Enhanced cycle stability of hybrid Li–air batteries with carbon nanofiber grown on carbon black

Abstract

The use of advanced carbon materials as an air electrode in hybrid Li–air batteries was thought to improve electrochemical performances such as cycle stability and a low voltage gap between discharge and charge. In this study, a carbon nanofiber grown on carbon black (CNF–CB) was prepared by a chemical vapor deposition (CVD) method at different temperatures (640–840 °C), and the electrochemical performance of the hybrid Li–air batteries based on the CNF–CB electrodes was investigated. The Li–air cell based on CNF–CB 740, with a cut-off voltage in the range of 2.5–4.2 V at 0.5 mA cm⁻² showed good cycle stability and demonstrated about 75 cycles (about 300 h) without an obvious increase in charge voltage.

---

1. Introduction

For the past few decades, interest in lithium–oxygen (Li–O₂) batteries has increased due to its extremely high energy density and gravimetric energy, which is three to four times that of Li-ion batteries.^1–5 Particularly, a non-aqueous electrolyte Li–air battery was widely investigated because it has the advantage of high energy densities (3504 W h kg⁻¹). However, the practical energy density of non-aqueous electrolyte Li–air batteries is limited by the deposition of the insoluble discharge product such as Li₂O₂ and LiO₂, which caused the cycle performance to fade due to clogging of the air pathway on the cathode. To overcome this problem, a hybrid Li–air battery system has been developed composed of a non-aqueous electrolyte on the anode side and an aqueous electrolyte on the cathode side, both electrolytes were isolated by Li conducted solid state separate.^6–10 Our group also has reported on the hybrid Li–air batteries since 2009.^11–14

In the basic electrolyte, the overall reaction of Li–air batteries is 2Li + H₂O + ½O₂ ↔ 2LiOH, with a theoretical cell potential of 3.43 V.^10–14 Such hybrid Li–air batteries have low-cost catalysts such as carbon materials or metal oxides M_xO_y (M = Ni, Cu, Mn, Co).^15–22 We also reported the hybrid Li–air batteries based on a graphene nanosheet, Mn₂O₃ supported carbon and reduced graphene oxide with Co(mqph) composite (RGO–Co(mqph)) as a cathode electrode. These displayed improved oxygen reduction reaction (ORR) activity during the discharge process and enhanced cycle stability.^23,24 However, the hybrid Li–air batteries are still in development to overcome the large discharge–charge overpotential gap and cycle stability. This cycle stability problem of hybrid Li–air batteries was related to the sluggish kinetics of ORR and poor durability of the cathode catalyst. Therefore, the development of an electrocatalyst that is highly active and stable for the ORR and oxygen evolution reaction (OER) was one of the most attractive challenges in hybrid Li–air batteries.

Huang et al. reported on the CNF composite with Co and Ni prepared by an electrospun method as a cathode material in Li–O₂ batteries that led to high cyclic stability and low initial overpotential. They suggested that graphitization of CNF contributed to its electrical conductivity and carbon stability.²⁵ Bhuvaneswari et al. reported that the improved electrochemical performance of a carbon nanofiber (CNF) composite with LiFePO₄ as a cathode in Li-ion batteries is attributed to the good conductivity of CNF.²⁶ Jang et al. investigated the electrochemical performance of CNF composed with graphite in Li-ion batteries.²⁷ They suggested that the specific morphology of CNF–graphite led to improved cyclability and high rate capability in Li-ion batteries. Moreover, they showed that CNF grown on graphite had a large surface area. Therefore, it is considered that the CNF composited with carbon materials may provide sufficiently large active sites as well as improve the kinetics of ORR due to the large surface area and high conductivity. In addition, the high electroconductivity of CNF could advance the durability of the cycle performance. However, the CNF composite with carbon materials has not yet been reported in Li–air batteries. Herein, CNF–CB composites were prepared and their electrochemical properties as an air-electrode for a hybrid Li–air battery were studied by examining their discharge–charge performance.

2. Experimental

2.1. Preparation of CNF–CB

CNF–CB composites were fabricated by CVD, as illustrated in Scheme 1. The FeNi supported CB were prepared from iron(III) nitrate nonahydrate (Fe(NO₃)₃·9H₂O) and nickel(II) nitrate hexahydrate (Ni(NO₃)₂·6H₂O) to provide nuclei for CNF on CB. First, 397.09 mg of Fe(NO₃)₃·9H₂O and 144 mg of Ni(NO₃)₂·6H₂O were dissolved in 250 mL of ethanol while stirring for 30 min at room temperature. Then, 1 g of CB was added to the Fe/Ni mixture solution and stirred for 2 h at room temperature. Ethanol was evaporated at 80 °C. Finally, the obtained FeNi@CB was dried in an oven at 100 °C overnight in the air condition.

