Plasma Enabled Fe₂O₃/Fe₃O₄ Nano-aggregates Anchored on Nitrogen-doped Graphene as Anode for Sodium-Ion Batteries

Abstract

Low electrical conductivity severely limits the application of Fe₂O₃ in lithium- and sodium-ion batteries. In respect of this, we design and fabricate Fe₂O₃/Fe₃O₄ nano-aggregates anchored on nitrogen-doped graphene as an anode for sodium-ion batteries with the assistance of microwave plasma. The highly conductive Fe₃O₄ in the composite can function as a highway of electron transport, and the voids and phase boundaries in the Fe₂O₃/Fe₃O₄ heterostructure facilitate Na⁺ ion diffusion into the nano-aggregates. Furthermore, the Fe–O–C bonds between the nano-aggregates and graphene not only stabilize the structural integrity, but also enhance the charge transfer. Consequently, the Fe₂O₃/Fe₃O₄/NG anode exhibits specific capacity up to 362 mAh g⁻¹ at 100 mA g⁻¹, excellent rate capability, and stable long-term cycling performance. This multi-component-based heterostructure design can be used in anode materials for lithium- and sodium-ion batteries, and potential opens a new path for energy storage electrodes.

1. Introduction

Insufficient lithium resources will seriously threaten the availability of future lithium-ion batteries (LIB). Owing to a similar working mechanism to LIBs, abundance, lower price, and environmental friendliness, sodium-ion batteries (SIBs) have gradually become a hotspot for energy storage [1,2,3,4]. Advances in SIB technology rely on the selection of suitable electrode materials to accommodate, insert, and extract Na⁺ ions with a larger radius than Li⁺ [5,6].

Among many metal oxides, the theoretical capacity of Fe₂O₃ is as high as ~1007 mAh g⁻¹, and its environmental friendliness, low cost, and abundant resources make it an ideal anode material for SIBs [7]. However, Fe₂O₃ undergoes a large volume expansion (200%) during charge-discharge, resulting in electrode pulverization and deteriorated electrical contact problems [8,9]. On the other hand, the low electrical conductivity of Fe₂O₃ (10⁻¹⁴ S cm⁻¹) and low ionic diffusivity limits the rate performance of batteries [10,11]. Common strategies are to design nanosized iron oxide, and to use conductive additives, such as graphene [9,12,13]. Another iron-based oxide, Fe₃O₄, has a slightly lower specific capacity (926 mAh g⁻¹) and much higher electrical conductivity (10²~10³ S cm⁻¹) than Fe₂O₃ [14]. Rationally compounding Fe₂O₃ and Fe₃O₄ may improve the conductivity, facilitate the redox charge transfer, and thus the rate performance. As has been demonstrated in supercapacitors, the conductive Fe₃O₄ and capacitive Fe₂O₃ synergistically produced excellent rate capability and cycling stability [15,16,17]. Recently, α-Fe₂O₃/Fe₃O₄ composite, Fe₂O₃/Fe₃O₄/FeCO₃ composite, and porous Fe₂O₃/Fe₃O₄@carbon have also been applied in LIBs and exhibited improved electrochemical performance in comparison with the Fe₂O₃ electrodes [18,19,20]. In these studies, the Fe₃O₄ plays the role of electron transport pathway, while the porous structure can facilitate fast ion transport. It is even claimed that the hetero-interfaces between the oxide components may provide an enhanced inner electric field which can assist the electron transfer and Li⁺ diffusion [19].

In this work, we designed Fe₂O₃/Fe₃O₄ nano-aggregates anchored on nitrogen-doped graphene (Fe₂O₃/Fe₃O₄/NG) as an anode of SIBs. Upon a microwave plasma process, the Fe₂O₃/Fe₃O₄/NG composite exhibited rich phase boundaries and voids. When working as an SIB anode material, the Fe₂O₃/Fe₃O₄/NG shows high specific capacity (362 mAh g⁻¹ at 100 mA g⁻¹), excellent cycling stability (84% capacity retention after 100 cycles at 1 A g⁻¹), and superior high-rate capability. The improved electrochemical performances are due to fast electron transport through Fe₃O₄, accelerated Na⁺ ion transport through Fe₂O₃/Fe₃O₄ phase boundaries, and voids in the nano-aggregates.

2. Materials and Methods

2.1. Materials Preparation

The graphene oxide (GO) used in this work was produced from natural graphite flakes (Acros, Geel, Belgium) by a modified Hummer’s method [21]. The aqueous GO dispersion was centrifuged at 11,000 rpm for 30 min and was redispersed in de-ionized (DI) water more than 3 times. By discarding the supernatant, GO suspension in N, N-dimethyl formamide (DMF, 98%, Sigma-Aldrich, St. Louis, MO, USA) with 2 mg mL⁻¹ was made.

