Demonstrating the steady performance of iron oxide composites over 2000 cycles at fast charge-rates for Li-ion batteries

The feasibility of using iron oxides as negative electrode materials for safe high-power Li-ion batteries is demonstrated by the carbon-coated FeOx/CNT composite synthesized by controlled pyrolysis of ferrocene, which delivered a specific capacity retention of 84% (445 mA h g(-1)) after 2000 cycles at 2000 mA g(-1) (4C).

improve their performance, namely the retention of storage capacity at fast C-rates. Therefore, iron oxide must retain its reversible specific charge capacity upon many cycles (41000) under demanding electrochemical conditions, which is very difficult to achieve due to the charge storage mechanism in iron oxides. 6 Metallic iron and lithium oxide are formed during the reduction of iron oxide. On the one hand, the formation of the ionic and electric insulating lithium oxide hinders the reversibility of the electrochemical process. On the other hand, the volume changes accompanying the reversible process lead to the pulverization of the electrode material. The detrimental impacts of these factors are magnified when operating at high C-rates due to the mechanical stress induced by the fast volume changes as well as the large overpotentials caused by the high current intensities. Herein, we present a facile synthesis strategy for carbon@FeO x composite materials with improved cycle stability at high C-rates. By controlled pyrolysis of ferrocene, a precursor that supplies both iron and carbon, at 550 and 650 1C in an autoclave, carbon-coated FeO x /carbon nanotube (CNT) composites were synthesized (Fig. S1, ESI †). The composites delivered an outstanding specific charge retention of up to 84% (445 mA h g À1 ) after 2000 cycles under fast cycling conditions (2000 mA g À1 and 4C). Barreiro et al. showed that single-walled CNTs can be obtained by the decomposition of ferrocene at 900 1C, 7 whereas Brandt and Balducci employed ferrocene as a precursor for the synthesis of the carbon@Fe 2 O 3 composite at 1050 1C which delivered good electrochemical performance as an active material for LIBs demonstrating that ferrocene is a suitable precursor for battery materials. 8 To achieve long-term stability at high C-rates, iron oxide based electrodes would benefit from (i) small FeO x particle sizes that can facilitate the electrochemical reaction and lower the mechanical stress, [9][10][11] (ii) graphitic carbon coating of FeO x for mechanical stability, [12][13][14] and (iii) interconnected CNTs enhancing the inter-particle electrical conductivity at high C-rates. [15][16][17] Since the role of the carbon coating (electrical conductivity and mechanical stability) is more important than that of CNTs, the pyrolysis of ferrocene was carried out at lower temperatures (550 and 650 1C) in the absence of precursor flow (autoclave) in an attempt to promote the formation of carbon coating. We focus the discussion on FeO x (650)/C because of its slightly better electrochemical performance than FeO x (550)/C, which is likely due to the enhanced graphitic character of carbon present in FeO x (650)/C. Note that the structural and electrochemical characterization of FeO x (550)/C is shown in the ESI. † Fig. 1a (and Fig. S2a and S3, ESI †) shows the formation of a short onedimensional carbon structure presumably due to the nucleation and growth of C species on metallic Fe nanoparticles resulting from the decomposition of ferrocene. Fig. 1a also suggests that a number of iron particles (brighter features) are embedded in carbon. Transmission electron microscopy (TEM) and highresolution TEM (HRTEM) micrographs ( Fig. 1b and c and Fig. S4, ESI †) confirmed that iron particles were covered by graphitic carbon featuring the typical (0002) planes with lattice parameters of a = 0.246 nm and c = 0.671 nm (space group = P6 3 /mmc).