Scheme 1 Schematic of the grown CNF–CB.

The powdered FeNi@CB catalyst was placed in a quartz boat at the center of a reactor tube in the furnace to grow CNF on the CB surface. After reduction in 20% H₂/He for 140 min, He gas was reflushed for 1 h before introduction of a C₂H₄ and He gas mixture for 30 min at each reaction temperature (640–840 °C). Then, He gas was flowed during cool down to ambient temperature. The prepared CNF–FeNi@CB was washed with 2 M HCl and distilled water to remove Fe and Ni metals. Finally, the CNF–CB was dried at the temperature of 80 °C. CNF–CB was denoted as CNF–CB 640, CNF–CB 740 and CNF–CB 840 by the CNF synthesis temperatures.

2.2. Assembly and testing of hybrid Li–air battery

The electrochemical test for hybrid Li–air batteries was described in our previous reports.^12,14,23 1 M LiClO₄ EC/DEC and 1 M LiOH were used as anodic and cathodic electrolytes, respectively. A Li_1+x+yAl_x(Ti, Ge)_2−xSiyP_3−yO₁₂ (LISICON) plate was used as an organic/inorganic electrolyte membrane to prevent intermixing of both solutions. For the preparation of the cathode, CNF–CB, acetylene black (AB) and 10 wt% polytetrafluoroethylene (PTFE) binder in a weight ratio of 80 : 5 : 15 were mixed and then pressed onto the Ni mesh. The discharge–charge performance was measured in a voltage range of 2.5–4.2 V at the current density of 0.5 mA cm⁻² every 2 h. The ORR activity was examined using a rotating disk electrode (RDE) in 1 M LiOH at room temperature under N₂/O₂ conditions. The catalysts were loaded on a glassy carbon electrode (0.285 cm²) with diluted (1 : 50 in methanol) 5 wt% Nafion solution (Aldrich). The RDE used a 3-electrode system where platinum (Pt) and a silver/silver chloride electrode (Ag/AgCl) were used as a counter electrode and reference electrode, respectively.

2.3. Characterization of CNF–CB

The specific surface area of all samples was measured by Brunauer–Emmett–Teller analysis (BET, TriStar 3000, Shimadzu). Raman spectra (Raman, NRS-1000 PT, Jasco) were used to compare the graphitization of CNF–CB. The morphology of CNF–CB was examined by a scanning electron microscope (SEM, SU 8000, Hitachi) and a transmission electron microscope (TEM, JEM 2100F, Hitachi). A commercial CB was used as a reference at the same conditions.

3. Result and discussion

The results of specific surface areas by nitrogen adsorption are given in Fig. 1(a). The specific surface area of CNF–CB was higher than that of CB. Particularly, CNF–CB 740 had a high specific surface area of 324 m² g⁻¹ due to the synthesized CNF.

Fig. 1 N₂ adsorption–desorption isotherms (a) and Raman spectroscopy (b) for CNF–CB 640, CNF–CB 740, CNF–CB 840 and CB.

The graphite structure of all CNF–CB and CB was also evaluated by Raman spectroscopy measurements, as shown in Fig. 1(b). All measured samples exhibited two distinct bands appearing at around 1324 cm⁻¹ (D-band) and 1597 cm⁻¹ (G-band). The D and G bands reflect the structure of sp³ and sp² hybridized carbon atoms, indicating disordered graphite and the ordered state of carbon material, respectively.^28–30 The degree of the graphitization on CNF–CB can be quantified by the intensity ratio of the D to G bands. The peak intensity ratio (I_D/I_G) was 1.68, 1.50, 2.14 and 1.86 for CNF–CB 640, CNF–CB 740, CNF–CB 840 and CB, respectively. The small I_D/I_G ratio of CNF–CB 740 suggests that CNF–CB 740 has the greatest degree of graphitization.

