Iron oxide/graphene composites as negative-electrode materials for lithium ion batteries – optimum particle size for stable performance

Abstract

We synthesized Fe₂O₃/graphene composites by a hydrothermal method. The effect of varying the pH in the range pH = 8–12 on the properties of the composites and their performance as negative-electrode materials in lithium ion batteries was investigated. The particle size increased with increasing pH. The best battery performance was obtained for composites prepared at pH = 10 (specific capacity exceeding 800 mA h g⁻¹ after 60 cycles) due to superior preservation of the electrode morphology compared to samples prepared at lower or higher pH values.

---

Introduction

Iron oxide and iron oxide/carbon composites are very promising materials for energy storage as lithium ion battery (LIB) or sodium ion battery (SIB) negative electrodes.^1–30 Iron oxide has advantages of being inexpensive, stable and non-toxic, in addition to having a high theoretical capacity.^2,5 The theoretical specific capacities of iron oxide LIB negative electrodes Fe₂O₃ and Fe₃O₄ are 1005 mA h g⁻¹ and 924 mA h g⁻¹, respectively, which are much higher compared to commonly used graphite negative electrodes (372 mA h g⁻¹).^{2,4,5,7,13–15,18}

However, similar to other metal oxides, pure iron oxide negative electrodes suffer from low electron conductivity and poor capacity retention due to large volume expansion (over 200%).^2,9,14 One of the methods to solve this problem is to use iron oxide–carbon composites. Composite materials involving various forms of carbon has been used in different types of batteries to date.^{1–27,31–33} A number of different iron oxide/carbon composite electrodes and different synthesis methods has been reported to date.^1–27 Different forms/morphologies of carbon have been used, such as graphene,^{3,6,8–11,13,15,16,19–21,23} reduced graphene oxide,^4,5,12,14,18 graphitic carbon microspheres,⁷ carbon nanotubes and graphene,^17,22etc. In particular, different Fe₂O₃/graphene composites have been reported to exhibit good LIB performance.^{3,6,8–11,13,15,16,19–21,23}

Among various possible synthesis methods for Fe₂O₃/graphene composites, the hydrothermal synthesis has an advantage of being a simple aqueous solution method typically conducted under mild conditions.² This method has been commonly used for synthesis of Fe₂O₃/graphene composite materials.^3,5,8,20 While it is known that the hydrothermal reaction conditions determine the morphology and composition of the prepared composite,² comprehensive studies of the effect of synthesis conditions on the composite morphology and battery performance have been scarce. Therefore, we investigated the effect of changing the pH during the hydrothermal synthesis to control the size of synthesized Fe₂O₃ nanoparticles in Fe₂O₃/graphene composites and evaluated the effect of the particle size on the battery performance. Despite the common expectation that smaller particle sizes would yield better performance due to larger surface area available, there is an optimal particle size for the best performance. Obtained results are discussed in detail.

Experimental

Synthesis of graphene oxide (GO)

GO was synthesized from commercial graphite (325 mesh) by a modified Hummers method.^9,12 Briefly, 1.5 g graphite powder and 9 g KMnO₄ were evenly mixed in a 500 mL round-bottomed flask. 20 mL H₃PO₃ and 180 mL concentrated H₂SO₄ were mixed in a beaker kept at 0 °C and stirred. After 20 min stirring, acid mixture was transferred into the flask slowly, while keeping the mixture below 50 °C. Then the flask was put into an oil bath at 50 °C and stirred overnight. After that, the solution was cooled down naturally. Subsequently 200 mL cold water was poured into the solution while stirring. H₂O₂ was then added into the solution drop by drop under stirring, until the color of the solution became brilliant yellow. After being stirred for another 1 hour, the mixture was washed and centrifuged with 0.05 M HCl (250 mL), de-ionized water and absolute ethanol respectively to remove other ions. Finally, the solid was dried in vacuum oven at 60 °C.