Fe₂O₃/NG was synthesized by a solvothermal method. Fe(NO₃)₃·9H₂O (840 mg, 99.9%, Sigma-Aldrich) was dissolved in the as-obtained GO suspension (100 mL). After stirring for 60 min, the mix was transferred into a Teflon-lined autoclave and was kept at 180 °C for 6 h. The as-obtained gel was rinsed in DI water for 24 h, then freeze-dried at −60 °C, 20 Pa for 24 h to generate Fe₂O₃/NG. The as-prepared bulk Fe₂O₃/NG was ground to powder with a mortar and pestle. After treating in a home-made microwave plasma fluidized bed (power of microwave = 400 W, Ar 40 sccm, H₂ 10 sccm, 20 min), Fe₂O₃/Fe₃O_4/NG was obtained.

2.2. Materials Characterization

The crystal structure and phases of the material were characterized by X-ray diffraction (XRD, Bruker AXS GmbH, Karlsruhe, Germany) using Cu-Kα radiation. Morphology was analyzed by a field emission scanning electron microscope (SEM, Hitachi S4800, Tokyo, Japan). The transmission electron microscope (TEM) and high-resolution transmission electron microscope (HRTEM) were performed on a Hitachi HT7700 (Tokyo, Japan) and Talos F200X (Waltham, MA, USA), respectively. Thermogravimetric analysis was performed on PE STA8000 (Waltham, MA, USA). X-ray photoelectron spectroscopy (XPS) was performed on KRATOS Axis Supra XPS system (Kratos, Manchester, UK). The measured binding energies were calibrated to the reference energy by C1s = 284.6 eV. Raman spectroscopy was measured using the RM2000 system (Renishaw, London, UK), using a laser source with 17 mW at 532 nm.

2.3. Electrochemical Measurements

The active material, carbon black, and polyvinylidene fluoride (PVDF, Sigma-Aldrich) were added to N-methyl-2-pyrrolidone (NMP, 98%, Sigma-Aldrich) at a mass ratio of 8: 1: 1, and fully stirred to obtain a slurry. The resulting slurry was applied to copper foil, and dried in a vacuum oven at 100 °C for 12 h to obtain negative electrodes. The total mass loadings on a copper foil (20 mm in diameter, 99.99%, Sigma-Aldrich) are 1.13 and 0.75 mg for the Fe₂O₃/Fe₃O₄/NG and Fe₂O₃/NG electrodes, respectively. They were electrochemically characterized in CR2016-type coin cells, using Na metal as the counter electrode and reference electrode, and glass microfiber (Whatman, Little Chalfont, UK) as the separator. Dissolving 1 M NaClO₄ (98%, Acros) in a mixed solvent of ethylene carbonate (EC, Acros) and diethyl carbonate (DEC, Acros) with 1:1 volume ratio, and adding 5 vol.% of fluoroethylene carbonate (FEC, Acros) additive, the electrolyte was prepared. Galvanostatic charge-discharge were performed on a LAND battery tester (Wuhan, China) in range of 0.01 to 3.0 V (vs. Na/Na⁺). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were performed using the CHI660E electrochemical workstation (Shanghai, China). All as-assembled half cells stood in a glove box for 12 h EIS tests without cycling. All capacities were calculated based on the mass of the active materials.

3. Results

The Fe₂O₃/Fe₃O₄/NG nanocomposite was obtained by a solvothermal synthesis of Fe₂O₃/NG followed by a microwave plasma treatment. Taking advantage of the activation ability of plasma, the Fe₂O₃ was partially reduced to the Fe₃O₄, and Fe₂O₃/Fe₃O₄ nanoaggregates were formed, as illustrated in Scheme 1 [22].

As shown in the SEM and TEM images (Figure 1a,b), Fe₂O₃ nanoparticles with size around 50 nm are well enveloped in graphene in the Fe₂O₃/NG composite. After plasma radiation, the morphology of the composite and the particle size distribution showed no significant change (Figure 1c and the inset). Under TEM observation, however, the previous spherical or ellipsoidal particles turned out to be in irregular shape. Meanwhile, dark and bright areas can be differentiated, of which the bright area is perhaps voids in the particles (Figure 1d). HRTEM reveals that the inhomogeneity of the particles stems from phase segregation of Fe₂O₃ and Fe₃O₄. In a close view of a particle, fringe spacing of 0.37 nm (corresponding to (012) plane of Fe₂O₃) and 0.29 nm (corresponding to the (220) plane of Fe₃O₄) can be found simultaneously (Figure 1e). The results confirmed that the Fe₂O₃ was partially reduced to Fe₃O₄ by energetic species in the plasma. Driven by phase segregation, Fe₂O₃/Fe₃O₄ nano-aggregates were eventually formed. The mass fraction of active materials ise about 87 and 80 wt.% for Fe₂O₃/Fe₃O₄/NG and Fe₂O₃/NG, respectively, as calculated from the thermal-gravimetric analysis and differential thermal analysis (TGA/DTA) results (Figure 1f).

