Introduction

Conversion-type transition metal dichalcogenides (TMDs), normally with a formula of MX2 (M = Mo, V, W, Re; X = S, Se), are promising anode materials for lithium/sodium ion batteries owing to their high theoretical capacities1,2,3,4,5. The electrochemical storage mechanism research of these materials have attracted tremendous attention as it is the critical footstone for rational structure and morphology design of electrode materials to improve electrochemical performances6,7,8,9. Generally, the initial discharging process of MX2 includes intercalation and conversion to form products M and A2X (A = Li, Na)10. In the following charging process, there are mainly two different pathways: MX2 can be regenerated during the charging process following Eq. (1)11, which means that the conversion reaction of MX2 is reversible; or the formed A2X rather than M is oxidized to X upon the charging process, and eventually, A2X/X becomes the main redox couple in the subsequent cycles according to Eq. (2), demonstrating the irreversibility of the conversion reaction in MX212. Obviously,

$${{{{{{\rm{M}}}}}}+{{{{{\rm{2A}}}}}}}_{2}{{{{{\rm{X}}}}}}\to {{{{{{\rm{MX}}}}}}}_{2}{+{{{{{\rm{4A}}}}}}}^{+}{+{{{{{\rm{4e}}}}}}}^{-}$$
(1)
$${{{{{{\rm{A}}}}}}}_{2}{{{{{\rm{X}}}}}}\to {{{{{\rm{X}}}}}}+2{{{{{{\rm{A}}}}}}}^{+}+2{{{{{{\rm{e}}}}}}}^{-}$$
(2)

taking MoSe2 as an example, the essence of the sodium storage based on the irreversible mechanism actually has become a Na-Se battery, which will suffer from the shuttle effect of polyselenides and poor structural stability, thus leading to rapid capacity attenuation13,14. Therefore, manipulating MX2 to follow a reversible reaction mechanism is highly necessary for achieving high electrochemical storage performance. When delving into the whole process in detail, it can be found that the key to the reversibility of the conversion reaction (Eq. 1) lies in whether Mo, the discharged product of MoSe2, is capable of reacting with Na2Se to re-form MoSe2, because Mo is electrochemically inert and the Na2Se has covalent characteristics15,16. Therefore, we assume that if we optimize the activity of the discharged products (Mo and Na2Se) to drive the reaction between them, can we achieve the reversible conversion in MoSe2? Nevertheless, this thinking is very challenging and has not yet been demonstrated in previous reports.

Recently, strain engineering in materials science has been deemed as an effective strategy to increase the intrinsic activity of the material by modifying its electronic properties, which is conductive to triggering the redox reaction17,18,19. For instance, Zhang et al. prepared tensile-strained Pd porous nanosheets and validated that the tensile strain could facilitate the conversion of Pd to PdO220, while Chen et al. highlighted that the interfacial strain on reactants could promote the reaction by making the reaction energetically more favorable21. Specifically, in theory, the spontaneity of the redox reaction is determined by the change in Gibbs free energy of the process (ΔG). In general, the smaller the ΔG between the products and the reactants, the more favorable the reaction is in thermodynamics. Moreover, the occurrence of the redox reaction also must conquer the kinetic energy barriers, which is closely associated with the ΔG, as expressed by following Marcus equation22:

$$\varDelta {G}^{{{\ddagger}} }=\frac{\lambda }{4}{(1+\frac{\varDelta G}{\lambda })}^{2}$$
(3)

Where ∆G‡ is the reaction activation energy and λ is the reaction reorganization energy (usually between 0.5 − 1.0). Thus, the activation energy of the electron transfer reaction can be determined by the ΔG. When ΔG > 0, a smaller ΔG usually means a lower reaction activation energy barrier, which also implies accelerated reaction kinetics (Fig. 1a). Thus, it can be concluded that the effective modulation of the ΔG can not only improve the reaction thermodynamics but also promote the reaction kinetics.

Fig. 1: Theoretical calculations.
figure 1

a Schematic diagram of tuning reactants to make the redox reaction more energetically favorable. b The variation of ΔG associated with the redox reaction under epitaxial strain (η) at a constant temperature. c The adsorption energies per 2-MI molecule on MoSe2 and Mo. d The atomic structure model of TS-Mo, Mo and the strain evolution from TS-MoSe2 to TS-Mo. e The ΔG values per formula unit of the reaction between Na2Se/Mo and MoSe2 under strained and unstrained conditions. f PDOS of Mo-4d orbitals for TS-Mo and Mo. g The adsorption energies of per Na2Se molecule on TS-Mo and Mo and the corresponding Na-Se distance evolution.

Full size image

Fortunately, it has been proved that, based on the quasi-harmonic density functional theory (DFT) calculations, the thermodynamically shifted Gibbs free energy (G) of redox reaction can be defined as a function of epitaxial strain (η) and temperature (T) according to the following relationship in Eq. (4)23,24,25:

$$\varDelta G=\varDelta E(\eta )+\varDelta F(\eta,T)+{\mu }_{M}(T)$$
(4)

Where µM is the chemical potential of the reactants, ΔE represents the zero-temperature contribution to the G, and ΔF represents the thermal contribution to the G. Based on this, to achieve a decreased value of ΔG and make the redox reaction energetically favorable, the key lies in applying lattice strain on the corresponding crystal (Fig. 1b). Therefore, it is highly expected that strain engineering may be an important enabler in the reversible transformation of sodiated MoSe2 during the charging process by reducing ΔG value of the reaction. However, few research works have put efforts to understand the effect of strain engineering on the lithium/sodium storage mechanism of conversion-type TMDs, owing to the lack of an efficient regulation methodology for the discharged products.

