dielectric or superconducting band gap is to be formed. However, Fig. 6 shows that the maximum concentration of charge carriers neither facilitates the establishment of the superconducting state; the dependence of the superconducting transition temperature on the charge carrier concentration has a characteristic dome-like shape. Apparently, parameters required for the superconducting state can be optimized by affecting the CDW parameters. As discussed above, intercalation and cationic substitutions suppress C–CDW, and 2H–Cu_xTaS₂ single crystals with low concentrations of intercalated copper become superconductors (T_c = 4.2 K at x = 0.03) [246], but no superconducting transition is observed at high concentrations of intercalated copper. With the suppression of CDW in Na_xTaS₂ crystals , T_c increases and reaches 4.4 K for Na_0.1TaS₂ [247]. In the series of Pd_xTaS₂ samples (0.01 ≤ x ≤ 0.08), the maximum value of T_c = 4.2 K is reached in Pd_0.04TaS₂. In this case, the charge density wave is gradually suppressed and disappears only in Pd_0.06TaS₂ [114]. CDW suppression was also observed upon doping with Fe in 1T–Fe_xTa_1–xS₂ single crystals (0 ≤ x ≤ 0.05) [136, 151]; superconductivity strongly depends on x in a narrow range, while the maximum T_c is 2.8 K at x = 0.02 [136]. This suggests that there superconductivity and CDW compete with each other, but complete suppression of the latter is not necessary for the former to reach maximal values.

Fig. 9

Schematic illustration of the changes in the CDW state in a 1T–TaS₂ nanocrystal under the action of a current pulse along the layers: the injected charge carriers at the pulse front transform the dielectric state of C–CDW into the metastable conducting state (Hidden CWD) consisting of C–CDW nanoclusters separated by conducting domain boundaries. The figure is cited from [237], ©Nature Communications, Springer Nature, 2016.

Conductive properties

In terms of electrical conductivity, most interesting are conducting dichalcogenides of group 5 metals due to various CDW manifestations and semiconductor TMDs of groups 4, 6, and 7 owing to the possible usage of thin crystals and monolayers in semiconductor nanoelectronics.

HfQ₂, ZrQ₂. In accordance with the previously considered electronic structure, bulk ZrS₂, HfS₂, ZrSe₂, and HfSe₂ crystals are indirect band gap semiconductors with the band gap varying from ~1.2-1.5 eV for disulfides to ~0.6-1.0 eV for diselenides [175]. The band gap increases by a few tenths of electronvolts as the crystal thickness decreases down to monolayers, but dichalcogenides of Zr and Hf, in contrast to MoS₂, remain indirect band gap semiconductors [175]. Strong dependence of the band gap width on the choice of chalcogen and existence of a continuous series of solid solutions ZrS_2–xSe_x and HfS_2–xSe_x open the possibility of preparing a semiconductor material with optimal characteristics for each specific device [123]. Few-layer crystals and monolayers of HfSe₂ and ZrSe₂ were among the first 2D materials to demonstrated a whole complex of properties technologically important for semiconductor micro- and nano-electronics; earlier available only by silicon: optimal band gap values (~1-1.5 eV), high current ratio in open and closed field-effect transistor modes (10⁶-10⁸), and physico-chemical compatibility with technologically suitable dielectrics such as HfO₂ and ZrO₂ oxides [190].

In addition to the direct use of Zr and Hf disulfides and diselenides as semiconductor materials, various possibilities of chemical modification of their properties were also considered. Intercalation of HfQ₂, ZrQ₂ by alkali metals leads to the addition of electrons into MQ₂ layers, conduction band filling, and the transition to the metallic state [69]. At low temperatures, metallic intercalates demonstrate transition to the superconducting state [174]. A semiconductor–metal transition was observed in ZrSe₂ single crystals also upon the intercalation with Cu [248]. In the case of polycrystalline samples HfS₂, HfSe₂, intercalation with Cu, Fe, and Cr significantly decreased electrical resistance [96], but the conductivity preserved its hopping character, apparently, due to the localization of charge carriers on the potential of defects. Interestingly, the semiconductor–metal transition in single crystal ZrSe₂ was initiated by the intercalation (not substitution) with only 1.3 at.% Re introduced into the structure in the Re²⁺ state [249].

In contrast to disulfides and diselenides, ZrTe₂ and HfTe₂ are semimetals due to the valence/conduction bands overlap [174, 203, 250, 251]. HfTe₂ is a semimetal with a relatively high concentration of charge carriers ~10²⁰-10²¹/cm³ [203, 250] and the mobility of hole carriers increasing from ~30 cm²/V·s at room temperature to ~1500 cm²/V·s at T = 2 K. Hydrostatic pressure (up to ~5 GPa), that increases interlayer interactions, expectedly increases the overlap between the bands and the concentration of charge carriers, while the electrical resistance decreases several-fold. For single crystals HfTe₂, the obtained resistivity values along the layers ρ_//(300 K) were 0.17-2.7 mΩ·cm and decreased upon cooling by 36-120 times depending on growth conditions and, consequently, the crystal quality (amount of impurities and defects) [250, 251]. Single crystals ZrTe₂ have very similar semi-metallic properties (electrical resistivity ρ_//(300 K) ~ 0.5 mΩ·cm) [251]. The authors of [252] reported the transition of ZrTe₂ to the superconducting state at T_c = 2 K.

