1 Introduction

Transition metal dichalcogenide (TMD) materials MX2 (M = transition metal; X = chalcogen) with lamellar structure have received broad attentions in the fields of batteries, supercapacitors, catalysts, and sensors due to their large layer distance and high surface area [1,2,3,4,5,6]. Among them, transition metal selenides (MSe2) have higher conductivity than the corresponding sulfides (MS2) and oxides (MO2). Moreover, the bond strength of M–Se is weaker than M–S or M–O, which is kinetically favorable for conversion reactions [7, 8]. Specifically, molybdenum diselenide (MoSe2) possesses large interlayer distance (0.65 nm, two times larger than that of commercial graphite) because of weak van der Waals forces and relatively high electronic conductivity (1 × 10−3 S m−1) triggered from the narrow band gap (1.1 eV) [9,10,11], making it a promising electrode material for lithium-ion batteries (LIBs). However, MoSe2 electrode still suffers from large volume variations, inherently low conductivity, and adverse reaction during cycling, which leads to severe capacity fading and inferior rate performance [8, 12,13,14,15].

To address aforementioned concerns, great efforts have been focused on nano-/microstructure design and carbon modification. In general, nanosized materials can effectively eliminate the effect of volume variations, providing a short ion diffusion length and offer large pseudo-capacitance, which lead to fast ionic migration and superior rate performance [16,17,18,19]. However, the large contact area between electrode materials and electrolyte may increase the side reaction and cause the dissolution of the electrode materials. Another effective strategy is making TMD and carbon composites, in which carbon serves as a soft matrix to buffer the volume variations and fast electron conductor [20,21,22,23]. Particularly, three-dimensional (3D) porous carbon skeletons are preferred because of their extra capability to tolerant the volume changes. Zhao et al. have achieved high-performance Co3O4 lithium-ion battery anode by employing 3D carbon network substrate [24]. Cao and co-workers synthesized 3D porous Mo2C/C architecture to improve the lithium storage capacity [25]. Although the electrochemical performances are improved greatly, the long-term cycling stability is rear reported. The difficult can be attributed to the weak bonding between the active material and carbon substrates, or the mismatch of their volume variations, which degrade the structural stability of the electrode upon long-term cycling.

Interface engineering can introduce distortions, dislocations, and lattice defects, with distinguished electronic structures, which are long-range disorder and, hence, can lower the activation barrier, thus boosting reaction kinetics [5, 26, 27]. More recently, through interface coupling by chemical bonds, previous theoretical calculation and practical experiments both have revealed that Mo–O–C or Mo–C bonds between carbon and MoSe2 can effectively enhance electronic conductivity and structure stability through the interface [28, 29]. However, the portion of the chemical bonding between the active material and substrate is still too limited to stabilize structure. Therefore, it is highly expected to increase the chemical bonds at the interface or develop new strategy to improve the stability of composites, in particular for long-term cycling applications. Introducing intermediate material to bridge the active material and the substrate presents to be a wise choice. Comparing to MoSe2, MoC owns intrinsic higher electrical conductivity and chemical stability [30,31,32]. The structural stability and charge transportation will be greatly improved by the incorporation of MoC into the composite, which bridges MoSe2 and carbon substrates by MoSe2–MoC–carbon connection.

Herein, we reported the temperature-induced one-step incorporation of MoC as an intermediate phase to bridge MoSe2- and N-doped three-dimensional carbon framework, which essentially improve the interfacial structural stability of the composite. Moreover, the in situ formed MoC can effectively inhibit the overgrowth of MoSe2 nanocrystals and constrain MoSe2 to nanodots with few layers and rich edge sites. The obtained materials guarantee bi-continuous fast electron/ion transport and highly reversible conversion reactions and enable fast reaction kinetics and good structural stability. As expected, 3D porous MoSe2/MoC/N–C electrodes deliver much improved long-term cycling stability. It retains 90% of its initial capacity after 5000 cycles at 2 A g−1. And the assembled LiFePO4//MoSe2/MoC/N–C full cell also exhibits a high reversible capacity and good cyclic stability (86% capacity retention after 300 cycles at 2 A g−1).

