Lithium-ion batteries (LIBs) possess several merits of high energy density, good safety, and excellent cycling stability, enabling broad application prospects in renewable energy storage and electric vehicles development. However, the currently commercialized graphite negative electrodes with limited theoretic capacity of 372 mAh/g inhibit the evolution of energy storage field [1-5]. Therefore, there is an ever-increasing demand to search for new negative electrode materials to substitute graphite.

Transition metal oxides (TMOs) [6-8] are considered as the promising alternative negative materials on account of their excellent theoretical capacity (~900-1000 mAh/g) and low cost. Fe₃O₄ based on conversion reactions possesses some distinctive merits, for instance, high capacity (924 mAh/g) and high abundance [9, 10]. Nonetheless, the large volume diversification given rise to the conversion of the oxide and the metallic states during cycling could bring about the fracture of electrode materials, thus leading to a fast decay of capacity and reduced cycle performance [11].

Graphene or reduced graphene oxide, with high conductivity and unique two-dimensional structure, has been widely used as a structure buffer medium for metal oxide-based negative materials [12-14]. By virtue of the morphology effect of Fe₃O₄, Kumar and coworkers showed that octahedral Fe₃O₄ nanocrystals on reduced graphene oxide nanosheets (rGO) could improve the electrochemical property (540 mAh/g at 100 mA/g after 120 cycles) in comparison to nanoparticles on rGO (270 mAh/g) [15]. The works mentioned above presented the high capacity of Fe₃O₄ negative electrode, but the volume change of Fe₃O₄ on lithiation/delithiation will result in its detachment from graphene and finally cause electrode pulverization to deteriorate the electrochemical performance. Very recently, Li et al. demonstrated N-doped carbon layer coated Fe₃O₄ nanoparticle networks displayed enhanced lithium storage performance at high current density (1125 mAh/g at 1 A/g) [16]. Therefore, the N-doped carbon layer is expected to alleviate the challenging pulverization of Fe₃O₄/graphene nanocomposites anodes.

Herein, we propose a utilization of N-carbon as a riveting layer to improve the structure stability and conductivity of Fe₃O₄/graphene (Fe₃O₄/G) nanosheets, thereby affording a noteworthy reversible capacity of 807 mAh/g at 100 mA/g, excellent rate performances with long cycling stability, i.e., 550 mAh/g at 1 A/g over 700 turns. The correlations between structure and property of N-carbon-riveted Fe₃O₄/G nanocomposites have been systematically studied, which proves that the carbon riveting layer has a crucial effect in enhancing lithium storage properties.

In this work, Hummers method was used to produce the graphene oxide (GO) [17, 18]. In a typical process, dissolve graphite flakes (3.0 g) and NaNO₃ (1.5 g) into H₂SO₄ and kept 0 ℃ in the ice bath. Then, KMnO₄ (15 g) was put into the solution slowly and maintain the reaction condition below 20 ℃ under stirring for 30 min. Next, the reaction was adjusted to about 95 ℃ for 15 min while water was added slowly, and the solution was stirred over 15 min to remove heat. After that, 3 mL of H₂O₂ (30 wt%) and 420 mL of deionized water were added. The final products were obtained after centrifugation, washing and freeze-drying.

Then, 0.075 g as-prepared GO was added into a beaker with 150 mL of ethylene glycol with ultrasound dispersion for 1 h, then dissolve 1.01 g Fe(NO₃)₃·9H₂O in the mixture which was put into Teflon-lined autoclaves with 100 mL capacity and maintained 150 ℃ for 2 h. The product of Fe₃O₄/reduced graphene oxide (Fe₃O₄/rGO) was prepared through centrifugation, washing three times using water, and freeze-drying. Then, Fe₃O₄/graphene (Fe₃O₄/G) was got after calcining at 700 ℃ for 1 h under Ar atmosphere.

In addition, 0.54 g of the Fe₃O₄/rGO was dissolved into a mixture of ethanol (40 mL) and pyrrole (0.586 mL). After ultrasonication for 20 min, 1 mol/L FeCl₃ (10 mL) was poured to the mixture and mixed for 12 h. After centrifugation, washing, freezedrying and calcining at 700 ℃ for 1 h in the Ar atmosphere, the final black powders were carbon/Fe₃O₄/grapheme (C/Fe₃O₄/G).