Similarly to the use of intercalation and doping to transform semiconducting TMDs into the metallic state, similar approaches can be used to modify the properties of semimetallic ZrTe₂ and HfTe₂. For example, it was shown in [252] that intercalation of ZrTe₂ with Cr not only leads to the expected electron doping, increased number of electrons in the conduction band, and gradual disappearance of holes in the valence band, but also reduces the overlap between the bands. As a result, single crystals Cr_0.4ZrTe₂ transformed from semi-metals to metals with a partially filled conduction band separated from the valence band by an indirect band gap. Interestingly, despite the impurity potential of chromium, the superconducting transition temperature in Cr_0.4ZrTe₂ increased up to 5.4 K [252]. Intercalation with Ni increases the superconducting transition temperature up to T_c = 4.1 K in Ni_0.5ZrTe₂ [253]. The transition of ZrTe_2–xSe_x to the semiconducting state associated with the appearance of an indirect band gap was induced by the isovalent substitution of the chalcogen atom [254].

NbQ₂, TaQ₂. At high temperatures, above the CDW formation point, dichalcogenides of Nb and Ta are quasi-2D metals with a half-filled conduction band whose resistivity values along the layers vary from several tens to several hundreds µΩ·cm [174, 227]. The CDW formation due to the Peierls instability (see above) leads to the formation of a band gap on a part of the Fermi surface and, consequently, abrupt decrease of the Pauli paramagnetic susceptibility due to the decreased density of states at the Fermi level, abrupt increase of electrical resistivity, changes in the Hall density of charge carriers and the thermal Seebeck coefficient. The magnitude of magnetic susceptibility jumps is determined by the percentage of the Fermi surface covered by the Peierls band gap. For electrical resistance, this dependence is more complicated since the reconfiguration of the Fermi surface affects not only the density of charge carriers but also the electron scattering parameters.

The Fermi surface of 2H and 4Hb polymorphs of NbS₂, NbSe₂, TaS₂, and TaSe₂ has a fairly complex topology. Their CDWs, which are formed at ~30-120 K mainly with the 3a×3a periodicity within a layer (see Refs. [174, 227] and references therein), lead to the formation of a band gap that covers only a small part of the Fermi surface. As a result, these compounds preserve metallic nature of electron transport and small electrical resistivity ρ_// at low temperatures (from one tenth to several µΩ·cm). At low temperatures, all these metallic compounds (bulk crystals or individual monolayers obtained, in particular, by the intercalation of organic molecules) become superconducting; the maximum T_c = 7.2 K was observed in NbSe₂ (Ref. [174] and references therein).

Much more substantial electronic changes are caused by CDWs in ditellurides NbTe₂, TaTe₂ [255-257]. Besides CDWs with a period of 3a×3a formed in ditellurides above room temperature, a commensurate superstructure $\sqrt{19}a\times \sqrt{19}a$ was observed (see Ref. [255] and references therein). As a result of the CDW influence, the electrical resistivity of NbTe₂, TaTe₂ significantly exceeds that in disulfides and diselenides with the 2H structure: for example, TaTe₂ single crystals exhibit ρ_//(300 K) ≈ 200 µΩ·cm and ρ_//(3.5 K) ≈ 110 µΩ·cm [256].

The formation of CDW is most well manifested in the electrical resistance of polymorphs 1T–TaS₂ and 1T–TaSe₂ where the CDW-induced dielectric band gap covers the most part of the Fermi surface to cause an abrupt (by 1-2 orders of magnitude) increase of the electrical resistivity (Fig. 8b). In 1T–TaSe₂, the IC–CDW appears above 600 K and transforms directly into the C–CDW with the $\sqrt{13}a\times \sqrt{13}a$ period at 473 K. Upon the transition into the commensurate state, the amplitude of structural modulation (ion displacements) grows sharply, the band gap increases, and the curve of electrical resistivity shows a sharp jump [174, 227]. The remaining charge carriers provide metallic behavior down to low temperatures with electrical resistance ρ_//(5 K) ≈ 100 µΩ·cm. In 1T–TaS₂, the C–CDW state is achieved in several steps with two jumps of the electrical resistivity (at ~350 K and 200 K). The mechanisms of CDW formation and the methods of transition control were discussed in detail above.