XRD analysis shows diffraction peaks at 24.13, 33.15, 35.61, 40.85, 49.48, 54.09, 62.45, and 63.9° for the Fe₂O₃/NG (Figure 2a), which are corresponding to the (012), (104), (110), (113), (024), (116), (214), and (300) planes of Fe₂O₃ (PDF No.33-0664), respectively. For the Fe₂O₃/Fe₃O₄/NG, additional peaks at 30.07, 37.05, and 47.13° can be found, which are corresponding to the (220), (222), and (331) planes of Fe₃O₄ (PDF No 99-0073), respectively, confirming coexistence of Fe₂O₃ and Fe₃O₄ phases (Figure 2a). The Raman D peak of RGO are upshifted from 1340 (RGO) to 1343 cm⁻¹ (NG) and G peak from 1573 to 1576 cm⁻¹ after N doping (Figure 2b). This may be due to the strain of the graphene basal plane caused by C–N bond, indicating the successful insertion of nitrogen heteroatoms [23]. Additionally, the unchanged Raman spectrum after microwave plasma process indicates stable N doping upon plasma (red curve in Figure 2b).

XPS survey scans discovered Fe 2p, O 1s, N 1s, and C 1s states in both Fe₂O₃/Fe₃O₄/NG and Fe₂O₃/NG (Figure 3a). In Figure 3b, the high-resolution Fe 2p spectra of the composites have two main peaks at 711.2 and 724.7 eV, corresponding to Fe 2p_3/2 and Fe 2p_1/2, respectively. In the Fe₂O₃/NG, the Fe 2p spectrum consists of a series of characteristic peaks of Fe³⁺ in Fe₂O₃, which are two spin energy separated components (Fe³⁺ (oh) at 711.1 and 725.1 eV), Fe 2p_3/2 satellite (719.7 eV), and Fe 2p_1/2 satellite (733.8 eV) signals, but no F⁰e or Fe²⁺ signal is observed. The Fe₂O₃/Fe₃O₄/NG has not only the same components with Fe₂O₃/NG, but also Fe²⁺ (oh), Fe³⁺(td), Fe²⁺(oh), and Fe³⁺(td) signals of Fe₃O₄ at 710.0, 713.0, 723.6, and 727.3 eV, respectively [17]. The XPS analysis clearly evidences the coexistence of Fe₂O₃ and Fe₃O₄ in the hybrid material. With the component area ratio of Fe²⁺ to Fe³⁺, the atomic ratio of Fe²⁺ to Fe³⁺ can be estimated to be 0.102. Combined with the TGA results, the mass loading of Fe₂O₃ and Fe₃O₄ are calculated to be 0.61 and 0.23 mg, respectively, as described in the Supporting Information for detail.

As shown in Figure 3c, the deconvolution of the O 1s profiles reveals three components, Fe–O (530.2 eV), Fe–O–C (531.2 eV), and C–O (533.8 eV). The areal percentage of the Fe–O–C bonds component increases from 53.5% for the Fe₂O₃/NG to 58% for the Fe₂O₃/Fe₃O₄/NG. The additional Fe–O–C bonds can not only promote the electron transport in the Fe₂O₃/Fe₃O₄/NG, but also stabilize the Fe₂O₃/Fe₃O₄ on the graphene sheet during charge/discharge, thereby improving the electrochemical kinetics and stability. [24] In Figure 3d, deconvolution of the C 1s profiles of the nanostructured hybrid material reveals four components including C–C (284.7 eV), C–N (285.2 eV), C–O(286.5 eV), and O–C=O (289.35 eV) [25,26], which remain unchanged before and after the plasma process, except for a slight areal ratio decrease of the O–C=O bonds from 22.78% for the Fe₂O₃/NG to 16.64% for the Fe₂O₃/Fe₃O₄/NG.