Several methods were chosen to characterize the physical and chemical properties of the as-synthesized materials. X-ray diffraction (XRD, Rigaku Smart lab) with Ni-filtered Cu-Kα radiation (λ =1.5406 Å) at a voltage of 40 kV and a current of 40 mA was utilized to characterize the phase of two products. The elemental information was collected on X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250, monochromatic Al-Kα radiation, 150 W). The morphology and size were examined by scanning electron microscopy (SEM, Zeiss sigma 300). The further detailed microstructure was recorded on transmission electron microscopy (TEM, JEM-2010) equipped with energy-dispersive X-ray spectroscopy (EDX) and high-resolution transmission electron microscopy (HRTEM, JEM-2010). The surface morphology and thickness were analyzed by atomic force microscope (AFM, Nanosurf NaioAFM). Raman was analyzed by Renishaw equipped with a 532 nm laser. Brunauer-Emmet-Teller (BET) testing via N₂ adsorption-desorption isotherms was estimated by Quantachrome Autosorb iQ3. Thermo-gravimetric analysis (TGA, NETZSCH, TG209) was carried out under airflow at a heating rate of 10 ℃/min.

In addition, the electrochemical performances were also investigated. The electrodes were fabricated by grinding the active material, carbon black and polyvinylidene fluoride (PVDF) (7:2:1 wt%) in a mortar to make a homogeneous paste. The paste was uniform smear on Cu foil with a mass of approximately 1.2–1.4 mg/cm². 1 mol/L solution of LiPF₆ in dimethyl carbonate, ethylene methyl carbonate, and ethylene carbonate (1:1:1 in volume) was chosen to be electrolyte. The battery was fabricated in Ar-filled glovebox (Mikrouna, China) using CR2025 cells with Li metal as reference electrode. Electrochemical impedance spectra (EIS) with frequency ranging from 0.01 kHz to 100 kHz at an applied voltage of 5 mV and cyclic voltammetry (CV) tests with sweep rate of 0.1 mV/s were evaluated on VSP multi-channel electrochemical workstation from 0.01 V to 3 V. Galvanostatic discharge/charge cycling was performed using Land CT2001A tester within the potential range of 0.01 V to 3 V.

The procedure of preparing C/Fe₃O₄/G and Fe₃O₄/G is shown in Scheme 1. The GO is used as the starting material, and Fe₃O₄ nanoparticles are grown on GO by a hydrothermal process to give rise the Fe₃O₄/rGO. Post annealing is performed to transform Fe₃O₄/rGO into Fe₃O₄/G. For the preparation of C/Fe₃O₄/G, polypyrrole (Ppy) is coated on Fe₃O₄/rGO to serves as the precursor for N-carbon, which is obtained by thermal annealing in Ar atmosphere. Compared with Fe₃O₄/G, the C/Fe₃O₄/G is expected to manifest improved electrochemical performance due to the carbon-coated nanoarchitecture, which can produce highly electrical conductive pathways and cushion the volume expansion of Fe₃O₄ effectively in the repetitive Li-ion insertion/de-insertion procedure.

The phase structures of C/Fe₃O₄/G and Fe₃O₄/G is examined by XRD (Fig. 1a), which show similar diffraction patterns. The diffraction peak at 26.3° denotes a minor phase originated from Fe₃C, which is due to carbothermal reduction of Fe₃O₄ to α-Fe and subsequently dissolving of the surrounding carbon atoms into the Fe lattice [19]. Other characteristic peaks are consistent with magnetite Fe₃O₄ (JCPDS No. 85–1436). The elemental composition is confirmed via XPS measurement. As depicted in Fig. 1b, the peaks of 711 and 725 eV correspond to Fe 2p_3/2 and Fe 2p_1/2 [20]. Furthermore, peaks centered at 716 and 730 eV are their satellites. Fig. 1c shows that the C 1s consists of four parts, C-Fe (283.8 eV), C=C (284.8 eV), C=N (285.6 eV) and C-N (287.6 eV) [21-23]. In Fig. 1d, the N 1s spectrum is fitted into four components of N species, i.e., pyridinic-N oxides, pyrrolic-N, graphitic-N and pyridinic-N at 405.7, 402.9, 400.5 and 398.3 eV [24], respectively. The ratio of the corresponding N species is estimated to be 1.36%, 3.31%, 58.09% and 37.24%. The high content of graphitic-N indicates the indeed doped of nitrogen in the C/Fe₃O₄/G composites [22]. The high ratio of pyridinic-N has been reported to improve the lithium storage performances [25].