Note that properties of 1T polymorphs cannot be explained only by considering the changes in the Fermi surface under the CDW action. Indeed, detailed ARPES (angle-resolved photoemission spectroscopy) studies of 1T–TaS₂ crystals [258] showed that, in agreement with calculations, the CDW with a $\sqrt{13}a\times \sqrt{13}a$ periodicity splits the 5d band of Ta into narrow subbands. The d¹ state and the 13-fold unit cell increase within the plane lead to the formation of six narrow filled subbands, six empty subbands, and one half-filled subband. The latter should have provided metallic conductivity; however, 1T–TaS₂ at low temperatures is dielectric. This behavior was explained (and experimentally confirmed) by strong correlations between electrons in the partially filled subband leading to the Mott transition [258]. In fact, in the C–CDW state in 1T–TaS₂, the cluster of 13 Ta ions (Fig. 8a) has only one electron located in its center. Coulomb electron correlations, forbidding two electrons to be in one cluster, “freeze” the order where the clusters form a periodic structure, with strictly one electron being localized in each cluster. Similar Mott dielectric states are well known in oxides with formally half-filled bands [259]. The violation of the Mott order, e.g. by the cationic substitution of Ta with Fe in 1T–Fe_xTa_1–xS₂, cancels the forbiddance of charge motion and leads to the formation of metallic and superconducting states [258]. Photoemission studies show that superconductivity coexists with CDW and arises in one of partially filled subbands out of 13 subbands formed by the CDW [258].

Conductive properties of NbQ₂ and TaQ₂ can be chemically modified using the same approaches as those for group 4 TMDs: intercalation with alkali, alkaline earth, and transition metals, organic molecules. For example, intercalation of silver and palladium into polymorphs 2H and 4Hb of TaS₂ can be used to modify their conductive properties and thee superconducting transition temperature [260]. In Pd-intercalated 2H–TaS₂, the superconducting state coexists and competes with the CDW; the maximal T_c = 4.2 K is reached in Pd_0.04TaS₂ [114].

The experimental data on the electron transport anisotropy associated with the layered crystal structure are very scarce because of the difficulty in preparing large single crystals without defects in their layer packing. However, it was noted that the anisotropy of group 5 TMD crystals is significantly lower than in their analogs of groups 4 and 6 due to the presence of an unpaired electron on the d shell. The anisotropy of electrical resistivity in stoichiometric crystals ranges from 10 to 100 and increases by orders of magnitude after the crystals are intercalated by metals or organic molecules [174, 227].

MoQ₂, WQ₂. Similarly to group 4 TMDs, bulk crystals MoS₂, WS₂, MoSe₂, and WSe₂ are indirect band gap semiconductors with a band gap of ~1.1-1.3 eV [185, 186, 216]; however, in contrast to the former, decreasing the crystal thickness to a monolayer not only increases the band gap up to ~1.9-2.0 eV but also transforms them into direct band gap semiconductors [185, 186, 261]. Consequently, the possible use of group 6 TMD monolayers, compared to those of group 4 TMDs, includes various devices related to electroluminescence and photovoltaics.

Like in the case of group 4 TMDs, the band gap of Mo and W disulfides and diselenides can be adjusted using its dependence on the choice of the metal and the chalcogen and by utilizing the possibility of preparing solid solutions. For example, it was shown that the band gap monotonously increases with x in the series Mo_1–xW_xSe₂ (0 ≤ x ≤ 1) [216]; additional increase can be achieved by substituting S with Se. At low temperatures, semiconducting compounds MoQ₂ and WQ₂ can be easily switched to metallic and superconducting states by intercalating them with metals or organic molecules (see above). The possibility of modifying electrical conductive properties by replacing Mo and W with group 5 and 7 elements is widely studied. For example, the study of Re-doped WSe₂ crystals showed that Re creates electronic levels in the forbidden band and switches p-type conductivity to n-type [262]. It was also demonstrated that electrical and optical anisotropy of single crystals of Mo and W dichalcogenides decreases by tens times with Re-doping [263, 264]. In turn, cationic substitution of Mo and W by Nb and Ta atoms, having one electron less on their outer shells, leads to the formation of acceptor impurities and increases the number of hole charge carriers [155, 265-267]. As the Nb concentration increases, the nature of electron transport changes to the metallic one.