Synthesis of Fe₂O₃/graphene composite

The synthesis of an iron oxide graphene composite was performed using a modified procedure from ref. 8. In a typical experiment, 70 mg of GO was dissolved in 35 mg DI water. After ultrasonication for 2 hours, 94.62 mg FeCl₃ was added into the GO dispersion and sonicated for 1 hour. Then the pH of the dispersion (measured by a pH meter) was adjusted to a desired value by adding drops of 0.5 M NaOH. After that, the solution was sonicated for another 30 min and transferred to autoclave and kept at 120 °C for 5 hours. After hydrothermal synthesis, the solid is washed with de-ionized water and ethanol three times respectively. Finally, the solid was collected by centrifugation and dried on a hotplate at a temperature of 50 °C overnight.

Characterization

X-ray diffraction (XRD) patterns were measured on a Bruker D8 advance diffractometer using Cu Kα radiation (λ = 0.154184 nm) as radiation source. Transmission electron microscopy (TEM) was carried out using a FEI Tecnai G2 20 S-TWIN Scanning Transmission Electron Microscope System. Thermogravimetric analysis (TGA) was performed by a Q50 Thermal Gravimetric Analyzer. Electrochemical measurements were completed in a coin-cell (CR2032). Working electrode was made by mixing active material (80%), Super-P@Li carbon black (7.5%), KS6 graphite (2.5%) and polyvinylidene fluoride (10%). The mixture was dried in vacuum at 90 °C for 12 hours before coating on Cu foil. The coated foil was dried at 120 °C in vacuum for 12 h to remove solvent. Then the disk-shaped electrodes with 14 mm diameter were cut and the coin-cells were assembled in a glove box filled with argon. The electrolyte (purchased from MTI Corporation) consisted 1.0 M LiPF₆ in a mixture of ethylene carbonate/dimethyl carbonate/diethyl carbonate (1 : 1 : 1 in volume). Lithium metal (15.6 mm in diameter, 0.25 mm in thickness) was used as a counter-electrode. The galvanostatic performance was measured by a Land battery test system (LAND-CT2001A) at different current densities of 100 mA g⁻¹ to 2 A g⁻¹ between 3.0 V and 0.005 V. In average, 2.5–3.0 mg active material was loaded on Cu foil. The specific capacities were calculated on the basis of the weight of the active material of each electrode. Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (C-V) measurements were both carried out using a BioLogic VMP3 electrochemical workstation. EIS measurements were conducted by employing an ac voltage of 5 mV amplitude in the frequency range of 0.01–100 kHz, while C-V was measured at the rate of 0.5 mV s⁻¹ between 0.005 V and 3.0 V.

Results and discussion

The morphology and crystallinity of the samples have been comprehensively characterized. TEM images of Fe₂O₃/graphene composites at pH = 8 and pH = 10 before cycling and after 1 cycle, 2 cycles, 5 cycles and 60 cycles are shown in Fig. 1. TEM images for other pH values are shown in ESI,† while the XRD results are shown in Fig. 2.

Fig. 1 TEM images of Fe₂O₃/graphene composites prepared at pH = 8 and pH = 10: initial, after 1 cycle, after 2 cycles, after 5 cycles and after 60 cycles.

Fig. 2 XRD pattern of Fe₂O₃/graphene composite prepared at different pH. The peaks are labelled according to JCPDS file 24-0072.

The particle size obtained at pH = 10 is much larger compared to that for pH = 8 before cycling. Particle size for all the samples increases with increasing pH from TEM images (see ESI†). In all the electrodes, changes in morphology are observed, with significant changes already evident after only a few cycles in some cases, such as pH = 8.

An increase in the Fe₂O₃ particle size with increasing pH can be clearly observed. However, crystallite size derived from XRD measurements shows nonlinear dependence on the pH value. The crystallite sizes for pH values 8, 9, 10, 11, and 12 were 33 nm, 34 nm, 36 nm, 38 nm, and 33 nm, respectively.

TGA of the composites prepared at different pH values is shown in Fig. 3. The percentages shown correspond to the graphene oxide content in the composites. The graphene oxide content is the highest for pH = 10 (39.2%), followed by pH = 8 (38.1%), pH = 9 and pH = 11 (36.5%), and pH = 12 (35.7%).

