Spontaneous self-intercalation of copper atoms into transition metal dichalcogenides

See allHide authors and affiliations

Science Advances  14 Feb 2020:
Vol. 6, no. 7, eaay4092
DOI: 10.1126/sciadv.aay4092


Intercalated transition metal dichalcogenides (TMDs) have attracted substantial interest due to their exciting electronic properties. Here, we report a unique approach where copper (Cu) atoms from bulk Cu solid intercalate spontaneously into van der Waals (vdW) gaps of group IV and V layered TMDs at room temperature and atmospheric pressure. This distinctive phenomenon is used to develop a strategy to synthesize Cu species–intercalated layered TMD compounds. A series of Cu-intercalated 2H-NbS2 compounds were obtained with homogeneous distribution of Cu intercalates in the form of monovalent Cu (I), occupying the tetrahedral sites coordinated by S atoms within the interlayer space of NbS2. The Fermi level of NbS2 shifts up because of the intercalation of Cu, resulting in the improvement of electrical conductivity in the z-direction. On the other hand, intercalation of Cu into vdW gaps of NbS2 systematically suppresses the superconducting transition temperature (Tc) and superconducting volume fraction.


Layered materials, particularly layered transition metal dichalcogenides (TMDs), have attracted extensive research interests due to their exotic two-dimensional (2D) physical properties and chemical reactivity. The layered TMDs have hexagonal layered structures in which a layer of metal atoms is covalently bonded to, and sandwiched between, two layers of chalcogens (X─M─X, X = S, Se, and Te, M = group IV to VII transition metals) (1, 2). The X─M─X layers are weakly bonded by van der Waals (vdW) forces, which permit intercalation of guest species, such as atoms, ions, and inorganic and organic molecules, between the sandwiched layers with unique physiochemical properties contrast to their parent TMDs. Specifically, the intercalated foreign species can tune electronic structure of TMDs leading to the change of bandgap and shift of Fermi energy, which, in turn, induces a variety of extraordinary electron and photon transport as well as thermal, thermoelectric, magnetic, catalytic, and electrochemical properties (3, 4). For example, superconducting dome emerged in alkali metal atom–intercalated MoS2 and TaS2 (5, 6). Among foreign species, the intercalation of metal atoms into TMDs can potentially form 2D metal-semiconductor heterostructures, which can develop truly 2D physics (e.g., 2D excitons, 2D magnets, commensurate-incommensurate transition, etc.) and heterostructure devices (e.g., such as tunneling transistors, light-emitting diodes, etc.) (79). Meanwhile, the metal atom–intercalated TMDs can also provide new possibilities in chemistry, especially in the area of catalysis and electrochemical energy storage (1013). For instance, the chemical reactivity of intercalated TMDs can be tuned by the work function through controlling the intercalated species and its concentration (12).

A large number of transition metal atoms could intercalate into group IV and V layered TMDs to form well-defined intercalation compounds, M′xMX2 (M′ is the intercalation metal different from the one in the host material, x intercalation state) (14, 15). Most of them were synthesized by the solid-state reaction or chemical iodine-vapor transport (CVT) methods, both at the elevated temperatures (650° to 1100°C) (1621). This high-temperature solid or vapor phase reaction forms homogeneous intercalation compounds; however, some defects are introduced into TMDs hosts, and meanwhile, the process is energy and time intensive. This high-temperature method often results in a low intercalation state of M′ (x < 0.65). Recently, a special example of solid-state metal capturing was reported to form cadmium (Cd)–intercalated layered tellurides at 80°C, ascribed to the effectiveness of the Lewis acid (Cd)–base (Te2−) reaction. However, this reaction also suffers from low intercalation state that is less than 0.3 (22). A low-temperature chemical method was developed to intercalate zero valent transition metal atoms into the layered materials with high content based on solution disproportionation redox reaction (9, 2327). This chemical method heavily relies on complex metal precursors and interlayer spacing of vdW gaps in layered materials. Meanwhile, it lacks of the accurate controllability of metal atom concentration and poor chemical homogeneity. To solve this issue, the electrochemical method used electrochemical potential gradient to homogeneously intercalate Cu into Bi2Se3 with maximum x up to 6.7 (Cu6.7Bi2Se3) at room temperature (28). However, the electrochemical intercalation needs to be implemented in a battery cell configuration, resulting in low efficiency and difficulty in scale up. So far, there is still a lack of an energy-efficient, concentration-tunable, and process-viable method to synthesize homogeneous intercalation compounds that can potentially form 2D/2D conductor/semiconductor heterostructures.