The structure and morphology of the CNF–CB were observed using SEM and TEM measurements. Fig. 2(a–c) displays the typical SEM images of CNF–CB 640, CNF–CB 740 and CNF–CB 840, respectively. The CNF–CB 640 and CNF–CB 740 exhibited CNF grown on the CB surface with an entangled and curled structure (Fig. 2(a and b)). Only a few fibers were observed on CNF–CB 840 in Fig. 2(c). This result reflects the importance of CNF synthesis temperatures. Such a correlation between the heat treatment temperature and structure of CNF was examined under TEM. Fig. 2(d–f) shows the TEM images of all CNF–CB samples. From the nanostructure (Fig. 2(d and e)), CNF–CB 640 and 740 samples provided selectively thin fibers of nearly 10 nm in diameter. The CNF–CB 640 and 740 have a tubular structure though CB particles were observed. On the other hand, the CNF of CNF–CB 840 showed no tubular structure and random directions in relation to the fiber axis (Fig. 2(f)). These results indicate that the formation of the CNF on the CB surface was influenced by the heat treatment temperatures of 640 °C and 740 °C that favor the formation of CNF.

Fig. 2 SEM and TEM images for CNF–CB 640 (a and d), CNF–CB 740 (b and e), and CNF–CB 840 (c and f).

Fig. 3 shows the cyclic voltammetry of CNF–CB 640, 740, 840 and CB at a scan rate of 20 mV s⁻¹ with a rotating rate of 500 rpm under 0.1 M KOH in N₂ and O₂. The onset potential of ORR was about −0.18, −0.12, −0.17 and −0.14 V vs. Ag/AgCl for CNF–CB 640, 740, 840 and CB, respectively. At an overpotential of −0.3 V, the current density of CNF–CB 640, 740, 840 and CB was −0.85, −1.22, −0.92 and −0.9 mA cm⁻², respectively. This indicated that CNF–CB 640 and CNF–CB 740 had a higher catalytic activity for ORR than that of CNF–CB 840 and CB.

Fig. 3 Cycle voltammetry (CV) of CNF–CB 640 (a), 740 (b), 840 (c) and CB (d) in the N₂ and O₂ at scan rate of 20 mV s⁻¹ with rotation rate of 500 rpm under 0.1 M KOH.

Fig. 4 shows the linear sweep voltammetry (LSV) curves for 20 wt% Pt–CB (a) and CNF–CB 740 (b) at a scan rate of 20 mV s⁻¹ under 0.1 M KOH along with the corresponding Kotechky–Levich (K–L) plots at different potentials (c and d). As shown in Fig. 4(a and b), the ORR of 20 wt% Pt–CB was diffusion controlled when the potential was less than −0.25 V. On the other hand, the CNF–CB 740 did not reach a mass transport limiting voltage in the potential region from −0.2 V to −0.4 V. The K–L equation was usually used to calculate the electrons transfer number during the ORR. The K–L equation was followed

i_l,c = 0.62nFAD^2/3₀w^1/2v^−1/6C^*₀

where i_k was the kinetic current, i_l,c was the limiting current, n the number of electrons transferred in the half reaction, F the Faraday constant (96 485C mol⁻¹), A the electrode area (0.285 cm²), D₀ the diffusion coefficient (1.9 × 10⁻⁵ cm² s⁻¹), w the angular rotation rate of the electrode, v the kinematic viscosity (0.01 cm² s⁻¹), and C^*₀ is the analyte concentration (1.2 × 10⁻⁶ mol cm³). The corresponding K–L plots of 20 wt% Pt–CB and CNF–CB 740 in Fig. 4(c and d) displayed good linearity and parallelism, indicating the first-order reaction kinetics for ORR. The electron transfer number of 20 wt% Pt–CB was calculated to be 3.9 at −0.4 V, which nearly approaches the 4e⁻ electron transfer number (n) of the Pt-based catalyst in aqueous electrolyte for ORR.³¹ Fig. 4(d) shows the K–L plots of the CNF–CB 740 according to their corresponding RDE curves. The electron transfer number was calculated to be 1.8 at −0.4 V, suggesting a 2e⁻ reduction process in an aqueous electrolyte for ORR. Similarly, the electron transfer number of CNF–CB 640, CNF–CB 840 and CB was found to be 1.8, 1.7 and 2.2, respectively (not shown).

Fig. 4 Linear sweep voltammetry (LSV) and Kotechky–Levich (K–L) plot of 20 wt% Pt–CB (a and c) and CNF–CB 740 (b and d), respectively.