The CV curves from the 2nd to the 4th cycle of the electrodes are shown in Figure 4a,b. In the cathodic sweeps, a peak around 1.2 V for the Fe₂O₃/Fe₃O₄/NG in Figure 4a can be assigned to the reduction of Fe²⁺/Fe³⁺ to Fe⁰ (Equations (1) and (2)), while the peak around 0.9 V for the Fe₂O₃/NG in Figure 4b corresponds to the reduction of Fe³⁺ to Fe⁰ (Equation (1)). In the anodic sweeps, broad anodic peaks around 1.5 V are ascribed to the oxidation of Fe⁰ to Fe³⁺ for both electrodes (insets in Figure 4a,b) [24,27]. Only the Fe₂O₃/Fe₃O₄/NG electrode has a small shoulder peak at 0.8 V (inset in Figure 4a), which is an indication of the oxidization of Fe⁰ to Fe²⁺ [19]. The lower peak-to-peak separation between the redox peaks and better coincidence of the sweeps of the Fe₂O₃/Fe₃O₄/NG suggest faster charge transfer kinetics and better reversibility.

Fe₂O₃ + 6 Na⁺ + 6 e⁻ ⟷ 2 Fe⁰ + 3 Na₂O

(1)

Fe₃O₄ + 8 Na⁺ + 8 e⁻ ⟷ 3 Fe⁰ + 4 Na₂O

(2)

Figure 4c,d shows charge–discharge curves of the two types of electrodes at a current density of 100 mA g⁻¹ in the voltage window of 0.05–3.0 V (vs. Na/Na⁺). The initial specific discharge and charge capacities of the Fe₂O₃/NG electrode are 931 and 334 mAh g⁻¹, respectively, with an initial coulombic efficiency of 36%. After the microwave plasma process, the obtained Fe₂O₃/Fe₃O₄/NG composite exhibits initial specific discharge capacity of 1004 mAh g⁻¹, charge capacity of 363 mAh g⁻¹, and an almost identical initial coulombic efficiency. The lost capacity mainly stems from the formation of SEI and probably degradation of the electrolyte. In the subsequent discharge–charge cycles, the Fe₂O₃/Fe₃O₄/NG electrode has significantly higher capacities than the Fe₂O₃/NG.

The rate performances of Fe₂O₃/NG and Fe₂O₃/Fe₃O₄/NG electrodes are compared in Figure 5a. With progressively growing current density from 100 to 1200 mA g⁻¹, the Fe₂O₃/Fe₃O₄/NG composite constantly exhibits higher discharge capacities than the Fe₂O₃/NG. After a series of cycles, a reversible discharge capacity of 305 mAh g⁻¹ at 100 mA g⁻¹ is reached for the Fe₂O₃/Fe₃O₄/NG, comparing with 239 mAh g⁻¹ for the Fe₂O₃/NG. The specific capacity contributed by Fe₂O₃ in the Fe₂O₃/Fe₃O₄/NG can be estimated by the XPS and TGA results. For example, the reversible capacity contributed by Fe₂O₃ after the rate performance testing at 100 mA g⁻¹ can be 307–358 mAh g⁻¹, which is higher than the 239 mAh g⁻¹ for the Fe₂O₃/NG. This evidently suggests a synergetic effect of Fe₂O₃ and Fe₃O₄ in the electrode. The calculation of capacity contributed by Fe₂O₃ is detailed in the Supporting Information.

In Figure 5b, the cycling performance of Fe₂O₃/NG and Fe₂O₃/Fe₃O₄/NG electrodes in 100 cycles are compared at a current density of 100 mA g⁻¹. After 100 cycles, the capacity of Fe₂O₃/Fe₃O₄/NG is 291 mAh g⁻¹, which is much higher than that of Fe₂O₃/NG 218 mAh g⁻¹. At a higher current density (1000 mA g⁻¹), Fe₂O₃/Fe₃O₄/NG can still discharge 158 mAh g⁻¹ after 100 cycles (capacity retention rate 84%, Figure 5c).

In Figure 5d, the electrochemical impedance spectra (EIS) Nyquist plots of Fe₂O₃/NG and Fe₂O₃/Fe₃O₄/NG electrodes present a depressed semicircle in the high-frequency region and a straight line in the low-frequency region. To extract the EIS parameters, an equivalent circuit is proposed, as displayed in Supplementary Figure S1. The EIS tests were performed before cycling of the electrodes, with open circuity voltages of 1.65 and 1.47 V for the Fe₂O₃/NG and Fe₂O₃/Fe₃O₄/NG, respectively. As listed in

Table 1. The fitted EIS parameters and diffusion coefficients of Fe₂O₃/NG and Fe₂O₃/Fe₃O₄/NG.