The morphology of as-prepared composites is investigated by SEM (Fig. S1 in Supporting information). Overall, both composites exhibit similar large size lamellar structure. From Figs. S1a and b, the Fe₃O₄ nanoparticles are uniformly loaded on the graphene sheet. However, no obvious Fe₃O₄ nanoparticles are observed in C/Fe₃O₄/G (Figs. S1c and d), which may be owing to the coverage of the carbon layers.

More details of morphological information are investigated by the TEM and HRTEM. From Fig. 2a, the homogeneously dispersed Fe₃O₄ nanoparticles on graphene display an average size of 30 nm, although some very big particle is also observed. However, for C/Fe₃O₄/G (Fig. 2b), the Fe₃O₄ nanoparticles are not obvious due to the coverage of carbon layers compared with Fe₃O₄/G. From the HRTEM image of C/Fe₃O₄/G in Fig. 2c, the (111) plane of Fe₃O₄ is clearly noticed with a lattice spacing of 0.48 nm, meanwhile, a carbon layer can be observed around the particle, which indicates the Fe₃O₄ nanoparticles are successfully loaded on the graphene and well enveloped by a thin carbon layer.

Typical STEM and EDX of C/Fe₃O₄/G composites are shown in Fig. 2d. The STEM in Fig. 2d reveals the particle size of 20-50 nm, and elemental mapping images demonstrate the co-existence and uniform distribution of Fe, O, N and C elements. Furthermore, the carbon coating layer in C/Fe₃O₄/G has been proved by the AFM test. As shown in Fig. 2e, the thickness of GO is around 3.7 nm, corresponding to 3–4 layers of graphene sheets. The average thickness of Fe₃O₄/G is measured to be 51 nm (Fig. 2f), which is smaller than that (57 nm) of C/Fe₃O₄/G (Fig. 2g). Fig. 2h is the thickness comparison chart of GO, Fe₃O₄/G and C/Fe₃O₄/G, which intuitively shows the thickness change. The increased thickness of C/Fe₃O₄/G compared to Fe₃O₄/G is due to an extra layer of carbon.

Both as-prepared composites are further analyzed by the Raman spectroscopy (Fig. S2a in Supporting information). The D band at ~1350 cm^-1 corresponds to the disordered structure or defects in carbon materials [26], and the G band of graphitic carbon is centered at 1587 cm^-1 [27, 28]. The calculated I_D/I_G for C/Fe₃O₄/G and Fe₃O₄/G are both less than 1, indicating the existence of a great amount of sp²-type carbon in the materials, which will improve the electronic conductivity of the materials [29]. Compared with Fe₃O₄/G, the value of I_D/I_G of C/Fe₃O₄/G is higher, demonstrating that the introduction of N-carbon increases the defects which may be due to the incorporation of N atoms to eliminate the side functional groups in the GO and causes the edges disordered [30, 31].

TGA has been conducted to study the thermal stability and to determine the mass content of Fe₃O₄ in two as-synthesized nanocomposites. The TGA was performed in air, therefore we assume the final product is Fe₂O₃ from the oxidation of Fe₃O₄. As shown in Fig. S2b (Supporting information), the final mass content of Fe₂O₃ is 85.5% and 39.7% in Fe₃O₄/G and C/Fe₃O₄/G, respectively. Based on the oxidation reaction Fe₃O₄ + O₂=6Fe₂O₃, the weight percentage of Fe₃O₄ in the two samples is calculated to be around 82.8% and 38.4%. It is obvious that the C/Fe₃O₄/G exhibits much higher weight loss in comparison to Fe₃O₄/G because of the combustion of carbon in C/Fe₃O₄/G. Taking the results above in consideration, the carbon layer has been successfully coated on the Fe₃O₄/G nanosheets. It is worth inspecting whether the N-carbon layers can contribute to an improved lithium storage performance.