Here, we report a new strategy to synthesize homogeneous Cu-intercalated TMD compounds with a high intercalant concentration at room temperature and atmospheric pressure by solid-state reaction. When interacting with the group IV and V layered TMDs (MX2, M = Ti, V, Ta, and Nb, X = S and Se), the bulk metallic Cu powder or foil could be spontaneously transformed into small size particles, followed by self-intercalating into the vdW gaps between sandwiched X─M─X slabs to yield ternary CuxMX2 (0 < x ≤ 1.2) compounds under ambient conditions. In contrast, the Cu atoms cannot intercalate into the group VI layered TMDs (M = Mo and W) under the identical conditions. The intercalant concentration (x) of Cu is accurately controllable by this spontaneous process with a maximum value of 1.2. A variety of physical characterizations consistently reveal that the intercalated Cu atoms are in the form of monovalent Cu (I) and locate in tetrahedral sites within the vdW gaps. The Fermi level of TMD (e.g., NbS2) shifts up because of the intercalation of Cu, resulting in the improvement of electrical conductivity in the z-direction. On the other hand, intercalation of Cu into vdW gaps of NbS2 systematically suppresses the superconducting transition temperature (Tc) and superconducting volume fraction.


Theoretical prediction of self-intercalation of Cu atoms into TMDs

The first-principles calculations were performed to study the change of Gibbs free energy (ΔG) after intercalation of Cu atoms into vdW gaps of TMDs. The relatively most stable phases of different TMDs were chosen as the host materials. They are 1T-TiS2, 1T-ZrS2, and 1T-HfS2 for group IV TMDs, 1T-VS2, 2H-NbS2, 3R-NbS2, 2H-NbSe2, and 1T-TaS2 for group V TMDs, and 2H-CrS2, 2H-MoS2, and 2H-WS2 for group VI TMDs. Under the dilute intercalation state of molar ratio of Cu:TMD = 1: 2, all the ΔG values are negative for intercalation of Cu atoms into vdW gaps of aforementioned 1T-TiS2 and group V TMDs, indicating that the intercalation is a spontaneous process (Fig. 1, fig. S1, and table S1). Among them, the ΔG for NbS2 shows the most negative value (−0.494 eV for 2H-NbS2 and 0.522 eV for 3R-NbS2). In contrast, the ΔG values are very positive for group VI TMDs, e.g., 0.357 eV for 2H-CrS2, 1.174 eV for 2H-MoS2, and 1.534 eV for 2H-WS2, suggesting that intercalation of Cu atoms into vdW gaps of group VI TMDs is thermodynamically nonspontaneous.

Fig. 1 Theoretical computing of self-intercalation of Cu atoms into vdW gaps of group IV-VI TMDs.

(A) Schematic of spontaneous self-intercalation and migration of Cu atoms in the vdW gap of 2H-NbS2 and the corresponding energy level. (B) Density functional theory (DFT) calculated Gibbs free energy change upon intercalation of Cu atoms into vdW gaps of TMDs at a dilute intercalation state (Cu:Nb = 1:2). The red solid line in the middle column for group V TMDs represents 2H-NbS2, while the line with red dashes and ling with red circles denote the 3R-NbS2 and 2H-NbSe2, respectively.

Since the self-intercalation of Cu atoms into NbS2 occurs most feasibly, the intercalation state of CuxNbS2 compound was further investigated. The Cu atom is initially intercalated at the edge of vdW gaps and then freely hops to the center of vdW gaps with a relatively low diffusion energy barrier of 0.41 eV (Fig. 1A), suggesting that vdW gaps can accommodate a Cu intercalation state up to x = 2 according to the number of tetrahedrons composed of sulfur atoms in the vdW gaps. However, this maximum intercalation state cannot be achieved because of restriction of intercalation energy change. As the molar ratio of Cu:2H-NbS2 gradually increases, ΔG shifts positively. The threshold intercalation state of Cu is 1.0 at which the ΔG remains negative (−0.170 eV). Further increasing the intercalation state to 1.2, ΔG becomes slightly positive (0.058 eV). The classic synthetic method of M′xMX2 at high temperatures often resulted in a low intercalate state (x < 0.65). To reveal the origin of low intercalation concentration at high temperatures, ab initio molecular dynamic (AIMD) simulation was first used to investigate the stability of different Cu coverages on the surface of 2H-NbS2 at 900°C, a typical temperature used for the synthesis of CuxNbS2 by the solid-state or CVT reaction route (fig. S2) (18, 20, 29). It revealed that some of Cu atoms were peeled off the surface of 2H-NbS2 at this high temperature, until the atomic ratio drops to 0.6:1.0 (Cu:Nb). This theoretical maximum intercalant concentration is close to a previous experimental value obtained in high-temperature synthesis (18, 19).

