Intercalated phases of transition metal dichalcogenides

Abstract

Two‐dimensional transition metal dichalcogenides (TMDs) play host to a wide range of novel topological states, such as quantum spin Hall insulators, superconductors, and Weyl semimetals. The rich polymorphism in TMDs suggests that phase engineering can be used to switch between different charge order states. Intercalation of atoms or molecules into the van der Waals gap of TMDs has emerged as a powerful approach to modify the properties of the material, leading to phase transition or the formation of substoichiometric phases via compositional tuning, thus broadening the electronic and optical landscape of these materials for a wide range of applications. Here, we review the current efforts in the preparation of intercalated TMD. The challenges and opportunities for intercalated TMDs to create a new device paradigm for material science are discussed.

INTRODUCTION

Transition metal dichalcogenides (TMDs), which are semiconductors of the type MX₂, where M is the metal and X the chalcogens, have attracted the attention of physicists since early 1923 when their structures were first proposed.¹ TMDs host a range of fascinating physical and chemical properties resulting from their rich electron–electron and electron–phonon interactions, dimensionality-dependent band structure, and weakly coupled layer structures.^2–4 TMDs provide a rich playground for fundamental studies, such as unconventional superconductivity (SC) and charge density waves (CDW).³ Controlling the twist angle between bilayers or changing between odd- and even-layer thickness can change the structural symmetry and band structure, yielding interesting spin, valley, and excitonic physics.^5–7

The ability of TMDs to serve as an intercalation host originates from its van der Waals (vdW) gap. The interlayer distance, defined as the chalcogen (X)-to-chalcogen (X) distance, is approximately 3–3.5 Å in TMD.⁸ Each unit cell consists of a sandwiched structure (X-M-X)—one transition metal atom plane (M) sandwiched by two chalcogenide atom planes (X). Based on the interlayer stacking registries, TMDs usually adopt three types of polymorphs: octahedral 1T, prismatic 2H, and prismatic 3R. The stacking sequence along c axis of 1T, 2H, and 3R polytypes are AbCAbC, AbABaB, and AbACaCBcB (the capital and lower case letters denote chalcogen and metal atoms), respectively. Different polymorphs can vary strongly in electronic properties. Taking group VIB TMD as an example, the d orbitals in the octahedrally coordinated 1T phase consist of degenerate $d_{\mathit{xy},\mathit{yz},\mathit{zx}}\left(t_{2g}\right)$ and $d_{x^2-y^2,z^2}\left(e_g\right)$ orbitals, the $t_{2g}$ orbital is partially filled, rendering 1T phase metallic.^9–11 By contrast, the trigonal prismatic 2H phase is semiconducting because the d orbital splits into three degenerate states $d_{z^2}$, $d_{x^2-y^2}$, and $d_{\mathit{xy},\mathit{yz}}$ with an energy gap of ~1 eV. The 1T phase can further distort to form the 1T' phase due to the clustering of Mo atoms into zigzag chains (2a × a).¹² The 1T' phase of TMDs is a low symmetry CDW phase, and can be regarded as a periodically distorted structure (Peierls distortion) of the high symmetry 1T phase.^13–16

A myriad of methods has been applied to modify the properties of TMDs.¹⁷ These include: substitutional doping,¹⁸ strain engineering,¹⁹ intercalation,^20,21 forming heterostructures,²² external electric/optical tuning,²³ size/dimensional tuning,²⁴ and defect engineering.²⁵ Extrinsic electric doping, such as ion gel gating, can modulate the quantum states of TMDs.^23,26 Intercalation is as old as graphite intercalation compounds (GIC), whereby both cations (alkali metals) and anions (iodide) can be inserted. The reversibility of Li-ion insertion into graphite enables the successful launch of commercial Li-ion batteries, one of the most widely used energy storage system in portable electronic gadgets.²⁷ The intercalation process occurs in stages, that is, not every layer is intercalated at first. It is important to recognize that intercalation changes the bond order in the interlayers from a vdW type to an ionic or covalently bonded type. For instance, the bonding between the alkali metal ion and graphite is ionic in nature, where KC₈ is a superconductor with a very low critical temperature T_c = 0.14 K.^28,29 GIC counterparts can be found in certain TMDs such as TiS₂.³⁰ TiS₂ goes through staged crystalline phase transitions upon potassium intercalation, the considerable activation barriers present at each stage results in poor reversibility during discharge and charge and rate capability when used as anodes in batteries. One strategy to improve battery performance in TMD-based intercalation compounds is to nanostructure the material to reduce the phase transition activation barriers.³¹ Tian et al.³⁰ chemically intercalated TiS₂ with K to transform the bulk crystal to nanocrystalline K_0.25TiS₂; the use of these nanocrystalline phase bypassed the multiple crystalline phase transitions during intercalation, and thereby improved the K/TiS₂ battery performances. Both foreign atoms/molecules or native atoms could be introduced into layer gaps and they can occupy the octahedral or trigonal vacancies sites, depending on the stacking structure of the TMD. When the concentration of intercalants cannot be controlled precisely, nonstoichiometric phases are formed. Intercalation of bulk TMD materials traced back to the 1970s, when the superconducting phase was induced in semiconducting 2H-MoS₂ after alkali and alkaline-earth atom intercalation with liquid ammonia method.³² In 1986, the first monolayer MoS₂ suspension was reported through intercalation with lithium followed by reaction with water.³³

Intercalation modifies the properties of TMD by a myriad of effects, including charge doping,^34–36 expansion of c axis lattice constants,^37–40 orbital hybridization,^41–43 and phonon scattering.^37,44,45 A wide range of properties could be promoted or tuned, from electrical conductivity,^46,47 optical modes,^48,49 magnetic order,^43,50 thermoelectricity,^45,51 catalytic activities,^52,53 to energy storage and conversion performances.^30,31 Moreover, the concentration of the intercalant can have a threshold effect in ferroic or charge orders. For example, Fe-intercalated TaS₂ could change from ferromagnetic (FM) to antiferromagnetic (AFM) order when Fe concentration is increased above 0.4.^54,55

The intercalation process can be reversible;  good reversibility translates to a small voltage gap in charge and discharge processes in intercalation type batteries. The intercalant concentration varies with electrochemical potential or temperature; this means that nonstoichiometric compounds are created, with properties that can vary widely, depending on composition. Intercalation provides a high doping concentration in 2D TMDs, especially for alkali metals, which can dope up to one electron per unit. In semimetallic or metallic TMD phases, modulating the phases requires a charge-doping level that is comparable to its initial carrier concentration; this is difficult to achieve with conventional gate-electric-field induced or even electron-double-layer surface gating using ionic liquid. Yuanbo Zhang and team utilized a gate electric field to drive lithium ions in and out of the layered 1T-TaS₂ and induced a Mott-insulator-to-metal transition in 1T-TaS₂ thin flakes at low temperatures; by this way, the resistance could be modulated by five orders of magnitude.⁵⁶ Besides changing electronic properties, intercalation also can cause the structural transition. In Cu-intercalated TiS₂, the occupation sites of Cu on tetrahedral or octahedral vacancies on the sulfur atoms determine the stability of the 2D (layered) or 3D (cubic) phase of TiS₂. At low temperature, octahedral site occupation dominates and Cu_xTiS₂ is in 2D (layered) phase. When the temperature is raised, tetrahedral site occupation gradually dominates and the 2D structure transforms to a 3D structure.⁵⁷ Under electrochemical intercalation, a layered compound can expand to many times its original volume when ions or molecules with diameters larger than the inherent vdW gaps are intercalated, resulting in the decoupling of layers and facilitating their exfoliation.^58,59

Solution phase exfoliation of bulk TMD offers advantages in terms of scalability and processability.^39,60 Li et al.⁶¹ reported a general method for the synthesis of highly crystalline 2D semimetallic TMD monolayers using electrochemical exfoliation method with organic ammonium cations solvated with propylene carbonate as cointercalants. The exfoliated NbSe₂ monolayers, which are usually unstable in air, are passivated by propylene carbonate by this exfoliation method, which allows subsequent device fabrication and wafer-scale printing. In the twisted vdW heterostructure formed by exfoliated NbSe₂ flakes, the critical superconducting current could be modulated by the magnetic field when a flux quantum fits an integer number of moiré cells.⁶¹

Figure 1 summarizes the scope of intercalation in terms of methods and properties. The layered structure of TMDs allows atoms, ions, or molecules to be intercalated into the vdW gaps, forming an intercalated phase we call intercalated TMD. The intercalants can be classified into three categories: (1) Alkali metal atoms, including Li, Na, K, Rb. We also include group II, group XIII, and group XIV elements in this group; (2) Transition metals, such as V, Cr, Mn, Fe, Co, Ni, and Cu; (3) Organic molecules, such as amines, pyridine, and ammonia. Due to the different atomic/molecular size and electronegativity of the intercalants relative to the host, different intercalation methods are needed, including vapor phase intercalation, melt-growth method, wet chemical intercalation disproportionation redox reaction, and electrochemical intercalation, which we will be introduced in Section 2. In Section 3, we describe how intercalation alters the electronic and crystalline structures of TMDs, giving rise to unconventional phases, new charge-ordered states, tunable plasmon modes, spin ordering, and improved electrochemical activities.

