Introduction

In recent years, lithium-ion batteries (LIBs) with a high voltage, high specific capacity, environmental friendliness, and low cost have been paid great attention to use as the best power source for an extensive range of portable electronic devices, power tools, and electric vehicles (Bogart et al. 2014; Ma et al. 2014; Zhang et al. 2014). The performance of LIBs greatly depends on the properties of electrodes. Graphite is currently served as the standard anode material for LIBs. Nevertheless, the graphite anodes are handicapped due to their relatively low theoretical capacity (372 mAh g−1) in LIBs (Zhang et al. 2012). Therefore, extensive efforts have been made to find alternative materials with higher capacities, including tin, silicon, and transitional metal oxide-based materials to replace graphite (Reddy et al. 2013; Qin et al. 2014; Yan et al. 2015; Xiao et al. 2014; Xie et al. 2013). However, these materials (e.g., Si, Sn, and metal oxides) show a large volume change during charge/discharge process and poor electrical conductivity, which lead to unsatisfactory cycling stability and rate performance, and seriously limit their applications.

Many approaches have been developed to solve the above problems. Anode materials with one-dimensional (1D) nanostructures have been extensively studied as they have great potential for enhancing electrical conductivity and relieving the volumetric change during cycling (Guo et al. 2008; Jiang et al. 2011). Particularly, 1D carbon nanofibers (CNFs) have demonstrated their superiority as an effective carbon matrix to load active materials to achieve high electrochemical performance for LIB application due to their super electronic conductivity, undamaged morphology during cycling, and independent characteristic (Agend et al. 2007; Bonino et al. 2011). In addition, CNFs can accommodate the dramatic volumetric expansion of active materials during lithiation/delithiation, and thus improve the capacity retention of active materials by preventing the nanoparticle pulverization (Su et al. 2014) and the loss of electric contact between active substances and the current collector. On the other hand, iron oxide (Fe3O4)-based nanostructured materials are attracting great attention as high-performance anode materials for LIBs due to their high theoretical capacity (924 mAh g−1), low toxicity, environmental friendliness, and the natural abundance of iron (Reddy et al. 2012; Wang et al. 2011; Wu et al. 2014; Lei et al. 2014; Liu et al. 2008; Choi et al. 2014). Recently, Fe3O4-based hybrid nanostructures, such as Fe3O4–graphene (Wang et al. 2011), Fe3O4–C composite (Wang et al. 2013), foam-like Fe3O4–carbon architectures (Wu et al. 2014), Fe3O4–carbon hollow particles (Lei et al. 2014), and Fe3O4–carbon core–shell nanorods (Liu et al. 2008) have been investigated as the potential architectures to improve the electrochemical performance in LIBs because the Fe3O4–carbon integrated materials could sustain their structural integrity during cycling (Xiong et al. 2012; Ma et al. 2012; Li et al. 2011).

In this work, we develop a simple solvothermal method followed by thermal annealing to fabricate 1D hierarchical Fe3O4/CNFs nanocomposites, which integrate the high-capacity Fe3O4 nanoparticles with highly conductive CNFs. Specifically, the CNFs serve as a supporting matrix for dispersing of Fe3O4 nanoparticles, resulting in large surface area and excellent electrical conductivity. The electrolyte can easily penetrate the porous CNFs network, leading to the increase of the contact area between electrolyte and active materials. Individual CNFs make direct contact with the current collectors, and provide a fast and efficient electron transfer pathway. As a result, the as-prepared Fe3O4/CNFs nanostructures show an improved electrochemical performance as anode for LIB application.