Herein, we demonstrate that tensile-strained MoSe2 (denoted as TS-MoSe2) can pass on the strain to its discharged product Mo (denoted as TS-Mo) by using 2-methylimidazole (2-MI) as a scaffold and rationalize the effect of strain engineering on its sodium storage mechanism. Both theoretical calculation and experimental results revealed that the tensile strain could activate inert Mo and Na2Se as well as reduce the ΔG value of their reaction. As a consequence, TS-MoSe2 displays a highly reversible conversion mechanism of sodium storage, while its counterpart (unstrained MoSe2) cannot be recovered during the charging process. Impressively, TS-MoSe2 achieves high rate capacities and excellent cycling stability over a wide temperature range. This work provides a direction for the research on the alkali-metal-ion energy storage of conversion-type TMDs, which is essential to the rational design of high-performance electrode materials.

Results

Theoretical predictions for the strain effect on the activity of Mo

The feasibility of a strain engineering strategy that affects the reversibility of the sodium storage mechanism of MoSe2 is assessed first by DFT calculations. It is worth mentioning that in our studies, 2-MI was chosen to exert the strain on the studied objects (Mo and MoSe2) by means of its strong ligand effect26, which can be further confirmed by its negative adsorption energies with Mo and MoSe2, respectively (Fig. 1c and Supplementary Fig. 1). The atomic structure model of TS-Mo was constructed by introducing 2-MI species into Mo crystal structure (here, the 2-MI is reduced in size to make the underlying Mo structure more visible), which exerts the strain on Mo through the interaction between Mo and N atoms (Fig. 1d). It has been well addressed that the strong adsorption of the ligand will induce tensile strain on the adsorption sites27,28, which significantly deform the structure to some extent and thereby might change the surface energy. Therefore, we first calculated the ΔG value of the studied reaction and the results clearly show that the reaction of TS-Mo and Na2Se to generate MoSe2 (TS-Mo + 2Na2Se → MoSe2 + 4Na+ + 4e−) delivers a ΔG value of 3.06 eV (Fig. 1e, Supplementary Fig. 2, and Supplementary Table 1), which is much smaller than that (3.95 eV) of the same reversible reaction based on unstrained Mo. Furthermore, for unstrained Mo, this ΔG value corresponding to the reversible reaction is higher than that of the irreversible reaction (Mo + Na2Se → Mo + Se + 2Na+ + 2e−) (Supplementary Fig. 3), indicating that the irreversible reaction may occur preferentially than the reversible one for unstrained Mo. According to the Marcus equation in Fig. 1a, the smaller ΔG values also mean a reduced reaction energy barrier, implying that strain engineering could evidently promote the reaction not only in thermodynamics but also in kinetics. Furthermore, TS-Mo also exhibits an enlarged Mo-Mo bond length after employing the tensile strain (Fig. 1d), owing to the electronegativity difference between Mo and N atoms. The increased distance weakens the atomic interaction between Mo atoms, which is conductive to improve its reactivity. Therefore, the electronic structure of TS-Mo was also evaluated by partially density of states (PDOS), as shown in Fig. 1f. Obviously, TS-Mo shows an upshift of the d-band center toward the Fermi Level compared to unstrained Mo. Consequently, TS-Mo achieves a more negative adsorption energy value (−6.58 eV) than unstrained Mo (−4.77 eV) when interacting with Na2Se, further providing a positive effect on the reaction between Mo and Na2Se, which can be further confirmed by an enlarged Na-Se distance (dNa-Se) of Na2Se adsorbed on TS-Mo (Fig. 1g and Supplementary Fig. 4). Taken together, it can be inferred that the strain engineering may show great potential in promoting the electrochemical reaction of Mo and Na2Se in energy storage devices. In our studies, Mo is the discharged product of MoSe2, and therefore, in order to obtain TS-Mo experimentally, we can only start from the MoSe2 and also expect that TS-MoSe2 can pass on its strain to Mo during the discharging process, thus possibly leading to the formation of TS-Mo.

Materials synthesis and characterizations

Based on the above theoretical calculations, we then prepared TS-MoSe2 by selenizing a Mo-precursor containing Mo source and 2-MI (Supplementary Fig. 5). Electron microscopy images clearly show that the resultant sample is featured with a hollow spherical structure with an average diameter of 150 nm (Fig. 2a and Supplementary Fig. 6), which inherits the spherical shape of the Mo-precursor (Supplementary Fig. 5d). The formation of the hollow structure follows the Kirkendall effect monitored by the time-dependent experiments (Supplementary Fig. 7). The specific surface area of the hollow TS-MoSe2 was determined to be 17.68 m2 g−1 according to the Brunauer–Emmett–Teller method, which is slightly higher than that of the unstrained MoSe2 (Supplementary Fig. 8). Additionally, further high-resolution transmission electron microscopy (HR-TEM) images display the typical features of few-layered MoSe2 and the interlayer spacing is approximately 0.66 nm (Fig. 2b, c), slightly larger than the intrinsic (002) plane value11,29. Furthermore, the energy-dispersive spectroscopy (EDS) elemental mappings and line-scan profiles reveal that besides the Mo and Se elements that are uniformly distributed throughout the structure (Fig. 2d and Supplementary Fig. 9), N and C elements were also detected and that this sample also shows more mass loss (Supplementary Fig. 10). All of these results indirectly indicate that there may be some 2-MI species in MoSe2. To further confirm this deduction, Fourier-transform infrared spectroscopy (FT-IR) was conducted. As shown in Supplementary Fig. 11, the bands at 1660, 1519, 1186, and 1030 cm−1 are attributed to the stretching vibration of N-H, the skeleton vibration of the imidazole ring, and the stretching vibrations of C-N and C=C bonds, respectively, while the band at 837 cm−1 is assigned to the vibration of C-H, proving the existence of the 2-MI species30. And the content of the 2-MI was determined to be about 4.10 wt% by CHN elemental analysis (Supplementary Table 2). Besides, X-ray photoelectron spectroscopy (XPS) measurements were performed to further study its existing form in MoSe2 (Supplementary Fig. 12). As shown in Supplementary Fig. 12a, the survey XPS spectrum also shows the presence of the C and N elements in good agreement with the previous EDS results. The high-resolution N 1 s XPS spectrum (Supplementary Fig. 12b) shows the presence of pyridinic-N (397.6 eV), pyrrolic-N (399.7 eV), and graphitic-N (401.5 eV) and the peak at 397.6 eV can be assigned to N-Mo bond31, suggesting the formation of coordination bond between 2-MI and MoSe2. This interaction may lead to the tensile strain between the MoSe2 layers, thus causing an expanded (002) interlayer spacing (Fig. 2c).