Figure 1. Schematic showing the different methods for the intercalation of 2D materials, as well as various properties that can be tuned or promoted in the intercalated phases

INTERCALATION OF TMDS

In this section, the most commonly used intercalation methods to form intercalated TMD are briefly discussed. Figure 2 illustrates the different intercalation processes.

Figure 2. Methods of TMD intercalations. (A) vapor-phase intercalation. A metal source and TMD crystal are sealed and heated in a vacuum tube, where vaporized metal atoms intercalate into vdW gaps of TMDs. (B) Self-intercalated TaS2 ultrathin film synthesized by MBE growth method. Reproduced with permission from Ref.43 Copyright 2020, Springer Nature. (C) Schematic of lithium chemical intercalation by n-butyllithium. Reproduced with permission from Ref.62 Copyright 2020, the Royal Society of Chemistry. (D) Schematic of disproportionate redox reaction for Cu intercalation. Reproduced with permission from Ref.63 Copyright 2012, American Chemical Society. (E) Molecular intercaltion in bulk BP by electrochemisty. Electrochemical current variation on the electrochemical bias. (Inset) SEM image of monolayer phosphorene molecular superlattice transistors. Scale bar = 5 μm. (F) Photoluminescence responses of monolayer, bilayer, and trilayer intercalated BP composites. (E,F) Reproduced with permission from Ref.38 Copyright 2018, Springer Nature. BP, black phosphorus; MBE, molecular beam epitaxy; SEM, scanning electron microscopy; TMD, transition metal dichalcogenide; vdW, van der Waals

Vapor phase methods

Intercalation is achieved by the diffusion of vaporized atoms into vdW gaps of TMD multilayers (Figure 2A). The composition tuning of intercalated TMD is achieved by adjusting the vapor pressures of intercalants. The intercalation state (i.e., concentration of intercalant relative to TMD) largely depends on the size and electronegativity of atoms.²¹ Small-sized alkali atoms are easier to intercalate by vapor-phase method compared to heavy atoms. Tan et al. evaporated lithium onto few-layer MoS₂ using a Li SAES getter source. Using secondary ion mass spectrometry, they found that the Li atoms were highly diffusive and intercalated to a depth of 50 nm.⁴⁹ The lithiated MoS₂ adopted the 1T' phase, which had a much stronger optical Kerr nonlinearity and higher optical transparency than the parent 2H phase, making it useful for nonlinear photonic applications.⁴⁹ In growth-intercalation methods, such as melt growth, chemical vapor transport, or molecular beam epitaxy (MBE) (Figure 2B), both the intercalants and TMD precursors are vaporized in a vacuum chamber. Single-crystalline Cu_xBi₂Se₃ was synthesized via melt growth method, where stoichiometric mixtures of Bi, Cu, and Se were melted and reacted in a sealed evacuated quartz tube.^64,65 Cu_xBi₂Se₃ showed an SC transition temperature that varied with the concentration of the intercalant, thus affording an excellent playground for studying the physics of topological superconductors. Recently, the self-intercalated TaS₂ phase was successfully grown by MBE method under a metal-rich chemical potential; the layered 2D compound was transformed into ultrathin, covalently bonded 3D materials with FM properties.⁴³

Solid phase diffusion

Solid phase diffusion generally relies on solid-state reaction at elevated temperatures and atmospheric pressure, and this has been used to intercalate tetragonal β-FeSe (T_c = 8 K) to form A_xFe_{2 − y}Se₂ (A = K, Rb, Cs, and Tl), whereby the T_c can be increased to 30 K⁶⁶; for even thinner layers, T_c can be increased to between 65 and 100 K.^67–69 One drawback of high-temperature solid-phase diffusion is that it often leads to inhomogeneity, microscale phase separation, and the creation of Fe vacancies, which compromise the superconducting state as maintaining it relies on keeping the integrity of the β-FeSe phase. Therefore, it is highly desirable to reduce the temperature used during the diffusion-intercalation process.⁷⁰ Ayajan reported a new solid-state reaction to synthesize homogeneous Cu-intercalated TMD compounds with a high intercalant concentration at room temperature and atmospheric pressure.⁴⁰ Essentially, the method involved mixing Cu powders with TMD powders in a stoichiometric manner in a stirred solution. Intercalant concentration as high as x = 1.2 could be achieved by this intercalation method. The intercalation method worked well with group IV and V layered TMDs (MX₂, M = Ti, V, Ta, and Nb, X = S and Se), but not with group VI TMDs (M = Mo or W).⁴⁰

Wet-chemical methods

Solvated electron solution based on alkali metal in liquid ammonia has been used for the intercalation of layered TMD materials, on account of the electron affinity of the host.⁷¹ Other than alkali metals, cointercalants, such as ammonia and metal-amides, are also intercalated between the layers.⁷² In the case of FeSe, for example, Burrad-Lucas carried out in situ powder X-ray and neutron diffraction experiments and confirmed the presence of ammonia-rich phase Li_0.6(NH₂)_0.2(NH₃)_1.8Fe₂Se₂ accommodating a double layer of NH₃ molecules between the FeSe slabs with T_c = 39 K.⁷³ Compared to the hydrothermal reaction with ammonia, n-butyllithium in hexane is now widely adopted as a reagent to prepare stoichiometric lithium intercalation complexes as the reaction can occur at ambient temperature without the intercalation of side product, so a cleaner product is obtained.^62,74 The intercalation of zero-valent transition metal had been carried out by solution-based disproportionation redox reaction.^63,75–77 For instance, the zero-valent copper can be intercalated in Bi₂Se₃ via the disproportionation redox reaction of a Cu⁺ precursor (tetrakis(acetonitrile)copper(I)hexafluorophosphate) in acetone (Figure 2D) based on the equation $2{\text{Cu}}^{+}\left(\text{aq}\right)\leftrightarrows {\text{Cu}}^0+{\text{Cu}}^{2+}\left(\text{aq}\right)$, where ${\text{Cu}}^{2+}$ is bound with solvent complex, thereby leaving free ${\text{Cu}}^0$ to be intercalated in Bi₂Se₃.^78,79 Such disproportionation reaction has been applied to intercalate a large range of zero-valent atoms, including Cu, Ag, Au, Co, Fe, In, Ni, Pd, and Sn atoms into the vdW gaps of TiS₂, Bi₂Se₃, Sb₂Te₃, In₂Se₃, GaSe.

Electrochemical methods

Electrochemical intercalation of metal ions is typically performed in an electrochemical cell with a metal anode (e.g., Li) and the host TMD as a cathode in the liquid electrolyte containing the metal ions (the intercalant). The electrochemical method allows the intercalation process to be controllable by the magnitude and polarity of the voltage applied. Deintercalation can be realized by applying the opposite bias. In situ characterization techniques can be combined with the electrochemical cell, where changes in optical and electrical responses of intercalated TMDs can be monitored in situ during the intercalation process.^80,81 Other than metal atoms or ions, charged organic molecules can also be intercalated. The electrochemical intercalation of organic molecules allows the creation of organic/inorganic superlattices with tunable chemical, electronic, and optical properties.^38,81 Using black phosphorus (BP) as the intercalation host, Duan's team demonstrated that intercalation with cetyl-trimethylammonium bromide (CTAB) produces phosphorene molecular superlattices in which the distance of the repeating unit cells can be tuned.³⁸ By regulating the electrochemical intercalation conditions, organic intercalants were separated periodically by monolayer, bilayer, or trilayerphosphorenes (Figure 2E,F), thus allowing a way to decouple interlayer interactions and tune the band gap of the material. Due to the passivation by organic molecules, FET devices fabricated on monolayer phosphorene molecular superlattice exhibited an on/off current ratio exceeding 10⁷, along with excellent mobility and superior stability. By varying the sizes, symmetries, and substituent groups of organic intercalants, it is possible to create a wide class of hybrid organic-2D TMD superlattice with tunable structure and electronic/optical properties.

