Electrochimica Acta Electrostatic heterocoagulation of
Electrochimica Acta 55 (2010) 5808–5812
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrostatic heterocoagulation of carbon nanotubes and LiCoO2 particles for a high-performance Li-ion cell Kyeu Yoon Sheem a , Minseok Sung a , Young Hee Lee b,∗ a
Samsung SDI Corporate R&D Center, Gongse-dong, Giheung-gu, Yongin-si, Gyeounggi-do 446-577, Republic of Korea Department of Physics, Department of Energy Science, Center for Nanotubes and Nanostructured Composites, Sungkyunkwan Advanced Institute of Nanotechnology, Sungkyunkwan University, Suwon 440-746, Republic of Korea b
a r t i c l e
i n f o
Article history: Received 7 April 2010 Received in revised form 5 May 2010 Accepted 5 May 2010 Available online 13 May 2010 Keywords: Thin multiwalled carbon nanotubes Heterocoagulation Swelling Electrode density Specific capacity
a b s t r a c t Nanotubes were coated on the surface of active LiCoO2 particles using electrostatic heterocoagulation to enhance the electrochemical properties of a Li-ion battery. Only 0.5 wt% of multiwalled carbon nanotubes (MWCNTs) was added as a conducting agent into the LiCoO2 cathode, which had a density of 4.0 g cm−3 . We found that our electrode that was prepared using heterocoagulation with 0.5 wt% of thin MWCNTs maintained a volumetric capacitance of 403 mAh cm−1 after 40 cycles from the initial 624 mAh cm−1 , compared with previous result of 310 mAh cm−1 obtained from simple mixing with 3 wt% MWCNTs. The high volumetric capacity with smaller swelling using less amount of MWCNTs was attributed to the self-assembled nanotube network formed between active particles during coagulation, which was maintained with volume expansion during cycle testing. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction High-energy density and long cycle life for secondary batteries are critical because of the diverse market for portable electronic devices. Among the different types of batteries, rechargeable Li-ion batteries have been widely used in portable electronic devices. The Li-ion battery is composed of a positive cathode (mostly LiCoO2 ) and a negative anode (mostly graphite). Although Si or Sn nanoparticles have been known to exhibit high-energy density when used as anode material [1,2], their cycle life is poor because of structural instability and needs to be improved. Another way of improving energy density is to increase the density of the cathode electrode. In general, the cathode of a lithium battery is composed of active material, conducting agent, and binder [3]. More specifically, the cathode typically consists of 96 wt% of active material, 2 wt% of binder, and 2 wt% of carbon blacks, which are used as a conducting agent [4]. Reducing the amount of highly compressed electrode and conducting agent (non-active material) will improve the energy density of the cell. There have been several efforts to construct a high-density electrode using conventional LiCoO2 and carbon nanotubes as an alternative conducting agent instead of carbon black [5–8]. In past efforts, carbon nanotubes were mixed with the cathode material
∗ Corresponding author. Tel.: +82 31 299 6507; fax: +82 31 290 5954. E-mail address: [email protected] (Y.H. Lee). 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.05.027
and high-density electrodes were fabricated, resulting in increased energy density and power density due to the high conductivity of carbon nanotubes [9]. Although carbon nanotubes have been known to exhibit high electrical conductivity, the network formation among cathode particles has been limited by their aggregation. Furthermore, volume expansion during cycle testing degraded energy density, and thus, the cell became unstable and sometimes deteriorated. The overall electrochemical performance was dependent on the degree of dispersion of nanotubes in the cathode. Therefore, uniform distribution of a minimum amount of carbon nanotubes among oxide particles is required for cell stability. The aim of this work was to find a new method of forming an effective network of carbon nanotubes between LiCoO2 particles to enhance the energy density of the cathode electrode while preserving the conducting network of the cathode electrode with minimal volume expansion during cycling. We used 0.5 wt% of carbon nanotubes as a conducting agent with an electrode density as large as 4.0 g cm−3 compared with 2–3 wt% of carbon black and a cathode density of 3.2–3.84 g cm−3 [10]. Carbon nanotubes were uniformly coated on LiCoO2 particles at the isoelectric point. The carbon nanotubes were uniformly coated on the surface of active particles by heterocoagulation at the isoelectric point. This method is advantageous compared with the conventional mixing method, particularly by reducing the amount of conducting agent while maintaining the stability of the conducting network.