2 Experimental

2.1 Synthesis of 3D Porous MoSe2/MoC@N–C and MoSe2/C

All chemicals were used as received without further purification. 0.4 g of (NH4)6Mo7O24·4H2O, 0.4 g of PVP and 1.2 g of NaCl were dissolved in 30 mL of distilled water. After stirring for 1 h at room temperature, the resulting transparent solution was quick-frozen with liquid nitrogen and further freeze-dried for 30 h at − 40 °C in vacuum to yield a precursor. Then, the precursor was mixed with selenium powders in a mass ratio of 5:1 and annealed under Ar atmosphere at 600 or 800 °C for 3 h with a heating rate of 5 °C min−1. After that, 3D porous MoSe2/C and MoSe2/MoC@N–C were obtained after dissolving NaCl in deionized water. For comparison, bare MoSe2 was prepared by annealing a mixture of ammonium molybdate tetrahydrate and selenium powders at 800 °C for 3 h with a heating rate of 5 °C min−1 under Ar atmosphere. MoC/C could be obtained by annealing the same precursor under Ar atmosphere at 750 °C for 3 h with a heating rate of 5 °C min−1 without Se powder.

2.2 Characterizations

The structure features of the samples were performed by X-ray diffraction (XRD, Rigaku D/max 2500), Raman microscope (Horiba Jobin–Yvon, Lab Ram Aramis), and X-ray photoelectron spectroscopy (XPS, AXIS-ULTRA DLD-600 W system). The morphologies and energy-dispersive X-ray (EDX) element mapping (SEM, Quanta FEG 250) were characterized by scanning electron microscopy and Transmission electron microscope (TEM, JEOL JEM-2100F). Specific surface areas were calculated by the multipoint Brunauer–Emmett–Teller (BET) method.

2.3 Electrode Fabrication and Electrochemical Measurement

Electrochemical performances were evaluated by using 2032-type coin half cells with metallic lithium served as the anode. Working electrodes were prepared by casting a slurry containing the active material (70 wt%), Super P (15 wt%) and sodium carboxymethyl cellulose (15 wt%) dispersed in distilled water on a clean Cu foil current collector. Each electrode in our experiment has an area of 1.1304 cm−2, and the loading of active material for each electrode was about 1.0–1.2 mg. All cells were assembled in a glove box (Mbraun, Germany) filled with ultrahigh-purity argon. For LIB assembly, polypropylene membrane and 1 M LiPF6 in ethylene carbonate (EC)–dimethyl carbonate (DMC)–diethyl carbonate (DEC) (1:1:1 in volume) were used as a separator and electrolyte, respectively. As for the assembly of full cell, a pre-lithiation procedure was carried on to compensate the loss of lithium during the initial cycle in half cell. To ensure the maximized material utilization and reasonably evaluate the electrochemical property of the MoSe2/MoC@N–C, the full cell in this work was assembled based on the capacity ratio of ≈ 1.2:1 between the LiFePO4 cathode and MoSe2/MoC/N–C anode.

The cyclic voltammetry (CV) measurements were taken on an electrochemical workstation (CHI660C) at a scan rate of 0.1 mV S−1 in the voltage range of 0.01–3 V (vs. Li+/Li). The electrochemical impedance spectrometry (EIS) data were collected on a ZAHNER-IM6ex electrochemical workstation (ZAHNER Co., Germany) in the frequency range of 100 kHz to 10 mHz. Land battery tester (Land CT 2001A, Wuhan, China) was employed to investigate the galvanostatic charge/discharge performances, and all tests were conducted at room temperature.