The specific surface area and pore structures of C/Fe₃O₄/G and Fe₃O₄/G have been tested by N₂ adsorption-desorption. The isotherms shown in Figs. S2c and d (Supporting information) reveal both samples present type-Ⅳ isotherm with an H1-type hysteresis loop, demonstrating a mesoporous structure. Mean pore diameter of C/Fe₃O₄/G and Fe₃O₄/G are about 3.935 and 3.937 nm by means of Barrett-Joyner-Halenda (BJH) tests, respectively. The BET values of C/Fe₃O₄/G is 125.571 m²/g, which has significantly increased compared with Fe₃O₄/G (47.153 m²/g). C/Fe₃O₄/G has higher surface area because of the coated carbon layer.

The electrochemical properties of C/Fe₃O₄/G and Fe₃O₄/G composites are first evaluated by the CV test. As illustrated in Fig. S3a (Supporting information), during the first cathodic cycle, the significant peak at 0.55 V is assigned to the formation of solid electrolyte interface (SEI) layer and electrolyte decomposition and the reduction reaction of Fe³⁺ and Fe²⁺ to Fe⁰ (Fe₃O₄ + 2Li⁺ + 2e^-→ Li²Fe₃O₄ and Li₂Fe₃O₄ + 6Li⁺ + 6e^-→3Fe⁰ + 4Li₂O) [32] which moves to about 0.8 V in the subsequent turns. The broad anodic peak at 1.6 V could be attributed to the oxidation of Fe⁰ to Fe³⁺ and Fe²⁺ (3Fe⁰ + 4Li₂O→Fe₃O₄ + 8Li⁺ + 8e^-), which moves to 1.7 V in the next curves [9]. Furthermore, the subsequent curves show a high degree of coincidence, indicating that the material has good electrochemical reversibility. For C/Fe₃O₄/G, the cathodic and anodic peaks intensity decreases a little after the 1^st cycle, showing the occurrence of an irreversible redox reaction. On the contrary, the peak intensity and integrated area decline significantly for Fe₃O₄/G in Fig. S3c (Supporting information), which demonstrates a great deal of electrochemical irreversible reaction taking place in the process of lithium storage.

Figs. S3b and d (Supporting information) display galvanostatic charge-discharge curves of C/Fe₃O₄/G and Fe₃O₄/G electrodes at 100 mA/g. The C/Fe₃O₄/G and Fe₃O₄/G exhibits starting discharge/ charge capacities of 1439/899, 1112/702 (mAh/g), accounting for a Coulombic Efficiency (CE) of 62.5% and 63.1%, respectively. The low initial CE is attributed to some irreversible reactions including the formation of the SEI film [33, 34]. In addition, C/Fe₃O₄/G has the lower initial CE than that of Fe₃O₄/G which is ascribed to the higher surface area and abundant content of carbon [21, 24]. Furthermore, the capacity of Fe₃O₄/G and C/Fe₃O₄/G only has been significantly reduced in the first two cycles and then becomes quite stable.

Fig. 3a compares the cycle performances of C/Fe₃O₄/G and Fe₃O₄/G electrodes at 100 mA/g. For Fe₃O₄/G, it delivers a high first discharge capacity of 1112 mAh/g and reduces to 540 mAh/g while the C/Fe₃O₄/G exhibits starting discharge capacity 1439 mAh/g and remains 807 mAh/g after 50 cycles, indicating its enhanced cycling stability. Based on the Eq. (1):

(1)

It can be calculated that the theoretical capacities of C/Fe₃O₄/G and Fe₃O₄/G are 584 and 828 mAh/g, respectively. The capacity of C/Fe₃O₄/G is higher than the theoretical capacity may due to a mixed Li-ion storage mechanism including diffusion and capacitive contribution [35, 36] and the presence of Fe₃C. Fe₃C can serve as a catalyst for the reversible formation/decomposition of some components in solid electrolyte interface (SEI) films in electrode and will bring extra capacity [23, 37, 38]. The capacity contribution (Cc) are calculated using Eq. (2) [39]:

(2)

where C(x) is the practical theoretical capacity of C(Fe₃O₄), C(carbon) and C(graphene). W(x) is the mass percent of Fe₃O₄, carbon and graphene. C(C/Fe₃O₄/G) is the stable capacity of C/Fe₃O₄/G at the 50^th cycles. According to the result of TGA, the proportion of carbon to C/Fe₃O₄/G is 53.62% and it can be concluded that the capacity contribution of carbon is 24.7%.