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
2. Experimental Because cell performance relies heavily on the dispersion of carbon nanotubes, we chose two types of nanotubes. Thin multiwalled carbon nanotubes (t-MWCNTs) were synthesized using chemical vapor deposition with FeMoMgO catalysts, which were prepared using a combustion method [11]. t-MWCNTs are composed of 2–6 walls and have a large specific surface area of 370 m2 g−1 . Hollow MWCNTs-100 (h-MWCNTs) were purchased from Nanokarbon Co., Korea. The diameter of h-MWCNTs is about 80 nm, and the specific surface area is about 30 m2 g−1 . Carbon black (Super-P) was purchased from MMM carbon, Belgium. The specific surface area of Super-P is about 60 m2 g−1 . The positive active material (LiCoO2 ), which had an average particle size of about 10 m, was purchased from Umicore Corp., Belgium. The samples were prepared using heterocoagulation and simple mixing [9,12]. To coagulate CNTs on active LiCoO2 particles, 0.5 wt% of gelatin was dissolved in 50 ml of deionized water and mixed with a known mass of LiCoO2 . The mixture was filtered and vacuum-dried at 110 ◦ C for 24 h. The surfaces of h-MWCNTs and t-MWCNTs are hydrophobic, thus, they are hardly soluble in water. Therefore, CNTs were dissolved in a mixture of nitric and sulfuric acid (volume ratio, 1:2) for 3 h and sonicated for 2 h. The CNTs were then filtered and washed with deionized water. The gelatin-coated active materials were then added to the CNT solution. To maximize the surface coating of active materials with CNTs, we titrated the solution with acetic acid to find the isoelectric point of the solution. The isoelectric point was located between pH 3 and 4. CNTs were coagulated on the surface of LiCoO2 and precipitated. After drying, the powder was heated up to 300 ◦ C under a N2 atmosphere to eliminate the gelatin and adsorbed water. For comparison, Super-P coated active materials were prepared by the conventional method [13]. To prepare electrodes, 2 wt% of polyvinylidene fluoride (PVDF) binder was dissolved in N-methyl-2-pyrrolidone (NMP), and then the active material (0.5 wt% of CNTs coated on 97.5 wt% of active material) was added. The manually mixed slurry was coated on only one side of an Al foil substrate and dried in a convection oven for 20 min at 120 ◦ C. Each conducting agent (t-MWCNTs, h-MWCNTs, and Super-P) and PVDF was simply mixed in NMP solution followed by mixing with LiCoO2 . The weight ratio of LiCoO2 , PVDF, and the conductive agent in the slurry was 97.5:0.5:2.0. By using this slurry, we obtained the positive electrode. The virgin electrode had rather low densities under 2.0 g cm−3 . These electrodes were further com-
Fig. 1. Absorbance of the functional groups on t-MWCNTs after acid treatment: (a) raw t-MWCNTs and (b) acid-treated t-MWCNTs.
5809
pressed with a rolling press to increase the density of the composite active mass up to 4.0 ± 0.05 g cm−3 . Compressing the electrode generally enhanced the electrical conductivity of the cathode [14,15]. Each electrode was cut into a circular disk with an area of 2 cm2 . The coin-type half-cells were assembled in an Ar-filled glove box. A porous polypropylene film was used as the separator. A 700 m thick Li-metal foil was used as the counter electrode. The electrolyte was 1 M LiPF6 solution in a 1:1 mixture (by volume) of ethylene carbonate (EC) and diethyl carbonate (DEC). Test cells were sealed in an ambient atmosphere after evacuation for 5 min to allow the electrolyte to homogeneously penetrate the electrode pores. Electrical testing was carried out using a Toyo Cycle Tester after aging the cells for 10 h. Initial charge–discharge cycles were carried out at a low rate of 0.2C for the initial cycle followed by cycling at a high rate of 1.0C for 40 cycles. ac impedance measurements after cycling tests
Fig. 2. SEM images of LiCoO2 particles coated by coagulation with (a) t-MWCNTs, (b) h-MWCNTs, and (c) Super-P.
5810
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
were uniformly coated on the surface of active materials without being aggregated, in contrast to the conventional simple mixing approach, in which the nanotubes are bridged and agglomerated between active particles [9]. Super-P was also agglomerated between active materials using a similar procedure [9]. The density of t-MWCNTs on the surface of active particles was higher than that of h-MWCNTs due to the larger number of nanotubes in the same weight density. T-MWCNTs were more tightly bound to the surface due to their floppiness, whereas h-MWCNTs often protruded from the LiCoO2 surface due to their stiffness. Fig. 3 shows the Zeta potential of acid-treated t-MWCNTs as a function of pH. The CNTs are well-dispersed in water because of the functional groups, such as hydroxyl and carboxyl groups, which are present on the surface of CNTs during acid treatment, as seen
Fig. 3. Changes in the Zeta potential of acid-treated t-MWCNTs (filled square) and raw LiCoO2 (open square). The vial shows the degree of dispersion and coagulation of t-MWCNTs and active particles.