Electrode	Rs (Ω)	Rct (Ω)	DNa+ (cm2 s−1)
Fe2O3/NG	6	210.6	1.65 × 10−12
Fe2O3/Fe3O4/NG	8.9	134.7	1.34 × 10−11

, both cells present similar values of electrolyte resistance (R_s). The Fe₂O₃/Fe₃O₄/NG electrode has a charge transfer resistance (R_ct: 134.6 Ω) that is only half of the Fe₂O₃/NG (210.7 Ω). The reduction of the semicircle (or R_ct) is due to more Fe-O-C bonds and the appearance of Fe₃O₄ in the electrode, indicating that the charge transfer kinetics in Fe₂O₃/Fe₃O₄/NG is faster. Moreover, the one order of magnitude higher diffusion coefficient of Na⁺ (D_Na⁺, obtained by fitting the EIS, see Supporting Information for detail) in the Fe₂O₃/Fe₃O₄/NG can be attributed to the fast Na⁺ diffusion through the phase boundaries and voids in the Fe₂O₃/Fe₃O₄ heterostructure.

The long-term cycling performance is elucidated by post-mortem observation, as shown in Figure 6. Serious cracks and particle swelling can be observed for the Fe₂O₃/NG electrode after 100 discharge-charge cycles at 1000 mA g⁻¹ (Figure 6a,b). By contrast, in the same condition, the Fe₂O₃/Fe₃O₄/NG electrode remained almost unchanged in terms of particle shape and size (Figure 6c,d). In a word, the comparison clearly demonstrates the stabilizing effect of the Fe₂O₃/Fe₃O₄/NG electrodes structure. Even after 300 cycles, the Fe₂O₃/Fe₃O₄/NG electrode could retain its particle size, although cracks occurred (Figure 6e,f).

4. Discussion

The outstanding sodium storage properties of the Fe₂O₃/Fe₃O₄/NG can be attributed to its unique structural features, as illustrated in Scheme 2. First, the highly conductive Fe₃O₄ improves electron transport in the hybrid Fe₂O₃/Fe₃O₄ nano-aggregates. Second, the phase boundaries and voids in the Fe₂O₃/Fe₃O₄ heterostructure provide fast diffusion channels for Na⁺ ions. Third, the robust interfacial interaction reinforced by Fe–O–C bonds can not only maintain the integrity of the electrode during the long-term cycles, but also provide a highway for electron transfer between the graphene and the Fe₂O₃/Fe₃O₄ nano-aggregates. Moreover, N-doping also assists the sodium storage, increasing the reversible capacity of materials. All these structural features endowed by the microwave plasma process lead to outstanding reversible capacity, good rate performance, and long-term cycling stability. As compared in

Table 2. Comparison of the specific capacities of iron oxide based anodes in literature and Fe₂O₃/Fe₃O₄/NG anode in this work.

Anodes	Current Density (mA g−1)				References
	Capacity (mAh g−1)
Fe2O3/Fe3O4/NG	100	200	1000	1200	This work
	362	300	185	174
Fe2O3/NG	50	100	200	1000	Meng 2017 [28]
	343	285	230	132
Fe2O3/C	50	100	200	1000	Zhang 2018 [29]
	364	291	245	150
Fe3O4/G	100	200	500	1000	Fu 2016 [30]
	310	225	180	140
Fe3O4@C/G	100	200	500	1000	Qi 2019 [31]
	375	300	254	200
Fe2O3@NC	200	500	1000	4000	Guo 2018 [32]
	289	253.7	221.5	167.8
Fe3O4/G/QD	100	200	500	1000	Liu 2016 [33]
	316	273	216	113
Fe3O4/C	100	200	500	1000	Wang 2017 [34]
	293	262	223	195
Fe3O4@N–C	80	240	400	800	Zhao 2019 [35]
	386	315	277	248

, the electrochemical performance of the Fe₂O₃/Fe₃O₄/NG is among the top-ranking of the reported iron oxide/graphene-based SIB anodes.

5. Conclusions

In summary, the Fe₂O₃/Fe₃O₄ nano-aggregates anchored on nitrogen-doped graphene as an anode for sodium-ion batteries were successfully prepared with the assistance of the microwave plasma process. The highly conductive Fe₃O₄ in the composite improves electron transport, and the voids and phase boundaries facilitate Na⁺ ion diffusion into the Fe₂O₃/Fe₃O₄ heterostructure. Moreover, the Fe–O–C bonds not only strength the structural robustness, but also electrically bridge the graphene and the Fe₂O₃/Fe₃O₄ nano-aggregates. Consequently, the Fe₂O₃/Fe₃O₄/NG anode exhibits outstanding electrochemical performance, i.e., high specific capacity, excellent rate capability, and stable long-term cycling performance. The design concept of this unique heterostructure can be extended to other energy storage applications based on metal oxides.