Fig. 2: Materials synthesis and characterizations.
figure 2

a TEM, b, c HR-TEM images, and d elemental mapping images of TS-MoSe2. e XRD patterns, f Raman spectra, and g The normalized Mo K-edge EXAFS spectra (circle) of TS-MoSe2 and MoSe2 as well as the corresponding EXAFS fitting curves (line). h Wavelet transform (WT) contour plots of MoSe2, TS-MoSe2, and MoN. i FT-IR spectra of 2-MI and the fully discharged and charged products of TS-MoSe2. j The normalized Mo K-edge EXAFS spectra of MoN and the fully discharged product of TS-MoSe2 (TS-MoSe2-D0.01) and MoSe2 (MoSe2-D0.01).

Full size image

To gain further information on the tensile strain in the as-prepared TS-MoSe2, a series of characterization techniques were used, such as X-ray diffraction (XRD), Raman spectrum, Fourier-transformed extended X-ray absorption fine structure (EXAFS) and XPS. As shown in Fig. 2e, the XRD pattern of TS-MoSe2 is well consistent with the standard card of MoSe2 (JCPDS card No. 29-0914) without any other crystalline impurities. However, compared to bulk MoSe2, it is worth noting that the (002) diffraction peak (c-axis) of the as-prepared sample moves to a lower angle while the (100) peak shifts to a higher angle, indicating that there is a lattice expansion along the (002) direction and in-plane compression in TS-MoSe232. More importantly, the quantitative insight of the strain can be obtained by the XRD pattern according to the following Eq. (5)33:

$${{{{{{\mathrm{strain}}}}}}}{\%}=\frac{{d}({{{{{{\mathrm{TS}}}}}}}{-}{{{{{{\mathrm{MoS}}}}}}}{{e}}_{{2}})-{d}{(}{{{{{{\mathrm{MoS}}}}}}}{{{{{{{\mathrm{e}}}}}}}}_{{2}}{)}}{{d}{(}{{{{{{\mathrm{MoS}}}}}}}{{{{{{{\mathrm{e}}}}}}}}_{{2}}{)}}\times 100\%$$
(5)

where d represents the spacing of the corresponding planes, which can be deduced by the Bragg equation. Consequently, TS-MoSe2 shows a tensile strain of about 6.34% along the c-axis and thereby an in-plane compressive strain of 3.15%. Furthermore, the distinct red shifts of E12 g, A1g, and B12 g in Raman spectrum of TS-MoSe2 also suggest the existence of the strain (Fig. 2f), as the tensile strain leads to an expanded interlayer spacing, which can weaken the interlayer interaction force and in turn decrease the frequencies of E12 g, A1g, and B12 g vibration modes32,34. EXAFS spectroscopy in Fig. 2g displays the bond length evolution in TS-MoSe2 and the shortened Mo-Se bond is clearly observed, which results from the in-plane compressive strain35. Besides, the curve fitting against the Fourier transforms of the EXAFS data for TS-MoSe2 (Supplementary Fig. 13 and Supplementary Table 3) further proves the coordination bond of Mo and N and the corresponding coordination number is 0.8. Additionally, the wavelet transform (WT) plot of the Mo K-edge EXAFS for TS-MoSe2 also presents a peak at 1.5 Å (Fig. 2h), which can be attributed to the dominance of the Mo-N scattering36. Meanwhile, the binding energies of Mo 3d5/2 and Mo 3d3/2 in the high-resolution Mo 3d XPS spectrum of TS-MoSe2 both shift to a low-energy side, suggesting that the strain may lead to the formation of 1T-MoSe2 (Supplementary Fig. 14)37. The above results indicate that the as-prepared TS-MoSe2 does exist the out-plane tensile strain and in-plane compressive strain. In addition, we also studied discharged and charged TS-MoSe2 by FT-IR (Fig. 2i) and found that the 2-MI species still exists, consistent with the result from Supplementary Fig. 11. Furthermore, the stability of the 2-MI molecule was further confirmed by linear sweep voltammetry (LSV) curves, in which no visible reduction peak ascribed to the 2-MI molecule is observed within the operating voltage window (Supplementary Fig. 15). Undoubtedly, the 2-MI species will continue to coordinate with the discharged product of TS-MoSe2 (i.e., Mo), as confirmed by the Mo-N bond in the EXAFS and XPS spectra (Fig. 2j and Supplementary Fig. 16). It is noticeable that the coordination effect may enable Mo also with the tensile strain, that is, the strain in TS-MoSe2 has been transferred to its discharged product. As depicted in Fig. 2j, TS-MoSe2-D0.01 displays a peak at ~2.69 Å, corresponding to the Mo-Mo bond in metallic Mo, which is slightly longer than that of the counterpart of unstrained MoSe2, implying the existence of the tensile strain in metallic Mo induced by the coordination between 2-MI and Mo.