Experimental

Synthesis

The CNFs were synthesized through a modified hydrothermal method using Te nanowire as the template according to the previous literature (Liang et al. 2012). Fe3O4/CNFs nanocomposites were prepared by a simple solvothermal process with subsequent annealing treatment. First, 10 mg of CNFs and 15 ml of FeCl3 (0.2 M) were dispersed in 120 ml of anhydrous ethanol by ultrasonic stirring. Then 4 ml of NH3·H2O was dropped into the above solution to produce uniform dispersion. The mixture was kept at 80 °C in a water bath with stirring for 24 h, and was subsequently transferred to a Teflon-lined autoclave and placed in an oven at 160 °C for 3 h. The resultant brown precursors were centrifuged, washed with distilled water and anhydrous ethanol, and were dried overnight at 60 °C. Finally, the precursors were calcined at 500 °C in Ar atmosphere for 3 h to obtain the Fe3O4/CNFs nanocomposites.

Characterization

The as-prepared samples were characterized using X-ray powder diffraction on a Philips PW3040/60 X-ray diffractometer with Cu Ka (λ = 1.54056 Å) radiation for phase identification. The content of CNFs in the Fe3O4/CNFs nanocomposites was ascertained by the thermogravimetric analysis measurement (TGA, Netzsch STA449C thermal analyzer). The morphologies were examined by scanning electron microscopy (SEM, Hitachi S-4800), and the microstructures were investigated by transmission electron microscopy (TEM, JEM 2010F).

Electrochemical test

The electrochemical performances of the Fe3O4/CNFs nanocomposites were investigated with two-electrode coin-type cells (CR2025). The working electrodes were prepared by dispersing the as-prepared Fe3O4/CNFs nanocomposites (75 wt%), and polyvinylidene fluorides (10 wt%) as binder, acetylene carbon black (15 wt%) as conducting additive in N-methyl-2-pyrrolidone solvent to form a slurry. The slurry was incorporated on nickel foam current collector and dried at 80 °C for 6 h in a vacuum oven. The loading density of Fe3O4/CNFs nanocomposites was about 1.8–2 mg cm−2. The test cells were assembled in an argon-filled glove box with the metallic lithium foil as both the counter and reference electrodes. The electrolyte solution was 1 M LiPF6 dissolved in a mixture of ethylene carbonate and dimethyl carbonate with a volume ratio of 1:1. The cyclic voltammetry (CV) was carried out on a CHI660C electrochemistry workstation in the potential window of 0.02–3.0 V at a scan rate of 0.1 mV s−1. The galvanostatic charge and discharge measurements were tested at a current density of 100 mA g−1 and the rate performance was conducted at different current densities (100, 200, 500, 1000, and 2000 mA g−1) on a battery test system (Neware Co. Ltd., Shenzhen) in the voltage range from 0.005 to 3 V. The specific capacity was calculated based on the total mass of Fe3O4/CNFs nanocomposites in coin-type cells. AC impedance measurements were carried out using a CHI604D electrochemical workstation. The amplitude was 50 mV and the frequency ranged from 100 kHz to 10 mHz. The impedance data were fitted using the ZsimpWin computer program.

Results and discussion

The XRD pattern of the as-prepared Fe3O4/CNFs nanocomposites is shown in Fig. 1. All the diffraction peaks can be indexed to the face-centered cubic Fe3O4 with the lattice constant of a = 8.394 Å (JCPDS No. 89-3854). No other impurity phase is observed in the XRD pattern, and the sharp diffraction peaks reflect the excellent crystallinity of Fe3O4, confirming the high-purity phase of the products.

Fig. 1
figure 1

XRD pattern of the Fe3O4/CNFs nanocomposites

Full size image

To estimate the amount of CNFs and Fe3O4 components in the nanocomposites, TGA was carried out in air with a heating rate of 10 °C/min from room temperature to 600 °C. TGA curve of the Fe3O4/CNFs nanocomposite was presented in Fig. 2. The slight weight loss between room temperature and 150 °C is about 1.0 %, corresponding to the evaporation of absorbed moisture in the sample, whereas the weight gain of about 1.3 % for Fe3O4/CNFs nanocomposites between 150 and 285 °C is likely the result of oxidation of the Fe3O4 in air (Zhu et al. 2011b). The sharp weight losses observed from 285 to 450 °C are mainly attributed to the oxidation and decomposition of CNFs. Calculated from the weight losses of CNFs and the mass gains of Fe3O4 due to oxidation, the amount of CNFs component in the nanocomposites is about 12.8 wt%.