CHARACTERIZATION OF INTERCALATED TMDS X-ray diffraction (XRD)

In the first instance, intercalation of foreign species expands the interlayer space; thus the XRD peaks originating from the planes normal to the c axis are expected to shift to lower angles. If the intercalation process proceeds in stages, the size of the repeating unit cell seen in XRD will change depending on the distance between the intercalated layers. A good illustration of this is in the solid-state intercalation of 2H-NbS₂ by copper atoms.⁴⁰ As shown in Figure 3A, as the Cu intercalation state (x) increases, the intensity of (002) peak for 2H-NbS₂ first decreases and then disappears when x rises to higher than 0.65, to be replaced by two new reflections at lower angles. The first new reflection peak is a broad peak originating from the intermediate phase with a high-order intercalation stage under the low intercalation state. The second peak appears at an even lower diffraction angle and becomes sharp when the Cu intercalation state increases to such an extent that the peak becomes a first-order peak assignable to the (002) plane of Cu_xNbS₂. With increasing Cu intercalation ratio, the interlayer distance between the Nb planes increases, and the new first-order peak shifts slightly to lower angles. Duan Xiangfeng and team intercalated BP with the CTAB molecule, and XRD was used to verify that the intercalation was successful. As can be seen in Figure 3B, the (004) and (002) peaks downshifted from 34° to 16° and from 17° to 8°, respectively, after intercalation, which indicates the interlayer distance has doubled from that of bulk BP.³⁸

Figure 3. Common characterization methods for intercalated TMD. (A) XRD characterization of Cu-intercalated NbS2 at different Cu concentrations. Reproduced with permission from Ref.40 Copyright 2020, American Association for the Advancement of Science. (B) XRD patterns comparing pristine bulk BP and monolayer phosphorene molecular superlattice (MPMS). Reproduced with permission from Ref.38 Copyright 2018, Springer Nature. (C, left) XRD characterizing phase transition in MoS2 by Li intercalation. (C, right) Zoom-in spectra on the (001) and (002) peaks of 1T' and 2H phase. Reproduced with permission from Ref.31 Copyright 2016, American Chemical Society. (D) and (E) STEM real space and FFT images of self-intercalated Ta7S12, respectively. (F) Schematics depicting an atomic model of Ta7S12. (G) and (H) Experimental and simulated cross-section view of Ta9S12, respectively. (D–H). Reproduced with permission from Ref.43 Copyright 2020, Springer Nature.  BP, black phosphorus; FFT, fast Fourier transform; STEM, scanning transmission electron microscope; TMD, transition metal dichalcogenide; vdW, van der Waals; XRD, X-ray diffraction

When the intercalation process causes a transition to another polymorph or polytype due to changes in intralayer or interlayer stacking registries, clear signatures of these transitions usually manifest in XRD. Figure 3C shows the XRD characterization of the phase conversion process of bulk MoS₂ powder during chemical intercalation of lithium naphthalenide.³¹ The (002) peak of 2H-MoS₂ at 14.4° downshifts to 14.2° after intercalation, which is assigned as the (001) peak of 1T′-MoS₂.⁸² As the Li concentration (x) increases from 0 to 0.3, (001) peak emerges and peak intensity of (002) peak reduces, indicating partial phase change from 2H to 1T' phase. As x increases to 0.7, the intensity of (001) peak is stronger than that of (002). Finally, when x = 1, only the (001) peak remains, indicating a complete transformation to 1T' phase.

Atomic imaging

The periodic structures formed by intercalant can be detected as superlattices by scanning transmission electron microscope (STEM). Zhao et al.⁴³ studied self-intercalated TaS₂ with STEM, and characterized the stoichiometry and composition-dependent superlattice. From Figure 3D,E, the periodic bright spots sandwiched between two TaS₂ monolayers had a periodicity of $\sqrt{3}\times \sqrt{3}$. These STEM images matched well with simulated images (Figure 3F) of Ta₇S₁₂—33.3% intercalation, evidencing that the $\sqrt{3}\times \sqrt{3}$ superstructures were formed by the intercalated Ta atoms. In addition, a cross-sectional STEM image directly proved the intercalation of Ta atoms in the vdW gaps (Figure 3G,H).

ELECTRONIC PROPERTIES OF INTERCALATED TMDS

The modulation of the electronic properties of TMDs by intercalants are due to two mechanisms: charge transfer to the host atoms from intercalants and lattice modulation.^32,83,84 Charge transfer from intercalated species to host lattice can modify the shape of Fermi surface and change the conductivity of host material, as well as induce semiconductor-to-metal transition, or metal-to superconductor transitions. Alkali metals like Li have the most negative reduction potential, thus they could dope up to one electron per formula unit of the host crystal. On the other hand, intercalant of larger size, such as organic molecules, increase the layer separation of TMD lattice. The increased layer distance results in weaker interlayer interactions and makes the physical properties of host materials approach the 2D limit.^38,85

Phase engineering

TMD polymorphs have different d band electron configurations and exhibit distinct electronic properties.⁸⁶ For example, quantum spin Hall effect⁸⁷ and dipolar ferroelectricity⁸⁸ have been predicted in 1T' and 1T MoS₂, respectively, although they are absent in the 2H phase. The most widely studied polymorph conversion is the semiconducting (2H)-to-metallic(1T) transition in MoS₂.

Generally, intercalation-induced phase conversion is promoted by electron transfer from guest species to host lattice. In the case of lithiated 2H-MoS₂ (Figure 4A), the injected electrons from Li⁺ reduce Mo⁴⁺ to Mo³⁺, and the intervalence charger transfer between Mo⁴⁺ and Mo³⁺ causes sulfur plane sliding, which results in the conversion between the two polymorphs.³⁴ The intercalant concentration determines the number of electrons transferred, which affects the polymorph conversion process. In thinner MoS₂, the critical injected electron concentration required for phase change is larger than that of thicker crystals, which is due to the decreased energy barrier of phase transition and stability of 2H phase upon electron doping in thicker layers.⁸⁹

Figure 4. Phase engineering of MoS2 by intercalation. (A) Mechanism of the phase transition from 2H to 1T by intercalation of alkali metal. Charge transfer between Mo3+ and Mo4+ induces phase conversion. Reproduced with permission from Ref.34 Copyright 2017, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. (B) Patterned 1T and 2H phases on a MoS2 monolayer flake characterized by electrostatic force microscopy. Scale bar = 1 µm. Reproduced with permission from Ref.46 Copyright 2014, Springer Nature. (C) Transfer curve of a 2H-MoS2 device with patterned 1T phase on the edge as contacts to metal leads. Reproduced with permission from Ref.47 Copyright 2014, author(s). (D) Superconducting transition temperatures of K interclated 2H, 1T and 1T' MoS2. Reproduced with permission from Ref.71 Copyright 2015, American Chemical Society. (E) TEM characterization of intercalated 2H-MoS2 with intercalation bias at 1.6, 1.1, 0.9, and 2.4 V, respectively. Reproduced with permission from Ref.90 Copyright 2017, The Royal Society of Chemistry. FFT, fast Fourier transform; TEM, transmission electron microscope

Phase engineering has emerged as a strategy to reduce contact resistance in device fabrication by converting the edges of the semiconducting 2H phase into metallic 1T-MoS₂. The use of 1T-MoS₂ as contacts on 2H-MoS₂ improved the device performances in terms of low subthreshold swing values below 100 mV per decade and a large on/off ratios of >10⁷ (Figure 4B,C).^46,47 K-intercalated MoS₂ by liquid-ammonia method showed superconducting phase for all MoS₂ phase: 2H, 1T, and 1T' phase with critical temperatures at 6.9, 2.8, and 4.6 K, respectively (Figure 4D).⁷¹

Compared with wet chemical intercalations, electrochemical methods are more efficient and provide better control of the phase conversion. The intercalant concentration, as well as the intercalation or deintercalation process, are controllable by changing the electrochemical potential. Figure 4E shows the different phases of MoS₂ at different K intercalation potential.⁹⁰ As the electrochemical intercalation potential changes from 1.6 V, to 1.1 V, 0.9 V and recharges to 2.4 V, the phase of MoS₂ changes in sequence from 2H, to 1T, dT (distorted 1T phase) to a mixture of 2H and 1T. Organic molecules, such as CTAB, can also be electrochemically intercalated into MoS₂, inducing semiconducting 2H to metallic 1T phase conversion (Figure 4E).