Investigation of reversible sodium storage mechanism in TS-MoSe2

Inspired by the positive influence that strain engineering has achieved on the redox reaction by the DFT calculations, we first performed ex situ XPS measurements to investigate the effect of the tensile strain on the sodium storage process. During the whole evolution process of discharging and charging, ten voltages were selected to evaluate the structural transformation of the TS-MoSe2 electrode. As shown in the Mo 3d XPS spectra (Fig. 3a), at the beginning of the discharging process (1.8 and 1.5 V), two main characteristic peaks at 228.83 and 231.93 eV that are related to 3d5/2 and 3d3/2 of Mo4+ in MoSe2 slightly shift towards the low binding energy, indicating the formation of the NaxMoSe2 intermediate. With further discharging (1.0 and 0.4 V), a component with lower binding energies at 227.43 (Mo 3d5/2) and 230.53 eV (Mo 3d3/2) appears and it can be assigned to metallic Mo38, suggesting that the NaxMoSe2 has partly transformed into metallic Mo. At a fully discharged state, the NaxMoSe2 completely disappears and only metallic Mo is detected. Correspondingly, the Se 3d peak at 54.5 eV first shifts to higher binding energy, and then restores to the original position, manifesting that Na2Se finally forms through the polyselenide Na2(Se)1+n (n > 1) during the discharging process (Fig. 3c)39. Afterward, in the following charging process, the peaks of both Mo 3d and Se 3d core levels can be fully recovered to their pristine state for TS-MoSe2, and in contrast, for unstrained MoSe2, metallic Mo is always present, and meanwhile, the elemental Se is eventually generated (Supplementary Figs. 17a, 18a). These changes can be observed more visually in corresponding 2D mapping images of the Mo 3d and Se 3d XPS spectra (Fig. 3b and Supplementary Figs. 17b, 18b, 19), which demonstrate that the strain engineering enables TS-MoSe2 to follow highly reversible sodium storage mechanism in the discharging and charging processes.

Fig. 3: Study on discharging and charging processes based on ex situ XPS and in situ Raman spectra.
figure 3

a–c ex situ Mo 3d XPS spectra (a) and corresponding mapping image(b), as well as Se 3d XPS spectra (c) of TS-MoSe2 during the initial discharging and charging processes. d Schematic illustration of in situ Raman measurement. e, f In situ Raman spectra (e) and corresponding mapping image (f) of TS-MoSe2 during the initial discharging and charging processes.

Full size image

The reversible sodium storage of TS-MoSe2 was further confirmed by in situ Raman (Fig. 3d). As shown in Fig. 3e, TS-MoSe2 exhibits a prominent peak at about 285 cm−1, corresponding to the E12 g vibration mode of MoSe2, and the peaks at 216 and 342 cm−1 belong to the Cu foil-derived oxide40. During the discharging process, the E12 g peak of MoSe2 shows a slight red shift along with a decrease of peak intensity, which may come from the lattice expansion and disorder increase of TS-MoSe2 induced by the intercalation of sodium ions. As the discharging process proceeds, the peak at 285 cm−1 disappears completely, indicating that MoSe2 is fully reduced. Reversibly, in the subsequent charging process, it is observed that the E12 g peak of MoSe2 is recovered again, indicating the regeneration of MoSe2. The phenomenon is also proved by the peak color change in the mapping image in Fig. 3f. Furthermore, to exclude the influence of testing errors, we repeated the in situ Raman testing and the experimental results are basically consistent (Supplementary Fig. 20).

To further verify the above results, ex situ Mo K-edge X-ray adsorption spectroscopy (XAS) of TS-MoSe2 was performed to track its valence state change and local atomic structure evolution during the electrochemical cycling. As depicted in Fig. 4a and Supplementary Fig. 21, during the discharging process, the absorption edge of Mo K-edge X-ray absorption near-edge structure (XANES) gradually shifts to a lower energy direction along with the insertion of sodium ions, manifesting that the valence state of Mo gradually decreases, namely, the reduction of MoSe2 to Mo. After that, the absorption edge returns to the higher energy state until it almost coincides with the absorption edge of the pristine TS-MoSe2 in the charging state (Fig. 4b and Supplementary Fig. 22). In addition, the corresponding wiggle/oscillatory features of the post-edge region of the pristine TS-MoSe2, fully discharged TS-MoSe2 (D0.01), fully charged TS-MoSe2 (C3.0), and Mo foil can also reflect the variation in the local structure of TS-MoSe2 during the electrochemical process. (Fig. 4c). Upon being fully discharged to 0.01 V, the appearance of the fingerprint feature of Mo foil at 20013.2, 20040.3, and 20084.1 eV supports the formation of metallic Mo16. In contrast, after the full charging, the aforesaid peaks almost recover to the original state of TS-MoSe2, while the Mo foil-related features disappear, indicating that the discharging and charging processes of TS-MoSe2 during the initial cycle are nearly fully reversible. It should be noted that the Mo K-edge XANES spectra of formed metallic Mo and the regenerated MoSe2 are slightly different from those of corresponding Mo foil and pristine TS-MoSe2, which may be caused by the ligand effect of imidazole and amorphous nature, respectively41,42. A similar change trend is also observed in the Se K-edge XANES (Supplementary Figs. 23, 24a–d). Specifically, during the initial discharging process, there are two obvious peaks located at 12661.08 and 12668.2 eV in the XANES spectra of TS-MoSe2, which can be assigned to MoSe243. Nevertheless, these two peaks disappear and a peak appears at 12666.5 V upon discharging to 0.01 V, which corresponds to the generation of the discharged product Na2Se. During the subsequent charging process, these peaks return to the original state, further indicating that the conversion reaction shows good reversibility.

Fig. 4: Study on discharging and charging processes based on ex situ XAS.
figure 4

Ex situ Mo K-edge XANES spectra of TS-MoSe2 during the first a discharged and b charged states. c Mo K-edge XANES spectra of pristine TS-MoSe2, fully discharged TS-MoSe2 (D0.01), fully charged TS-MoSe2 (C3.0), and Mo foil. d Evolution of Mo K-edge EXAFS during electrochemical cycling. e The intensity evolution of the Mo-Se peak in TS-MoSe2 (2.11 Å, representing the concentration of TS-MoSe2) and the Mo-Mo peak in metallic Mo (2.69 Å, representing the concentration of Mo) during electrochemical cycling. Mo K-edge XANES and EXAFS spectra of TS-MoSe2 f and MoSe2 g after the first, second, and fifth cycles.