Fig. 2
figure 2

TGA curve of the Fe3O4/CNFs nanocomposites

Full size image

The morphology and microstructure of Fe3O4/CNFs nanocomposites were investigated using SEM and TEM. Figure 3a shows a SEM image of the pristine CNFs with an average diameter of 300 nm; the lengths of CNFs range from several to tens of micrometers. The entangled CNFs interconnect with each other to a high degree through numerous junctions, which is advantageous for the electrochemical lithium storage performances due to the formation of 3D conductive network. Figure 3b, c shows SEM images of Fe3O4/CNFs nanocomposites. Fe3O4 nanoparticles are homogenously dispersed on the surface of CNFs. The fine dispersion and tight attachment of the Fe3O4 nanoparticles (80–100 nm) on individual CNFs is further revealed by TEM as shown in Fig. 3d, e. Figure 3f presents a HRTEM image of a Fe3O4 nanoparticle on CNFs (the inset is the fast Fourier transform pattern). The lattice fringes with spacing of 0.30, 0.21, and 0.17 nm can be assigned to (220), (004), and (224) planes of cubic Fe3O4, respectively, which is in agreement with the XRD result. Figure 3g, j shows a TEM image of Fe3O4/CNFs and the mapping of C, Fe, and O elements within this Fe3O4/CNFs nanocomposite, which further confirms the architecture of Fe3O4 nanoparticles dispersed on CNFs surface.

Fig. 3
figure 3

a SEM image of the pristine CNFs. b, c SEM images of Fe3O4/CNFs nanocomposites. d, e TEM images of Fe3O4/CNFs nanocomposites. f HRTEM image of Fe3O4/CNFs nanocomposites and the corresponding FFT pattern (inset). g–j TEM image of Fe3O4/CNFs and corresponding mapping of carbon, iron, and oxygen elements

Full size image

The electrochemical behavior and performance of the Fe3O4/CNFs nanocomposites were evaluated by CV and galvanostatic charge–discharge using two-electrode CR2025 coin-type cells. Figure 4a shows the CV curves of Fe3O4/CNFs nanocomposites electrode. A reduction peak at about 0.6 V is observed in the first cathodic scan for the Fe3O4/CNFs nanocomposites, which can be attributed to the reduction of Fe3+ and Fe2+ to Fe0 and the formation of solid electrolyte interface (SEI) layer (Lei et al. 2014). Meanwhile, two anodic peaks at about 1.67 and 1.90 V correspond to the reversible oxidation of Fe0 to Fe2+/Fe3+, which agrees well with early studies (Wu et al. 2014; Lian et al. 2010). During the subsequent cycles, both cathodic and anodic peaks are positively shifted, which is attributed to the structural change of the electrode materials induced by the first lithiation process due to the irreversibility. The CV curves are found to be coincided after the first cycle, which indicates the good electrochemical reversibility and stability of Fe3O4/CNFs nanocomposites.

Fig. 4
figure 4

Electrochemical measurement of Fe3O4/CNFs nanocomposite electrodes. a CV curves at a scan rate of 0.1 mV s−1. b Voltage profiles for the selected galvanostatic cycles at a current density of 100 mA g−1. c Reversible charge/discharge capacities against cycle number at a current density of 100 mA g−1. d Rate performance of the electrode at the current of 100, 200, 500, 1000, 2000, and 100 mA g−1

Full size image

Figure 4b shows the galvanostatic discharge–charge curves of the Fe3O4/CNFs nanocomposite electrodes between 0.005 and 3 V at a current density of 100 mA g−1. In the first discharge step, a long voltage plateau can be observed at about 0.75 V, followed by a sloping curve down to the cut voltage of 5 mV, which are the typical characteristics of voltage profile for Fe3O4 (Wu et al. 2014; He et al. 2010). From the second cycle, this discharge plateau becomes much shorter and shifts to higher voltage, indicating the different lithiation behaviors between the first and the subsequent cycles; there are irreversible reactions during the first cycle, such as the formation of SEI layer and irreversible phase conversion (Su et al. 2013). Meanwhile, the main plateau from 1.6 to 1.9 V during charging can be attributed to the oxidation of Fe0 to Fe2+ and Fe3+ (He et al. 2010; Xia et al. 2013). These results agree well with the CV analysis.