Experimental verification of Cu self-intercalation

Fig. 2A illustrates the process of spontaneous intercalation of Cu atoms into layered TMDs at room temperature. The Cu powders (1 or 5 μm) first mixed with group IV, V, and VI layered TMD powders (see morphology and crystal structure of TMDs in figs. S3 and S4) in hexane under ambient conditions, followed by magnetic stirring for different hours depending on the type of TMDs and the molar ratio of Cu over TMD. Figure 2B and fig. S4 clearly display that the first powder x-ray diffraction (PXRD) peaks, including (002) for 2H-NbS2 and 2H-NbSe2, (001) for 1T-TiS2, 1T-TaS2, and 1T-VS2, and (003) for 3R-NbS2, shift to a lower angle region. In contrast, the (002) peaks for both 2H-MoS2 and 2H-WS2 almost remain at the same locations. The PXRD results indicate that Cu species could intercalate into the vdW gaps of group IV and V TMDs (e.g., 2H-NbS2, 2H-NbSe2, 3R-NbS2, 1T-TiS2, 1T-TaS2, and 1T-VS2), resulting in the enlarged interlayer d-spacing. Cu species, however, fail to intercalate into vdW gaps of group VI TMDs (e.g., 2H-MoS2 and 2H-WS2). These results corroborate the density functional theory (DFT) predictions.

Fig. 2 Fabrication of Cu-intercalated layered TMDs under room temperature.

(A) Schematic of the spontaneous intercalation of Cu atoms into interlayer space of layered TMDs at room temperature (RT) and ambient pressure. (B) XRD patterns of different TMDs after Cu self-intercalation. a.u., arbitrary units. (C) XRD patterns of CuxNbS2 with various Cu intercalation states.

The micrometer-scaled Cu powders self-assemble into nanostructures preceding the self-intercalation process when interacting with the TMDs. The initial Cu powders in micrometer size are fragmented into nano size and further down to smaller size when they contact the surface of 2H-NbS2 as shown in the scanning electron microscopy (SEM) images (fig. S5), scanning transmission electron microscopy (STEM), and x-ray energy-dispersive spectroscopy (XEDS) elemental mapping (figs. S6 and S7). Then, the atomic-scaled Cu intercalates into the vdW gaps at the edges, followed by hopping to the middle of vdW gaps as the DFT modeling suggested (Fig. 1). Although the Cu atom cannot intercalate into the vdW gaps of group VI TMDs (2H-MoS2 and 2H-WS2), the nanosized phenomena were also observed on the surfaces of 2H-MoS2 and 2H-WS2, which were affirmed by the PXRD, SEM, TEM, and XEDS elemental mapping characterizations (figs. S8 to S15). These results reveal a unique phenomenon that the bulk Cu powder can be spontaneously transformed into nano/atomic-sized Cu species on the surface of TMDs under ambient conditions, which is largely attributed to the fact that the bond dissociation energy of Cu─Cu (176 kJ mol−1) is far less than the binding energy of Cu─S (276 kJ mol−1) (30, 31).

Structure evolution of CuxNbS2 compounds

To further understand the self-intercalation process, we investigated Cu-intercalated 2H-NbS2 in detail. In a mixture of Cu and 2H-NbS2 powders with a molar ratio of 1.2:1, the Cu powders visibly vanish after stirring for 5 days (fig. S5), while the intercalation process simultaneously occurs to form a nominal CuxNbS2 compound as indicated by the down shift of diffraction peak (002) to a smaller angle (Fig. 2B and fig. S16). The structural evolution upon intercalation under different Cu concentrations was further monitored in detail by PXRD measurements (Fig. 2C and fig. S17). For pristine 2H-NbS2 with the lattice parameter of c = 1.189 nm, the reflection at 2Θ = 14.76° can be indexed as (002) plane. As the Cu intercalation state x gradually increases, the intensity of (002) peak for 2H-NbS2 first decreases and then disappears when x rises to higher than 0.65, whereas two new reflections start to grow (Fig. 2C). The first new reflection is a broad peak originally emerging at 2Θ = 14.26° at x = 0.1. The position of this peak keeps shifting to a lower diffraction angle as x increases. This broad reflection could be assigned to an intermediate phase having a high-order intercalation stage under the low intercalation state, and consequently, it cannot be indexed with integral (hkl) indices (32). In contrast, the second new reflection at a relatively lower diffraction angle propagates, accompanied by the decrease of peak intensity of the 2H-NbS2 phase. This new reflection peak suddenly becomes sharp at x = 0.3 largely because of the transition of intercalation stage from high order to first order (33). This sharp reflection can be indexed as (002) plane of CuxNbS2, of which the c lattice constant enlarges as the intercalation state intensifies. Compared to that of 2H-NbS2, the c lattice constant increases to 1.307 nm for Cu0.3NbS2 and further to 1.321 nm for Cu1.2NbS2 (fig. S18). Accordingly, the neighboring Nb–Nb distance along the c axis (dNb–Nb = c/2) increases from 0.597 nm for 2H-NbS2 to 0.661 nm for Cu1.2NbS2, owing mainly to the expanding of interlayer spacing upon Cu intercalation since the sandwiched NbS2 slab height only changes slightly (19, 34). The swelling of dNb-Nb leads to the gradual shift of (002) peak of CuxNbS2 to a lower diffraction angle until x = 1.2 upon which the diffraction peaks of Cu completely disappears (fig. S17), suggesting that a saturation of Cu intercalation state would be reached at this point. The homogeneous Cu1.2NbS2 has a much higher intercalation state than the previous ones synthesized at high temperatures (1621). Our experimental saturation of Cu intercalation state is very close to that from the DFT calculations (table S1). Beyond the saturated intercalation state, we found that excess Cu powders were transformed into atomic-scale Cu covering the edges and outmost basal planes of CuxNbS2 (figs. S6 and S7). The exact Cu content in those CuxNbS2 (x ≤ 1.2) compounds was analyzed quantitatively by electron probe microanalyzer (EPMA), as shown in fig. S19 and table S2, which is in accordance with the feed ratio. It is worthy to point out that, when using CuO and Cu2O as the Cu sources, no Cu intercalation into 2H-NbS2 was observed under the identical experimental conditions as PXRD results show no peak shift (fig. S20). These results demonstrated that only zero-valent Cu can be used as source for intercalation of Cu atoms into the interlayer space of group IV and V TMDs.