Since metallic iron is not active for Li-ion storage, 18-20 the autoclave was cautiously opened at 350 1C during cooling to oxidize the iron particles. Note that drastic overpressure was not observed due to the lack of solvent. HRTEM micrographs of the composite revealed that there were two types of iron particles, namely solid iron oxides and core/shell (Fe/FeO x ) structures. The Fast Fourier Transformation (FFT) of the selected area in Fig. 2a (and Fig. S5, ESI †) confirmed that one type of particle consists of a face-centered cubic a-Fe 2 O 3 phase (hematite) (space group = Fd% 3m, lattice parameter a = 0.833 nm) while Fig. 2b shows that a number of particles possess a core/ shell structure, some of which acted as catalysts for CNT growth (Fig. S7, ESI †). FFT analysis of the selected area ( Fig. 2b.3) indicates that the shell consists of a-Fe 2 O 3 . Since the nature of the particle core could not be elucidated by FFT due to insufficient details of the core for larger particles (Fig. 2b) and the presence of Moiré fringes for smaller ones (Fig. S6, ESI †), STEM electron energy-loss spectra (EELS) were employed to further characterize the core/shell structure. Additionally, there is a broad peak at B26.01 arising from the (002) reflection of the graphitic carbon structure. The formation of graphitic carbon was also confirmed by simultaneous differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) (Fig. S9, ESI †). To remove remaining amorphous carbon and decrease the content of metallic iron, FeO x (650)/C was calcined in air at 300 1C for 2 hours (mild oxidation), and the resultant sample was referred to as FeO x (650)/C_ox. Differential scanning calorimetry (DSC) thermograms showed that amorphous carbon was eliminated after calcining as evidenced by the lack of the pseudo-peak at 340 1C for FeO x (650)/C_ox (Fig. S9, ESI †), whereas XRD revealed that the   reflections of Fe 0 became significantly weakened upon calcining as seen in traces B (Fig. 3a). The carbon content of FeO x (650)/C and FeO x (650)/C_ox was determined by elemental analysis to be 43 wt% and 30 wt%, respectively. Utilizing the Scherrer equation for relating the coherently scattering domains to Bragg peak widths: L = kl/B cos(y), in which k = 0.89 for spherical particles and B is the full angular width at half-maximum of the peak in radians, the average crystallite sizes in FeO x (650)/C was determined to be around 20 nm using the (311) reflection. Based on this estimation, no change in the particle size was observed during the thermal post-processing. The mean particle diameter was estimated to be 20 AE 14 nm based on TEM imaging of 100 different particles (Fig. S10, ESI †), which is in good agreement with the XRD results. Nitrogen adsorption/desorption measurements of FeO x (650)/C (Fig. S11, ESI †) indicate a mesoporous character of the sample as well as a high specific surface area (66 m 2 g À1 ), which is beneficial for electrolyte accessibility and rapid Li-ion diffusivity. [21][22][23][24][25] The presence of maghemite, magnetite and hematite in the sample prior to and after calcining was also verified using two vibrational spectroscopy techniques (FTIR and Raman) ( Fig. S12 and S13, ESI †).
To provide insight into the surface composition of FeO x (650)/C, X-ray photoelectron spectroscopy (XPS) was employed. There are three elements detected in the samples, i.e., Fe, O and C (Fig. S14a, ESI †). The C 1s region (Fig. S14b, ESI †) shows a dominant peak at around 284.6 eV regardless of calcining, corresponding to graphitic carbon. Deconvolution of the spectra manifests three additional small peaks at B285.4 and 286.7 and 288.9 eV, which are assigned to carbon singly bound to oxygen in phenols and ethers (i.e. C-O), carbon doubly bound to oxygen in ketones and quinones (i.e. CQO), carbon bound to two oxygen atoms in carboxyls, carboxylic anhydrides and esters (i.e. -COO), respectively. The main graphitic C peak comprises B75% of the spectrum for the sample prior to calcining. Upon calcination, the graphitic C content decreased to B67% due to the mild oxidation of the C structure. Despite this, the value is still much higher than the one reported for graphene oxide (31% graphitic C). 26 The Fe 2p 3/2 region (Fig. S14c, ESI †) consists of a peak centred at B711.1 eV typical of Fe 3+ . A lower binding energy Fe 2p shoulder is also observed, indicating the Fe 2+ species. The broad satellites centred at B714.5 and B719.1 eV are characteristic of Fe 2+ and Fe 3+ species, respectively. The O 1s spectrum (Fig. 3b) was deconvoluted into three peaks at around 529.8, 532.0 and 533.4 eV. The peak at 529.8 eV originates from the lattice oxygen of iron oxides, the intensity of which was enhanced upon calcination. The peaks at higher binding energies may correspond to O doubly (B532.0 eV) and singly (B533.4 eV) bound to carbon, consistent with the results derived from the C 1s spectra.