However, most phase engineering using intercalation generate mixed phase of 2H and 1T over micron-sized regions, and pure 1T- or 1T' phase are difficult to synthesize.^91,92 The 1T or 1T' phase produced by wet chemical intercalations is metastable, which hinders further investigation of its intrinsic properties and device applications. Air stable and phase-pure 1T' phase MoS₂ could be synthesized by using lithium hydride instead of lithium as the intercalant.⁴⁹ The lithium hydride (LiH) intercalant was generated from the hydrogenation of lithium. Owing to the greater stability of LiH than Li, the 1T' phase induced by LiH intercalant was stable in air. LiH acted as a good electron donor and created strong dipole−dipole interactions within the MoS₂ interlayers, thus improving the stability of the 1T' phase.

Correlated states

Rich electron-electron and electron–phonon interactions in TMDs lead to intriguing correlated states, such as CDW and SC. Because of the significance of quantum fluctuations and the complex nature of the collective electronic states, SC in the 2D limit is one of the most fascinating topics in condensed matter physics.^26,93 CDWs are periodic charge modulations associated with lattice distortions. At low temperatures, CDW conductors change from normal metal state to CDW state by opening a gap at the Fermi level. Interestingly, most CDW TMD conductors have either an intrinsic SC phase or an extrinsic one induced by external doping or pressure.^56,94,95 The interplay between SC and CDW has been a long-standing question in condensed matter physics.^26,96 Studying the interactions between these correlated states help provide an insight into the relation between these quantum phenomena, as well as CDW and SC formation mechanism. In this section, we explain how intercalation modulates CDW and SC states.

Intercalation could melt the CDW phase, allowing the SC state to emerge in TMD crystals. Theoretical simulations predicted that extra charges (both electrons and holes) would affect electronic structures and phonon dispersion in TMDs, resulting in decreased CDW transition temperatures or suppressed CDW state (Figure 5A).^97,101 CDW could compete with or be a precursor to SC, depending on the specific electronic structure, chemical combination and crystal structure of a particular compound.^102–104 Cu-intercalated 1T-TiSe₂ was synthesized by the melt growth method. The CDW phase of Cu_xTiSe₂ was continuously suppressed upon Cu intercalation and the SC state emerged when the Cu concentration x reached 0.04 (Figure 5B)⁹⁸; SC dome appeared as a function of Cu concentration. XRD study observed coexistence of SC state and incommensurate CDW in the phase diagram of Cu_xTiSe₂, suggesting that the origin of SC may result from incommensurate CDW (Figure 5C).⁹⁹

Figure 5. CDW and SC state modulated by intercalation concentrations. (A) Simulated evolution of total electron-phonon coupling constant (λ) and CDW transition temperature (Tc) of monolayer 1T-TiSe2 upon electron and hole doping. Reproduced with permission from Ref.97 Copyright 2017, American Physical Society. (B) and (C) Phase diagram of CuxTiSe2 as a function of Cu concentration obtained from transport and XRD studies. Reproduced with permission from Ref.98 Copyright 2006, Springer Nature. Reproduced with permission from Ref.99 Copyright 2017, American Physical Society. (D) Temperature-resistance curves at different gate voltages in a 14-nm 1T-TaS2 device. The hysteris loops indicate CDW transitions. Arrow shows the trend of increasing gate voltages. (E) Phase diagram as a function of gate voltage (doping) of a 14-nm 1T-TaS2 flake. (D,E) Reproduced with permission from Ref.56 Copyright 2015, Springer Nature. (F) Phase diagram as a function of doping level in 2H-TaS2. (G) Calculated acoustic phonon dispersions in 2H-TaS2 at varied levels of hybridization with the substrate, Γ (HWHM of the electronic broadening), and charge doping, -x (in units of injected electrons per Ta atom). x [less than] 0 means electron doping. (F, G) Reproduced with permission from Ref.100 Copyright 2019, American Chemical Society. CDW, charge density wave; HWHM, half width at half maximum; SC, superconducting; XRD, X-ray diffraction

On the other hand, the competition of CDW and SC was discovered in ion-gel gated Li-intercalated 1T-TaS₂. Li-intercalated TaS₂ showed suppressed commensurate CDW phase followed by emergence of SC (Figure 5D,E).⁵⁶ A SC dome emerged as a function of gate voltage, which determined the intercalation concentration. The intercalation by ion gel gating was reversible and offered good control of intercalation/doping process. This intercalation was coupled to electrostatic gating, thus it could be potentially implemented in practical devices.

Intercalation can also change the CDW periodicity. STM studies found that the periodicity of CDW in 2H-TaS₂ shifted from 3 × 3 to 2 × 2 upon Li intercalation, and finally CDW phase disappeared under high Li-doping concentration (Figure 5F).¹⁰⁰ Calculations showed that charge doping from Li atoms disturbed phonon dispersions, which changed CDW structures (Figure 5G).

Intercalation can induce or enhance the SC phase in TMDs. Pristine 2H-MoS₂ is not superconducting. Potassium (K) intercalated 2H-MoS₂ yielded a mixed phase of 2H to 1T and 1T', and the superconducting transitions of all phases were detected at 6.9, 2.8, and 4.6 K, respectively (Figure 4D).⁷¹ In addition to alkali metals, molecular intercalants also can tune the SC transition temperatures of host TMDs. Wang et al.⁸⁵ intercalated 2H-TaS₂ with cetyltrimethylammonium ions (CTA⁺) by electrochemical method; T_c was enhanced from 0.8 K in pristine 2H-TaS₂ to 3.7 K in TaS₂(CTA⁺)_x (Figure 6B,C). The phase diagram of TaS₂(CTA⁺)_x is shown in Figure 6B, where an SC dome manifests in the plot of T_c versus CTA⁺ concentration. In pure 2H-TaS₂, electrons or holes dominate the transport properties in different temperature ranges, according to the sign of the Hall coefficient; Hall coefficient has the same sign as the charge of the dominant carrier type. However, for TaS₂(CTA⁺)_x (x = 0.1, 0.2), the Hall coefficient decreased linearly and remained negative throughout the whole temperature range (Figure 6D); this is indicative of the electron doping to the host lattice from CTA⁺, which enhanced SC critical temperature. Another factor of the enhanced T_c was the expansion of the interlayer spacing after intercalation CTA⁺, which caused the electronic structure of 2H-TaS₂ approaching the 2D limit, in which SC was enhanced on account of the stronger electron–phonon interactions for low dimensional system.¹⁰⁵

Figure 6. SC enhancement in 2H-TaS2 after intercalation of CTA+. (A) The crystal structure model of 2H-TaS2 before and after CTA+ intercalation. (B) The superconductivity dome of TaS2 (CTA+)x with Tc obtained from temperature-dependent resistance and magnetic susceptibility data. (C) Temperature-resistivity curves of TaS2 (CTA+)x (x = 0, 0.1, 0.2, 0.3, respectively). (D) Hall coefficients as a function of temperature of 2H-TaS2 and TaS2(CTA+)x (x = 0.1, 0.2). (Inset) Hall resistance shows a linear dependence on the magnetic field in TaS2 (CTA+)0.1. (A-D) Reproduced with permission from Ref.85 Copyright 2018, the Author(s). CTA+, cetyltrimethylammonium ion; SC, superconductivity

Topological superconductors

Topological superconductors, with bulk superconducting gap and Majorana fermion states on the surface or edge, are highly sought after quantum materials owing to their possible applications in fault-tolerant quantum computers.¹⁰⁶ Superconducting proximity effect at the interface¹⁰⁷ and electron doping by intercalation¹⁰⁸ are two strategies to induce SC in topological materials.