In situ PXRD study of the spontaneous intercalation process

To directly observe the spontaneous process of Cu intercalation, we carried out in situ PXRD to investigate the structure change with time. First, the micrometer-sized 2H-NbS2 and micrometer Cu powders were mixed in hexane under stirring. The function of stirring is to enhance the Cu nano-sizing over the 2H-NbS2 surface. The mixture was periodically sampled for ex situ PXRD up to stirring duration of 66 hours (fig. S21). After that, this sample was continuously characterized by in situ PXRD without any disturbance (figs. S22 and S23). We clearly observed that the intensity of (002) peak for 2H-NbS2 decreased while the intensity of (002) peak for CuxNbS2 increased, both in a slower rate in the first 66 hours. After that, the decreasing/increasing rate was accelerated because Cu intercalation was facilitated when nanosized Cu species had covered both the basal plane and edge surfaces of 2H-NbS2. To further observe the whole spontaneous intercalation process, in situ time-resolved PXRD was used to detect the Cu intercalation in a sample of nanosized NbS2 sheets supported on a Cu foil (thickness, 100 μm) without any disturbance. After 5 min (scan number 3), the CuxNbS2 phase was clearly observed. As the time increased to 110 min (scan number 100), the (002) reflection peak of 2H-NbS2 completely disappeared (fig. S24). These results illustrate that the Cu intercalation is a fast process as long as 2H-NbS2 and Cu source have large contact interface.

Crystal and chemical structure of Cu intercalant in CuxNbS2

The structure of Cu intercalant in Cu1.2NbS2 was initially characterized by the projected electron density mapping along the [110] direction of 2H-NbS2 derived from the PXRD pattern. The atom distribution as examined by maximum-entropy method (MEM)/Rietveld analyses confirmed that the host lattice is changed from 2H-NbS2 to 2H-MoS2 structure upon Cu intercalation (fig. S1). Maximum-entropy method (MEM)/Rietveld analyses also reveal that the charge density distribution is elongated toward the S sites surrounding the Cu sites (Fig. 3A). The Cu atoms occupy the tetrahedral sites coordinated by S atoms within the interlayer space (Fig. 3, A and B). As a result, the maximum intercalation state can reach x = 2.0 in the simulated structure. In reality, intercalation energy change, ΔG, transits to more positive as the Cu intercalation concentration increases. Thus, the practical highest intercalation state is always less than 2 (maximum x = 1.2 in our CuxNbS2). To directly observe the Cu atoms between TMD sandwiched layers, we used high-resolution STEM for cross-sectional imaging. A STEM cross-sectional image was taken along the [110] direction to directly verify the intercalation of Cu atoms into vdW gaps and determine the Cu occupation structure. An intercalant layer with a zig-zag lattice can be clearly seen in Cu1.2NbS2 compared to pristine 2H-NbS2 (Fig. 3, C and D), showing the tetrahedral occupation of Cu atom in the interlayer space, which corroborates the atom distribution result obtained from PXRD (Fig. 3A). Compared to that of pure 2H-NbS2 (Fig. 3E), the corresponding atom intensity profile for Cu1.2NbS2 also evidently displays an additional layer of atoms between the two sandwiched S─Nb─S layers (Fig. 3F). The atom number for the foreign atom is smaller than Nb as implied by its relatively lower atom intensity, which is consistent with intercalant Cu. Note that Cu cluster or particles were not observed on the surface of Cu1.2NbS2 along the [001] direction (fig. S25).

Fig. 3 Structural characterization of NbS2 and CuxNbS2.