for 40 cycles were carried out in a three-electrode beaker cell using a SOLARTRON impedance analyzer (SI1260). A beaker-cell stack assembly consisted of the electrode disk as a working electrode with the coated side facing an over-sized Li-counter electrode, and a small Li-foil reference electrode located at the uncoated side of the working electrode. The whole beaker cell was sealed airtight. Impedance measurements were carried out in the frequency range of 100 kHz to 10 mHz. 3. Results and discussion Fig. 1 shows Fourier-transformed infrared (FTIR) spectra of the electrode materials. The pristine t-MWCNTs did not show an appreciable number of residual functional groups, whereas the acid-treated t-MWCNTs showed several functional groups, such as carboxyl and hydroxyl groups. This suggests that some oxygen atoms were introduced that form strong bonds to the carbon network of nanotubes during acid treatment [16–18]. Those polar functional groups made the t-MWCNTs more hydrophobic and thus more accessible to water because of hydrogen bonding, which explains why the t-MWCNTs became water-soluble. Fig. 2 shows the morphology of (a) t-MWCNTs, (b) h-MWCNTs, and (c) Super-P coated on LiCoO2 . t-MWCNTs and h-MWCNTs
Fig. 4. Comparison of cycle performance at a rate of 1C. Square, triangle, and circle indicate Super-P, h-MWCNTs, and t-MWCNTs, respectively. Open and filled symbols indicate simple mixing and coagulation methods, respectively. The open rhombus indicates MWNTs [9].
Fig. 5. ac impedance plots of all the electrodes at 3.8–3.9 V vs. Li after cycling test: (a) full plots and (b) a blown-up view of the high-frequency part. (c) The change of each cathode film resistance after cycling.
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
5811
Fig. 6. SEM images of the electrodes after 40 cycles. (a) Super-P mixed cathode, (b) Super-P coated LiCoO2 , (c) h-MWCNTs mixed cathode, (d) h-MWCNTs coated LiCoO2 , (e) t-MWCNTs mixed cathode, and (f) t-MWCNTs coated LiCoO2 .
in Fig. 3. The negative Zeta potential over a wide range of alkali and weak acid regions results from the presence of functional groups such as hydroxyl and carboxyl groups. These negatively charged sites are protonated by protons in the strong acid region and thus neutralized. This eventually leads to heterocoagulation with LiCoO2 particles because of van der Waals interactions and subsequently to precipitation, as shown in the left vial of the inset in Fig. 3. Uniform coating of CNTs on the active particles (Fig. 2a and b) is in contrast with the previous simple mixing method, in which the CNT aggregates are bridged between active particles [9]. The advantage of this approach is that the heterocoagulated active LiCoO2 particles that have the least amount of conducting agent (CNTs in our case) can demonstrate high cycle performance.
Fig. 4 shows the cycle life for 40 cycles at a rate of 1C. All of the electrodes contained only 0.5 wt% conducting agent. The initial volumetric specific discharge capacity of the cathode with t-MWCNTs at a rate of 0.2C was 624 mAh cm−1 (equivalently 158 mAh g−1 ), which was the highest of all the samples. This value is larger than 546 mAh cm−1 , which was obtained from the 3 wt% cathode that was prepared by simple mixing [9]. This indicates that the electrode of uniformly coated t-MWCNTs on LiCoO2 particles is less swollen than other electrodes during electrolyte soaking. Further cycling was done at a rate of 1C, and the volumetric specific capacitance dropped rapidly independent of the samples. It is more intriguing that the degradation of the t-MWCNTs-coated electrode is minimal after 40 cycles compared with others. In particular, the specific
5812
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
capacitance of the t-MWCNTs-coated electrode is 403 mAh cm−1 after 40 cycles. This value is much higher than 310 mAh cm−1 , which is the specific capacitance of 3 wt% MWCNTs obtained by simple mixing [9]. The robustness of t-MWCNTs electrodes is due to the uniform coating of t-MWCNTs on LiCoO2 particles, which is shown in scanning electron microscope images in Fig. 2. The ac impedance of the different kinds of electrodes was measured in a three-electrode beaker-type cell at 3.8–3.9 V vs. Li after 40 cycles. The impedance curves in Fig. 5 consist of two rather distorted semicircles. The high-frequency semicircles might be associated with impedance due to particle-to-particle contacts in the electrode composite, such as cathode/electrolyte interfacial impedance. The low-frequency semicircles are associated with charge transfer impedance [19,20]. The second (larger) semicircle at low frequency corresponds to the open-circuit voltage (OCV) of the cell. In other