Bi₂Se₃ is an archetypical topological insulator. Cu_xBi₂Se₃ bulk crystal grown by melt-growth method showed SC at 3.8 K for $0.12\le x\le 0.15$ (Figure 7A).^{64,65,111–113} However, because the superconducting volume fraction (∼50%) of Cu_xBi₂Se₃ was relatively low, it is ambiguous if Cu_xBi₂Se₃ was a real topological superconductor or not. Liu et al.¹⁰⁹ synthesized the intercalated phase Sr_xBi₂Se₃ by melt growth method and Sr_xBi₂Se₃ achieved an SC phase at 2.5 K when x reached 0.062 (Figure 7B); the T_c was dependent on Sr concentration. Bulk SC was confirmed by a high-shielding volume fraction of 91.5%. Quantum oscillation measurements were performed to prove the existence of the surface state in Sr_xBi₂Se₃, evidencing that Sr_xBi₂Se₃ compounds were true topological superconductors (Figure 7C). STM studies gained insight into how the surface Dirac electrons were driven into Cooper pairs, shown by the obvious weakening of the quantum oscillations resulted from Landau levels within the superconducting gap.¹⁰⁸

Figure 7. Inducing superconductivity in topological TMDs by intercalation. (A) The temperature-dependent resistivity curve of a Cu0.12Bi2Se3 crystal. (Lower inset) The superconducting transition temperature is 3.8 K. (Upper inset) The magnetoresistance data collected at 1.8 K with the direction of the magnetic field parallel to the c axis. The third inset: The Seebeck coefficients obtained from CuxBi2Se3 and Bi2-xCuxSe3. Reproduced with permission from Ref.64 Copyright 2010, The American Physical Society. (B) Onset transition temperature (Tconset), zero-resistivity temperature (Tc0), and the shielding fraction plotted as a function of Sr content at 0.5 K. (C) Plots of the oscillatory component of the Hall conductance as a function of at 1/B at 0.35 K. (B,C) Reproduced with permission from Ref.109 Copyright 2015, American Chemical Society. (D) SC transition temperature Tc against intercalation concentration x. (E) Magnetoresistance of K0.78WTe2 measured at different magnetic fields parallel to c axis. (D,E) Reproduced with permission from Ref.110 Copyright 2018, American Chemical Society. TMD, transition metal dichalcogenide

Other than topological insulators, topological semimetals have stimulated great interest because of the prospect for topological SC. Bulk T_d WTe₂ and 1T' MoTe₂ are 3D type-II Weyl semimetal.^114,115 In Weyl semimetals, the conduction and valence bands contact only at Weyl points in the Brillouin zone and are protected against gap formation by crystalline symmetry or time-reversal symmetry. Bulk WTe₂ reached a maximum critical temperature (T_c) of 7 K at around 16.8 GPa.^116,117 Monolayer WTe₂ is a 2D topological insulator with conducting helical edge channels.^118,119 Intrinsic SC can be induced in monolayer WTe₂ by electrostatic gating at 0.82 K.^120,121 Bulk and monolayer MoTe₂ has intrinsic SC at 0.25 and 8 K, respectively.^122,123 K-intercalated WTe₂ synthetsized by liquid ammonia method shows improved superconducting transition temperature at 2.6 K.¹¹⁰ Figure 7D presents transport evidence of SC state in K_xWTe₂ compounds with different K concentration. The absence of phase change and the presence of positive magnetoresistance (MR; Figure 7E) indicate that the topology of WTe₂ persists after intercalation, suggesting K_xWTe₂ could be a topological superconductor.¹²⁴ Zhang et al.³⁶ intercalated organic cations in WTe₂ and MoTe₂ with an electrochemical reaction method using ionic liquids [C_nMIm]⁺ [TFSI]⁺ (1-alkyl-3-methylimidazolium-bis(trifluoromethylsulfonyl)imide) as reacting agent.³⁶ The interlayer coupling of the WTe₂ and MoTe₂ crystals are weakened by the organic intercalants, leading to an enhanced SC critical temperature at 7.0 K for intercalated MoTe₂, and 2.3 K for intercalated WTe₂. Therefore, organic cation intercalation can control the dimensionality of layered materials. The organic-inorganic hybrid lattices also show greatly improved stability compared to pristine flakes.

OPTICAL PROPERTIES OF INTERCALATED TMDS Light transmittance and photoluminescence (PL)

The optical properties of TMDs can be changed by intercalation due to changes in the band gap and dielectric constant. 1T phase MoS₂ nanoflake converted from 2H phase after intercalation showed higher optical transparency (up to 90%; Figure 8A–C).¹²⁵ The imaginary parts of the refractive index calculated from the transmission data were much smaller in lithiated MoS₂ than in pristine MoS₂, suggesting that light absorption was decreased after lithiation (Figure 8A,B). Bi₂Se₃ showed much-enhanced light transmission in the wavelength range from 400 to 900 nm after Cu intercalation by wet chemical reactions,^126,127 where the Fermi level was lifted into the conduction band. Electrons could only be excited to the first unoccupied level above the Fermi level. This blue shift of absorption edge is the Burstein–Moss effect, as illustrated in Figure 8D. As shown in Figure 8E, after Cu intercalation, the imaginary part of the refractive index of Bi₂Se₃ reduced, corresponding to the reduced absorption in the wavelength of 400–900 nm.

Figure 8. Tuning optical transmission of TMDs by intercalation. (A) Band structure model of 2H-MoS2 at the K-point in the Brillouin zone before and after Li intercalation. (B) The refractive indexes versus wavelengths were calculated from the thickness-dependent optical transmission measurements of 2H-MoS2 (top panel) and 1T-Lix = 1MoS2 (bottom panel), respectively. The real part and imaginary part of the refractive indexes are indicated by the filled blue squares and hollow red circles. (C) Optical transmission from 400 to 800 nm of pristine (red) and Li-intercalated (blue) MoS2 flake. (A-C) Reproduced with permission from Ref.125 Copyright 2015, American Chemical Society. (D) Illustration for the Burstein–Moss shift caused by the injected free electrons by intercalation. EF, Fermi level, Eg, intrinsic bandgap width. (E) Refractive indices of Bi2Se3 before and after Cu intercalation, calculated from the experimental data. Black squares represent the real parts and red squares are the imaginary parts. (D,E) Reproduced with permission from Ref.126 Copyright 2014, Springer Nature. (F) Closed aperture Z scan measurements of lithiated MoS2 with varied input laser power. At a lower pumping power, lithiated MoS2 showed saturable absorption and self-focusing behavior. Reproduced with permission from Ref.49 Copyright 2017, American Chemical Society. TMD, transition metal dichalcogenide

Tan et al.⁴⁹ synthesized phase pure, ambient stable 1T'-MoS₂ flakes by intercalation with lithium hydride. The broken inversion symmetry in 1T′ phase, coupled with carrier-dependent renormalization of electron−hole interactions, leads to much larger optical Kerr nonlinearity (Figure 8F). The stability, high transparency and nonlinear two-photon absorption loss of 1T'-Li_xMoS₂ are potentially useful for applications in waveguides and ultrafast switches in integrated photonic circuits.

The indirect-to-direct band gap transition in monolayer group VIB TMDs gives rise to enhanced PL emission. Wang et al.¹²⁸ tuned PL intensity in 2H-MoS₂ flakes with Li⁺, Na⁺, and K⁺ ions by electrochemical intercalation. Reversible PL response modulations were realized through phase transitions induced by alkali metal intercalation, with short response and recovery time (Figure 9A). As 1T-MoS₂ is metallic, the 2H to 1T phase causes PL quenching. The sensitivity of PL to the electron concentration in MoS₂ were exploited for the detection of enzymatic activities and cell viabilities in biological systems. One example was based on the K⁺ uptake ability of viable yeast cells relative to nonviable cells. When the viable cells were coated with MoS₂ nanoflakes, K⁺ ions intercalated the MoS₂ nanoflakes and quenched their PL. Nonviable yeast cells were not capable of uptaking K⁺ ions, so the PL from the MoS₂ nanoflakes coating the cells was not affected.¹³⁰

Figure 9. Tuning PL intensity and plasmonic resonance modes of TMDs by intercalation. (A) Electrochemical intercalation control of PL response in the 2D MoS2 nanoflakes. Reproduced with permission from Ref.128 Copyright 2013, American Chemical Society. (B) Red shift of plasmon resonance peak in 2H-NbSe2 as K concentration increases from 0 to 0.63. Reproduced with permission from Ref.129 Copyright 2016, American Physical Society. (C) Evolution of UV−vis absorbance spectra of a 2D MoS2 nanoflake under varied electrochemical intercalation potential, where the pristine 2D MoS2 works as the differential reference. (D) Theoretical simulation for the correlation between the plasmon resonance peak positions and Li+ ion concentration in 2D MoS2 nanoflakes for both intercalated 2H and 1T phases. (C,D) Reproduced with permission from Ref.48 Copyright 2015, American Chemical Society. PL, photoluminescence; TMD, transition metal dichalcogenide; UV−vis, ultraviolet–visible