Fig. 3 Thermogravimetric analysis of composites prepared at different pH.

There are significant differences in cycling and rate performances of composites prepared at different pH, as shown in Fig. 4. For negative electrodes prepared at pH = 8, specific capacity of 1480 mA h g⁻¹ in the first cycle drops to 903 mA h g⁻¹ in the second cycle and 613 mA h g⁻¹ in the 60^th cycle at 100 mA g⁻¹ charge/discharge rate. At charge/discharge rates of 0.5 A g⁻¹ and 1 A g⁻¹, the specific capacities are 1029 mA h g⁻¹ and 934 mA h g⁻¹ in first cycle, 579 mA h g⁻¹ and 511 mA h g⁻¹ at 2^nd cycle, and 318 mA h g⁻¹ and 277 mA h g⁻¹ at 60^th cycle respectively. In terms of the rate performance, for the rate 0.1 A g⁻¹ the specific capacity drops to 771 mA h g⁻¹ by 5^th cycle. For the rates 1 A g⁻¹ and 2 A g⁻¹ specific capacities of 503 mA h g⁻¹ and 429 mA h g⁻¹ are obtained, respectively. The sample prepared at pH = 11 exhibits similarly inferior performance, with specific capacity of 717 mA h g⁻¹ at 5^th cycle at 0.1 A g⁻¹, while for the rates 1 A g⁻¹ and 2 A g⁻¹ specific capacities of 405 mA h g⁻¹ and 307 mA h g⁻¹ are obtained. For this sample, initial capacity of 1408 mA h g⁻¹ drops to 847 mA h g⁻¹ by the second cycle, and 590 mA h g⁻¹ by 60^th cycle. In addition, the specific capacities at 0.5 A g⁻¹ charge/discharge rate are 563 mA h g⁻¹ and 322 mA h g⁻¹ at 2^nd cycle and 60^th cycle, respectively, while the corresponding capacities are 767 mA h g⁻¹ and 266 mA h g⁻¹ at 1 A g⁻¹ charge/discharge rate.

Fig. 4 Cycling performance of Fe₂O₃/graphene composite for different pH at current density of (a) 100 mA g⁻¹; (b) 0.5 A g⁻¹; (c) 1 A g⁻¹; (d) rate capability of Fe₂O₃/graphene for different pH at charge/discharge rates from 100 mA g⁻¹ to 2 A g⁻¹.

The best performance is obtained for the sample prepared at pH = 10, with initial capacity drop from the first to second cycle 1336 mA h g⁻¹ to 1049 mA h g⁻¹ (Fig. 4(a)). The capacity at 60^th cycle remains at 806 mA h g⁻¹. From Fig. 4(b) and (c), it is obvious that pH = 10 also performed the best at higher charge/discharge rates, with specific capacities of 432 mA h g⁻¹ and 289 mA h g⁻¹ at 60^th cycle for charge/discharge rates of 0.5 A g⁻¹ and 1 A g⁻¹, respectively. For the rate performance, specific capacities of 707 mA h g⁻¹ and 620 mA h g⁻¹ are obtained at rates 1 A g⁻¹ and 2 A g⁻¹. For pH values 9 and 12, similar performance is obtained, with slightly higher specific capacities for pH = 9. At 60 cycles, specific capacities of 756 mA h g⁻¹ and 739 mA h g⁻¹ are obtained for pH = 9 and 12, respectively, at a charge/discharge rate of 100 mA g⁻¹ while the specific capacities at the rates of 1 A g⁻¹ and 2 A g⁻¹ are 614 mA h g⁻¹ and 537 mA h g⁻¹ for pH = 9 and 582 mA h g⁻¹ and 505 mA h g⁻¹ for pH = 12.

The loss of capacity with cycling is likely related to the change in morphology. The worst performing materials (those prepared at pH = 8 and pH = 11) also exhibit the most significant change in morphology before and after cycling. It can be observed from the TEM images after cycling that some pulverization occurs, which leads to the decrease of cycling performance. But the degree of pulverization is dependent on the pH during synthesis, and thus it may be related to starting particle size or graphene content (since higher graphene content may result in better buffering of volume changes).