words, the diameter of the semicircle decreases with increasing potential [21]. Coated t-MWCNTs show similar (or slightly smaller) resistance compared with mixed t-MWCNTs. The similar behavior was also confirmed by the smaller semicircle in Fig. 5, which indicates similar contact resistance between particles. The contact resistance and series resistance are only dependent on the types of nanotubes, not the coating method. Nevertheless, cycle testing showed differences in degradation, as shown in Fig. 4. Changes in the electrode morphology with cycling were studied carefully by inspecting the SEM images of the electrode surface to obtain more information about cycle performance and impedance changes. Some cracks or gaps between active materials were observed in the aged electrodes shown in Fig. 6. These cracks and air gaps after cycling seem to be the primary factor that enhances the resistance of the cathode. The agglomerated conducting materials within active materials are used as electrical conducting bridges in the early stage of cycling. As a result of prolonged cycle testing, the contraction/swelling of active materials cause electrode expansion. Cracks and gaps between particles developed despite the elasticity of the conducting network. A small amount of super-P (0.5 wt%) was insufficient to bridge all of the cracks independent of the coating method, as shown in Fig. 6a and b. The resiliency of the nanotube type-conducting network is favorable for maintaining electrical conductivity of the swelling electrode. Because h-MWCNTs have a larger diameter of about 100 nm, fewer MWCNTs participate in the bridging, as shown in Fig. 6c and d. We emphasize that coated t-MWCNTs are capable of forming bridges more efficiently than mixed t-MWCNTs, as shown in Fig. 6e and f. Greater floppiness of tMWCNTs is another advantage for maintaining the resiliency of the electrode during cycling compared with h-MWCNTs, particularly with a small amount of conducting agent.
4. Conclusion Our results show that CNTs can be easily coagulated onto the surface of LiCoO2 active particles. With the heterocoagulation method controlled by the isoelectric point, low weight percent of t-MWCNTs can be a constructed and used as conducting agents in high-density cathodes in Li-ion cells. A large specific surface area and the high flexibility of t-MWCNTs help the cathode form a strong adhesion and robust conducting network to maintain its electrical conductivity effectively during the cycling test. Our results show that heterocoagulation is a useful approach for the coating of carbon nanotubes uniformly on the surface of active particles, which reduces the expansion of electrodes and prohibits the entanglement of nanotubes. Acknowledgments This project was supported by KICOS through a grant provided by MOEST in 2007 (No. 2007-00202), by the STAR-faculty project, WCU program through the NRF funded by the MOEST (R31-2008000-10029-0), and the NRF through CNNC at SKKU. References [1] Y.P. Wu, E. Rahm, R. Holze, Electrochim. Acta 47 (2002) 3491. [2] J. Sarradin, N. Benjelloun, G. Taillades, M. Ribes, J. Power Sources 97–98 (2001) 208. [3] K. Du, H. Zhang, L. Qi, Electrochim. Acta 50 (2004) 211. [4] J. Kim, B. Kim, J. Lee, J. Cho, B. Park, J. Power Sources 139 (2005) 289. [5] Q. Lin, J. Electrochem. Soc. 151 (2004) A1115. [6] X. Li, F. Kang, W. Shen, Carbon 44 (2006) 1298. [7] X. Li, F. Kang, X. Bai, W. Shen, Electrochem. Commun. 9 (2007) 663. [8] Q.-T. Zhang, M.-Z. Qu, H. Niu, Z.-L. Yu, New Carbon Mater. 22 (2007) 361. [9] K.Y. Sheem, Y.H. Lee, H.S. Lim, J. Power Sources 158 (2006) 1425. [10] S. Dakuya, D. Masatoshi, M. Yoshikumi, Patent, JP 2001-00047891 (2001). [11] H.J. Jeong, K.K. Kim, S.Y. Jeong, M.H. Park, C.W. Yang, Y.H. Lee, J. Phys. Chem. B 108 (2004) 46. [12] J. Sun, L. Gao, Carbon 41 (2003) 1063. [13] R. Dominko, M. Gabersˇcˇek, J. Drofenik, M. Bele, J. Jamnik, Electrochim. Acta 48 (2003) 3709. [14] J. Fan, P.S. Fedkiw, J. Power Sources 72 (1998) 165. [15] R.S. Rubino, H. Gan, E.S. Takeuchi, J. Electrochem. Soc. 148 (2001) A10. [16] M. Aizawa, M.S.P. Shaffer, Chem. Phys. Lett. 368 (2003) 121. [17] Y. Liang, H. Zhang, B. Yi, Z. Zhang, Z. Tan, Carbon 43 (2005) 3144. [18] S.C. Ray, T.I.T. Okpalugo, P. Papakonstantinou, C.W. Bao, H.M. Tsai, J.W. Chiou, J.C. Jan, W.F. Pong, J.A. McLaughlin, W.J. Wang, Thin Solid Films 482 (2005) 242. [19] Y.M. Choi, S.I. Pyun, Solid State Ionics 99 (1997) 173. [20] D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi, H.J. Kim, Electrochim. Acta 47 (2002) 1899. [21] M.D. Levi, G. Salitra, B. Markovsky, D. Aurbach, U. Heider, L. Heider, J. Electrochem. Soc. 146 (1999) 1279.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrostatic heterocoagulation of carbon nanotubes and LiCoO2 particles for a high-performance Li-ion cell Kyeu Yoon Sheem a , Minseok Sung a , Young Hee Lee b,∗ a
