Introduction

With the rapid development of the world economy and the shortage of nonrenewable resources such as fossil fuels, finding alternative green energy has become an urgent research topic for researchers [1]. Lithium-ion batteries (LIBs) have attracted extensive attention and in-depth research in light of their advantages of high energy density, large output power, cycle stability, no memory effect, long lifespan, and environmental friendliness [2]. Today, LIBs have become the dominant power source in our lives, from small portable electronic devices to large electric and hybrid electric vehicles [3, 4]. However, graphite, as a widely used anode material, has ideal cycling performance, and its low theoretical specific capacity seriously affects the performance of LIBs [5]. Based on this, researchers have constantly looked for excellent anode materials in recent years. Transition metal oxides (TMOs) as electrode materials have attracted significant attention for their rich resources and high capacity [6,7,8]. For example Fe2O3 [9], Fe3O4 [10], NiO [11], CuO [12], MnO2 [13], and various TMOs (M = Ni, Co, Cu, Ti, etc.) [14, 15] have been intensively studied in LIBs. Among these metal oxides, Co3O4 nanomaterials are the most prominent anode materials for LIBs in light of their high theoretical specific capacities, diversity of oxidation valence, and environmental friendliness. Co3O4 can be obtained from natural abundant and inexpensive resources [16, 17]. However, there are also some severe drawbacks that are similar to those of other pristine TMOs, such as perishing cycle stability and poor ionic conductivity.

To overcome the unilateral problem of materials, combining the advantages of different materials, using a carbon layer to wrap cobalt tetroxide is a breakthrough attempt [18]. Hollow structured micro-/nanospheres, due to their unique electrochemical properties, such as low density, large specific surface area, good thermal stability, strong surface permeability, and relatively large internal space, are attracting increasing research [19,20,21,22,23,24]. Specifically, yolk-shell nanostructure materials as a special structure of LIB anode materials have received robust attention. In contrast to the general core-shell structure, the yolk-shell nanostructure materials with a movable space inside the protecting shell can effectively adjust the spatial variation compared to hollow nanostructures [25]. As a promising precursor at present, metallic organic frameworks (MOFs) have highly ordered permanent pore structures [26] and have been commonly prepared for hollow structures after thermal decomposition [27, 28]. Through high-temperature pyrolysis, abundant organic ligands in MOFs become well-suited carbon sources for the synthesis of porous carbon and metal-doped carbon [29]. With this consideration in mind, Co3O4 yolk-shell hollow spheres covered by carbon are considered to be suitable for use in the upgrade of LIBs. For example, Chen and coworkers synthesized Co3O4 hollow nanofiber through template-based engineering, and Lou and coworkers designed a multistep approach to prepare hierarchical tubular structures, both of which showed somewhat improved reversible specific capacity [30,31,32]. Some MOFs, such as MOF-5, MOF-74, ZIF-8, ZIF-11, ZIF-67, and ZIF-68, have been used as templates for fabricating nanoporous carbons [26, 33, 34]. Through calcining MOF precursors in controlled atmospheres, Mai and coworkers designed hierarchical yolk-shell Co3O4/C dodecahedrons which showed excellent cycling stability [30, 35]. However, the above materials reported are more single-shell structures obtained through complicated synthesis approaches. Hollow multishell structures (HoMSs) are defined as those built up with at least two shells and two corresponding internal voids. Unlike single-shell materials, they are suggested intensively as electrodes [19, 36,37,38,39,40,41,42]. Higher volume energy density and better structural stability have helped them to become potential electrodes in energy storage applications [43,44,45]. Recently, Altin S, Yaşar S, and coworkers using ZIF-12 as the precursor synthesized unique transition metal-doped carbon composite materials, improving the electrochemical performance of the lithium-ion batteries [26, 46]. However, to date, accurately and effectively designing multishell hollow sphere structures is still a challenge.

Compared with ball milling [47], hydrothermal [48], and solvothermal methods [49], the molecular precursor pyrolysis strategy can accurately regulate the crystal phase, composition, and morphology of the target product by changing the experimental parameters [50]. In this paper, with spherical ZIF-12 as a molecular precursor for pyrolysis, two carbon materials CHS@C and CHS@CG were obtained. The preparation method is simple, universal, and repeatable. Because of the excellent electrical conductivity of graphene and the interfacial synergistic effect between Co3O4 and rGO, the prepared hollow nanostructure microspheres CHS-400 show excellent Li-storage capacity. The specific capacities reach 645 and 831 mAh·g−1, respectively, after 400 cycles at the rate of 0.2 C, and in the constant charging and discharging process, the Coulombic efficiency reaches nearly 100%, exhibiting a good reversible cycle. Both materials have excellent structural stability and excellent conductivity. By comparison, CHS@CG has better cyclic stability in LIBs due to the unique yolk-shell architecture filled with and covered by graphene sheets. The above effective design ideas for synthetic methods will be conducive to the development of electrodes for LIBs and the preparation of other TMOs of electrode materials and other energy storage devices.