The properties of MoTe₂ and WTe₂ qualitatively differ from those of disulfides and diselenides due to their ability to crystallize not only as the 2H polymorph but also as 1T′ and Td polymorphs [177-181, 268]. Similarly to disulfides and diselenides, 2H–MoTe₂ is a semiconductor with a band gap of ~0.9-1.0 eV and a charge carriers mobility of ~40 cm²/V·s at 300 K [179-181]. However, bulk crystals structural 1T′ and Td modifications of MoTe₂ and Td–WTe₂ are semimetals. According to early research, the band overlap in semi-metallic phases is ~0.5 eV [177]; however, revised data showed that it is as small as ~0.05-0.06 eV [178-181]. Structurally, 1T′ –phase of MoTe₂ and WTe₂ can be considered as distorted 1T, and the nature of this distortion is related to the Peierls instability of the hypothetical 1T phase and to superstructural distortions leading to the formation of Mo–Mo or W–W chains. The resulting Fermi surface is composed of electronic and hole pockets along the axis corresponding to the direction of M–M chains. The concentration of charge carriers in MoTe₂ and WTe₂ is relatively low (~10²¹/cm³) with an electrical resistivity at room temperature of ~600-700 µΩ·cm in Td–WTe₂ [178] and ~400-1000 µΩ·cm in 1T′–MoTe₂ [179, 196, 268]. With decreasing temperature, the electrical resistivity decreases by hundreds of times: e.g., the mobility of charge carriers in 1T′–MoTe₂ increases from ~1 cm²/V·s to 4000 cm²/V·s [179]. Similarly to other TMDs, the degree of band overlap in 1T′ and Td phases depends on thickness: the overlap disappears in thin crystals (less than 10 layers) and the semi-metallic state switches to the semiconducting one.

It was shown that all three MoTe₂ polymorphs have very close energies. The reversible structural transition between structural modifications 2H and 1T′ occurs in the MoTe₂ crystal at 500-600 °C [179], yet under the action of a ~0.05% tensile deformation it takes place at 100°C, and under the action of a ~0.2% deformation it occurs at room temperature. As a result of such reversible transition, the electrical resistivity changes by four orders of magnitude [180].

In recent years, MoTe₂ and WTe₂ have attracted much attention due their to unusual electron transport properties. Compared to other non-magnetic materials, WTe₂ showed exceptional magnetoresistance values: its low-temperature electrical resistivity increases by 1.3·10⁵ times in the 60 T magnetic field [178]; huge magnetoresistance was also observed in MoTe₂. The superconducting transition temperature T_c in semimetallic MoTe₂ crystals increases up to 8.2 K (by 80 times) under pressure [181].

Unusual properties of Td phases of MoTe₂ and WTe₂ are explained by the theoretically predicted “Weyl semimetal” state which added to the list of exotic and potentially practically useful electronic states such as: “topological insulators”, “topological superconductors”, “Dirac fermions” in graphene whose discussion goes beyond the scope of this review.

TcQ₂, ReQ₂. Group 7 TMDs are semiconductors with band gaps ~0.7-1.1 eV and ~1.0-1.4 eV for TcQ₂ and ReQ₂, respectively [182-184, 269]. Similarly to most other TMDs, the band gap decreases in the series S → Se → Te. Although first papers reported contradictory conclusions about the forbidden band structure, photoluminescence studies [183] showed that at least ReS₂ is a direct band gap semiconductor.

The properties of TcQ₂ and ReQ₂ are very close to other semiconducting TMDs, except for two significant features.

a. The semiconducting state in TcQ₂ and ReQ₂ is caused by M–M interactions that increase the unit cell size and lead to the formation of metal double-chains within the layer planes. As a result, electronic and optical properties of TcQ₂ and ReQ₂ exhibit significant in-plane anisotropy [182-184, 269].

b. In fact, the chains formed by metal ions close interionic interactions within the layer thus drastically weakening the interlayer bonds [183]. As a result, the electronic structure of TcQ₂ and ReQ₂, in contrast to other TMDs, is virtually independent of the crystal thickness, including the case of monolayers: weakly connected MQ₂ layers in these crystals already behave as virtually isolated monolayers. Therefore, the use of TcQ₂ and ReQ₂ does not require taking into account the dependence of their electronic properties on the crystal thickness and interlayer interactions.

Chemical modification of TcQ₂ and ReQ₂ is performed with standard approaches for TMDs. In particular, the studies of cationic substitutions in Mo_xRe_(1–x)Se₂ and W_xRe_(1–x)Se₂ showed that cationic substitutions not only increased the number of charge carriers and electrical conductivity but also reduced the in-plane anisotropy, apparently distorting the structure of Re–Re chains [270, 271].

Thermoelectric properties. Due to low thermal conductivity values κ, TMDs are attractive for thermoelectric applications. Among the TMDs considered in this review, thermoelectric properties were most well studied for molybdenum and tungsten dichalcogenides and their thermoelectric properties. Electrical conductivity of MoS₂ was increased to the required level by doping with Cu [272], Re, and Nb [197, 266]; however, no desired values of the thermoelectric figure of merit (ZT) have been achieved so far. Films prepared by mechanical exfoliation of MoS₂ crystal an subsequent deposition had a ZT value of ≈ 2·10^–4 at room temperature (the required value being ZT > 1 [273]), apparently, due to the defects at the boundaries of crystallites forming the film. The contacts between the crystallites were improved using annealing [274] and the method of introducing Cu into the defect areas [275] in order to increase the mobility of charge carriers and the Seebeck coefficient. Since one of the limitations for the use of flexible TMD films is weak charge transfer along c axis, the authors of [276] prepared a 3D structure of MoS₂ layers interpenetrated by carbon nanotubes (CNTs). Increasing the CNT content increased electrical conductivity by several orders of magnitude and decreased thermal conductivity. The maximal ZT = 0.17 at room temperature was obtained for the 20% CNT content.