Samsung SDI Corporate R&D Center, Gongse-dong, Giheung-gu, Yongin-si, Gyeounggi-do 446-577, Republic of Korea Department of Physics, Department of Energy Science, Center for Nanotubes and Nanostructured Composites, Sungkyunkwan Advanced Institute of Nanotechnology, Sungkyunkwan University, Suwon 440-746, Republic of Korea b
a r t i c l e
i n f o
Article history: Received 7 April 2010 Received in revised form 5 May 2010 Accepted 5 May 2010 Available online 13 May 2010 Keywords: Thin multiwalled carbon nanotubes Heterocoagulation Swelling Electrode density Specific capacity
a b s t r a c t Nanotubes were coated on the surface of active LiCoO2 particles using electrostatic heterocoagulation to enhance the electrochemical properties of a Li-ion battery. Only 0.5 wt% of multiwalled carbon nanotubes (MWCNTs) was added as a conducting agent into the LiCoO2 cathode, which had a density of 4.0 g cm−3 . We found that our electrode that was prepared using heterocoagulation with 0.5 wt% of thin MWCNTs maintained a volumetric capacitance of 403 mAh cm−1 after 40 cycles from the initial 624 mAh cm−1 , compared with previous result of 310 mAh cm−1 obtained from simple mixing with 3 wt% MWCNTs. The high volumetric capacity with smaller swelling using less amount of MWCNTs was attributed to the self-assembled nanotube network formed between active particles during coagulation, which was maintained with volume expansion during cycle testing. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction High-energy density and long cycle life for secondary batteries are critical because of the diverse market for portable electronic devices. Among the different types of batteries, rechargeable Li-ion batteries have been widely used in portable electronic devices. The Li-ion battery is composed of a positive cathode (mostly LiCoO2 ) and a negative anode (mostly graphite). Although Si or Sn nanoparticles have been known to exhibit high-energy density when used as anode material [1,2], their cycle life is poor because of structural instability and needs to be improved. Another way of improving energy density is to increase the density of the cathode electrode. In general, the cathode of a lithium battery is composed of active material, conducting agent, and binder [3]. More specifically, the cathode typically consists of 96 wt% of active material, 2 wt% of binder, and 2 wt% of carbon blacks, which are used as a conducting agent [4]. Reducing the amount of highly compressed electrode and conducting agent (non-active material) will improve the energy density of the cell. There have been several efforts to construct a high-density electrode using conventional LiCoO2 and carbon nanotubes as an alternative conducting agent instead of carbon black [5–8]. In past efforts, carbon nanotubes were mixed with the cathode material
∗ Corresponding author. Tel.: +82 31 299 6507; fax: +82 31 290 5954. E-mail address: [email protected] (Y.H. Lee). 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.05.027
and high-density electrodes were fabricated, resulting in increased energy density and power density due to the high conductivity of carbon nanotubes [9]. Although carbon nanotubes have been known to exhibit high electrical conductivity, the network formation among cathode particles has been limited by their aggregation. Furthermore, volume expansion during cycle testing degraded energy density, and thus, the cell became unstable and sometimes deteriorated. The overall electrochemical performance was dependent on the degree of dispersion of nanotubes in the cathode. Therefore, uniform distribution of a minimum amount of carbon nanotubes among oxide particles is required for cell stability. The aim of this work was to find a new method of forming an effective network of carbon nanotubes between LiCoO2 particles to enhance the energy density of the cathode electrode while preserving the conducting network of the cathode electrode with minimal volume expansion during cycling. We used 0.5 wt% of carbon nanotubes as a conducting agent with an electrode density as large as 4.0 g cm−3 compared with 2–3 wt% of carbon black and a cathode density of 3.2–3.84 g cm−3 [10]. Carbon nanotubes were uniformly coated on LiCoO2 particles at the isoelectric point. The carbon nanotubes were uniformly coated on the surface of active particles by heterocoagulation at the isoelectric point. This method is advantageous compared with the conventional mixing method, particularly by reducing the amount of conducting agent while maintaining the stability of the conducting network.