To investigate the electrochemical performance in more detail, C-V measurements have been performed and the obtained results are shown in Fig. 5. It can be observed that in all cases similar behaviour is observed, with the 1^st cycle exhibiting different peak positions indicating irreversible changes. For Fe₂O₃/graphene negative electrodes, electrochemical reactions can be described as Fe₂O₃ + 6e⁻ + 6Li⁺ ↔ 2Fe⁰ + 3Li₂O^{3,5,6,9,12,13} or as a two-step reaction Fe₂O₃ + 2e⁻ + 2Li⁺ ↔ Li₂(Fe₂O₃) and Li₂(Fe₂O₃) + 4e⁻ + 4Li⁺ ↔ 2Fe⁰ + 3Li₂O.¹⁶ In all cases, we can observe cathodic feature at ∼0.005 V and anodic feature at ∼0.24 V which correspond to the intercalation and deintercalation of lithium into graphene, respectively.^3,8 Broad anodic peak at ∼1.9 V corresponds to the reversible oxidation of Fe⁰ to Fe³⁺,^{3,5,9,15,19–23} and the position of this peak remains stable with cycling.

Fig. 5 Cyclic voltammetry curves of the Fe₂O₃/graphene composite negative electrodes for different pH.

In the first cycle, we can observe two cathodic peaks ∼0.55–0.6 V and at ∼1.57–1.67 V, which can be attributed to the reduction of Fe³⁺, the formation of Li₂O and the formation of solid–electrolyte interface (SEI) layer.^{3,5,15–17,19,20} In the subsequent cycles, only the lower potential peak is observed and it shifts towards higher potentials of 0.66–0.7 V. Such shifts have also been previously observed.^{5,8,9,16,17,20–23} and attributed to irreversible changes such as irreversible phase transformation^9,17 or the formation of SEI layer.¹⁶

To further characterize the electrochemical performance, electrochemical impedance spectroscopy (EIS) has been performed and the obtained results are shown in Fig. 6 and summarized in Table 1. In the Nyquist plots, the straight line corresponds to Warburg resistance.⁴R_ct denotes charge transfer resistance of the negative electrode,⁶ while R_s is the ohmic resistance.⁸ Constant phase elements represent SEI film and contact problems and the capacitance of the double layer.⁸ For all composites, we can observe that the changes in the ohmic resistance are small, with some increase observed after 3 cycles. Initial R_ct values are similar for all the negative electrodes except the one prepared at pH = 12, where a smaller initial R_ct value is observed. After 3 cycles, the smallest change in R_ct value is observed for pH = 10, in agreement with small changes in morphology and good capacity retention. For the majority of the electrodes, R_ct increases after cycling, similar to previous work.¹³ For negative electrodes exhibiting poor capacity retention, pH = 8 and pH = 11, some decrease in R_ct is observed. This is likely due to the fact that these electrodes exhibit larger degree of pulverization, resulting in smaller particle sizes and consequently faster diffusion of Li⁺,²⁰ as expected for a larger surface area. However, with continued cycling excessive pulverization and changes in morphology become detrimental to capacity retention.

Fig. 6 (a) Equivalent circuit of Fe₂O₃/graphene electrodes and obtained Nyquist plots (b) initial and (c) after 3 cycles.

Table 1 EIS fitting parameters corresponding to the equivalent circuit shown in Fig. 5

pH		R s	R ct
8	Initial	1.02	48.85
	After 3 cycles	1.94	39.71
9	Initial	1.72	50.8
	After 3 cycles	2.40	63.82
10	Initial	0.70	50.3
	After 3 cycles	1.38	52.7
11	Initial	1.04	56.39
	After 3 cycles	1.46	45.11
12	Initial	1.21	35.81
	After 3 cycles	2.25	44.63