(A) Projected electron density mapping of Cu distribution in 2H-NbS2 in the [110] direction by PXRD, revealing that the Cu atoms occupy the tetrahedral sites within the interlayer space. (B) 3D ball-and-stick model of CuxNbS2 structure. (C and D) STEM image of 2H-NbS2 and Cu1.2NbS2, respectively. The inset images in (C) and (D) are atomic structural models of 2H-NbS2 and Cu1.2NbS2, respectively. (E and F) Atom intensity profiles as outlined in dashed boxes in (C) and (D).

The STEM image shows that the dNb-Nb is 0.65 nm for Cu1.2NbS2, agreeing well with 0.66 nm calculated from the (002) diffraction peak (Fig. 3E). The intercalation of Cu increases the Nb-Nb distance paralleling to the c axis by ΔdNb-Nb = 0.063 nm in Cu1.2NbS2. The observed expansion of dNb-Nb upon Cu intercalation is much larger than that caused by intercalation of the preceding 3d transition metal atoms. For example, there is only expansion of ΔdNb-Nb = 0.037 and 0.016 nm for Mn1/3NbS2 and Fe1/3NbS2, respectively (19). The intercalation compounds of Co1/3 NbS2 and Ni1/3 NbS2 show dNb-Nb quite similar to those of pure 2H-NbS2 (19). The sandwiched S─Nb─S slab height (0.314 nm) has negligible change upon intercalation as previously reported due to neglected change of Nb─S bond length (19). The comparatively large increase of dNb-Nb upon intercalation of Cu is due to the tetrahedral coordination of the intercalated Cu atoms by the S atoms. Insertion of Cu atoms into tetrahedral sites requires a larger expansion of vdW gap than that of the intercalation of 3d metal atoms into octahedral sites (19, 35).

To fully understand the local electronic and geometric features of Cu atoms in Cu1.2NbS2, x-ray absorption fine spectroscopy (XAFS), including x-ray near-edge spectroscopy (XANES) and extended XAFS (EXAFS), which are sensitive to the local electronic and geometric nature, respectively, was performed (Fig. 4). As shown in Fig. 4A, the XANES reveals that the absorption edge of Cu1.2NbS2 is quite similar to that of Cu2S while lying between that of Cu and CuS, suggesting that the intercalated Cu atom is in the form of monovalent Cu (I). The chemical status of Cu (I) in the Cu1.2NbS2 prepared under ambient conditions agrees well with the oxidation state of Cu in the early reports on CuxNbS2, CuxNbSe2, and CuxTiSe2 synthesized by CVT and solid-state stoichiometric elemental reaction methods at the elevated temperature (19, 21, 36). The intercalated Cu acts as an electron donor, which transfers electrons to the host TMDs (37). It was confirmed by comparing the band structure between NbS2 and CuxNbS2 (Fig. 5, A and B). The difference in the white-line profile between Cu1.2NbS2 and Cu2S can be assigned to the unique local structure of Cu1.2NbS2. Figure 4B shows the EXAFS spectra of Cu1.2NbS2 and the references without phase correction. The bond distance located at ~1.82 and ~1.9 Å in CuS and Cu2S reference samples, respectively, is ascribed to the Cu─S contribution, while the peak located at ~2.3 Å in the Cu foil can be assigned to Cu─Cu bond (38). However, the widened peak features of Cu1.2NbS2, especially the peak located at ~2.6 Å, are slightly different from above references. To extract quantitative structural parameters for Cu1.2NbS2, the EXAFS fitting analysis was performed both in K-space (Fig. 4C) and R-space (Fig. 4D). Four kinds of fitting contributions are considered, including Cu─Cu, Cu─S1, Cu─S2, and Cu─Nb. Two kinds of Cu─S bond distances located at 2.12 and 2.27 Å in Cu1.2NbS2 (table S3) further confirm that the Cu atoms sit in the tetrahedral sites within the vdW gaps due to its similar bond length to that of tetrahedrally coordinated binary Cu─S compounds (2.30 Å) while quite shorter than Cu─S bond length (2.60 Å) for Cu octahedral occupation (19). In addition, the Cu─S coordination numbers are 0.7 for Cu─S1 and 2.3 for Cu─S2, indicating the unsaturated environmental of center Cu atoms (table S3).

Fig. 4 XAFS characterization of Cu chemical state.

(A) XANES at the Cu K-edge for Cu1.2NbS2 and corresponding reference samples. (B) Fourier transforms of extended fine structure at the Cu K-edge for Cu1.2NbS2 and corresponding reference samples. (C and D) Experimental and fitting curves for Cu EXAFS in Cu1.2NbS2 at the K- and R-space, respectively.

Fig. 5 Electronic and superconductivity properties of CuxNbS2.