Results and discussion

Scheme 1 illustrates the detailed formation process of CHS@C and CHS@CG. Carbon-wrapped Co3O4 hollow sphere structures (CHS@C and CHS@CG) are obtained after ZIF-12-1.0 g PVP calcination at 400 °C in an air atmosphere. Due to the addition of PVP, ZIFs have changed from a previously angular structure into a spherical framework compound with metal Co2+ as the center and ligand assembly (Fig. S1, Fig. S2). Due to the excellent thermal conductivity of graphene, the rapid heat transfer in the pyrolysis process leads to the formation of a double-layer Co3O4 yolk-shell hollow sphere structure (CHS@CG) with the addition of graphene. Without the addition of graphene, a multilayer Co3O4 hollow sphere structure (CHS@C) will form due to the gradual heat transfer process. Carbon in the structure of both materials comes from the pyrolysis of the ligand benzimidazole, and the CHS@CG material is also filled with and covered by graphene sheets from the pyrolysis of graphene. Both materials have excellent structural stability and excellent conductivity. In comparison, CHS@CG has better cyclic stability in LIBs due to the role of graphene.

Scheme 1
scheme 1

Synthesis route and structure of CHS@C and CHS@CG

Full size image

The microstructure of Co-MOFs is observed by TEM images. ZIF-12-XX g-PVP possesses a different morphology and structure with different amounts of PVP. ZIF-12 does not clearly change at 0.5 g of PVP (Fig. S3), the weight of PVP increases to 1 g, and the shape is changed and becomes sphere-like (Fig. S2). This phenomenon is not an accident; even if the amount of PVP is increased to 1.5 g, the shape is still sphere-like (Fig. S4), so we selected 1 g PVP conditions to perform more individual studies.

The yolk-shell-structured hollow carbon spheres (CHS@C and CHS@CG) are obtained after calcination at 400 °C in an air atmosphere. They have a well-retained sphere-like structure even after heating. ZIF-12-PVP (1.0 g) is exposed to air at 400 °C to obtain Co3O4 yolk-shell hollow nanospheres with multilayer structure CHS@C, and ZIF-12-PVP (1.0 g) with graphene is heated to form double-layer yolk-shell hollow nanospheres CHS@CG. The edge of CHS@CG (Fig. 1a–d) is less lucid than CHS@C (Fig. 1e–h) after 400 °C. The reason for the formation of the edge structure is that the graphene parcels ZIF-12-1.0 g-PVP and turns into carbon wrapped in the surface of the hollow spheres after pyrolysis. Therefore, there are layers of carbon around the edges (lattice spacing d = 0.33 nm). After pyrolysis of ZIFs in air, Co3O4 is mainly formed (lattice spacing d = 0.28 nm) [51], which is consistent with the expectation. The microstructures of ZIF-12 and ZIF-12-GO without PVP are shown in Fig. S5. Cobalt oxide exists as Co3O4 above 400 °C, and the polyhedron of Co3O4 nanoparticles grows on GO in ZIF-12-GO-400.

Figure 1
figure 1

TEM images of a–d CHS@C and e–h CHS@CG

Full size image

The phase purity and composition of the samples were checked with X-ray diffraction (XRD). Figure S6a-c shows the XRD pattern of ZIF series materials. The phase structures of ZIF-12, ZIF-12 with 0.5 g, 1.0 g, 1.5 g, 2.0 g PVP and ZIF-12-GO with 0.5 g, 1.0 g, 1.5 g, 2.0 g PVP have little discernible difference. Therefore, the ZIFs could be defined as the ZIF-12-XX series (JCPDS No.80–0382) [52]. By comparing Fig. S6b with S6a, it can be seen that there is a broad peak at approximately 20° for ZIF-12-GO, ZIF-12-0.5 g PVP-GO, ZIF-12-1.5 g PVP-GO, and ZIF-12-2.0 g PVP-GO, which indicates the existence of graphene. The diffraction peaks at 19.0°, 31.5°, 36.8°, 38.7°, 44.8°, 55.6°, 59.4°, and 65.3° corresponding to the (111), (220), (311), (222), (400), (422), (511), and (440) lattice planes are characteristic diffraction peaks of Co3O4 (Fig. 2a) (JCPDS No. 42–1467). This shows that the CHS@C and CHS@CG composites are obtained by annealing the ZIF-12 composites at 400 °C in air. The existing form of Co is Co3O4, which is completely consistent with the peak of pure Co(NO3)2·6H2O at 400 °C in air, so it could be indicated that the existing metal oxides are Co3O4. ZIF-12-400 and ZIF-12-GO-400 (without PVP) have the same effect (Fig. S6d).