Fig. 5(a) shows the discharge–charge curves from the 1^st and 75^th cycles of a hybrid Li–air cell based on CNF–CB 740 electrodes at a current density of 0.5 mA cm⁻² with discharge and charge every 2 h. The discharge voltage of CNF–CB 740 at the 1^st cycle was approximately 2.88 V vs. Li/Li⁺, and the discharge voltage gradually decreased with cycling. After 75 cycles, the discharge voltage was reached at 2.5 V and the difference in the discharge voltage between the 1^st and the 75^th cycles was 0.38 V. However, the difference in the charge voltage between the 1^st and the 75^th cycles was only 0.12 V. This result indicated that the cycle corrosion in the discharge process was more serious than in the charge process. Fig. 5(b) shows the discharge and charge voltages obtained for all measured samples during cycling at a range from 2.5 to 4.2 V. The discharge voltage of CNF–CB 640, CNF–CB 840 and CB at the 1^st cycle was 2.75, 2.75 and 2.77 V, respectively. Compared to the discharge voltage of 2.88 V at the 1^st cycle for CNF–CB 740, the 1^st discharge voltage of CNF–CB 640, CNF–CB 840 and CB was small at 0.013 V, indicating that the best ORR performance is that for CNF–CB 740. With cycling, the discharge voltage of the CNF–CB 640 and CNF–CB 840 decreased and the discharge voltage was reached at 2.5 V vs. Li/Li⁺ after 44 cycles. However, the CB showed a discharge voltage of 2.52 V after 18 cycles. It is thus considered that the CNF grown on CB led to improve the cycle stability in hybrid Li–air batteries. The CNF–CB 740 enabled long term cycling performance of hybrid Li–air batteries over 300 h, which is a much longer time than those of reported Li–air batteries with a carbon cathode (120 h, 200 h).^23,32

Fig. 5 Discharge–charge curves of CNF–CB 740 (a) and the discharge–charge voltage of all measured carbon materials (b).

Fig. 6 shows the discharge rate performance of hybrid Li–air batteries with CNF–CB 740 and CB electrodes at various current densities up to 1.5 mA cm⁻². The operating voltage of CNF–CB 740 and CB was at 3.12 and 3.06 V vs. Li/Li⁺ at 0.001 mA cm⁻², respectively. At 1.5 mA cm⁻², the cell potential of the CNF–CB 740 was as high as 2.44 vs. Li/Li⁺ which is about 300 mV better than that of CB. With increasing current densities, a linear decrease of the operating voltage of both samples is clearly observed in Fig. 6(a and b). We estimated the internal resistance of both samples by analyzing the I–V curve in Fig. 6(c and d). The internal resistance for the CNF–CB 740 and CB was estimated to be about 446 and 666 Ω cm⁻², respectively. It was also calculated that the internal resistance of CNF–CB 640 and CNF–CB 840 was about 733 and 600 Ω cm⁻², respectively. This result means that the low interfacial resistance of CNF–CB 740 provided the improvements in the activity of ORR and the cycle stability in hybrid Li–air batteries.

Fig. 6 Discharge rate performance of CNF–CB 740 (a) and CB (b) at different current densities up to 1.5 mA cm⁻² and the current–voltage curve of CNF–CB 740 (c) and CB (d).

4. Conclusions

A detailed mechanism of ORR for the CNF–CB 740 is currently not clear. However, we have found that the CNF–CB 740 exhibited good high activity of ORR in alkaline media. This good performance of CNF–CB 740 as a catalyst to reduce oxygen may be attributed to the presence of active site as shown by the specific surface area. Moreover, the CNF–CB 740 enabled long term cycling performance of hybrid Li–air batteries over 300 h. This enhanced cycle stability could be attributed to the good graphitization of CNF, which is suggested by Raman data. This is the first time using CNF–CB as a cathode of hybrid Li–O₂ batteries, and these results will encourage further investigation using optimized CNF–CB as a promising candidate to replace conventional carbon materials in hybrid Li–O₂ batteries.