3 Results and Discussion

3.1 Formation Mechanism and Structure Characterization of Materials

The synthetic mechanism of the 3D porous MoSe2/MoC/N–C and advantages of the structure are shown in Scheme 1. As a nitrogen-rich carbon network source, polyvinyl pyrrolidone (PVP) is a nonionic surfactant that can adsorb molybdate ions to facilitate the in situ formation of Mo–C bond. When the precursor annealed at high temperature, molybdenum can react with carbon to form MoC. Selenium powder is used to form MoSe2. And NaCl is employed as a template to create pores due to its environmental friendliness, cheapness, and high melting point. The obtained porous structure with 3D electrically conductive pathways possesses both large surface areas and abundant mass transportation channels, guaranteeing the good contact between electrolyte and electrode interface. More importantly, as shown in Scheme 1b, different from previous reports of MoSe2/C composites, MoSe2 and carbon substrate were bridged by highly conductive MoC interphase, forming a tri-phase MoC/N–C heterostructure where the MoSe2/MoC/N–C connection can boost fast charge kinetics and highly reversible conversion, leading to improved long-term cycling stability. This MoSe2/MoC/N–C connection also induced the growth of MoSe2 to terrace-terminated mode (illustrated in Scheme S1), resulting few-layered MoSe2 nanodots, and thus, greatly contributed to pseudo-capacitance effect and guaranteed the fast the ion diffusion kinetics, which synergistically boosts lithium storage capability.

Scheme 1
scheme 1

Schematic illustration of a the formation process of interconnected 3D porous structure of obtained MoSe2/MoC/N–C nanocomposite and b model of interface bridging between MoSe2- and N-doped carbon network

More specific characterizations of material are comprehensively presented within Figs. 1 and 2. As shown in Fig. 1a (XRD patterns), the as-prepared sample annealed under 800 °C exhibited two additional diffraction peaks located at 35.7° and 48.7° compared to MoSe2 phase (JCPDS No. 29-0914, 2H-MoSe2), which are corresponding to MoC phase (JCPDS No. 45-1015). Specifically, compared to bare MoSe2 and MoSe2/C, the broadened diffraction peaks in MoSe2/MoC/N–C sample indicate its relatively small crystal size. And the weaker intensity of (002) peak indicates rich defects along [002] direction [28]. Unlike the crystallinity normally increase with the rising of the annealing temperature (800 °C), the intensity and crystallites of MoSe2 in the composite decreased. The XRD patterns suggest the growth of MoSe2 may be changed. No detection of XRD peaks for Mo metal indicates the easier formation of MoC and its stable state at 800 °C.

Fig. 1
figure 1

a XRD patterns and b Raman spectrum of bare MoSe2, MoSe2/C, and MoSe2/MoC/N–C; XPS spectra of MoSe2/MoC/N–C: c Mo 3d, d Se 3d, e N 1 s, and f C 1 s

Fig. 2
figure 2

SEM images of a MoSe2/C and b MoSe2/MoC/N–C; c statistical data of surface area; df TEM and g HRTEM images of MoSe2/C; hj TEM and k HRTEM images of MoSe2/MoC/N–C (inset: FFT of the selected area); l elemental mappings of MoSe2/MoC/N–C