(A) The band structure of 2H-NbS2. (B) The band structure of 2H-CuNbS2. The color change indicates the states contributed by Cu orbitals. (C) Normalized susceptibility for CuxNbS2 with different Cu doping levels (0 ≤ x ≤ 0.65). (D) Superconducting phase diagram for CuxNbS2.

Electronic and superconducting property

In general, chemical doping/intercalation and application of external pressure were used to tune the superconducting properties, since these two methods can vary the principal parameters determining the superconducting properties, i.e., the electronic density of states at the Fermi energy, the characteristic phonon frequency, and the coupling constant of electrons and phonons (39). The investigation on the pressure effect on the crystal structure and superconductivity in NbSexTe2−x found that the superconducting transition temperature (Tc) increased under higher pressure, and the crystal structure remained stable under pressure up to 39 GPa (40). The Cu doping density was used as a tuning parameter to investigate the CuxTiSe2 system and successfully induced a new superconducting state in this system (21). In this work, we used Cu intercalation as a probe to tune the superconducting properties of NbS2 system, which help shed light on the role of electron density and interlayer spacing in the superconductivity of the layered materials. The Cu intercalation results in the increase of electron density in CuNbS2, which subsequently leads to an upshift of Fermi level (Fig. 5, A and B). This is confirmed by weight of Cu contribution analysis, where the states below the Femi level are mainly composed of Cu orbitals. Meanwhile, note that a band goes through a Fermi level between Γ and A, indicating conductivity in the z-direction before and after Cu intercalation. Further DFT calculation shows that intercalation of Cu into 2H-NbS2 leads to a significant increase of charge density between adjacent layers along the z-direction, which provides channel for electron mobility, indicating increase in conductivity in the z-direction (fig. S26). The electrical conductivity in the z-direction of 2H-NbS2 and CuxNbS2 was measured by current-voltage (I-V) profiles (fig. S27). The current of CuxNbS2 is threefold higher than that of 2H-NbS2. Because the electron density on Cu is reduced, the increase in total conductivity in the z-direction of CuxNbS2 implies enhanced electron mobility, which is consistent with the DFT calculations and XAFS result. In contrast, intercalation of Cu into vdW gaps of 2H-NbS2 suppresses the superconducting transition temperature (Tc) and superconducting volume fraction systematically as Cu intercalate state increases, and thus, c constant increases as well (Fig. 5, C and D). The Tc decreases as the Cu concentration increases, consistent with the Cu-intercalated NbS2 synthesized by high-temperature CVT method (41). CuxNbS2 becomes nonsuperconducting when x is above 0.6, which is accorded with the PXRD result (Fig. 2C), i.e., the 2H-NbS2 phase completely disappears when x is ≥0.65. This dependence of Tc on c lattice constant for 2H-NbS2 has been similarly observed by changing the c constant via tuning pressure (42). The Tc increased as a higher pressure was applied to compress 2H-NbS2 in the c-axis direction (42). The Cu interaction may induce changes of the Fermi surface in the opposite way as the physical pressure does. On the other hand, the sharp difference of Tc and superconducting between pure 2H-NbS2 and our synthesized CuxNbS2 verify the successful intercalation of Cu atoms into vdW gaps of 2H-NbS2.


In summary, we present a strategy to synthesize homogeneous Cu-intercalated TMD compounds with high Cu concentration by spontaneous self-intercalation of Cu atoms from bulk Cu, based on theoretical prediction and experimental validation. Among TMDs, the bulk metallic Cu could be transformed to Cu nanosized species in a thermodynamically downhill way, and then, Cu species self-intercalate into the vdW gap of the group IV and V layered TMDs, i.e., 2H-NbS2, 3R-NbS2, 2H-NbSe2, 1T-TiS2, 1T-TaS2, and 1T-VS2, at room temperature and atmospheric pressure. In contrast, no Cu species are intercalated into the group VI TMDs, i.e., 2H-MoS2 and 2H-WS2, under the same conditions. The maximum Cu concentration in CuxNbS2 is as high as x = 1.2, which, so far, is the highest intercalation state in Cu-intercalated TMD compounds. A 2D analog Cu atom sheet in between the TMD sandwiched layers was clearly observed in the cross-sectional STEM image. Further analysis using XAFS and band structure calculations demonstrated that the intercalated Cu in the tetrahedral coordination acts as an electron donor and the TMD as an electron acceptor, which leads to form monovalent Cu (I). In situ x-ray diffraction (XRD) demonstrated that the process of Cu self-intercalation is spontaneous. Our work opens a new avenue to synthesis of metal-intercalated TMD compounds with a high intercalated state in a process-viable and effective way. The potential application of those compounds remains to be explored.