Notes and references

1. J. L. Shui, N. K. Karan, M. Balasubramanian, S. Y. Li and D. J. Liu, J. Am. Chem. Soc., 2012, 134, 16654–16661.
2. V. Viswanathan, K. S. Thygesen, J. S. Hummelshoj, J. K. Norskov, G. Girishkumar, B. D. McCloskey and A. C. Luntz, J. Chem. Phys., 2011, 135, 214704.
3. M. Wu, J. Y. Jo, S. J. Kim, Y. K. Kang, H. T. Jung and H. K. Jung, RSC Adv., 2016, 6, 23467–23470.
4. Y. L. Cao, F. C. Lv, S. C. Yu, J. Xu, X. Yang and Z. G. Lu, Nanotechnology, 2016, 27, 135703.
5. Y. Zhang, X. Zhang, J. Wang, W. C. McKee, Y. Xu and Z. Peng, J. Phys. Chem. C, 2016, 120, 3690–3698.
6. A. Kraytsberg and E. Ein-Eli, J. Power Sources, 2011, 196, 886–893.
7. H. He, W. Niu, N. M. Asl, J. Salim, R. Chen and Y. Kim, Electrochim. Acta, 2012, 67, 87–94.
8. L. Li, X. Zhao, Y. Fu and A. Manthiram, Phys. Chem. Chem. Phys., 2012, 14, 12737–12740.
9. Y. Li, Z. Huang, K. Huang, D. Carnahan and Y. Xing, Energy Environ. Sci., 2013, 6, 3339–3345.
10. Y. Li, K. Huang and Y. Xing, Electrochim. Acta, 2012, 81, 20–24.
11. H. Zhou, Y. Wang, H. Li and P. He, ChemSusChem, 2010, 3, 1009–1019.
12. P. He, Y. Wang and H. Zhou, Electrochem. Commun., 2010, 12, 1686–1689.
13. P. He, Y. Wang and H. Zhou, Chem. Commun., 2011, 47, 10701–10703.
14. Y. Wang, R. Ohnishi, E. Yoo, P. He, J. Kubota, K. Domen and H. Zhou, J. Mater. Chem., 2012, 22, 15549–15555.
15. S. Wang, S. Dong, J. Wang, L. Zhang, P. Han, C. Zhang, X. Wang, K. Zhang, Z. Lan and G. Cui, J. Mater. Chem., 2012, 22, 21051–21056.
16. L. Wang, X. Zhao, Y. Lu, M. Xu, D. Zhang, R. S. Ruoff, K. J. Stevenson and J. B. Goodenough, J. Electrochem. Soc., 2011, 158, A1379–A1382.
17. C. Sun, F. Li, C. Ma, Y. Wang, Y. Ren, W. Yang, Z. Ma, J. Li, Y. Chen, Y. Kim and L. Chen, J. Mater. Chem. A, 2014, 2, 7188–7196.
18. L. Li and A. Manthiram, Nano Energy, 2014, 9, 94–100.
19. J. S. Lee, G. S. Park, H. I. Lee, S. T. Kim, R. Cao, M. Liu and J. P. Cho, Nano Lett., 2011, 11, 5362–5366.
20. J. Wu, H. W. Park, A. Yu, D. Higgins and Z. Chen, J. Phys. Chem. C, 2012, 116, 9427–9432.
21. E. M. Benbow, S. P. Kelly, L. Zhao, J. W. Reutenuer and S. L. Suib, J. Phys. Chem. C, 2011, 115, 22009–22017.
22. L. Li, S. Liu and A. Manthiram, Nano Energy, 2015, 12, 852–860.
23. E. Yoo and H. Zhou, ACS Nano, 2011, 5, 3020–3026.
24. C. H. Ahn, T. Okada, M. Ishida, E. Yoo and H. Zhou, J. Power Sources, 2016, 307, 474–480.
25. J. Huang, B. Zhang, Y. Y. Xie, W. W. K. Lye, Z.-L. Xu, S. Abouali, M. Akbari Garakani, J.-Q. Huang, T.-Y. Zhang, B. Huang and J.-K. Kim, Carbon, 2016, 100, 329–336.
26. M. S. Bhuvaneswari, N. N. Bramnk, D. Ensling, H. Ehrenberg and W. Jaegermann, J. Power Sources, 2008, 180, 553–560.
27. S. M. Jang, J. Miyawaki, M. Tsuji, I. Mochida, S. H. Yoon and K. Fei-yu, New Res. Carbon Mater., 2010, 25, 89–96.
28. M. S. Dresselhaus, A. Jorio, M. Hofmann, G. Dresselhaus and R. Saito, Nano Lett., 2010, 10, 751–758.
29. T. Jawhari, A. Roid and J. Casado, Carbon, 1995, 33, 1561–1565.
30. Y. Wang, S. Serrano and J. J. Santiago-Aviles, Synth. Met., 2003, 138, 423–427.
31. L. Li, S. Liu and A. Manthiram, Nano Energy, 2015, 12, 852–860.
32. L. Li and A. Manthiram, Adv. Energy Mater., 2014, 4, 1301795.

---