The performance of the composites was evaluated as active materials for LIBs in three-electrode half-cell configuration to ensure that the counter-electrode did not limit the electrochemical performance. Note that the carbon content was adjusted to be the same in all samples with addition of carbon black. Fig. 4a shows the potential profiles of commercial Fe 3 O 4 , Fe 2 O 3 , FeO x (650)/C and FeO x (650)/C_ox in the first electrochemical cycle between 3.0 and 0.1 V vs. Li/Li + . The lower cut-off potential was intentionally set to 0.1 V to increase the safety at high C-rates. All four samples displayed the characteristic behavior of Li-ion storage in iron oxides. [4][5][6][8][9][10][11][12][13][14][15][16][17][21][22][23][24][25] The lower initial reversible capacity of both composites suggests the presence of metallic iron, which, together with the larger surface area, led to lower coulombic efficiency in the first cycle (Fig. S15, ESI †). Electrolyte additives, prelithiation or preactivation may be used to mitigate low coulombic efficiency in the first cycle. [27][28][29][30] However, this issue is beyond the purpose of this work, which focuses on demonstrating the long cycle life achieved using the synthesized composites at high C-rates. At 1000 mA g À1 , commercial iron oxide electrodes were unable to retain the reversible capacity (Fig. 4b), which was dropping below 50% after only 10 cycles. Indeed, the main limitation of iron oxides, even for the nanostructured ones (BET surface area of ca. 40 m 2 g À1 for the commercial Fe 2 O 3 and Fe 3 O 4 ), is its short cycle life, especially at high C-rates. In contrast, both composites show excellent cycle life retaining 87% of the initial reversible capacity after 120 cycles. In addition, the coulombic efficiency of the composites after the first cycle outweighed those of the commercial ones indicating higher Li-ion storage reversibility of the composites. Even at 2000 mA g À1 (4C), both composites delivered remarkable electrochemical performance, e.g. FeO x (650)/C_ox (Fig. 4c) storage 445 mA h g À1 (84% retention) after 2000 cycles. The slightly higher capacity of the oxidized composite may be attributed to the lower metallic iron content, while the higher capacity retention after 2000 cycles (84% vs. 63%) could have resulted due to the removal of the low conducting amorphous carbon. To the best of our knowledge, FeO x (650)/C_ox delivers the best long term cyclability observed for iron oxide based electrodes at high C-rates comparable to the remarkable cycle life recently reported for FeOOH at 2.5C. 31 Electrochemical impedance spectroscopy (EIS) revealed easy charge transfer for FeO x (650)/C_ox (Fig. S16, ESI †). In contrast to commercial iron oxides, the EIS spectra of FeO x (650)/C_ox did not evolve significantly upon cycling which was in good agreement with its excellent cyclability (more details in the ESI †). For comparison, commercial CNTs before and after oxidation (air-annealed at 300 1C) were tested (Fig. S17, ESI †). At 2000 mA g À1 , both samples stored reversibly only ca. 130 mA h g À1 between 3.0 and 0.1 V. The electrochemical performances of composites prepared at 550 1C are also presented in the ESI † ( Fig. S18 and S19). Here, the capacity retention of the oxidized samples decreased to 63% after 2000 cycles at 2000 mA g À1 (4C). The relatively lower graphitic character of the carbon coating obtained at 550 1C is likely responsible for the poorer electrochemical performance. The specific charge of FeO x (650)/C_ox was also evaluated at lower C-rates after prolonged cycling (Fig. S20, ESI †) revealing an excellent C-rate retention. The specific charge at 0.5C also shows that the average state of oxidation of Fe in FeO x is not +3, but lower. A value of X = 0.63 (+1.25) was electrochemically estimated (ESI †).
In conclusion, we showed that iron oxide based electrodes can operate at high C-rates and with long cycle life demonstrating the feasibility of using iron oxides as negative electrode materials for safer high-power Li-ion batteries. We synthesized carbon-coated FeO x /CNTs composites by controlled pyrolysis of ferrocene, which delivered a specific capacity retention of 84% (445 mA h g À1 ) after 2000 cycles at 2000 mA g À1 (4C).
Financial support from the Deutsche Forschungsgemeinschaft in the framework of the Cluster of Excellence RESOLV (EXC1069), the Beijing University of Chemical Technology (No. oic-201503005, buctrc201 and No. JD1505) and the Generalitat de Catalunya 2014 SGR 1638 is gratefully acknowledged.