All reagents and solvents were purchased from chemical reagent companies and used without further purification. Niobium (1 to 5 μm and 325 mesh), vanadium (99.9%, 325 mech), tantalum (99.9%, 2 μm) powder were purchased from Alfa Aesar. Titanium (99.99%, 300 mesh), Cu (1 and 5 μm), molybdenum disulfide (99.99%), tungsten disulfide (99.99%), and selenium (99.99%) powder were purchased from Aladdin Reagents Co. Ltd. Sulfur (99.5%) was purchased from Sinopharm Chemical Reagent Co. Ltd. N-hexane (high-performance liquid chromatography) was purchased from Fisher Chemical.


SEM images of the samples were obtained by a Sigma field-emission scanning electron microscope (Zeiss Ltd.) operated at 20 kV. XRD and in situ XRD analyses were performed on a SmartLab (Rigaku) with filtered Cu Kα radiation (Rigaku D/max-2500, λ = 1.5405 Å). In situ XRD was performed by sequential scans, with each scan collected at a scanning rate of 50° min−1. TEM images, STEM images, and EDS were acquired using an FEI Tecnai F-20 microscope. Samples were dispersed in ethanol and dropped onto molybdenum grids and then attached to the double-tilted sample holder. To confirm the compositional distribution and molar ratio of Cu:NbS2 of nano-CuxNbS2, EDS was carried out to map the distribution of Cu, Nb, and S under the STEM mode. EPMA (JXA-8530F Plus) was applied to analyze the concentration of Cu in the intercalated TMD compounds. The synchrotron XAFS was collected at Beamline 11-ID-C in Advanced Photon Sources, Argonne National Laboratory. Magnetization measurements were performed using Magnetic Property Measurement System by Quantum Design. The electrical conductivity in the z-direction of 2H-NbS2 and CuxNbS2 were measured using a Keithley 4200 parameter analyzer.

Synthesis of TMDs

TMD powders were synthesized by the solid-state reaction of metal (Ti, Nb, V, and Ta, 325 mesh) and sulfur powders (or selenium, 4.0 atomic % excess). The mixture of powders was ground in a mortar for 30 min and then put into a quartz tube in an Ar-filled glove box [H2O < 1 parts per million (ppm) and O2 < 1 ppm]. After moving the tube out of the glove box, it was evacuated to 0.1 torr and then sealed. The samples were heated slowly to 850°C and kept there for different durations depending on the sample, e.g., 2 hours for 2H-NbS2 and 7 days for 3R-NbS2, 2H-NbSe2, 1T-TiS2, 1T-TaS2, and 1T-VS2. Afterward, the samples were cooled down naturally to room temperature.

Synthesis of Cu-intercalated TMDs

Cu-intercalated TMDs were synthesized by mixing Cu powder with MX2 (MX2 = 2H-NbS2, 3R-NbS2, 2H-NbSe2, 1T-TiS2, 1T-TaS2, 1T-VS2, 2H-WS2, and 2H-MoS2) powder in a stoichiometric manner in 2 ml of n-hexane under stirring for different times in an Ar-filled glove box [H2O < 1 ppm and O2 < 1 ppm].

Computational detail

All calculations were performed using the spin-polarized DFT as implemented in Vienna ab initio simulation package (VASP). The core-valence interactions and electron exchange-correlation function were described by projected augmented wave methods and Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation (GGA), respectively. The cutoff energy for the plane-wave basis set of 400 eV was used. A 12 × 12 × 4, 5 × 5 × 4, and 5 × 5 × 1 Γ-centered Monkhorst-Park grids of K-point was for single-cell, super-cell, and surface structural optimization. A vacuum thickness of 20 Å was adopted to avoid interaction between two adjacent periodic images. The conjugate gradient algorithm was used, and the convergence threshold was reached until forces on every atom were less than 0.01 eV/Å. The Becke88 optimization (optB88) was used to accurately account for nonlocal vdW force. The diffusion pathway and the corresponding energy barrier were calculated by the climbing image nudged elastic method. The reaction of Cu intercalated into TMDs (MX2) is defined by the following equation: MX2 + nCu = CunMX2. The change of Gibbs free energy (ΔG) is calculated as ΔG = [G(CunMX2) − G(MX2) − n*G(Cu)]/n, where G(CunMX2) and G(MX2) are the ground-state energy of TMDs after and before Cu intercalation calculated by first-principles calculation, respectively. G(Cu) is the energy required by taking one Cu atom from bulk Cu. n is the number of intercalated Cu atom per TMD unit cell. The ΔG is negative, indicating that Cu atom prefers to intercalate into vdW gaps of TMDs rather than forming bulk Cu. AIMD of Cu on monolayer NbS2 of which the vacuum thickness was set 15 Å was carried out using VASP with GGA-PBE. The total simulation time of all different coverages of Cu was 10 ps with each time step of 2 fs. A canonical ensemble (NVT) was simulated using the algorithm of Nosé (43), while the plane wave energy cutoff was set to 400 eV with 2 × 2 × 1 Monkhorst-pack grids of K-point.