Figure 2
figure 2

a XRD patterns of CHS@C and CHS@CG; b Raman spectra of CHS@C and CHS@CG; TG curves of c ZIF-12-PVP, ZIF-12-PVP-GO and d CHS@C, CHS@CG; N2 adsorption–desorption isotherms of e CHS@C and f CHS@CG at 77 K

Full size image

Figure 2b is the Raman spectra of CHS@C and CHS@CG. The diffraction peaks approximately 191, 466, 509, 600, and 670 cm−1 are revealed in CHS@C and CHS@CG. These are characteristic peaks of Co3O4, in accordance with the five Raman active modes (F2g, Eg, F2g, F2g, A1g) of the spinel Co3O4 phase [53]. It is further proven that the synthesized material is pure Co3O4 without any impurities. From the graph, we can see that the position of the A1g peak in Co3O4 (670 cm−1) is slightly lower than that of CHS@C and CHS@CG (680 cm−1) because of the trace N-doped carbon coating on the nanocomposites [19]. All of these results are consistent with the XRD results.

To obtain an appropriate calcination temperature from converting the precursor into Co3O4, TGA–DSC was carried out on the precursors (ZIF-12-PVP and ZIF-12-PVP-GO), as shown in Fig. 2c, in an air atmosphere. From the curves, we can see there is a slight mass change from 40 °C because of the sublimation of the residual moisture in the composites. A small mass loss appears after 200 °C, which may derive from the decomposition of a few ligands. From 400 °C, there is a significant mass loss due to the pyrolytic decomposition of ligands and the oxidation of Co(NO3)2 to Co3O4 [54]. From 600 °C, there is a plateau of the mass equilibrium where the main component is Co3O4. Therefore, 400 °C is selected as the temperature of pyrolysis in air [36]. Figure 2d shows the TG curves of CHS@C and CHS@CG. From the TG curves, it can be seen that 0.1% of the sample mass lost occurred between 100 and 200 °C. The loss of TG quality may be due to the moisture in the sample sublimation remaining. From 330 and 340 °C to 460 and 470 °C, the obvious mass loss is attributed to the decomposition of carbon shells and graphene in the composites. By calculation, the content of Co3O4 in CHS@C is 89.3 wt%, and in CHS@CG, it is 93.6 wt% [55].

The specific surface area and pore structure of CHS@C and CHS@CG were obtained from the N2 adsorption–desorption isotherms measured at 77 K (Fig. 2e, f). From the adsorption branch of the isotherm curve, the specific surface area was calculated using the BET method. The BET specific surface areas are 32 and 24 m2·g−1 for CHS@C and CHS@CG, respectively [56]. The SBET of CHS@CG is slightly smaller than that of CHS@C, which can be explained by the fact that the specific surface area of graphene shells in CHS@CG is lower than that of amorphous carbon in CHS@C after pyrolysis. This result is perfectly consistent with the electron microscopy. An appropriate specific surface area will have a positive effect on the performance in LIBs, because the moderate specific surface area in LIBs is conducive to the insertion and extraction of lithium ions [57].

To further determine the valences and the chemical composition of CHS@C and CHS@CG hollow nanospheres, X-ray photoelectron spectroscopy (XPS) measurements were taken (Fig. 3). As shown in Fig. 3e, f, the XPS full survey scan spectra confirm the presence of C, N, Co, and O atoms in the CHS@C and CHS@CG samples. The high-resolution spectra of C1s and O1s are shown in Fig. 3a–d. In the C1s spectra, the peaks at 284.4 eV, 285.1 eV, and 288.7 eV correspond to C − C, C − O, and C = O groups, respectively. In the O1s spectra, from the peaks of Co = O (529.7 eV), Co − O (530.0 eV), and O = C (532.7 eV), it can be concluded that carbon exists in the pyrolysis process of the ZIF series, and the existing carbon is inserted into the middle of Co3O4 to form the carbon-doped Co3O4 nanohollow sphere structure, which contributes to the conductivity of the composite nanohollow sphere and improves its electrochemical performance. From the N1s spectra of the CHS@C and CHS@CG in Fig.S7a and S7c, the characteristic peaks of N1s can be attributed to the pyrrolic-N and trace graphitic-N [46]. Figure S7b and S7d shows the Co2p spectra of CHS@C and CHS@CG, in which the peaks at 797.1 eV and 782.3 eV correspond to the presence of Co2p1/2 and Co2p3/2, respectively, proving that cobalt exists in the form of the Co3O4 state, which is consistent with reports in the literature [58, 59].