The highest electrical conductivity increase in tungsten disulfide and diselenide was achieved using the acceptor substitution by group 5 elements (V, Nb, Ta) [267]. The authors of Refs. [155, 156, 265, 277] described the effect of substitutions in cationic and anionic sublattices and double substitutions in WQ₂. The maximal value ZT = 0.26 was obtained for sample W_0.98Nb_0.02Se_1.7S_0.3 at 650 K. In composite samples WS₂/MCNTs (multi-walled CNTs) ZT = 0.22 at T = 800 K was obtained for the CNT content ~0.75 wt.% [278].

Magnetism

Stoichiometric TMDs are weakly magnetic materials, since neither metal ions nor chalcogen ions have localized magnetic moments. In semiconducting TMD of groups 4 and 6 elements, the transition metal ions occur either in the d⁰ state with no d electrons at all or in the d² state with a pair of electrons on the ${{d}_{{{z}^{2}}}}$ orbital (see above), all other orbitals being empty. In semiconducting group 7 TMDs with metal ions in the d³ state, the unit cell quadrupling leads to the formation of six completely filled subbands, which also excludes the magnetism of d shell electrons. In turn, in the case of metallic TMDs, mobile electrons in the conduction band can demonstrate only weak Pauli paramagnetism whose magnitude provides data on the density of states at the Fermi level. This is widely used to characterize phase transitions associated with the CDW formation when the density of states is decreased due to the emerging band gap (Fig. 10) [227].

The possibility to intercalate various magnetic ions with unfilled d or f shells into the interlayer space of TMDs was widely used to “design” various magnetic materials ${{{M}'}_{x}}M{{Q}_{\text{2}}}$ consisting of alternating magnetic layers M′_x and non-magnetic MQ₂ layers (see refs. [96, 174, 279-281] and references therein). The over-exchange interaction of M′ ions via M′–Q bonds and indirect RKKY (Ruderman–Kittel–Kasuya–Yosida) interaction via conduction electrons of the MQ₂ layer lead to the formation of antiferromagnetic states in ordered intercalates ${{{M}'}_{1/3}}M{{Q}_{\text{2}}}$ and ${{{M}'}_{1/4}}M{{Q}_{\text{2}}}$ as well as various cluster spin glasses in the case of intermediate compositions and multicomponent doping. Magnetic intercalates attracted much interest due to the possibility of studying the physics of quasi-2D magnets and the interaction between the magnetic order and the electronic subsystem of MQ₂ layers.

Fig. 10

Impact of CDW transitions on magnetic susceptibilities of various TaS₂ polymorphs. The figure is cited from [227], ©Advances in Physics, Taylor & Francis, 1975.

High interest in the research and application of thin TMD crystals and monolayers has raised a problem of obtaining magnetic states directly in the electronic subsystem of MQ₂ layers. In particular, the preparation of semiconducting and metal monolayers with spin-polarized charge carriers is of exceptional interest for spintronics. One possible approach is cationic substitutions. For example, the ferromagnetic state of ZrS₂ single crystals is obtained by doping with iron [282]. Detailed theoretical calculations prove the possibility of obtaining FM semiconducting and FM semimetallic states upon cationic substitution with various 3d metals [218-220, 283, 284]. An interesting method for obtaining ferromagnetic states was proposed in [178] where it was shown that a certain tensile strain applied to the monolayers of Zr and Hf dichalcogenides, the hole carriers doped into the monolayer should exhibit spontaneous ferromagnetic polarization.