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
2. Experimental Because cell performance relies heavily on the dispersion of carbon nanotubes, we chose two types of nanotubes. Thin multiwalled carbon nanotubes (t-MWCNTs) were synthesized using chemical vapor deposition with FeMoMgO catalysts, which were prepared using a combustion method [11]. t-MWCNTs are composed of 2–6 walls and have a large specific surface area of 370 m2 g−1 . Hollow MWCNTs-100 (h-MWCNTs) were purchased from Nanokarbon Co., Korea. The diameter of h-MWCNTs is about 80 nm, and the specific surface area is about 30 m2 g−1 . Carbon black (Super-P) was purchased from MMM carbon, Belgium. The specific surface area of Super-P is about 60 m2 g−1 . The positive active material (LiCoO2 ), which had an average particle size of about 10 m, was purchased from Umicore Corp., Belgium. The samples were prepared using heterocoagulation and simple mixing [9,12]. To coagulate CNTs on active LiCoO2 particles, 0.5 wt% of gelatin was dissolved in 50 ml of deionized water and mixed with a known mass of LiCoO2 . The mixture was filtered and vacuum-dried at 110 ◦ C for 24 h. The surfaces of h-MWCNTs and t-MWCNTs are hydrophobic, thus, they are hardly soluble in water. Therefore, CNTs were dissolved in a mixture of nitric and sulfuric acid (volume ratio, 1:2) for 3 h and sonicated for 2 h. The CNTs were then filtered and washed with deionized water. The gelatin-coated active materials were then added to the CNT solution. To maximize the surface coating of active materials with CNTs, we titrated the solution with acetic acid to find the isoelectric point of the solution. The isoelectric point was located between pH 3 and 4. CNTs were coagulated on the surface of LiCoO2 and precipitated. After drying, the powder was heated up to 300 ◦ C under a N2 atmosphere to eliminate the gelatin and adsorbed water. For comparison, Super-P coated active materials were prepared by the conventional method [13]. To prepare electrodes, 2 wt% of polyvinylidene fluoride (PVDF) binder was dissolved in N-methyl-2-pyrrolidone (NMP), and then the active material (0.5 wt% of CNTs coated on 97.5 wt% of active material) was added. The manually mixed slurry was coated on only one side of an Al foil substrate and dried in a convection oven for 20 min at 120 ◦ C. Each conducting agent (t-MWCNTs, h-MWCNTs, and Super-P) and PVDF was simply mixed in NMP solution followed by mixing with LiCoO2 . The weight ratio of LiCoO2 , PVDF, and the conductive agent in the slurry was 97.5:0.5:2.0. By using this slurry, we obtained the positive electrode. The virgin electrode had rather low densities under 2.0 g cm−3 . These electrodes were further com-
Fig. 1. Absorbance of the functional groups on t-MWCNTs after acid treatment: (a) raw t-MWCNTs and (b) acid-treated t-MWCNTs.
5809
pressed with a rolling press to increase the density of the composite active mass up to 4.0 ± 0.05 g cm−3 . Compressing the electrode generally enhanced the electrical conductivity of the cathode [14,15]. Each electrode was cut into a circular disk with an area of 2 cm2 . The coin-type half-cells were assembled in an Ar-filled glove box. A porous polypropylene film was used as the separator. A 700 m thick Li-metal foil was used as the counter electrode. The electrolyte was 1 M LiPF6 solution in a 1:1 mixture (by volume) of ethylene carbonate (EC) and diethyl carbonate (DEC). Test cells were sealed in an ambient atmosphere after evacuation for 5 min to allow the electrolyte to homogeneously penetrate the electrode pores. Electrical testing was carried out using a Toyo Cycle Tester after aging the cells for 10 h. Initial charge–discharge cycles were carried out at a low rate of 0.2C for the initial cycle followed by cycling at a high rate of 1.0C for 40 cycles. ac impedance measurements after cycling tests
Fig. 2. SEM images of LiCoO2 particles coated by coagulation with (a) t-MWCNTs, (b) h-MWCNTs, and (c) Super-P.