---

Thus, we can conclude that while small particles offer a good performance initially, it does not necessarily result in good capacity retention. The fact that the larger particles may result in a better battery performance and that there is an optimal particle size for good capacity retention with cycling has been reported previously for different oxide materials,^34–37 such as Cu₂O,³⁵ ZnMn₂O₄,^34,36 and CoMn₂O₄.³⁷ Similar to ZnMn₂O₄,³⁴ we find that there is an optimal particle size. With increasing pH, the particle size increases, but the crystallinity does not necessarily improve. The initial particle size and crystallinity would affect the division of metal particles upon reduction,³⁵ which is consistent with the fact that there is a considerable variation in the preservation of negative electrode morphology after cycling among different samples. This is dependent not only on the size but also on crystallinity of the particles, since the relationship between pH and particle size is linear, while the relationship between pH and crystallite size and LIB performance is nonlinear. This is possibly due to the fact that there are also differences in graphene content among the samples, which could result in small differences in buffering of volume changes. The samples exhibiting good capacity retention and rate performance also exhibit smaller morphology difference before and after cycling.

Conclusions

We investigated the effect of Fe₂O₃ particle size on the LIB performance of Fe₂O₃/graphene composite negative electrodes. The particle size was varied by changing the pH of the solution for the hydrothermal synthesis. The particle size was found to increase with increasing pH, while the crystallinity dependence on pH was nonlinear. There was an optimal particle size and/or crystallinity resulting in the best capacity retention during cycling, consistent with small electrode morphology changes after cycling. For the optimal pH value (pH = 10), specific capacity exceeding 800 mA h g⁻¹ was obtained after 60 cycles, while for the worst performing sample (pH = 11) specific capacity after 60 cycles was below 600 mA h g⁻¹.

Acknowledgements

Financial support from the National Science Fund of China (Project No. 21403103), the ​ Strategic Research Theme on Clean Energy and Seed Funding for Basic Research Grant (of the University of Hong Kong) is acknowledged. The authors would like to thank Prof. K. Y. Chan for the use of the electrochemical workstation.