Full size image

Furthermore, the EXAFS spectra were applied to reveal the local structural changes of TS-MoSe2 during the initial discharging and charging processes. As shown in Fig. 4d, the Mo K-edge EXAFS spectra of the pristine TS-MoSe2 exhibit two obvious peaks at 2.11 and 3.09 Å, corresponding to Mo-Se interaction in the first coordination shell and Mo–Mo interaction, respectively44,45. With the intercalation of sodium ions, a peak appears at 2.69 Å accompanied by an increase of peak intensity, corresponding to the Mo–Mo bond in metallic Mo, which further confirms the generation of Mo during the discharging process6. The concentration changes of the TS-MoSe2 and its discharged product Mo can be monitored by tracking the intensity changes of the corresponding peaks (Fig. 4e)46. Obviously, during the discharging process, the Mo-Se peak that belongs to TS-MoSe2 gradually decreases in intensity, while the Mo-Mo peak (metallic Mo) continues to increase. Similarly, the corresponding Se K-edge EXAFS spectra (Supplementary Fig. 24e) also witnessed the gradual transformation of the Se-Mo bond (MoSe2) to the Se-Na bond (Na2Se) upon discharging. In the subsequent charging process, the Mo-Mo (metallic Mo) and Se-Na peaks gradually disappear, while Mo-Se/Se-Mo and Se-Se (TS-MoSe2) peaks become stronger. These observations further prove the excellent electrochemical reversibility of TS-MoSe2. In addition, the Mo and Se K-edge XANES and EXAFS spectra of TS-MoSe2 at D0.01 and C3.0 during the second and fifth cycles were also recorded, further confirming the reversible conversion of TS-MoSe2 in the subsequent cycles (Fig. 4f and Supplementary Figs. 25–27). The whole sodium storage process of TS-MoSe2 is illustrated in Supplementary Fig. 28, which goes through an intercalation and conversion reaction during the charging process and then the generated Mo and Na2Se are reversibly converted into MoSe2. By contrast, the unstrained MoSe2 exhibits a different conversion mechanism compared with the TS-MoSe2, as disclosed by its Mo K-edge XANES and EXAFS spectra. As shown in Fig. 4g and Supplementary Fig. 29, after charging at C3.0, the corresponding Mo-Mo (metallic Mo) peak at 20013.2 eV always exists, indicating that resultant metallic Mo did not participate in the subsequent reaction. In other words, the discharged product (metallic Mo) of MoSe2 cannot regenerate MoSe2 again in the charging process, and its sodium storage mechanism is irreversible. Therefore, combining ex situ XPS and XAS with in situ Raman spectra, it can be concluded that the strong interaction between the ligand and metal surface induces surface strain and subsequent surface reconstruction47,48, which plays a significant role in the activation of Mo and thereby promotes the reversible sodium storage of MoSe2, that is, TS-MoSe2 exhibits a reversible sodium storage mechanism following Eq. (1), while unstrained MoSe2 is irreversible as shown in Eq. (2).

Electrochemical performance

In general, the reversible structure evolution will contribute to the improvement of battery performance. Thus, the electrochemical performances of TS-MoSe2 and MoSe2 as the anodes for SIBs were evaluated. Figure 5a shows cyclic voltammogram (CV) profiles of TS-MoSe2 for the first four cycles within a potential range (V vs. Na/Na+) of 0.01−3.0 V. The first cathodic scan presents two pronounced peaks at around 1.32 and 0.56 V, respectively, which are attributed to the intercalation of Na+ into the MoSe2 and the conversion reaction to form Na2Se and metallic Mo nanograins. Meanwhile, the broad reduction peak ranging from 0.01 to 0.5 V in the first cycle is related to the electrolyte decomposition along with the formation of solid electrolyte interface (SEI) film49,50. During the following anodic scan, the distinct peak at 1.75 V, accompanied by a shoulder at 2.15 V, is ascribed to the conversion reaction between Mo and Na2Se to form MoSe2. These CV profiles overlap very well after the initial cycle, indicating the admirable reversibility and cyclic stability during the cycling process. By contrast, MoSe2 (Supplementary Fig. 30) shows different peak positions in its CV curve of the first cathodic sweep. The absence of the peak located at 1.32 V suggests that the intercalation reaction hardly occurs in the initial cathodic process, which could be attributed to its relatively small interlayer distance and larger diffusion energy barrier51. Figure 5b shows their capacity vs. voltage (dQ/dV vs. V) plots at different selected cycles, where the redox peaks of TS-MoSe2 corresponding to the reversible intercalation and conversion reaction hardly change in intensity even after 100 cycles. However, in sharp contrast, the redox peaks of the unstrained MoSe2 electrode almost disappear after 100 cycles, which may ascribe to the loss of active materials owing to the shuttling of polyselenides52. Furthermore, the representative galvanostatic charge and discharge voltage curves of the TS-MoSe2 and MoSe2 anode in Supplementary Fig. 31 agree well with the above CV and dQ/dV results. Figure 5c and Supplementary Fig. 32 show the cycling performance of TS-MoSe2 at the current density of 0.1 A g−1. TS-MoSe2 still maintains a specific capacity of 610 mA h g−1 and an areal capacity of 0.36 mA h cm−2 after 100 cycles, much higher than that of the MoSe2 counterpart (350 mA h g−1, 0.09 mA h cm−2), and that its Coulombic efficiencies are near 100% over 100 cycles, further implying the good cycling stability. TS-MoSe2 also exhibits excellent rate performance. As shown in Fig. 5d, TS-MoSe2 delivers reversible discharge capacities of 652, 604, 562, 533, 502, 460, and 408 mA h g−1 at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g−1, respectively. Notably, when the current rate is switched back to 0.05 A g−1, the specific capacity recovers to about 665 mA h g−1. The cycle and rate performances are superior to those of most of the reported MoSe2-based nanomaterials (Fig. 5e). Additionally, we also tested the cycling stability and rate performance of the TS-MoSe2 electrode with increased loadings of the active materials. As shown in Supplementary Fig. 33, only a slight capacity reduction is observed. Besides, even with relatively high mass ratios of the active materials (the mass ratios of the active materials: carbon: binder are 7:2:1 and 8:1:1, respectively), TS-MoSe2 still exhibits good cycling and rate performance. Meanwhile, Fig. 5f exhibits the rate performance of TS-MoSe2 in the temperature range of 50 to −30 °C was also tested. Surprisingly, TS-MoSe2 displays admirable adaptability to the temperature. When the temperature is as low as −30 °C, TS-MoSe2 still remains a reversible capacity as high as 380 mA h g−1 at 0.1 A g−1 after 100 cycles (Fig. 5g). In sharp contrast, the reversible capacity of MoSe2 at −30 °C is only 128 mA h g−1 after 100 cycles. In view of the superior sodium storage performance of TS-MoSe2 in the half cells, the Na-ion full cells were further assembled with homemade Na3V2(PO4)2O2F (NVPOF) as a cathode to preliminarily assess its practicability as an anode for SIBs (Supplementary Figs. 34, 35). As shown in Supplementary Fig. 35b, c, the full cells exhibit good cycling performance and the capacity can still maintain 434.9 mA h g−1 after 200 cycles at 0.2 A g−1 (based on the mass of the anode). The full cells also present superior rate capabilities (Supplementary Fig. 35d), in which about 70.2% of the capacity can be retained even when the current density increases by 50-folds from 0.1 to 5 A g−1. The good rate capabilities endow the full cells with a specific energy of 108.6 Wh kg−1 at a power density of 19.0 W kg−1, and even 74.1 Wh kg−1 at a power density of 648.5 W kg−1 (based on the total mass of the electrode materials), which are comparable or superior to those of many reported full cells (Supplementary Fig. 35e).