Fig. 3b displays the rate capabilities of the two electrodes at various current densities. C/Fe₃O₄/G can deliver stable average capacities of 950, 803, 681, 550 and 442 mAh/g at 100, 200, 500, 1000 and 2000 mA/g, respectively, while Fe₃O₄/G only exhibits 629, 540, 450, 380 and 304 mAh/g. As the current density returned to 100 mA/g, the reversible capacity of the C/Fe₃O₄/G reaches 835 mAh/g, demonstrating the superior reversibility. To further explore the C/Fe₃O₄/G cycling performance, the battery that was used for rate test was further tested at high current density of 1 A/g. A stable capacity of around 550 mAh/g is maintained steady after 700 cycles (Fig. 3c).

To further probe the electrode reaction kinetics of Fe₃O₄/G and C/Fe₃O₄/G, EIS measurements are carried out after 50 cycles. The Nyquist plots and fitting results are displayed in Fig. 3d, Fe₃O₄/G and C/Fe₃O₄/G display sloping line at low-frequency part and incomplete semicircle in the high-frequency part which represent the Warburg type resistance [40, 41] and Li⁺ diffusion procedure, respectively [42]. The corresponding semicircle diameter of the Fe₃O₄/G electrode is larger compared to C/Fe₃O₄/G. The resistance of SEI (R_f) and charge-transfer (R_c) are calculated to be 6.3 and 34.9 Ω for C/Fe₃O₄/G while Fe₃O₄/G corresponds to 29.9 and 135.9 Ω, respectively, indicating the lower charge transfer resistance related to the additional conductive carbon layer [21, 43].

Based on Eq. (3) [44], the Warburg factor (σ) is the coefficient in the linear relationship between Z_re and ω^-1/2, Zre is the real part of the impedance, ω is the angular frequency.

(3)

According to Fig. 3e, the s of C/Fe₃O₄/G and Fe₃O₄/G are 24.92 and 72.10, respectively.

The Lithium-ion diffusion coefficient (D_Li⁺) are expressed using Eq. (4) [21, 45]:

(4)

where R is the gas constant, T is the absolute temperature, A is the contact area between the electrode and the electrolyte, n is the number of electrons transferred in the chemical reaction, F is the Faraday constant and C is the concentration of lithium ions in the electrolyte. The D_Li⁺ of the C/Fe₃O₄/G (0.22 × 10^-10 cm²/s) is higher than that of Fe₃O₄/G (0.26 × 10^-11 cm²/s). It indicates that the carbon increases the Li⁺ conduction and shortens the Li⁺ diffusion length [46, 47].

The above results show that enhanced cycle stability of C/Fe₃O₄/G is not only because of its high capacity but also because of its excellent conductivity of graphene and N-carbon. Therefore, the improved performances of C/Fe₃O₄/G compared to Fe₃O₄/G are reasonably originate from the function of N-carbon, which has good conductivity and can confine the location of Fe₃O₄ to prevent its agglomeration and pulverization, as well as to improve the lithium transportation kinetics.

To sum up, C/Fe₃O₄/G composites were synthesized by a simple hydrothermal and post-annealing procedure. Fe₃O₄/G interacts with N-carbon which makes C/Fe₃O₄/G possess good structural and phys-chemical advantages. The graphene can supports that Fe₃O₄ nanoparticles uniformly anchored on its surface which cooperates with carbon could effectively improveelectrical conductivity, meanwhile, relieve stress changes caused by volume expansion. Furthermore, the N-carbon layers serve a riveting layer tightly confined Fe₃O₄ nanoparticles to the surface of graphene, so that a much higher resistivity to material pulverization and electrode crack is obtained. Consequently, the C/Fe₃O₄/G delivers 807 mAh/g after 50 cycles at 0.1 A/g and even at 1 A/g, a stable capacity of 550 mAh/g after 700 cycles can still maintain. This study provides a general and reliable strategy for preparing other composites with high-performance for LIBs.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work is financially supported by the National Natural Science Foundation of China (Nos. 51602167, 51972182 and 61971252), Shandong Provincial Science Foundation (No. ZR2017JL021) and Key Research and Development Program (No. 2018GGX102033), and Qingdao Applied Fundamental Research Project (No. 17-1-1-81-jch) and "Distinguished Taishan Scholar" Project.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi: https://doi.org/10.1016/j.cclet.2020.02.006.