Figure 4c shows the cycling performance of Fe3O4/CNFs electrodes at the constant current density of 100 mA g−1. The initial discharge and charge capacity of Fe3O4/CNFs nanocomposites is 1694 and 1063 mAh g−1, which is larger than the theoretical capacity of Fe3O4 (926 mAh g−1). The extra capacity of the electrodes compared with the theoretic capacity results from the formation of solid electrolyte interface SEI film and possibly interfacial Li+ ion storage during the discharge process (Chou et al. 2010; Débart et al. 2001). The Coulomb efficiency in the first cycle is about 63 %. The initial capacity loss of 37 % mainly results from the diverse irreversible process, like electrolyte decomposition and inevitable formation of SEI layer, and irreversible phase conversion (Su et al. 2013), which are similar to most metal oxide anode materials (Xu et al. 2012; Wu et al. 2012; Zhu et al. 2011a; Zhang et al. 2008). The Coulomb efficiency reaches 93 % in the second cycle and remains relatively stable at 97 % in the subsequent cycles. The reversible discharge capacity of Fe3O4/CNFs electrode remains 684 mAh g−1 after 55 cycles at 100 mA g−1.

The rate performance of Fe3O4/CNFs nanocomposite has been evaluated. As we can see in Fig. 4d, the specific capacity of the Fe3O4/CNFs nanocomposites is 930, 700, 511, 326, and 211 mAh g−1 after each ten cycles for the current density of 100, 200, 500, 1000, and 2000 mA g−1, respectively. When the current density is reduced back to 100 mA g−1, the Fe3O4/CNFs electrode can recover a capacity of 757 mAh g−1. The results indicate that the Fe3O4/CNFs nanocomposites can tolerate high current cycling and have good retention of capacity. Although the performance of Fe3O4/CNFs nanocomposites is not as good as the Fe3O4–CNT (Wu et al. 2013), hierarchical Fe3O4 microsphere/graphene nanosheet composite (Wang et al. 2014), and 600 °C carbonized C/Fe3O4 nanofibers (Wang et al. 2008), but superior to Fe3O4/carbon core–shell nanorods (Li et al. 2011), Fe3O4 nanocubes (Cao et al. 2011), Fe3O4/Fe/carbon composite (Zhao et al. 2012), 500 °C carbonized C/Fe3O4 nanofibers (Wang et al. 2008), and Fe3O4 nanotube (Xia et al. 2013). The excellent electrochemical behaviors can be attributed to several aspects: (1) The CNFs after annealing in Ar at high temperature have enhanced conductivity. (2) The interconnected 1D Fe3O4/CNFs nanocomposites can provide 3D continuous conducting pathways for Li+ and e− transports within the electrode. (3) During the charge/discharge process, the space among CNFs can accommodate the serious volume change of Fe3O4 nanoparticles and thus enhance the structural stability of electrodes.

Electrochemical impedance spectroscopy (EIS) of Fe3O4/CNFs nanocomposites was investigated to understand the kinetics that influences the electrochemical reaction toward lithium. The EIS measurements were performed before the first cycling and after the 55th cycling test. As shown in Fig. 5a, the Nyquist plots consist of a semicircle in the high- and middle-frequency regions, and a straight line in the low-frequency region. The classic equivalent circuit as shown in Fig. 5b is adopted to model the electrochemical system, where R e represents the resistance of the electrolyte, and R sl and C sl are designated for the resistance of the migration and capacity of the layer in the high-frequency semicircle. R ct and C dl are associated with the charge transfer resistance and a double-layer capacitance in the medium-frequency semicircle, and Z W represents the diffusion-controlled Warburg impedance. The main parameters calculated according to the equivalent circuit are presented in Table 1. The charge transfer resistance (R ct, 0.439 Ω) after the 55th cycles is significantly smaller than the R ct (291 Ω) before cycling. The reduced R ct indicates more sufficient contact between active materials and electrolyte with cycling (Wu et al. 2014). The relationship between real impedance Z r and the reciprocal root square of the lower angular frequency ω−0.5 for the nanocomposites before the first cycling and after the 55th cycling test is displayed in Fig. 5c. Linear behaviors with different slopes are observed in the two curves. We can achieve the Warburg factor σ and calculate the diffusion coefficient D and the exchange current density i o according to Eqs. (1) and (2) (Shenouda and Liu 2010; Xiang et al. 2011a), and the values are listed in Table 1.