As for the Raman characterization, all the samples present two bands at around 240 and 280 cm−1 (Fig. 1b), which are assigned to the A 1g and E 12g stretching of MoSe2, respectively. The A 1g -related peak is attributed to the exclusive peak of MoSe2 2H structure, preferred for edge-terminated mode, while the E 12g peak is assigned to the characteristic peaks of MoSe2 1T structure, favored for terrace-terminated MoSe2 [13, 33, 34]. Thus, 1T/2H phases of MoSe2 were consisted in three synthesized samples. Moreover, the increased intensity ratio of A 1g /E 12g from bare MoSe2 to MoSe2/MoC/N–C suggests the decreased thickness and increased exposed edge sites of MoSe2 in the interconnected 3D porous structure. The D and G bands locating at 1375 and 1590 cm−1 confirm the presence of disordered and graphitic carbon structures within interconnected 3D porous superstructure, respectively [12, 35, 36]. The content of carbon was calculated as 34.6% in the MoSe2/C composite, determined by thermal gravimetric (TG) analysis as shown in Fig. S1, and the residual product (MoO3) is verified by the XRD pattern in the inset. Furthermore, the XPS spectrum shows the existence of Mo, Se, C, and N elements (Fig. S2). The Mo 3d XPS spectrum (Fig. 1c) shows three doublets: 229.5 and 232.4 eV assigned to Mo–Se bonds, 233.3 and 235.7 eV corresponding to Mo–O–C, and 228.6 and 231.7 eV corresponding to Mo–C bonding [13, 24, 33, 37], confirming the formation of MoC in the MoSe2/MoC/N–C nanocomposite, which is consistent with the XRD results. More importantly, the formation of Mo–O–C and Mo–C bonds could be demonstrated, which implies the strong electronic coupling at the interface between MoC, MoSe2and 3D porous carbon. The formation of Mo–O–C bonds was mainly due to the combination of Mo ion with the carbonyl group (–C=O) in the PVP. Previous experiments and theoretical calculations have shown that Mo–C or Mo–O–C bonds in MoSe2/C nanocomposite could facilitate the ion and electron transportation in the composite [28, 29]. For the Se 3d XPS spectra, two typical peaks at 55.5 and 54.8 eV correspond to the 3d3/2 and 3d5/2 peaks of Se2− in the MoSe2/MoC/N–C nanocomposite (Fig. 1d). The C 1s profile in Fig. 1f can be deconvoluted into three individual peaks at 285.5 eV (C–N), 284.7 eV (C–C and C=C), and 287.1 eV (C–O, C=O). The formation of C–N bonds demonstrated the nitrogen doping in the carbon substrate. Furthermore, the N 1s peaks are partially overlapped with Mo 3p peaks (at 395.1 eV) and could be deconvoluted as graphitic N at 400.9 eV, pyrrolic-N at 399.5 eV, and pyridinic-N at 398.3 eV [13, 24, 33, 37]. The nitrogen doping could introduce more defects and electroactive sites, resulting in good electron transportation in the composite [38].

As depicted in Fig. S3, pure MoSe2 agglomerates into irregular bulks about 20 μm in width. However, as expected, both MoSe2/C and MoSe2/MoC/N–C possessed a highly interconnected three-dimensional porous network with microstructure (Fig. 2). Interestingly, compared to MoSe2/C, MoSe2/MoC/N–C possesses more uniform pore distribution and thinner network thickness (Fig. 2a, b), which can be further confirmed by TEM observations. The uniform pores (about 80 nm) were created by dissolution of NaCl and the release of gases generated during the selenization. As a result, MoSe2/MoC/N–C composite possesses a large Brunauer–Emmett–Teller specific surface area (SBET) of 34.9 m2 g−1 and MoSe2@N–C possesses specific surface area of 21.5 m2 g−1, both of which are much larger than 7.9 m2 g−1 of bare MoSe2 (Fig. 2c). The detailed information of pore size distribution of three samples is shown in Fig. S4. The high SBET and the porous structure are favorable to ion diffusion and structural stability upon cycling.

The semitransparent TEM images further indicate the ultrathin network structure of composites (Fig. 2d–k). When we zoom in, for MoSe2/C sample, thick nanosheets with various sizes were attached to or embedded in the network (Fig. 2e, f), which were proved as MoSe2 (Fig. 2g). In general, agglomeration is more likely to occur in high-temperature reactions. However, MoSe2 crystals in MoSe2/MoC/N–C sample (800 °C) are smaller and more dispersed than those in MoSe2/N–C (600 °C). In comparison, for MoSe2/MoC/N–C, the interconnected 3D porous architecture (Fig. 2h) is more obvious than that of MoSe2/C. Moreover, a large number of small particles with size of 5–20 nm are uniformly distributed in smooth N–C network (Fig. 2i, j). These small sizes are favorable to the lithium-ion intercalation/deintercalation. Figure 2j further indicates that the thin carbon layer totally coats the small MoSe2/MoC nanocrystal. The N–C capsule can effectively protect nanosized MoSe2 from adverse reactions and facilitate fast electron transport.