Supplementary material for this article is available at

Fig. S1. Schematic of structures of TMDs and Cu-intercalated TMDs.

Fig. S2. AIMD simulations of the stability of different coverages of Cu on NbS2 at 900°C.

Fig. S3. Morphologies of TMDs.

Fig. S4. PXRD patterns of different TMDs after reaction with Cu.

Fig. S5. Morphology evolution of Cu and CuxNbS2 with different reaction duration.

Fig. S6. Elemental distribution of nanometer CuxNbS2.

Fig. S7. Elemental distribution of micrometer CuxNbS2.

Fig. S8. Morphology evolution of Cu and 2H-WS2 with the stirring time.

Fig. S9. PXRD patterns of Cu and 2H-WS2 mixed powders by an atomic ratio of 1.2:1.0 under different stirring time.

Fig. S10. Elemental distribution of Cu and 2H-WS2 mixed powder by atomic ratio of 1.2:1.0 after stirring for 10 days.

Fig. S11. TEM images of Cu and WS2 mixed powder by an atomic ratio of 1.2:1.0 after stirring for 10 days.

Fig. S12. Morphology evolution of Cu and 2H-MoS2 with the stirring time.

Fig. S13. PXRD patterns of Cu and 2H-MoS2 mixed powders by an atomic ratio of 1.2:1.0 under different stirring time.

Fig. S14. Morphology, composition, and size distribution of Cu and 2H-MoS2 after mixing for 10 days.

Fig. S15. Elemental distribution of Cu and MoS2 mixed powder by an atomic ratio of 1.2:1.0 after stirring for 10 days.

Fig. S16. PXRD patterns of Cu1.2NbS2 with different time under stirring.

Fig. S17. PXRD patterns of CuxNbS2 by adding different Cu contents and after stirring for the same 5 days.

Fig. S18. The c lattice constant as a function of Cu concentration in CuxNbS2.

Fig. S19. Electron probe microanalysis of CuxNbS2 with different Cu content.

Fig. S20. PXRD patterns of NbS2 with different Cu sources.

Fig. S21. Ex situ PXRD patterns of microsized NbS2 and Cu mixed powder (atomic ratio 1:1.5) after stirring time in hexane.

Fig. S22. In situ PXRD patterns of microsized CuxNbS2 (atomic ratio Cu:NbS2 = 1.5:1) from 66 to 118 hours.

Fig. S23. The intensity of (002) peak ratio between NbS2 and CuxNbS2 in figs. S21 and S22.

Fig. S24. Time-resolved in situ PXRD measurement of nanosized NbS2 coating on the Cu sheet.

Fig. S25. High-resolution high-angle annular dark field (HAADF)–STEM images of Cu1.2NbS2 along the [001] direction.

Fig. S26. Theoretical computing of the charge density of NbS2 and CuNbS2.

Fig. S27. Cross-sectional morphology and vertical electrical conductivity of 2H-NbS2 and CuxNbS2.

Table S1. Gibbs free energy change of Cu intercalation with different Cu intercalation states.

Table S2. Elemental ratios in CuxNbS2 analyzed by electron probe microanalysis.

Table S3. The structural parameters of Cu1.2NbS2.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank H. Cong, Q. Liu, X. Xu, Y. Zhou, H. Deng, and S. Chen (Wuhan University) for invaluable assistance. We thank S.-G. Sun and Z.-Y. Zhou (Xiamen University) for fruitful discussions. Funding: This work was sponsored by National Nature Science Foundation of China (21773176, 21403157, and 11874036) and the Fundamental Research Funds for the Central Universities (2042017kf0232, 2042019kf0200). F.-S.K. acknowledges the New Faculty Startup Fund of Wuhan University, Large-scale Instrument and Equipment Sharing Foundation of Wuhan University. This work was also partly supported by the U.S. Air Force Office of Scientific Research under award number FA9550-14-1-0268. P.M.A. thanks the U.S. Air Force Office of Scientific Research for support under award number FA9550-18-1-0072. This work was also partly supported by the National Key Research and Development Program of China (no. 2017YFB0701600), the Local Innovative and Research Teams Project of Guangdong Pearl River Talents Program (no. 2017BT01N111), and Shenzhen Projects for Basic Research (no. JCYJ20170412171430026). Tianjin and Guangzhou Supercomputing Center are also acknowledged for allowing the use of computational resources. Author contributions: F.-S.K., J.W., and P.M.A. conceived the project and supervised the experiment. J.W. and F.-S.K. wrote the manuscript with the assistance from the other authors. X.-C.L. and F.-S.K. performed the synthesis and most of the material characterizations. S.Z., X.Z., J.L., and Y.H. carried out the theoretical calculations. X.S. did the XAFS analysis. Y.-X.W. did the STEM experiment. L.D. and C.-W.C. performed the superconductivity measurement. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article