Figure 3
figure 3

XPS spectra of a, c C1s, b, d O1s in CHS@C and CHS@CG, and e, f full survey of CHS@C and CHS@CG

Full size image

Figure 4a, b presents cyclic voltammetry (CV) curves of CHS@C and CHS@CG in the voltage range of 0.01–3.0 V at a scanning rate of 0.1 mV·s−1 for the initial three cycles. As shown in the figure, the two potentials of 0.85 V and 2.08 V correspond to the cathodic peak and anodic peak of Li+ in the Co3O4-C material during insertion and extraction, respectively, namely, the reduction peak and oxidation peak. Figure 4a shows that the current intensity of the CHS@C composite decreases significantly in the first three cycles, and the first current is significantly stronger than the second and third cycles, while the current intensity of the CHS@CG (Fig. 4b) composite is more stable than that of CHS@C, and the difference between the first current intensity and the second and third cycles is not significant, suggesting good reversibility and stability [60]. This phenomenon shows that the carbon-wrapped form of CHS@CG has better stability than the carbon-free CHS@C, and the outer layer carbon shell provides protection for its volume change in the charging and discharging process [61].

Figure 4
figure 4

CV curves of a CHS@C and b CHS@CG at 0.1 mV·s−1 for the initial three cycles; charge/discharge voltage profiles of c CHS@C and d CHS@CG samples at the rate of 0.2 C with four cycles

Full size image

Figure 4c, d shows the galvanostatic charge–discharge curves of CHS@C and CHS@CG in the voltage range of 0.01–3.0 V versus Li/Li+ at the rate of 0.2 C. As seen from Fig. 4d, the CHS@CG discharge specific capacity of the first cycle (1110 mAh·g−1) is higher than the theoretical value and the second cycle (846 mAh·g−1). The initial capacity loss can be ascribed to the decomposition of the electrolyte and the formation of a solid electrolyte interface (SEI) membrane [62]. As shown in the figure, the voltage curve exhibits a good charge–discharge platform in 1–4 cycles and displays good cycling stability. In Fig. 4c, the CHS@C discharge specific capacity of the first cycle (1256 mAh·g−1) is higher than the theoretical value and the second cycle (835 mAh·g−1), and the voltage curve has a good charge–discharge platform. Due to the hollow spherical structure, the two samples have the desired reversibility, because the electrolyte can penetrate into the hollow cavity, making the insertion and extraction of lithium ions more rapid [63]. From the figure we can see that CHS@CG has a much better reversibility. Combined with the above analysis, the CHS@C and CHS@CG lithium storage mechanisms can be illustrated by the following equations:

$${\text{8Li}} \leftrightarrow {\text{8Li}}^{ + } + {\text{ 8e}}^{ - }$$
(1)
$${\text{Co}}_{{3}} {\text{O}}_{{4}} + {\text{ 8Li}}^{ + } + {\text{ 8e}}^{ - } \leftrightarrow {\text{3Co}} + {\text{ 4Li}}_{{2}} {\text{O}}$$
(2)
$${\text{Co}}_{{3}} {\text{O}}_{{4}} + {\text{ 8Li}} \leftrightarrow {\text{3Co}} + {\text{ 4Li}}_{{2}} {\text{O}}$$
(3)

Multiplier performance is one of the important parameters to evaluate the performance of lithium-ion batteries. Figure 5a,b shows the rate performance of the two electrode materials at various rates with 10 charging and discharging tests. As demonstrated in the graphs, the discharge specific capacity is attenuated as the current density increases. The specific capacity decays significantly when the current density rises to 1 C and 2 C. This can be attributed to two reasons: one is the insufficient diffusion coefficient of lithium ions, and the other is volume change in the process of lithium-ion insertion and extraction under a high current density [64]. As the rate returns to the initial 0.2 C, the specific capacity is recovered to the original specific capacities. This indicates that the hollow sphere structures can tolerate changes of such high rates, show robust rate performance, and demonstrate good reaction kinetics of Co3O4 hollow sphere structures [65]. Compared with Fig. 5a and 5b, CHS@CG has outstanding performance. The possible reason is that the double-layer yolk-shell hollow nanosphere CHS@CG enables the insertion and extraction of lithium ions to be faster, and the good thermal conductivity of graphene makes CHS@CG have better crystallinity than CHS@C, which improves the stability of CHS@CG.