5810
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
were uniformly coated on the surface of active materials without being aggregated, in contrast to the conventional simple mixing approach, in which the nanotubes are bridged and agglomerated between active particles [9]. Super-P was also agglomerated between active materials using a similar procedure [9]. The density of t-MWCNTs on the surface of active particles was higher than that of h-MWCNTs due to the larger number of nanotubes in the same weight density. T-MWCNTs were more tightly bound to the surface due to their floppiness, whereas h-MWCNTs often protruded from the LiCoO2 surface due to their stiffness. Fig. 3 shows the Zeta potential of acid-treated t-MWCNTs as a function of pH. The CNTs are well-dispersed in water because of the functional groups, such as hydroxyl and carboxyl groups, which are present on the surface of CNTs during acid treatment, as seen
Fig. 3. Changes in the Zeta potential of acid-treated t-MWCNTs (filled square) and raw LiCoO2 (open square). The vial shows the degree of dispersion and coagulation of t-MWCNTs and active particles.
for 40 cycles were carried out in a three-electrode beaker cell using a SOLARTRON impedance analyzer (SI1260). A beaker-cell stack assembly consisted of the electrode disk as a working electrode with the coated side facing an over-sized Li-counter electrode, and a small Li-foil reference electrode located at the uncoated side of the working electrode. The whole beaker cell was sealed airtight. Impedance measurements were carried out in the frequency range of 100 kHz to 10 mHz. 3. Results and discussion Fig. 1 shows Fourier-transformed infrared (FTIR) spectra of the electrode materials. The pristine t-MWCNTs did not show an appreciable number of residual functional groups, whereas the acid-treated t-MWCNTs showed several functional groups, such as carboxyl and hydroxyl groups. This suggests that some oxygen atoms were introduced that form strong bonds to the carbon network of nanotubes during acid treatment [16–18]. Those polar functional groups made the t-MWCNTs more hydrophobic and thus more accessible to water because of hydrogen bonding, which explains why the t-MWCNTs became water-soluble. Fig. 2 shows the morphology of (a) t-MWCNTs, (b) h-MWCNTs, and (c) Super-P coated on LiCoO2 . t-MWCNTs and h-MWCNTs
Fig. 4. Comparison of cycle performance at a rate of 1C. Square, triangle, and circle indicate Super-P, h-MWCNTs, and t-MWCNTs, respectively. Open and filled symbols indicate simple mixing and coagulation methods, respectively. The open rhombus indicates MWNTs [9].
Fig. 5. ac impedance plots of all the electrodes at 3.8–3.9 V vs. Li after cycling test: (a) full plots and (b) a blown-up view of the high-frequency part. (c) The change of each cathode film resistance after cycling.
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
5811
Fig. 6. SEM images of the electrodes after 40 cycles. (a) Super-P mixed cathode, (b) Super-P coated LiCoO2 , (c) h-MWCNTs mixed cathode, (d) h-MWCNTs coated LiCoO2 , (e) t-MWCNTs mixed cathode, and (f) t-MWCNTs coated LiCoO2 .
in Fig. 3. The negative Zeta potential over a wide range of alkali and weak acid regions results from the presence of functional groups such as hydroxyl and carboxyl groups. These negatively charged sites are protonated by protons in the strong acid region and thus neutralized. This eventually leads to heterocoagulation with LiCoO2 particles because of van der Waals interactions and subsequently to precipitation, as shown in the left vial of the inset in Fig. 3. Uniform coating of CNTs on the active particles (Fig. 2a and b) is in contrast with the previous simple mixing method, in which the CNT aggregates are bridged between active particles [9]. The advantage of this approach is that the heterocoagulated active LiCoO2 particles that have the least amount of conducting agent (CNTs in our case) can demonstrate high cycle performance.