Raman spectroscopy

Raman spectroscopy has been very useful in the study of layered TMDs. Upon the transition from bulk samples to few-layer ones and upon the decrease in the number of layers, the samples demonstrate, among other things, symmetry changes while forbidden vibrations become allowed, possibly causing shifts in the positions of modes and the appearance of new modes. The Raman spectra of Mo and W dichalcogenides are most sensitive to the number of layers. The main vibrational modes of MoS₂ are A_1g 408 cm^–1 and $\text{E}_{2g}^{1}$ 383 cm^–1. The decrease in the number of layers leads to the blue shift of A_1g to 403 cm^–1 and the red shift of $\text{E}_{2g}^{1}$ to ~385 cm^–1 and to the band broadening [285]. MoSe₂ shows a similar band shifts as the number of layers decreases. However, A_1g splits into two bands at a thickness of 3-4 layers and into three bands at a thickness of five layers. The spectral positions of two peaks for three MoSe₂ layers are located almost symmetrically with respect to the monolayer line. The observed splittings range from 2.4 cm^–1 to 3.2 cm^–1. This effect is caused by the so-called Davydov splitting due to the presence of more than one molecule in the MoSe₂ unit cell. At a thickness of two layers, mode $\text{B}_{2g}^{1}$ appears at 353 cm^–1, which was not active in the bulk sample [286]. Another interesting effect was observed in the case of applying dichalcogenide layers on a substrate. For example, mode E_1g, which is usually forbidden in the Raman configuration, was observed in supported TaSe₂ layers and disappears in suspended layers, thus suggesting that this mode may be activated due to symmetry breaking caused by the interaction with the substrate. A systematic low-temperature Raman study revealed some phonon features related to the NC–CDW formation in monolayer layer and double layer 2H–TaSe₂, namely, softening of a wide second-order Raman mode near the phase transition point as a result of strong electron-phonon coupling (Kohn anomaly) [287]. The Raman spectra of dichalcogenides ZrSe₂ [288], HfS₂ [289, 290], HfSe₂ [290, 291], ReS₂ [183, 292], ReSe₂ [293] show no significant changes upon the transition from bulk to nanostructured few-layer samples. Note however that rhenium dichalcogenides exhibit significant in-plane anisotropy of optical properties caused by the chains of Re ions in their structure [293]. For dichalcogenides ZrTe₂ [294], HfTe₂ [295], TaTe₂ [295], and TcS₂ [296], some spectroscopic data were reported without considering the effect of sample thickness. Table 3 summarizes the data on TMDs whose Raman spectra depend on the sample thickness along with the general changes exhibited by the spectra upon the transition from bulk to few-layer samples.

Chemical modification (doping, intercalation) causes more significant changes in Raman spectra. As a result of isovalent substitution in the anion sublattice of sulfur atoms with selenium, the frequency of Raman modes A_1g(M–Q) decreases as the content of the heavier Se atom increases, i.e. there is an evident red shift of A_1g in M–Se and blue shift in M–S accompanied by corresponding intensities changes in TMDs: ZrS_2–xSe_x [123], HfS_2–xSe_x [308], 1T–TaS_2–xSe_x [309], MoS_2–xSe_x [310], WS_2–xSe_x [311]. A similar behavior is exhibited also by Te-substited chalcogenides, provided that the 2H modification, e.g. MoTe_xSe_2–x, is preserved [312]. At increasing x, the A_1g mode is red shifted from 240 cm^–1 to 230 cm^–1 and is asymmetrically broadened towards low frequencies; the $\text{E}_{2g}^{1}$ frequency is not changed and only shows the band broadening. The changes of phonon energy in mixed rhenium dichalcogenide ReS_2–xSe_x were studied by optical methods [313].

Table 3 General Changes in the Raman Spectra with Decreasing Number of TMD Layers

The effect of isovalent substitution in the cationic sublattice was studied on the example of Mo_1–xW_xQ₂ in bulk [314] and few-layer samples [134, 315]; all the samples showed the same tendency. The A_1g modes exhibit a single-mode behavior and $\text{E}_{2g}^{1}$ exhibit a double-mode behavior for the entire series caused not only by sulfur vibrations but also by the metal atom vibrations. The double-mode behavior is caused by the difference between the weights of vibrating cations Mo and W. At increasing x, A_1g is red shifted and its intensity increases, whereas $\text{E}_{2g}^{1}$ is blue shifted and splits into two modes. The Raman spectrum of Nb_0.5Ta_0.5Q₂ (Q = S or Se) is a superposition of spectra of corresponding NbQ₂ and TaQ₂ [316].

The Raman shift upon non-isovalent substitution with atoms from other groups can be explained by the reduction of bond lengths and by the changes in electron-phonon interactions. For example, doping ZrS₂ with iron makes modes E_g (246.78 cm^–1) and A_1g (332.23 cm^–1) shift to 257.5 cm^–1 and 343.43 cm^–1, respectively, and the Raman shift is a result of reduced metal–chalcogen bond lengths [282]. Non-isovalent doping of the p/n type leads to the blue/red shift of A_1g due to the changes in electron-phonon interactions, while the frequency of E_2g should undergo almost no changes [317]. This was shown on the example of Nb_xMo_1–xS₂ in [318, 319]. Two optical phonon modes corresponding to A_1g and E_2g were observed for a Nb_xW_1–xS₂ monolayer at 413 cm^–1 and 353 cm^–1 with a small blue shift compared to pure WS₂. Doping WS₂ with Nb leads to the formation of a large number of hole charge carriers that interact with incoming photons, thus possibly causing blue shifts of Raman peaks, similarly to the changes observed in ReS₂ doped with Mo [320].