The high-resolution TEM (HRTEM) image (Fig. 2g, k) reveals that nanodots in MoSe2/MoC/N–C nanocomposite is composed of only a few layers (about 10 layers), much less than that of MoSe2/C. Furthermore, the layer distance of MoSe2/MoC/N–C (0.676 nm) is also larger than that of MoSe2/C (0.647 nm), which facilitate fast Li-ion transportations. Notably, the interlayer distances of 0.25 nm marked in Fig. 2k is assigned to the distance spacing of (100) plane of MoC crystal, which is in close contact with MoSe2 crystals. And the corresponding fast Fourier transform (FFT) result (inset in Fig. 2k) revealed the (110) and (100) plane of the MoSe2 and MoC, respectively, further confirming the formation of MoSe2/MoC heterostructure. Because the carbon source of MoC is from the carbonized PVP (N–C substrate), the in situ formed MoC is bonded to carbon framework, which can be confirmed by Mo-C and Mo–O-C bonding from XPS (Fig. 1c). Therefore, it can be deduced that the intermediate MoC bridges N–C framework and MoSe2 through MoSe2/MoC/N–C connection, strengthening the interfacial coupling in tri-phase MoSe2/MoC/N–C superstructure, which enable fast charge transportation and good structural stability. And defects in MoSe2/MoC/N–C multiphase boundaries with distinguished electronic structures could lower the activation barrier, thus boosting reaction kinetics. In addition, around nanodots, some graphitic carbon fringes about 0.37 nm are observed, as indicated by the yellow dotted line (Fig. 2k). The graphitic carbon in the composite can provide conductive pathways for rapid electron transfer even at high current densities.

By comparing the structure of MoSe2/C and MoSe2/MoC/N–C composites (Fig. 2), it can be concluded that the in situ formed intermediate MoC could effectively inhibit the overgrowth of MoSe2 nanocrystals and expand their layer spacing. During the heating process, partial Mo reacted with Se to form MoSe2 first. When the reaction temperature rose to 800 °C, Mo and carbon reacted to form dispersion MoC, while MoSe2 crystals continue to nucleate and grow, forming MoSe2/MoC/C heterojunction. This MoSe2/MoC/C connection induced the growth of MoSe2 to terrace-terminated mode (illustrated in Scheme S1), resulting in few-layered and small-sized MoSe2 nanodots. Furthermore, the in situ formed MoC can pin heterostructure, preventing crystals form and agglomerating. The dark-field scanning TEM and the corresponding elemental mapping images of MoSe2/MoC/N–C reveal that Mo, Se, O, N, and C are evenly distributed on the ultrathin carbon wall (Figs. 2l and S12). The distribution of Mo and Se concentrate on the location of nanodots, indicating the formation of MoSe2 again, while the uniform distribution of N throughout the sample demonstrates that the 3D porous framework consists of N-doped carbon.

3.2 Electrochemical Lithium-Ion Storage Performance and Reaction Kinetics

The MoSe2/MoC/N–C composite was assembled to half and full cells to investigate their electrochemical performances. The cyclic voltammetry (CV) curves obtained during the initial 3 cycles (at a scanning rate of 0.1 mV s−1) exhibit multiple redox reactions for MoSe2/MoC/N–C electrode (Fig. 3a). During the first cathodic sweep, four obvious reduction peaks at 0.605, 0.782, 1.175, and 2.138 V can be clearly observed. The small peak at 1.175 V can be assigned to the reduction of MoC, and other peaks are well documented for MoSe2 [14, 31, 39]. These peaks are also consistent with the results in CV curves of MoSe2/C and MoC/C counterparts in Figs. S7b and S8c. The reduction peak at 0.782 V can be ascribed to the insertion of Li ion into MoSe2 to form LixMoSe2, and the subsequent peak at 0.605 V can be ascribed to the reduction of LixMoSe2 to Mo and Li2Se [28, 44]. In the first cycle, two oxidation peaks at 1.428 and 2.138 V correspond to a partial oxidation of Mo metal to form MoSe2 and oxidation of Li2Se to Se, respectively [14]. In the next cycles, reduction peaks at 0.605 and 0.782 V disappeared, but new peak at 1.847 V can be obtained for the conversion of LixMoSe2 to Mo and Li2Se [41,42,43,44], while two oxidation peaks at 2.138 and 1.428 V remain constant. The shift of peaks of MoSe2 involved with the irreversible reaction [43,44,45,46,47]. However, the peak at 1.175 V corresponding to MoC remains constant (the same as this peak at CV curves of MoC/C in Fig. S8c), suggesting the reversible conversion reactions of MoC [31, 39]. Figure S6 presents charge/discharge profiles of selected cycles of MoSe2/MoC/N–C electrode; after initial cycle, two platforms at around 1.8 and 1.2 V can be ascribed to the contribution of MoSe2 and MoC in the composite electrode, respectively. The result is the same as peak voltages in CV curves. The initial charge and discharge capacities are 812 and 1014 mAh g−1, respectively, corresponding to a coulombic efficiency (CE) of 80.1%, which is higher than most reported MoSe2 electrodes [14, 24, 40, 41]. The irreversible capacity in the initial cycle may come from the formation of a SEI film [24]. From the second cycle, all charge/discharge curves were close to overlapping, indicating the excellent electrode reversibility.