Fig. 5: Electrochemical performance.
figure 5

a CV curves of TS-MoSe2 between 0.01 and 3.0 V at a potential sweep speed of 0.1 mV s−1. b dQ/dV plots of TS-MoSe2 and MoSe2. c Cycling and d rate performances of TS-MoSe2 and MoSe2. e Comparison of the rate capacities of TS-MoSe2 with a series of reported MoSe2-based anodes. f Cycling performances of TS-MoSe2 and MoSe2 at different temperatures from 50 to −30 °C. g Cycling performances of TS-MoSe2 at −10, −30 °C and MoSe2 at −30 °C.

Full size image

Electrochemical kinetics analysis

To deeply understand the excellent reaction kinetics of TS-MoSe2 as the anode for SIBs, the temperature-dependent electrochemical impedance spectroscopy (EIS) spectra of TS-MoSe2 and MoSe2 were investigated (Fig. 6a and Supplementary Fig. 36). The Nyquist plots exhibit a high-frequency semicircle and a low-frequency sloping line, which refer to the charge transfer resistance (Rct) at the electrolyte interface and the Na+ diffusion resistance in the electrode, respectively53. The fitted parameters of TS-MoSe2 and MoSe2 are shown in Supplementary Table 4. Obviously, the Rct values of TS-MoSe2 are always lower than those of its counterparts under all the test temperatures, indicating that the tensile strain contributes to accelerating the electron transfer rate of TS-MoSe2. Then, we further analyzed the diffusion in the low-frequency region by calculating the diffusion coefficient of Na+ (DNa) (the details can be found in Supporting Information). The DNa is inversely proportional to the Warburg factor σ value and the σ can be obtained by fitting the real part Z’ of the electrochemical impedance spectroscopy with ω−1/2. As shown in Fig. 6b and Supplementary Fig. 37, TS-MoSe2 exhibits a much smaller σ value than MoSe2, suggesting its faster Na+ diffusion rate. Moreover, based on the excellent ion diffusion kinetics of TS-MoSe2, we further calculated its apparent activation energy (Ea) of sodium ion diffusion according to Arrhenius equations54,55. As displayed in Fig. 6c, the Ea value of TS-MoSe2 is determined to be 36.99 kJ mol−1, which is smaller than that of MoSe2, manifesting that the tensile strain could evidently lower the reaction activation energy and thus accelerate the reaction kinetics56. Besides, according to the correlation of the phase angle with the characteristic frequency, the corresponding time constant of the sample was also studied using the formula τ0 = 1/f0, where τ0 is the minimum time required to release all the energy with an efficiency >50%. The smaller the value of τ0, the more conducive to rapid ion diffusion and transmission, and f0 is the characteristic frequency when the phase angle is −45°. As shown in Fig. 6d, the time constant of TS-MoSe2 was calculated to be 2.9 s, which is significantly lower than that of MoSe2 (10 s). The fast frequency response of TS-MoSe2 further provides evidence for its smaller charge transfer resistance and better Na+ diffusion/transportation dynamics57,58.

Fig. 6: Electrochemical kinetics analysis.
figure 6

a EIS spectra of TS-MoSe2 at different temperatures after 5 cycles. The inset is an equivalent circuit used to simulate EIS spectra. b σ values at different temperatures calculated from EIS curves. c Arrhenius plots of ln (T/Rct) vs. 1/T in TS-MoSe2 and MoSe2 electrodes. d Bode plots of TS-MoSe2 and MoSe2 electrodes. e Tafel plots of TS-MoSe2 and MoSe2 electrodes during the anodic scan. f The sodium ion diffusion coefficient vs. De/Intercalation state of TS-MoSe2 and MoSe2 electrodes during the discharging/charging process after 5 cycles. g The adsorption sites of sodium ions and the corresponding adsorption energies. h The diffusion pathway of sodium ions and the corresponding diffusion energy barrier.