Plasmons

Plasmons arise from the collective excitation of carriers in the metallic band. Metallic 2D materials, such as NbSe₂ or TaSe₂, can support plasmonic modes in the infrared.¹³¹ Due to its inherent subwavelength nature and spatial profile, surface plasmons can greatly accumulate the optical field and energy on the nanoscale and dramatically enhance various light–matter interactions. Plasmonics with 2D materials can find applications in the fields of thermoplasmonics, biosensing, and plasma-wave Terahertz detection.^132–134 In addition, a distinct advantage compared to conventional metals like gold or silver is that the plasmonic properties can be easily controlled by doping or electrical voltage.^48,131,135

The plasmon modes in TMDs could be modulated by tuning the free electron density, thus charge transfer by intercalants serves as a strategy for tuning plasmon modes in intercalated TMDs. The energy required to excite plasmon modes is described by ${\omega }_{\mathrm{p}}^2=n{e}^2/{\epsilon }_{0}m$, where n is the conduction electron density, e denotes the elementary charge, m the electron effective mass, and ${\epsilon }_0$ the free space permittivity. K-intercalated NbSe₂ demonstrated a shift of plasmon mode from ~0.9 eV at the intercalation state of x = 0 to ~0.55 eV at x = 0.63 (Figure 9B).¹²⁹ Similar red shift of plasmon mode was observed in K-doped 2H-TaSe₂.^136,137 Pristine 2H-TaSe₂ or NbSe₂ can be characterized by a rare case of half-filled conduction bands, and injection of either electrons or holes will shift the Fermi level upward or downward, leading to reduced conduction electron density n due to electron-hole symmetry, and thus lower the plasmon energy.¹³⁸ The plasmon frequency can be used to characterize the doping level during intercalation.

Plasmon modes in MoS₂ can be tuned from UV to the visible range by the electrochemical intercalation of Li.⁴⁸ As shown in Figure 9C, for an applied electrochemical potential between 0 and −4 V, the electron density in 2H-MoS₂ is too low to support plasmon modes. At −6 V, the electron density n increases to support two plasmon modes at 380 and 714 nm, corresponding to 1T and 2H phase, respectively. Finally, at −10 V, the phase is converted entirely to 1T phase, the 714 nm peak disappears and only 380 nm peak remains. Theoretical simulations reveal that plasmon resonance peak energy is correlated to the doping level in both 1T and 2H MoS₂, as shown in Figure 9D.

Another strategy of tuning plasmon energy in intercalated TMDs is to modulate the dielectric constant or refractive index of guest species. Surface plasmon obeys the following dispersion relation: $k\left(\omega \right)=\frac{\omega }{c}\sqrt{\frac{{\epsilon }_1{\epsilon }_2{\mu }_1{\mu }_2}{{\epsilon }_1{\mu }_1+{\epsilon }_2{\mu }_2}}$, $n=\sqrt{\epsilon \mu }$, where $k\left(\omega \right)$ is the wave vector, $\epsilon$ is the relative permittivity, and $\mu$ is the relative permeability of the material (1: the dielectric medium, 2: metal), while $\omega$ is the angular frequency, $c$ is the speed of light in a vacuum, and n is the refractive index of the material. The intensities and shape of plasmon modes in Bi₂Se₃ nanoribbons could be modulated by intercalation of pyridine-, nitrobenzene-, and dodecylamine molecules.¹³⁹ The different plasmonic properties in different organic Bi₂Se₃ composites originated from the varying dielectric constants of the molecules.¹³⁹ The diverse design of organic molecules provides infinite choices to functionalize TMDs and tune their optical properties.

MAGNETIC PROPERTIES OF INTERCALATED TMDS

2D magnets are exciting because they enable fundamental studies of magnetism in the 2D limit and are potentially easier to integrate into planar device structures compared to bulk magnets.¹⁴⁰ In addition, the electrical control of magnetic devices using 2D magnets should be more efficient than that of bulk crystals because the most effective means to switch magnetic parameters by electric field are interface-based, for instance, spin-transfer torques and voltage-controlled magnetic anisotropy.¹⁴¹ 2D magnets can also expand the functionalities of 2D electronic devices¹⁴⁰ by supporting magnetic proximity effects in a 2D heterostructure,¹⁴² affording a means to control the valley degree of freedom in adjacent 2D material.^143,144 However, most 2D materials are either nonmagnetic or the Curie temperature is too low for practical applications.² One way to induce magnetism in TMDs is by intercalating magnetic elements. In this section, we review intercalated TMD phases that possess AFM or FM order. We also describe how the guest atoms affect the spin order of the host lattice, with a few unresolved questions. The applications of intercalated TMDs in spintronics are also discussed, along with the possibilities for multiferroics.

The intercalation of magnetic atoms, such as Fe, Co, and Ni into TMDs, can induce AFM/FM orders. Magnetic intercalated TMDs are mainly synthesized by iodine vapor transport growth.¹⁴⁵ Fe-intercalated 2H-TaS₂ and 1T-TiS₂ possess out-of-plane FM order, in which an MR of 150% is achievable at certain Fe concentrations (Figure 10B,C).¹⁴⁶ The out-of-plane easy axis is attributed to the large, unquenched, out-of-plane orbital magnetic moments of Fe atoms,^147,148 and the large MR arises from spin-disorder scattering in the strong spin-orbit coupling environment.¹⁴⁹ By contrast, other TMDs intercalated with Fe element show AFM order. For instance, Fe_xNbS₂ shows AFM order at all intercalation concentrations.^145,150 As regard to the question why Fe_xTaS₂ and Fe_xTiS₂ show out of plane FM whereas most other TMDs do not, a possible mechanism may lie in the Ruderman-Kittel-Kasuya-Yosida (RKKY) type Fe-Fe interactions in different d orbital electronic configurations.^151,152

Figure 10. Tuning magnetic properties of Fe-intercalated TiS2. (A) Schematics of the crystal structure of Fe-intercalated TiS2. (B) Magnetization curve of Fe0.197TiS2 under horizontal (closed, circle), and perpendicular (closed, square) magnetic field, and in-plane MR under perpendicular (open, circle) magnetic field at T = 2 K. (C) Phase diagram of FexTiS2 presenting Tc (top, left axis, down triangle), Weiss temperature θW (top, right axis, up triangle), switching field Hs (bottom, left axis, square), and MR (bottom, right axis, circle) verse intercalation concentration x. (D) possible intercalation positions and corresponding superstructure models of Fe intercalated TiS2. From top to bottom, the periodicity of superstructures are 2 × 2, 3×3, 2 × 2 + 3×3, and ion cluster phase. (E) Phase diagram of Fe-RKKY interaction model in ferromagnetic (J1 = 1) and antiferromagnetic (J1 = −1) state. (A‒E). Reproduced with permission from Ref.146 Copyright 2019, American Physical Society. MR, magnetoresistance; RKKY, Ruderman-Kittel-Kasuya-Yosida

The magnetic critical transition temperatures and even spin ordering of intercalated TMDs can be modified by the intercalation concentration. The Curie temperature of Fe_xTaS₂ decreases monotonically as Fe concentration increases, for example, from 160 to 70 K when x increases from 0.25 to 0.28^149,153; when x further increases above 0.4, Fe_xTaS₂ changes from FM to AFM order with a Néel temperature of 85 K.⁵⁵ Meanwhile, the periodicity of superstructures change from 2$a_0$ at x = 1/4 to ${\sqrt{3}a}_0$ at x = 1/3, where $a_0$ is host lattice constant.¹⁴⁵ The critical temperature and spin order depend on the structure of the supercell, which is concentration-dependent. Calculations based on the inter Fe-RKKY interaction model predict several possible supercells and the corresponding phase diagram for FM and AFM ordering, as shown in Figure 10D,E.¹⁴⁵

The ability to induce FM order is not only limited to FM intercalants. Zhao et al.⁴³ showed that even nonmagnetic atoms could produce magnetism in a broad class of TMDs for specific intercalation state, and they named this class of materials “ic-2D” materials. The self-intercalation of Ta into TaS₂ thin film occurs under metal-rich chemical potential during MBE or CVD growth (Figure 11A), and the native metal atoms occupy the octahedral vacancies in the vdW gap. When the intercalation ratio increases successively from 25%, 33%, 50%, 66.7%, to 100%, superstructures appear to change from $2\times \sqrt{3}$, $\sqrt{3}\times \sqrt{3}$, glassy phase, Kagome lattice to a fully intercalated 1 × 1 phase (Figure 11B). For 33% Ta intercalation—Ta₇Se₁₂, an out-of-plane FM order is discovered by anomalous Hall effect. The double-exchange mechanism is triggered by charge transfer from intercalated Ta to TaS₂ lattice. Inserted Ta atoms induce additional spin-split bands across the Fermi level, resulting in a magnetic state. DFT calculations show that the magnetic moments are localized on the d orbitals of the intercalated Ta atom, as evidenced by the calculated orbital-resolved spin-up and spin-down band structures. At a low intercalation state, the charges transfer from inserted Ta to host atoms is stronger, with a consequently stronger induced magnetic moment on the intercalant, than at a high intercalation state. The importance of this finding is that it suggests that ferromagnetism could be developed in a wide class of self-intercalated TMDs, which opens up vast opportunities for 2D magnets research.