Figure 5
figure 5

Cycling performance at various rates and corresponding Coulombic efficiency profiles of a CHS@C and b CHS@CG; cycling performance at the rate of 0.2 C and corresponding Coulombic efficiency profiles of c CHS@C and d CHS@CG; EIS spectra of CHS@CG and CHS@C e–g after 0, 100, and 500 cycles

Full size image

As seen from the constant current charge–discharge diagram at the rate of 0.2 C (Fig. 5c, d), the reversible specific capacity is almost unchanged as the number of cycles increases, and the battery still has a high charge–discharge capacity even after 500 cycles, showing the excellent cycle stability of lithium-ion batteries. After 400 cycles, the discharge specific capacities of CHS@C and CHS@CG are 645 mAh·g−1 and 831 mAh·g−1, respectively, which are 74% and 95% of the theoretical value (872 mAh·g−1). Relative to CHS@C composites, CHS@CG carbon shows better stability, because most of the outer layer of carbon from carbonated graphene provides a volume change buffer space in the process of lithium-ion transport, making its structure without being destroyed, and the carbon in the outermost layer can also constrain the volume change [66]. The Coulombic efficiencies of the two electrodes are stable and nearly 100%. The sample in this research shows decent lithium storage performance compared with the other Co-bases and Co3O4 anode materials (Table 1).

Table 1 Electrochemical performance comparison of the CHS@C and CHS@CG with the literature
Full size table

Figure 5e–g shows the electrochemical impedance spectra (EIS) of CHS@C and CHS@CG during various charge and discharge processes. The curve consists of a semicircle in the high-middle frequency range and a slant line in the low-frequency range. The slant line represents the diffusion degree of lithium ions. The larger the slope of the slant line is, the faster the diffusion of lithium ions and the better the electrochemical properties of the material. The semicircular arc represents the resistance of charge transfer in the middle- and high-frequency regions. The higher the activity of the electrode material is, the smaller the diameter of the semicircular arc. As shown in the picture, without the electrode reaction, the migration resistance of CHS@CG is only 43 Ω, and the transfer resistances are 90 Ω and 145 Ω after 100 cycles and 500 cycles, respectively. Correspondingly, the migration resistance of CHS@C is 56 Ω without the electrode reaction, and the transfer resistances are 195 Ω and 230 Ω after 100 cycles and 500 cycles, respectively [75], which are all larger than those of CHS@CG. The result shows that the addition of rGO nanosheets improves the electrical conductivity of the composites and greatly speeds up ion/electron transport in the process of lithiation/delithiation [76].

The Li+ insertion/extraction processes of CHS@CG are schematically illustrated in Scheme 2. During the first discharge, an irreversible chemical reaction takes place in electrode material, and Li2O, Co, and LixC nanodispersed metallic particles are formed in the spherical shell and between the spherical shell and the carbon layer. In subsequent cycles, the electrode material shows good redox reversibility, and the nanodispersed metallic particles (Co) can effectively make extra Li2O reversibly convert to Li+, which can maintain a high capacity during the charge–discharge cycles [77, 78]. Furthermore, the carbon layer outside limits the volume change of the hollow nanospheres and ensures good cycling stability.

Scheme 2
scheme 2

Li+ insertion/extraction mechanism for CHS@CG

Full size image

The as-prepared CHS-400 materials exhibit excellent Li-storage capacity, structure stability, and rate capability due to the good electrical conductivity and the synergistic effect of the interface among Co3O4, graphene, and pyrolytic carbon [50]. The synthesis method has the advantages of simple operation steps, strong repeatability, and versatility.

Conclusion

In conclusion, two carbon-wrapped Co3O4 hollow sphere structures were obtained from ZIF-12-PVP via molecular precursor pyrolysis in an air atmosphere. The main phase of the CHS@C and CHS@CG heat-treated at 400 °C was found Co3O4 and carbonized ZIF-12. Due to the large specific surface area, abundant active sites, shortened electron transport path, and sufficient cavities in the shell, the composite anode materials exhibit superior electrochemical performances. The reversible specific capacities reach up to 645 and 831 mAh·g−1 after 400 cycles at the rate of 0.2 C. In the constant charging and discharging process, the Coulombic efficiency reaches nearly 100%, exhibiting a good reversible cycle. The preparation method is simple, universal, and repeatable. This superior synthesis strategy will provide a new means of increasing the storage capacity of battery electrode materials and promising applications in the next-generation safer high-performance energy-storage systems.