Full size image

Furthermore, Na+ diffusion kinetics was also evaluated by the linear relationship between the redox peak current (Ip) and the sweep speed (v1/2) based on the CV curves at different scan rates (Supplementary Fig. 38). According to the Randles–Sevcik formula, the slope of the fitted Ip ~ v1/2 is proportional to the DNa (see details in Supporting Information). As presented in Supplementary Fig. 39, the slopes of TS-MoSe2 at the oxidation peak and reduction peak are 1.81 and –1.30, which are greater than those of MoSe2 (1.24/–0.92), in accordance with those obtained from EIS. Furthermore, Tafel plots of TS-MoSe2 and MoSe2 were used to further study their reaction kinetics (Fig. 6e). As the overpotential (η) approaches to zero, the plot deviates sharply from a linear behavior and can be extrapolated to an interception of log i0. Based on the Butler Volmer model, the standard rate constant (k0) of an electrochemical reaction is proportional to its exchange current (i0)59. Clearly, TS-MoSe2 displays a higher i0 value during the anodic scan compared with MoSe2, implying the faster oxidative kinetics of the TS-MoSe2. The galvanostatic intermittent titration technique (GITT) was further performed to access the Na+ diffusion kinetics of TS-MoSe2 and MoSe2 upon cycling (Fig. 6f and Supplementary Fig. 40)60. Clearly, the calculated DNa values of TS-MoSe2 are larger than those of MoSe2 at most of the discharging/charging states, while in some regions, their DNa values almost overlap. Based on the foregoing analyses, TS-MoSe2 experiences the in-/de-tercalation and conversion reactions (generally, the former has higher DNa due to weaker interlayer van der Waals forces3,61), while for MoSe2, Se/Na2Se becomes the sole redox couple after the initial cycling that only occurs the conversion reaction (Se + 2Na+ + 2e− ↔ Na2Se). Thus, the conversion process of the two cases involves similar intermediate phases, thereby resulting in almost the same DNa values.

To further confirm the influence of strain engineering on the diffusion kinetics of sodium ions, DFT calculations were carried out. Two typical adsorption sites were considered in Fig. 6g: the top of the Mo/Se atom, that is, the Top site (T); the hollow position in the center of the six-membered ring, which is the Hollow position (H). Then, the adsorption energy of Na+ (Ead) on TS-MoSe2 and MoSe2 was simulated in Fig. 6g, in which the positive Ead values at the T site suggest that the optimized adsorption site lies in the H site. Additionally, when adsorbed at the H position, the larger Ead of TS-MoSe2 for sodium ions than MoSe2 indicates that the tensile strain can increase the adsorption of the material to sodium ions. Finally, the diffusion path of sodium ions between two adjacent adsorption sites was simulated (the inset in Fig. 6h and Supplementary Fig. 41), and the corresponding diffusion energy barrier was calculated. From Fig. 6h, it can be seen that the diffusion energy barrier of sodium ions in TS-MoSe2 is 0.75 eV, lower than that in MoSe2. The calculation results manifest that the tensile strain can accelerate the dynamics of sodium ions by improving the adsorption energy to sodium ions as well as reducing its diffusion energy barrier, which is consistent with the above experimental results.

Discussion

In summary, we have demonstrated that the constructed TS-MoSe2 can transfer the strain gene to its discharged product Mo and uncovered the decisive role of the tensile strain in regulating Gibbs free energy change of the redox chemistry in the charging process to promote the efficient reversible conversion reaction. By electrochemical in situ Raman, ex situ XPS, and XAS, as well as DFT calculations, it was proved that the tensile strain could improve the activity of Mo, thus resulting in a reversible sodium storage mechanism, which endows TS-MoSe2 with favorable reaction kinetics and thereby highly reversible capacity even in a wide temperature range. Our work provides insights into the electrochemical storage mechanism of conversion-type TMDs, which is essential in improving their electrochemical performance.

Methods

Materials

Ethanol (≥99.7%) and methanol (≥99. 5%) were obtained from Beijing Tongguang Fine Chemical Company. Phosphomolybdic acid hydrate (POM), 2-methylimidazole (2-MI), selenium powder, and bulk MoSe2 were purchased from Aladdin reagents. Hydrazine hydrate (≥80 wt%) was bought from Sinopharm Chemical Reagent Co., Ltd. All chemicals used in the experiments were analytical grade without further purification.

Synthesis of Mo-precursor

In a typical synthesis, 0.03 mmol of phosphomolybdic acid hydrate (POM) was dissolved in 200 mL of ethanol. Then, 200 mL of ethanol containing 6.4 mmol of 2-methylimidazole (2-MI) was dropwise added into the above solution to form a yellow and transparent solution under vigorous stirring. After stirring for 24 h at room temperature, a light green precipitate was collected by centrifuging and washing with ethanol and then redispersed in 50 mL of ethanol/methanol (V/V = 4:1).

Synthesis of TS-MoSe2

Typically, 0.5 mmol of selenium powder was added to 2.0 mL of hydrazine hydrate (80 wt%) in a separate flask, then the colorless solution rapidly changed dark brown and was kept under atmospheric conditions at least for one night. Subsequently, the resultant hydrazine hydrate‒Se solution was slowly added to the aforementioned Mo-precursor solution under vigorous stirring. After that, the mixed solution was transferred into a Teflon-lined stainless steel autoclave and kept at 180 °C for 12 h. After being cooled to room temperature naturally, the generated precipitate was centrifuged and washed with ethanol and water several times and then dried at 60 °C.

Materials characterizations

The morphology and structure of the as-prepared products were characterized by field-emission scanning electron microscopy (FE-SEM; JSM-6490LV) and transmission electron microscopy (TEM; JEOL JEM-2010) as well as high-resolution transmission electron microscopy (HR-TEM). Elemental mapping images were recorded using energy-dispersive X-ray spectroscopy (EDX) attached to TEM. The crystal phase of the products was detected by powder X-ray diffraction (XRD, Bruker D8 Advance) in the range of 5−80° (2θ) with a scanning step of 10o min−1. Raman spectrum was obtained on a Renishaw inVia Raman spectrometer with a laser of 532 nm. X-ray photoelectron spectroscopy (XPS) was measured on an ESCALAB 250 spectrometer (Perkin–Elmer). The elemental contents of C, H, and N were obtained on a Thermo Scientific Flash 2000 CHN-analyzer. Fourier-transform infrared (FT-IR) spectra were collected on a Bruker VECTOR 22 spectrometer. X-ray adsorption spectroscopy (XAS) measurements of the power samples were measured in transmission mode at the 1W1B station in Beijing Synchrotron Radiation Facility (BSRF).