Fig. 3
figure 3

a Initial three cyclic voltammograms cycles of the MoSe2/MoC/N–C electrode at a scan rate of 0.1 mV s−1. b Cycling performance (0.1 A g−1), c rate performances, and d long-term cycling performances (2 A g−1) of bare MoSe2, MoSe2/C, and MoSe2/MoC/N–C electrodes between 0.01 and 3 V versus Li/Li+. e Comparison plot of rate performance between this work and previously published MoSe2/C composite-related works. f Schematic illustration of the LiFePO4//MoSe2/MoC/N–C full-cell configuration. g Charge/discharge profiles of the full cell. h Cycling performance of the full cell at 0.1 A g−1

Figure 3b compares the cycling performances of three samples at 0.1 A g−1. MoSe2/MoC/N–C nanocomposite exhibits the highest specific capacity of around 800 mAh g−1 compared to MoSe2/C (650 mAh g−1) and bare MoSe2 (440 mAh g−1). This result demonstrates that constructing interconnected 3D porous carbon-modified nanocomposite can indeed improve the capability of storage Li ion. By the way, in the early stages of the cycling (Fig. 3b), the capacities of MoSe2/C and MoSe2/MoC/N–C rise slightly, which may be related to the activation of the electrode material. MoSe2/MoC/N–C shows the best rate performance (Fig. 3c). The electrode delivered average capacities of 773, 720, 680, 648, 634, 622, and 575 mAh g−1 at 0.1, 0.2, 0.5, 1.0, 1.5 2.0, and 4.0 A g−1, respectively, and it can fully recover and even displays the higher capacity when the current density switched back to 0.1 A g−1. However, MoSe2/C and pure MoSe2 undergo apparent capacity loss at high rates in different degrees. The obtained high specific capacity and good rate capability of the MoSe2/MoC/N–C electrode also show superiority compared to many other carbon-modified MoSe2 anode materials (Fig. 3e) [41,42,43,44,45,46,47], which demonstrates the advantage of fast interfacial charge transfer through MoSe2/MoC/N–C connection again.

What makes the MoSe2/MoC/N–C electrode more attractive is its ultralong cycle life with the capacity retention of 90% and a high capacity of 535 mAh g−1 after 5000 cycles at 2 A g−1 (Fig. 3d). In sharp contrast, for MoSe2/C electrode, a capacity of 308 mAh g−1 can be maintained just 1000 cycles, and bare MoSe2 even undergoes a sharply fading after 295 cycles.

To confirm the role of MoC in lithium storage, three-dimensional porous MoC/C was prepared through a similar routine of preparing MoSe2/MoC/N–C. The SEM image (Fig. S8b) shows that MoC/C possesses the similar morphology to MoSe2/MoC/C composites (Fig. 2b). The cyclic voltammetry (CV) curves and the cycling performance of MoC/C composite as anode for LIBs were further carried out to study the lithium storage properties of MoC (Fig. S8c, d). As a result, it exhibits the capacity close to that of MoSe2/MoC/N–C and good electrochemical reversibility. Meanwhile, the weight ratio between MoSe2 and MoC was estimate as 51:2 (molar ratio of 54:5) according to Table S1, indicating that MoC possess relatively low contents in the electrodes. Therefore, the direct capacity contribution from MoC to the whole electrode is relatively limited. However, MoC plays an important role in the interfacial structure regulation, including more stable chemical binding and reduced MoSe2 particle size, which is more correlated with the rate performance and structural stability. MoSe2/MoC/N–C shows a higher capacity than MoSe2/C mainly because its smaller MoSe2 nanodots can provide more lithium storage sites. The MoC connected MoSe2 and N–C network through C/MoC/MoSe2 multiphase boundaries, enabling fast electron/ion transport and structural stability through interface, resulting in good rate performance and structural stability.