References

1. J. Tucek, K. C. Kemp, K. S. Kim and R. Zboril, ACS Nano, 2014, 8, 7571–7612.
2. Z. Y. Wang and C.-J. Liu, Nano Energy, 2015, 11, 277–293.
3. L. Li, G. M. Zhou, Z. Weng, X.-Y. Shan, F. Li and H.-M. Cheng, Carbon, 2014, 67, 500–507.
4. X. Y. Li, Y. Y. Ma, L. Qin, Z. Y. Zhang, Z. Zhang, Y.-Z. Zheng and Y. Q. Qu, J. Mater. Chem. A, 2015, 3, 2158–2165.
5. X. M. Dong, L. Li, C. J. Zhao, H.-K. Liu and Z. P. Guo, J. Mater. Chem. A, 2014, 2, 9844–9850.
6. Y. Yang, X. J. Fan, G. Casillas, Z. W. Peng, G. D. Ruan, G. Wang, M. J. Yacaman and J. M. Tour, ACS Nano, 2014, 8, 3939–3946.
7. H. W. Zhang, X. R. Sun, X. D. Huang and L. Zhou, Nanoscale, 2015, 7, 3270–3275.
8. C. Wu, H. Zhang, Y.-X. Wu, Q.-C. Zhuang, L.-L. Tian and X.-X. Zhang, Electrochim. Acta, 2014, 134, 18–27.
9. S. K. Liu, Z. X. Chen, K. Xie, Y. J. Li, J. Xu and C. M. Zheng, J. Mater. Chem. A, 2014, 2, 13942–13948.
10. X. Wang, W. Tian, D. Q. Liu, C. Y. Zhi, Y. Bando and D. Golberg, Nano Energy, 2013, 2, 257–267.
11. Z. L. Jian, B. Zhao, P. Liu, F. J. Li, M. B. Zheng, M. W. Chen, Y. Shi and H. S. Zhou, Chem. Commun., 2014, 50, 1215–1217.
12. X. J. Zhu, Y. W. Zhu, S. Murali, M. D. Stoller and R. S. Ruoff, ACS Nano, 2011, 5, 3333–3338.
13. G. Wang, H. Wang, S. B. Cai, J. T. Bai, Z. Y. Ren and J. B. Bai, J. Power Sources, 2013, 239, 37–44.
14. J. Qu, Y.-X. Yin, Y.-Q. Wang, Y. Yan, Y.-G. Guo and W.-G. Song, ACS Appl. Mater. Interfaces, 2013, 5, 3932–3936.
15. J. Kan and Y. Wong, Sci. Rep., 2013, 3, 3502.
16. X. Zhu, X. Y. Song, X. L. Ma and G. Q. Ning, ACS Appl. Mater. Interfaces, 2014, 6, 7189–7197.
17. S. Q. Chen, P. Bao and G. X. Wang, Nano Energy, 2013, 2, 425–434.
18. Q. Zhou, Z. B. Zhao, Z. Y. Wang, Y. F. Dong, X. Z. Wang, Y. Gogotsi and J. S. Qiu, Nanoscale, 2014, 6, 2286–2291.
19. M. Du, C. H. Xu, J. Sun and L. Gao, Electrochim. Acta, 2012, 80, 302–307.
20. T. Hu, M. Xie, J. Zhong, H.-T. Sun, X. Sun, S. Scott, S. M. George, C.-S. Liu and J. Lian, Carbon, 2014, 70, 141–147.
21. R. H. Wang, C. H. Xu, J. Sun and L. Gao, Sci. Rep., 2014, 4, 7171.
22. M. H. Chen, J. L. Liu, D. L. Chao, J. Wang, J. H. Yin, J. Y. Lin, H. J. Fan and Z. X. Shen, Nano Energy, 2014, 9, 364–372.
23. J. Ye, J. Zhang, F. X. Wang, Q. M. Su and G. H. Du, Electrochim. Acta, 2013, 113, 212–217.
24. Y. Q. Zou, J. Kan and Y. Wang, J. Phys. Chem. C, 2011, 115, 20747–20753.
25. S. L. Chou, J. Z. Wang, Z. X. Chen, H. K. Liu and S. X. Dou, Nanotechnology, 2011, 22, 265401.
26. W. J. Yu, P. X. Hou, L. L. Zhang, F. Li, C. Liu and H. M. Cheng, Chem. Commun., 2010, 46, 8576–8578.
27. H. Liu, G. X. Wang, J. S. Park, J. Z. Wang, H. K. Liu and C. Zhang, Electrochim. Acta, 2009, 54, 1733–1736.
28. Z. C. Wu, K. Yu, S. D. Zhang and Y. Xie, J. Phys. Chem. C, 2008, 112, 11307–11313.
29. J. Sarradin, M. Ribes, A. Guessous and K. Elkacemi, Solid State Ionics, 1998, 112, 35–40.
30. J. Sarradin, A. Guessous and M. Ribes, J. Power Sources, 1996, 62, 149–154.
31. Y. Chen, S. T. Lu, X. H. Wu and J. Liu, J. Phys. Chem. C, 2015, 119, 10288–10294.
32. S. T. Lu, Y. Chen, X. H. Wu, Z. D. Wang and Y. Li, Sci. Rep., 2014, 4, 4629.
33. W. Qin, B. D. Fang, S. T. Lu, Z. D. Wang, Y. Chen, X. H. Wu and L. Han, RSC Adv., 2015, 5, 13153–13156.
34. F. M. Courtel, H. Duncan, Y. Abu-Lebdeh and I. J. Davidson, J. Mater. Chem., 2011, 21, 10206–10218.
35. P. Poizot, S. Laruelle, S. Grugeon, L. Dupont and J. M. Tarascon, Nature, 2000, 407, 496–499.
36. A. R. Armstrong, C. Lyness, P. M. Panchmatia, M. S. Islam and P. G. Bruce, Nat. Mater., 2011, 10, 223–229.
37. M. Bijelić, X. Liu, Q. Sun, A. B. Djurišić, M. H. Xie, A. M. C. Ng, C. Suchomski, I. Djerdj, Ž. Skoko and J. Popović, J. Mater. Chem. A, 2015, 3, 14759–14767.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra19852f

---