There are examples of some other Raman spectra of non-isovalently substituted TMDs: MoS₂ doped with Nb [318, 319], Re [321, 322], Sb [323], Pd [324], Ni [325]; Re_xMo_1–xSe₂ [326], WS₂, doped with Nb [327], V [328], Au [329]; WSe₂ doped with Co [330]; WQ₂ (Q = S, Se, Te) doped with Nb (our works, JSC 2017 and 2020).

Raman spectroscopy of intercalated compounds can reveal phonon energy shifts of active phonon modes A_1g and E_g, disappearance of double-phonon modes, and appearance of new modes with low wavenumbers caused by the fact that the presence of an intercalant superlattice reorganizes the Brillouin zone [331]. A shift of phonon modes A_1g and E_g was observed upon the intercalation of TMDs both by alkali metal atoms and by transition metals: M_xZrSe₂ (M = Li, K, Cs) [202], Li_xHfQ₂ [203], Fe_xNbS₂ [332], Fe_xNbSe₂ [333], Fe_xTaSe₂ [334], Mn_xMoSe₂ [334].

APPLICATIONS: CATALYSIS, SENSORS, DETECTORS

Due to unique properties, in particular, the possibility of preparing stable crystals of monomolecular thickness, TMDs are promising materials for electronics, spintronics, optoelectronics and as sensor materials. Thin-layer dichalcogenides of Zr and Hf are studying in as semiconductor materials for electronics [190, 335], solar panels (ZrS₂ [109]), photodetectors, matrices for Li⁺- and Na⁺- ion batteries, photocatalysts [336], and thermoelectric materials [32]. NbQ₂ (Q = S, Se) and its modifications are used for energy storage (Li-ion batteries) [337], as conductive electrodes for nanoelectronic devices, including field effect transistors [338]. NbSe₂ can be used as a material for superconducting photodetectors [339] and monopole RF antennas [340]. 2H–TaS₂ is studied in HER electrocatalysis [341, 342]. 1T–TaS₂ was proposed as a material that can be switched between metastable (conductive and dielectric) CDW states by applying electric field or optical excitation [229, 343]. MQ₂ (M = Mo, W; Q = S, Se) are used in energy storage devices (Li⁺, Na-ion batteries), sensors (NO, glucose, DNA, small molecules), photoelectronic and piezoelectric devices, biosensors, transistors [186, 188, 189, 344-346].

Morphology changes and chemical modification provide fine tuning of practically important TMD properties to be used more effectively. For example, the ratio of currents in open and closed states of field-effect transistors made of Nb-doped MoS₂ significantly varies depending on the flake thickness, and even ultra-thin heavily doped 2D semiconductors cannot be completely depleted and behave like a 3D material when used in the FET geometry [347].

Electrocatalysis, photocatalysis, hydrogen evolution reactions

Over the past decade, TMDs have attracted increasing attention as electrode materials for electrocatalytic decomposition of water. It was shown on the example of MoS₂ that, even though the basal plane of MoS₂ is catalytically inert, its sulfided Mo edges are active in hydrogen evolution reactions (HERs) with a suitable energy ΔG_H = 0.1 V, which is close to that of several efficient HER electrocatalysts, including Pt, nitrogenase, and hydrogenase [348]. Electrocatalytic properties of TMD materials are improved by two main approaches: increasing the number of active centers with the help of structural engineering (“extrinsic” approach) and modifying chemical composition to reduce ΔG_H (“intrinsic” approach) [349].

The “extrinsic” approach of TMD modification includes morphology change, i.e. size and shape of catalyst particles and the addition of electrically conductive substrates such as graphene. The effect of morphology on the catalytic activity in HERs was studied by experimental and calculation methods on the example of MoS₂: dependence of catalytic activity on the number of layers [2, 350], hollow MoS₂ nanoparticles doped with oxygen [351], MoS₂ flakes on carbon [352] and graphene nanosheets [353, 354], monolayers doped with transition metals [35], etc.

The “intrinsic” approach to increase catalytic activity is aimed at changing the ΔG_H value either by isovalent substitution in cationic and anionic sublattices or by non-isovalent substitution (doping) in the cationic sublattice, which also affects electrical conductivity.

Catalytic activity of isovalently substituted TMDs is fairly well studied for Mo and W: MoS_2–xSe_x [356], MoSe_2–xTe_x [357], WS_2–xSe_x [358], Mo_xW_1−x(S_ySe_1−y)₂ [359]. There are examples of other TMDs. Thus, the calculated band gap of ZrS₂ and Janus ZrSSe monolayers is 1.58 eV and 1.08 eV, respectively; therefore, they can be used in as photocatalysts, whereas ZrSe₂ has a small band gap of 0.84 eV and hence can be possibly used as an electrocatalyst. Calculations of the hydrogen adsorption energy and the Gibbs free energy showed that Janus ZrSSe exhibits better catalytic activity than ZrS₂ and ZrSe₂ [360]. The catalytic activity of ReS_xSe_(2–x) composites with substituted anionic sublattices on carbon fiber paper (CFP) were studied in HER reactions. Doping ReSe₂ with sulfur atoms increases the number of active sites and enhances electrical conductivity; electrocatalyst ReSe_1.78S_0.22/CFP provides better HER characteristics than ReSe₂ with a low overvoltage of 123 mV to achieve a current density of 10 mA/cm² as well as long-term stability for 30 h [361].