Figure 11. Self-intercalated TMD grown by MBE. (A) Schematic showing growth of self-intercalated 2Ha-stacked TaxSy. (B) High resolution STEM imaging of self-intercalated TaxSy with different composition, ranging from 25% Ta-intercalated Ta9S16, 50% Ta-intercalated Ta10S16, 66.7% Ta-intercalated Ta8Se12 to 100% Ta-intercalated Ta9Se12 crystals. Scale bar = 2 and 0.5 nm for upper panels and zoomed images, respectively. (A,B) Reproduced with permission from Ref.43 Copyright 2020, Springer Nature. (C) The cross-sectional STEM image of a multilayer Ta1 + xSe2 epitaxial film grown by MBE. Red arrows mark the self-intercalated Ta atoms between the TaSe2 layers. The left panel shows a schematic structure of the 3R phase unit cells. Reproduced with permission from Ref.154 Copyright 2020, American Chemical Society. (D) LEED and STM pattern of VTe2 mono- (top panels) and multilayer (bottom panels) films grown by MBE. Top panels: 1 × 1 pattern on monolayer 1T VTe2. Bottom panels: 2 × 1 reconstruction on bilayer VTe2 with 1/2 layer of V intercalated. Reproduced with permission from Ref.155 Copyright 2020, American Chemical Society. LEED, low-energy electron diffraction; MBE, molecular beam epitaxy; STEM, scanning transmission electron microscope; STM, scanning tunneling microscopy; TMD, transition metal dichalcogenide

The self-intercalation of TMD to form ic-2D TMD reported by Zhao et al.³⁶ has been verified by several groups. Iwasa and coworkers synthesized self-intercalated 3R Ta_{1 + x}Se₂ by MBE method,¹⁵⁴ in which the intercalated Ta atoms in Ta_{1 + x}Se₂ stabilized the 3R phase. Cross-sectional STEM revealed the presence of a layer of intercalated Ta atoms between the TaSe₂ layers (Figure 11C). The 3R Ta_{1 + x}Se₂ system exhibited exotic superconducting properties, including the highest T_c of 3.0 K among all the TaSe₂ polymorphs and giant in-plane upper critical fields. The 3R phase is a unique material system because both in-plane and out-of-plane lattice symmetry are broken, which may lead to intriguing spin-dependent phenomena even in bulk form. Lasek et al.¹⁵⁵ synthesized monolayer to multilayer self-intercalated early transition metal (Ti, V, and Cr) tellurides by MBE. The concentration of the intercalated atoms can be tuned from 1/4, 1/3, to 1/2, forming superstructures with the periodicity of 2 × 2, ($\sqrt{3}\times \sqrt{3}$) R 30°, or 2 × 1, as observed by STM. Figure 11D shows low-energy electron diffraction and STM patterns showed 1 × 1 atomic lattice of monolayer 1T VTe₂ and reconstruction of 2 × 1 in 1/2 intercalated VTe₂.

The convenience of self-intercalation as a means to tune the properties of TMD arises from the ease of implementing it in CVD or MBE growth systems, where the chemical potential of the metal species relative to the chalcogen can be changed readily. The ability to tune the properties of ic-2D materials by changing the intercalation concentration affords a finer modulation compared to the metal-semiconductor type transition in stacking-layer induced phase change. Other than crystalline phases, a glassy alloy type state may form at a specific intercalation state where short-range order forms due to the interplay between geometric frustration and Coulomb interactions. The field of ic-2D is still in its infancy and a large family of ic-2D material remains to be discovered.

Electric field control of magnetism boosts the development of next-generation low-power spintronics and microelectronics, such as “electric-writing and magnetic-reading” or “electric-writing and reading” memory devices. AFMs are particularly attractive in spin-transfer torque memory devices due to their negligible magnetic moments, making them robust to magnetic perturbations.⁵⁰ AFMs also allow ultrafast switching at THz speed, which is orders of magnitude higher than the GHz switching of ferromagnets.¹⁵⁶ Spin-orbit torque switching of AFMs has the potential to operate in timescale and energy scale of picosecond and attojoule, respectively, which are superior to the FM counterparts by 2–3 orders of magnitudes. Some magnetically intercalated TMDs are AFMs, and are good candidates to investigate magneto-electric coupling. Fe-intercalated NbS₂ crystal, Fe_1/3NbS₂ undergoes AFM order transition at 42 K.⁵⁰ Nair et al.⁵⁰ successfully reoriented AFM order in Fe_1/3NbS₂ by current pulses as low as 10⁻⁴ A/cm², with switching durations at 10 μs at low temperature. Other than Fe_1/3NbS₂, it is expected that other magnetic intercalated TMD exists that can show room temperature AFM/FM switching.

Multiferroics are functional materials with more than one ferroic order, such as ferromagnetism and ferroelectricity. These materials render switchable magnetic moment through coupling between ferroelectricity and ferromagnetism, which is ideal to realize “electric writing and magnetic reading” for high-density data storage. However, multiferroics rarely exist in nature due to the opposite d-orbital occupations imposed by ferroelectricity and ferromagnetism.¹⁵⁷ It is interesting to explore the potential of intercalated TMD as multiferroics, for example, an intrinsically 2D ferroelectric material can also be made FM by intercalation, and that magnetoelectric coupling occurs when switching of the ferroelectric polarization simultaneously switches the spin state. First-principle calculations predict that the intercalation of transition metal ions such as Ni, Cu, Ag, Cd into TMDs (e.g., MoS₂, Bi₂Se₃, or CrS₂) can generate both ferromagnetism and ferroelectricity, and that these could be stable in the ambient.^42,158 In the initial intercalation state of AB-stacked Cr-intercalated MoS₂, Cr ion is tetrahedrally coordinated with one bond to the upper layer and three bonds to the lower layer; this causes the vertical distance between Cr ion and the top layer to be longer than the distance of Cr ion to the bottom layer. The polarization direction is downward at the initial state. The displacement of Cr ions causes switching of vertical polarization with a low-energy barrier of 0.36 eV per Cr ion. At the same time, Cr_xMoS₂ is FM with a magnetic moment of 0.3 µB per Cr ion. The spin-density distribution is coupled to electrical polarization, according to DFT calculations. Experiments are needed to verify if magneto-electric coupling can occur according to theoretical predictions.

THERMOELECTRIC (TE) PROPERTIES OF INTERCALATED TMDS

Thermoelectricity is the direct conversion of electricity to heat and vice versa. TE technology can directly convert waste heat energy into clean electric power; therefore, it improves the efficiency of energy consumption and alleviates energy and environmental problems.¹⁵⁹ The TE performance of a material is characterized by the figure of merit $\text{ZT}=\mathrm{\sigma }{S}^{2}T/\rho$, where $S$, $\mathrm{\sigma }$, $\rho$, and $T$ are the Seebeck coefficient, electrical conductivity, thermal conductivity, and absolute temperature, respectively.²¹ From the equation above, to improve ZT, one needs to reduce thermal conductivity and increase electrical conductivity. TMDs with high mobility have shown good TE performances, and their TE properties can be further improved by intercalation. The mechanism whereby intercalation improved thermoelectricity is mainly due to two factors. First, the guest species introduced in host material can perturb phonon propagation, which reduces thermal conductivity.⁴⁵ Second, intercalants transfer charges to TMDs and improve their electrical conductivity.