Encouraged by the excellent performance MoSe2/MoC/N–C electrode in half cell, we assembled a lithium-ion full by pairing with the commercial LiFePO4 cathode (Fig. 3f). As shown in Fig. 3g, the full cell displays a complete discharge plateau at about 2.3 V with a reversible capacity of about 267 mAh g−1 at 2 A g−1. Meanwhile, its energy density was calculated as 93.9 Wh kg−1 at the high power density of 698.5 W kg−1. It is still a great challenge to obtain stable cycling performance of the full-cell batteries based on conversion reaction electrodes because of the low coulombic efficiency and large consumption of lithium sources during the initial cycles to form SEI layers. However, the LiFePO4//MoSe2/MoC/N–C full cell exhibits good capacity retention of 86% after 300 cycles (Fig. 3h), indicating the good cyclic stability.

The reaction kinetics of MoSe2/MoC/N–C electrode was systematically studied by CV measurements, galvanostatic intermittent titration technique (GITT) and EIS spectroscopy. The electrochemical kinetics in the as-prepared electrodes could be considered as capacitive-dominated and diffusion-controlled from the CV curves (Figs. 4a and S9). After detailed calculations as detailed descriptions in the Supporting Information, the higher b value of the MoSe2/MoC/N–C electrode (0.90 vs. 0.69 of the MoSe2/C at the cathodic peak) reveals a faster ionic transportation which resulted in a better rate performance (Fig. 4b) [48,49,50,51]. Figure 4c shows the calculated voltage profile for the capacitive current (colored region) at the scan rate of 1.2 mV s−1. As a result, 74% of the total capacity is calculated as the capacitive contribution for the MoSe2/MoC/N–C electrode, higher than 64% of the MoSe2/C. Furthermore, Fig. 4d compares the calculated capacitive contribution of two electrodes at various scan rates. The ratios of capacitive contribution of MoSe2/MoC/N–C electrode are always higher than those of MoSe2/C, with a maximum value of 90% at 2 mV s−1. The large pseudo-capacitive contribution would play a critical role to be benefit for ionic transportation in the electrode [48, 49], which boosting high-rate Li-ion storage capability.

Fig. 4
figure 4

Quantitative capacitive analysis of lithium storage behavior. a CV curves at different scan rates of the MoSe2/MoC/N–C electrode. b Relationship between logarithm cathodic peak current and logarithm scan rates. c Capacitive contribution (red wine for MoSe2/MoC/N–C and navy blue for MoSe2/C) and diffusion contribution (gray) at 1.2 mV s−1. d Normalized contribution ratio of capacitive capacities at different scan rates; GITT curves and the corresponding Li-ion diffusion coefficient at e the 2nd discharge process and f the 2nd charge process. TEM and corresponding HRTEM images of g, h MoSe2/C and i, j MoSe2/MoC/N–C electrodes after 100 cycles at 2 A g−1 (full charge state)

The diffusion-controlled reaction kinetics of samples were further analyzed in depth by GITT (Fig. S11). As shown in Fig. 4e, f, the diffusion coefficient of MoSe2/MoC/N–C electrode is about 3.93e−11 to 1.55e−10 cm2 s−1, higher than those of MoSe2/C and bare MoSe2, suggesting that expanded layer distance, rich edge defect MoSe2 nanocrystal, and MoSe2/MoC/N–C connection definitely facilitate fast Li-ion diffusion kinetics. The electrochemical impedance spectra (EIS) results (Fig. S10 and Table S2) demonstrated that charge transfer resistances (Rct) of MoSe2/MoC/N–C (43 Ω) are smaller than those of MoSe2/C (72 Ω) and bare MoSe2 (557 Ω), suggesting that MoSe2/MoC/N–C shows the fastest interface kinetics. All above analysis suggested that the optimized interconnected 3D porous N–C framework and stronger electronic coupling at interface through MoSe2/MoC/N–C connection facilitate reaction kinetics.