Fig. 5
figure 5

a Nyquist plots of the Fe3O4/CNFs electrode obtained by applying a sine wave with amplitude of 5.0 mV over the frequency range from 100 kHz to 0.01 Hz. b Equivalent circuit for Fe3O4/CNFs electrode/electrolyte interface. c Relationship between Z r and ω −0.5 in the frequency region of 0.1–0.01 Hz

Full size image
Table 1 Electrochemical impedance parameters of Fe3O4/CNFs nanocomposites before cycling and after the 55th cycle
Full size table
$$D = \, 0. 5 { }(RT/AF^{ 2} \sigma C)^{ 2}$$
(1)
$$i^{\text{o}} = RT/nFR_{ct}$$
(2)

where n is the number of electrons per molecule during reaction; A is the active surface area of the electrode; R is the gas constant; T is the absolute temperature; F is the Faraday constant; and C is the molar concentration of Li+ ions. From Fig. 5c, we can see that the Warburg impedance coefficient (σ) is 375.18 Ω s−0.5 after 55 cycles, and it has a lower value than that (610.33 Ω s−0.5) before cycling. The obtained diffusion coefficient (D = 1.05 × 10−12 cm2 s−1) after cycling explains the higher mobility of Li+ ions than that before cycling (D = 3.96 × 10−13 cm2 s−1). Moreover, the exchange current density (i o) is 7.31 × 10−3 mA cm−2 after cycling, which is higher than 1.10 × 10−5 mA cm−2 before cycling. It is generally believed that the process of charge transfer occurs mainly on the active materials/electrolyte interfaces, so the surface morphology and microstructure of the electrode materials are the most important factors affecting the electrochemical impedance during the discharge/charge process (Xiang et al. 2011b). The conversion reaction occurred during the first lithiation/delithiation process and the electrode volume expansion/shrinkage make Fe3O4 particles convert to ultrafine nanograins (Su et al. 2013), which can increase the surface area of the active materials, shorten Li+ diffusion route, and facilitate charge transfer on active materials/electrolyte interfaces. In other words, the structural modification in the Fe3O4/CNFs electrode owing to the lithiation/delithiation processes leads to the decrease of charge transfer resistance and significant enhancement of the diffusion coefficient and exchange current density. So the charge transfer reaction after cycling is much stronger than that before cycling. The higher diffusion coefficient and exchange current density and lower charge transfer resistance after the 55th cycle account for the stable cycling and excellent rate performances of the Fe3O4/CNFs nanocomposites.

Conclusion

In summary, we have successfully synthesized Fe3O4/CNFs nanocomposites by a facile two-step method involving the solvothermal synthesis and subsequent thermal annealing. The CNFs are found to be attached by many uniform Fe3O4 nanoparticles around 80–100 nm. The as-prepared Fe3O4/CNFs nanocomposites demonstrate extraordinary electrochemical performance, such as high capacity, long cycling stability (684 mAh g−1 after 55 cycles at 100 mA g−1), and excellent rate capability. The greatly improved electrochemical performance can be attributed to the interconnected 1D nanostructures that provide efficient pathways for Li-ion diffusion and electron transfer, and can also accommodate the volumetric change of Fe3O4 nanoparticles during cycling. Our results reveal a new approach to Fe3O4/CNFs nanocomposites with great potential as anode material in LIBs.