The catalytic activity of doped TMDs in HER reactions was shown both by calculation and experimental methods. For example, introducing Na⁺, Ca²⁺, Ni²⁺, Co²⁺ cations between the layers of 1T–MoS₂ is an effective strategy to reduce overvoltage in hydrogen evolution reactions. Thus, the initial potential of 1T–MoS₂ with intercalated ions decreases by ~60 mV in acidic environment compared with the initial 1T–MoS₂ potential (beginning at ~180 mV) [362]. A presence of Fe, Co, Ni, Pd, and Pt dopants in the TcS₂ monolayer affects the profiles of electron density of states which may be important for HER. It was found out that ΔG_H in the case of vacancy complexes Ni, Pd, and Pt is very close to the thermoneutral point ΔG_H = 0 eV. This indicates that the activity of TcS₂ in HERs can be significantly improved by Ni, Pd, and Pt doping [363]. Doping ZrQ₂ with Nb, Pt, and W atoms adjusts the electronic properties of these systems by switching their nature to the metallic one; as a result, its catalytic activity is improved compared to the semiconductor catalyst. Calculations of the Gibbs free energy showed that Nb–ZrSe₂ is the best material for HER, the next ones are Pt–ZrS₂, Nb–ZrSSe, Nb–ZrS₂, and ZrSSe, and the introduction of W atoms does not lead to significant improvements [360]. A ReS₂ sample doped with 2% of Co atoms was found highly effective both in acidic (149 mV) and in alkaline electrolytes (240 mV): its current density reached 10 mA/cm², and it demonstrated excellent long-term durability in hydrogen evolution reactions [364].

Sorption, gas sensors, and detectors

In the field of gas sensors, current studies focus on the development of miniaturized high-performance gas sensors operating at room temperature. Layered semiconductor TMDs have attracted particular attention, in particular, due to the possibility to prepare flexible and compact sensors [365, 366]. Characteristics of MoS₂ based gas sensors (Fig. 11) with proposed mechanisms are discussed in the context of a wide range of different morphologies/nanostructures, nanocomposites, van der Waals heterostructures, and photoactivation effects [367].

The ability of doped TMDs, in particular MoS₂, to adsorb various molecules was studied by Calculation methods (Table 4).

It was established that the defectiveness of TMD structures with respect to chalcogen may improve the sensor response. Thus, DFT calculations showed that defected monolayers of ZrS₂ with S vacancies demonstrate better gas adsorption (NO, NO₂, CO, SO₂, N₂, H₂, O₂, CO₂) than pure ZrS₂ structures. Among these gases, CO₂, CO, and N₂ molecules are physically adsorbed on the S vacancies of defected ZrS₂ monolayers, while O₂, NO, SO₂ molecules are chemically adsorbed in the areas that are closed for S vacancies of ZrS₂ monolayers. Also, an NO₂ molecule can be separated by an S vacancies of the defected ZrS₂ monolayer. The results of this work suggest that defected ZrS₂ monolayers with S vacancies can be used as adsorbents in gas sensors to determine the content of CO₂, CO and are particularly sensitive to N₂ [380].

It was shown on the example of tungsten dichalcogenides that doping and functionalization also improve the sensor response. Adsorption of gas molecules CO₂, NO₂, and SO₂ on a WSe₂ monolayer doped with Pd, Ag, Au, Pt was studied in terms of adsorption energy, charge transfer, and band structures. It was shown that NO₂ adsorption on a Ag–WSe₂ monolayer is energetically favorable and that chemisorption is achieved [381]. Functionalization of the 2D surface of WS₂ nanoflakes proved a possibility of fabricating flexible CO gas sensors with a low power consumption [382].

Fig. 11

Schematic illustration of strategies for the fabrication of the MoS₂ based gas sensor. The figure is cited from [367], ©Advanced Materials Technologies, John Wiley & Sons, Ltd, 2020.

Table 4 Ability of Doped MoS₂ Samples to Adsorb Various Molecules

The possibility of combining TMDs possessing unique sorbent and electrophysical properties with other functional materials opens up opportunities of designing FET-based gas sensors. To prepare a thin film transistor gas sensor based on the p-n heterojunction, a WS₂ monolayer grown by the CVD process was transferred onto an indium–gallium–zinc-oxide (IGZO) film [383]. The authors of [383] reported that the resulting device had the best NO₂ gas sensor response compared to all the gas sensors based on transition metal dichalcogenide materials.