Ex situ TEM images were employed to investigate the structural integrity of MoSe2/MoC/N–C electrode after certain cycles at 2 A g−1. Compared to the ex situ TEM images of as-synthesized MoSe2/C (Fig. S13A1 and A2), the MoSe2 flakes in the MoSe2/C electrode (Fig. 4g or Fig. S13A3) were broken into small pieces and form aggregates after 100 cycles, which may further lead to electrode pulverization and indicate its inferior structure stability. However, at the same cycles, MoSe2/MoC/N–C electrode (Fig. 4i or Fig S13B3) displayed the good preservation of interconnected 3D porous network structure, and the small nanocrystals encapsulated in the carbon could still be observed, indicating the excellent structural stability of MoSe2/MoC/N–C nanocomposite. Compared to the 10-layered structure in Fig. 2k (or Fig. S13B2), the HRTEM image (Fig. 4j or Fig. S13B4) demonstrates that MoSe2 nanodots have been exfoliated into less layers (about 4 layers) and smaller size (about 5 nm). The interlayer distance also expanded to 0.751 nm after cycling. In addition, large areas of graphitized carbon were observed with an enlarged spacing (about 0.411 nm), slightly larger than the initial value of 0.370 nm for the pristine carbon. The expanded interlayer distance of MoSe2 and carbon could be caused by the repeated insertion/extraction of Li ions [28, 36]. The smaller MoSe2 nanocrystal can further provide Li storage sites, benefiting the capacity of electrode. And the increased graphitization carbon layer enhanced the electronic conductivity. These changes of electrodes material during cycling result in climbing capacities. To further examine the reinforced structure stability of MoSe2/MoC/N–C, the TEM images of MoSe2/MoC/N–C electrode after 500 cycles are compared in Fig. S13B5–B7. Figure S13B5 confirms that the MoSe2/MoC/N–C electrode material retains integrated. As shown in Fig. S13B6 and B7, small shadow dots were still distributed in the carbon framework in MoSe2/MoC/N–C electrode, implying the products were also evenly constrained in carbon framework. These results prove that MoSe2/MoC/N–C connection can constrain the in situ generated MoSe2 and facilitate the highly reversible conversion during cycling.

4 Conclusion

In conclusion, 3D porous MoSe2/MoC/N–C nanocomposite was synthesized by a temperature-induced method. The interface bridging between MoSe2 and carbon framework was tuned by MoSe2/MoC/N–C connection, greatly improves the structural stability and electronic conductivity and reduces the interfacial charge resistance. Moreover, it also exhibits many features favorable for lithium-ion storage, such as MoSe2 nanodots encapsulated into 3D connected carbon network to improve charge/ion transfer kinetics and minimize the effect of volume expansion; expanded interlayer spacing of MoSe2 to promote lithium-ion diffusion; large electroactive surface area from N–C and rich edge defects of MoSe2; and reinforced structure stability by intermediate in situ MoC. As a result, the as-prepared electrode delivered ultralong and stable cycling performance with a specific capacity of 618 mAh g−1 after 5000 cycles (90% capacity retention) at 2 A g−1. It also shows promising potential in LiFePO4//MoSe2/MoC/N–C full cell (86% capacity retention at 2 A g−1 after 300 cycles). Inspired by the harvest of superior cell performance, in situ engineering corresponding metal carbide as interphase to connect metal dichalcogenide and carbon matrix can be developed as an innovative and effective strategy to achieve high-electrochemical-activity materials.

5 Supporting Information

Experimental section, supplementary scheme diagram of interface modification, TG, XRD, BET, SEM, EIS, CV, GITT and equations associated with this article are given in the online version or from the author.