Ex situ XPS characterizations: firstly, the battery was discharged and charged up to the required potential using a LAND workstation at a current density of 0.02 A g−1. Then, the battery was disassembled in a glove box to collect the electrode sheet. Afterward, the resulting electrode sheet was washed with dimethyl carbonate (DMC) to remove any residual salts. Finally, the tested electrode sheet was transported with a vacuum transfer module from the glove box to the XPS test system to avoid component changes when exposed to air. Furthermore, the electrode sheet was also etched with an Ar+ ion beam before the test to further avoid interference from surface SEI.

In situ Raman spectroscopy: The in situ Raman was measured using a Renishaw inVia Raman spectrometer with a laser of 532 nm. The electrochemical cells were adapted from CR2025 coin cells: a 4 mm hole was drilled in the top and then sealed with a thin cover glass slide with epoxy. The homemade coin cells were cycled at 100 mA g−1 between 0.01 and 3.0 V.

Ex situ XAS: Electrodes at different discharge/charge conditions were disassembled and sealed in the glove box. All samples above were measured in fluorescence mode at ambient temperature. And the unreacted MoSe2 detected in EXAFS spectra, which is inevitable, has been subtracted.

Electrochemical measurements

The electrochemical behavior of the as-synthesized products was carried out by using two electrode CR2025 coin cells. All cells were assembled in an argon-filled glove box. The working electrode was composed of 60 wt% active material, 30 wt% carbon black, and 10 wt% sodium carboxymethyl cellulose binder (CMC), which were mixed homogeneously with deionized water and the resultant slurry was pasted onto a copper foil current collector. Then the coated electrode was dried at 120 °C overnight. The thickness, diameter, and area of the electrode were 20–30 μm, 14 mm, and 1.54 cm−2, respectively. The cells were assembled in an Ar-filled glove box and the mass loading of the active material is around 0.8−1.2 mg cm−2. Sodium metal was used as counter and reference electrodes, and glass fiber paper (GF/C, Whatman, diameter: 16 mm) was used as the separator. The electrolyte was a solution of 1 M NaClO4 in a 1:1 volume mixture of ethylene carbonate (EC)/dimethyl carbonate (DMC) to which 5 wt% fluoroethylene carbonate (FEC) was added. Galvanostatic charging-discharging curves of the cell were performed on a LAND CT2001A cell at different current densities in the voltage range of 0.01−3.00 V vs. Na+/Na at room temperature. Furthermore, Electrochemical impedance spectroscopy (EIS) was performed on an electrochemical station (CHI-660) in the frequency range of 100 kHz to 0.01 Hz at the open circuit voltage. In order to effectively avoid errors introduced during the testing process, 3 specimens were tested for each type of battery performance evaluation to ensure that they have almost identical results. The electrochemical stability of the 2-MI molecule was evaluated in a three-electrode configuration using a glassy carbon electrode as the working electrode, a platinum sheet as the counter electrode, and Ag/AgCl (in saturated KCl aqueous solution) as the reference electrode. The potential vs. Ag/AgCl was converted into potentials vs. standard Na+/Na, under the assumption that the potential of the Ag/AgCl electrode was 3.326 V vs. Na+/Na. For the Na-ion full cells, the cathode was made of homemade NVPOF, carbon black, and poly(vinylidene fluoride) (PVDF) in a weight ratio of 8:1:1 on an aluminum foil. The mass loadings of TS-MoSe2 and NVPOF were about 0.85 and 4.05 mg cm−2, respectively. The capacity ratio of anode/cathode was controlled to be around 1.05:1. Meanwhile, the anode was electrochemically activated for three cycles before it was used in the full cells.

Calculation details

Density functional theory (DFT) calculations were carried out with the Cambridge Sequential Total Energy Package (CASTEP)62 based on the plane-wave-pseudopotential approach. The exchange and correlation interactions were described by the Perdew–Burke–Ernzerhof (PBE) functional combined with the generalized gradient approximation (GGA)63. The Grimme’s semi-empirical DFT-D64 correction was contained for the computations to ensure a better description of the electron interaction in a long range. The cut-off energy was set to 750 eV by using the Norm-conserving pseudopotential65,66,67. All atomic positions are fully relaxed during optimization (convergence thresholds of maximum displacement <0.001 Å, maximum force <0.03 eV/Å, and the energy difference <1.0 × 10–5 eV/atom).

The linear response density functional perturbation theory (DFPT)68,69,70 implemented in the CASTEP code was used to obtain the entropies and zero-point energy in the quasi-harmonic approximation. The Gibbs free energies for all the reactions were calculated at 298.15 K, and the calculation formula was defined as:

$$G={E}_{{{{{{\mathrm{DFT}}}}}}}-{{{{{\rm{TS}}}}}}+{E}_{{{{{{\rm{ZPE}}}}}}}$$
(6)

where EDFT, TS, and EZPE stand for the DFT energy, entropy contribution, and zero-point energy, respectively.

The diffusion of Na was evaluated by searching the plausible migration paths and identifying their transition states with the lowest diffusion energy barrier. Transition state searching calculations was performed according to the generalized synchronous transit method implemented in the CASTEP, in which the LST/QST algorithm combined the linear (LST) and quadratic synchronous transit (QST) methods with conjugate gradient (CG) refinements71.

The adsorption energies (Ead) for the Na atom on the TS-MoSe2/MoSe2 are calculated by the following equation:

$${E}_{{{{{{\rm{ad}}}}}}}={E}_{{{{{{{\rm{MoSe}}}}}}}_{2}/{{{{{\rm{cluster}}}}}}-{{{{{\rm{adsorbate}}}}}}}-{E}_{{{{{{{\rm{MoSe}}}}}}}_{2}/{{{{{\rm{cluster}}}}}}}-{E}_{{{{{{\rm{adsorbate}}}}}}}$$
(7)

where EMoSe2/Cluster-adsorbate stands for the total energy of Na adsorbed on the TS-MoSe2/MoSe2 and Na2Se adsorbed on the Mo15/Mo15-MI clusters, EMoSe2/Cluster represents the energy of the TS-MoSe2/MoSe2 and Mo15/Mo15-MI clusters, and Eadsorbate is the energy of Na/Na2Se.