Intercalation can lower the thermal conductivity of the TE material by increasing phonon scattering. TiS₂ is a 2D anisotropic TE material. Intercalation of TiS₂ has resulted in better TE performances, and intercalants, such as Cu,^160,161 Co,^162,163 Nb,¹⁶⁴ Nd,¹⁶⁵ Ag,¹⁶⁶ Ta,¹⁶⁷ Ti,^168,169 and organic molecules,^44,170 have been investigated. Among the various intercalated TiS₂ materials, an organic-TiS₂ hybrid material in the form of TiS₂/[(hexylammonium)_x(H₂O)_y(DMSO)_z] is particularly interesting as a n-type and flexible TE material. TiS₂/[(hexylammonium)_x(H₂O)_y(DMSO)_z] was synthesized by a facile two-step intercalation method combing of electrochemical intercalation and solvent exchange (Figure 12A).⁴⁴ The thermal conductivity of TiS₂ nanocomposites significantly decreased from $4.45\mathrm{W}/\left(\text{mK}\right)$ in pristine TiS₂ to $0.12\mathrm{W}/\left(\text{mK}\right)$ in the intercalated phase. As a result, the ZT value increased from 0.08 to 0.28 at 373 K (Figure 12B). Theoretical calculations revealed that the decreased thermal conductivity resulted from the broadened acoustic dispersions, which indicated strong phonon scattering after intercalation. The strong phonon scattering was attributed to the increased interfacial coupling between host and guest atoms, that is, enhanced electrostatic interactions in the vdW gaps. In addition, the elastically strained crystal planes created after intercalation decreased the phonon mean free path (Figure 12C). Zhu et al.³⁷ showed that both in-plane and out-of-plane thermal conductivity were tunable in MoS₂ by electrochemical lithiation (Figure 12D). Most strikingly, the anisotropic ratio of thermal conductivity increased from 52 (x = 0) to 110 (x = 0.34) in Li_xMoS₂, as lithiation reduced the phonon mean-free path to a greater degree in the out-of-plane direction than in the in-plane direction.³⁷

Figure 12. Thermal conductivity of intercalated TMDs. (A) Schematic displaying synthesis of TiS2-based inorganic/organic superlattices. TiS2[(HA)x(DMSO)y] superlattice was synthesized by electrochemical intercalation. A bilayer structure of the hexylammonium ions formed in the vdW gap due to DMSO stabilization. TiS2[(HA)x(H2O)y(DMSO)z] was synthesized by the solvent exchange process after immersion in water, with the hexylammonium ions in a monolayer configuration. (B) Comparison of in-plane thermoelectric figure of merit, ZT of pristine TiS2 and inorganic/organic TiS2 superlattices. (C) STEM characterization of the wavy structure of TiS2[(HA)x(H2O)y(DMSO)z] hybrid composite. (A–C) Reproduced with permission from Ref.44 Copyright 2015, Springer Nature. (D) In-plane and out-of-plane thermal conductivity measurements of lithiated MoS2 samples. Reproduced with permission from Ref.37 Copyright 2016, Springer Nature. DMSO, dimethyl sulfoxide; STEM, scanning transmission electron microscope; TMD, transition metal dichalcogenide; vdW, van der Waals

Further, cointercalation/doping was found to enhance electrical conductivity without compromising the Seebeck coefficient, leading to enhanced power factor and Z value. In a recent study, Cu was intercalated into the vdW gap of SnSe₂ and Br was doped into lattice simultaneously by vapor phase reactions.⁵¹ The Cu intercalant and Br dopant interacted strongly in SnSe₂ lattices and delocalized the electrons around Sn-Se covalent bonds (Figure 13A). The Cu-Br pair acted as “electric bridges” between the SnSe₂ layers. As a result, carrier concentration and mobility increased, leading to the enhanced electrical conductivity of SnSe₂. The Seebeck coefficient of SnSe₂ was maintained after doping, together with the increased electric conductivity, it enabled a high power factor of 12 µW/(cm·K²) from 500 to 773 K and a high ZT value of 0.67 at 773 K (Figure 13B,C).

Figure 13. (A) Modulating thermal electrical properties in SnSe2 by intercalation of Cu and doping of Br atoms. Mapping of electron localization function along the 〈100〉 zone axis for SnSe1.99Br0.01 (top left) and SnCu0.01Se1.99Br0.01 (top right). Green, orange, purple, and red spheres indicate the Sn, Se, Br, and Cu atoms, respectively. Bottom panel: charge transfer from Cu to Br atoms in SnCu0.01Se1.99Br0.01. Cyan ellipsoids indicate electron loss. (B,C) Comparison of power factor and ZT in pristine and intercalated SnSe2. (A–C) Reproduced with permission from Ref.51 Copyright 2019, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. (D) Simulated cross-plane thermal conductance of a 10 nm thick lithiated MoS2 film against c axis strain (%), relative to the pristine unlithiated 2H-MoS2 (solid black circle). (E) Plot of thermal conductance G and Li concentration χ as a function of electrochemical voltage, from which large hysteresis is observed between the charge and discharge curves. (D,E) Reproduced with permission from Ref.171 Copyright 2018, Springer Nature

One ultimate goal of tuning thermal conductance is to build thermoelectric devices with mechanical robustness and fast switching speed. Recently, lithiated MoS₂ synthesized by electrochemical method was applied as a thermal transistor.¹⁷¹ The thermal transistor constructed using 10 nm-MoS₂ nanosheets operated with an on/off ratio of nearly one order between delithinated and lithiated states (Figure 13E). The mechanism for the modulation of thermal conductance was explained to be caused by a combination of the following factors: enhanced phonon scattering from Li rattler modes, phonon softening because of strain along the c axis, stacking disorder due to a mixture of 2H and 1T phase, and mesoscopic disorder (Figure 13D).

SUMMARIZING REMARKS

Layered TMDs and other 2D materials that possess a vdW gap can be intercalated by atoms, ions, or molecules to form new compounds. Intercalation induces phase transitions, produce new correlated electronic states, creates magnetic ground states, or shifts of phonon modes. The study of intercalated phases in topological superconductors helps to construct the phase diagram of correlated electron systems. Besides the doping effect, intercalation-induced lattice distortions can also modulate electronic⁵⁶ or piezoelectric properties due to the breaking of in-plane and out-of-plane symmetry, but this remains little studied.¹⁷²

Besides foreign atoms or ions, it was recently discovered that native metal atoms could be intercalated, leading to a large class of nonstoichiometric compounds.⁴³ At a high intercalation state of metal ions or atoms, the vdW gap is closed and a compound that is covalently bonded in both in-plane and out-of-plane directions is formed. However, there are technological bottlenecks with regard to the use of intercalation as a way to tune electronic properties in semiconductors. Although compositional tuning of the host-to-intercalant ratios provides a handle to tune phase transitions and other properties, the precise control of the intercalation state is challenging because of the inhomogeneous spread of intercalants arising from sluggish diffusion kinetics and phase segregation of the intercalant. Deintercalation can also occur when electric field is applied, or if the intercalated material is heated, thus there is a limited voltage or temperature window for the operation of devices based on the intercalation phase. Another issue that precludes practical application is the stability of the intercalated phase. Most alkali-metal intercalated TMDs are not stable in air; this means any properties promoted by charge transfer from the intercalants are also unstable. In contrast, organic-based intercalants and host TMDs form a robust hybrid structure, thus more efforts should be directed at discovering the potentials of self-assembled molecular arrays and 2D materials to form layer-by-layer hybrid organic-inorganic superstructures.¹⁷³ The attractiveness of such a system comes from the fact that a bulk material can be rendered truly 2-D by virtue of the decoupling of interlayer interactions by the organic layers, thus it is interesting to investigate if layer-dependent aniferroelectric or AFM properties can be modulated by intercalation. More studies in these areas are needed.

One topic that has received scant attention is the issue of defect pairing. The presence of single vacancies or antisite vacancies may enhance the intercalation rate because the intercalant atom and a vacancy or antisite can form a defect pair. Intercalation also allows the atomic centers responsible for the ferroic order to be physically separated. For example, the intercalant can be an FM atom, while the ferroelectricity arises from the displacement of some other atoms in the host. Magneto-electric coupling in such a system has not been investigated, and the intercalated phases that are coupled to magnetically active defect sites provide an excellent platform for this investigation.

ACKNOWLEDGMENT

Kian Ping Loh acknowledges the support from Singapore's Ministry of Education Tier 3 grant: “Two-dimensional crystal quantum exciton photonics (MOE2018-T3-1-005).” The authors would like to acknowledge the Shenzhen Peacock Plan (KQTD201653112042971).