Fig. 4 shows the cycle life for 40 cycles at a rate of 1C. All of the electrodes contained only 0.5 wt% conducting agent. The initial volumetric specific discharge capacity of the cathode with t-MWCNTs at a rate of 0.2C was 624 mAh cm−1 (equivalently 158 mAh g−1 ), which was the highest of all the samples. This value is larger than 546 mAh cm−1 , which was obtained from the 3 wt% cathode that was prepared by simple mixing [9]. This indicates that the electrode of uniformly coated t-MWCNTs on LiCoO2 particles is less swollen than other electrodes during electrolyte soaking. Further cycling was done at a rate of 1C, and the volumetric specific capacitance dropped rapidly independent of the samples. It is more intriguing that the degradation of the t-MWCNTs-coated electrode is minimal after 40 cycles compared with others. In particular, the specific
5812
K.Y. Sheem et al. / Electrochimica Acta 55 (2010) 5808–5812
capacitance of the t-MWCNTs-coated electrode is 403 mAh cm−1 after 40 cycles. This value is much higher than 310 mAh cm−1 , which is the specific capacitance of 3 wt% MWCNTs obtained by simple mixing [9]. The robustness of t-MWCNTs electrodes is due to the uniform coating of t-MWCNTs on LiCoO2 particles, which is shown in scanning electron microscope images in Fig. 2. The ac impedance of the different kinds of electrodes was measured in a three-electrode beaker-type cell at 3.8–3.9 V vs. Li after 40 cycles. The impedance curves in Fig. 5 consist of two rather distorted semicircles. The high-frequency semicircles might be associated with impedance due to particle-to-particle contacts in the electrode composite, such as cathode/electrolyte interfacial impedance. The low-frequency semicircles are associated with charge transfer impedance [19,20]. The second (larger) semicircle at low frequency corresponds to the open-circuit voltage (OCV) of the cell. In other words, the diameter of the semicircle decreases with increasing potential [21]. Coated t-MWCNTs show similar (or slightly smaller) resistance compared with mixed t-MWCNTs. The similar behavior was also confirmed by the smaller semicircle in Fig. 5, which indicates similar contact resistance between particles. The contact resistance and series resistance are only dependent on the types of nanotubes, not the coating method. Nevertheless, cycle testing showed differences in degradation, as shown in Fig. 4. Changes in the electrode morphology with cycling were studied carefully by inspecting the SEM images of the electrode surface to obtain more information about cycle performance and impedance changes. Some cracks or gaps between active materials were observed in the aged electrodes shown in Fig. 6. These cracks and air gaps after cycling seem to be the primary factor that enhances the resistance of the cathode. The agglomerated conducting materials within active materials are used as electrical conducting bridges in the early stage of cycling. As a result of prolonged cycle testing, the contraction/swelling of active materials cause electrode expansion. Cracks and gaps between particles developed despite the elasticity of the conducting network. A small amount of super-P (0.5 wt%) was insufficient to bridge all of the cracks independent of the coating method, as shown in Fig. 6a and b. The resiliency of the nanotube type-conducting network is favorable for maintaining electrical conductivity of the swelling electrode. Because h-MWCNTs have a larger diameter of about 100 nm, fewer MWCNTs participate in the bridging, as shown in Fig. 6c and d. We emphasize that coated t-MWCNTs are capable of forming bridges more efficiently than mixed t-MWCNTs, as shown in Fig. 6e and f. Greater floppiness of tMWCNTs is another advantage for maintaining the resiliency of the electrode during cycling compared with h-MWCNTs, particularly with a small amount of conducting agent.
4. Conclusion Our results show that CNTs can be easily coagulated onto the surface of LiCoO2 active particles. With the heterocoagulation method controlled by the isoelectric point, low weight percent of t-MWCNTs can be a constructed and used as conducting agents in high-density cathodes in Li-ion cells. A large specific surface area and the high flexibility of t-MWCNTs help the cathode form a strong adhesion and robust conducting network to maintain its electrical conductivity effectively during the cycling test. Our results show that heterocoagulation is a useful approach for the coating of carbon nanotubes uniformly on the surface of active particles, which reduces the expansion of electrodes and prohibits the entanglement of nanotubes. Acknowledgments This project was supported by KICOS through a grant provided by MOEST in 2007 (No. 2007-00202), by the STAR-faculty project, WCU program through the NRF funded by the MOEST (R31-2008000-10029-0), and the NRF through CNNC at SKKU. References [1] Y.P. Wu, E. Rahm, R. Holze, Electrochim. Acta 47 (2002) 3491. [2] J. Sarradin, N. Benjelloun, G. Taillades, M. Ribes, J. Power Sources 97–98 (2001) 208. [3] K. Du, H. Zhang, L. Qi, Electrochim. Acta 50 (2004) 211. [4] J. Kim, B. Kim, J. Lee, J. Cho, B. Park, J. Power Sources 139 (2005) 289. [5] Q. Lin, J. Electrochem. Soc. 151 (2004) A1115. [6] X. Li, F. Kang, W. Shen, Carbon 44 (2006) 1298. [7] X. Li, F. Kang, X. Bai, W. Shen, Electrochem. Commun. 9 (2007) 663. [8] Q.-T. Zhang, M.-Z. Qu, H. Niu, Z.-L. Yu, New Carbon Mater. 22 (2007) 361. [9] K.Y. Sheem, Y.H. Lee, H.S. Lim, J. Power Sources 158 (2006) 1425. [10] S. Dakuya, D. Masatoshi, M. Yoshikumi, Patent, JP 2001-00047891 (2001). [11] H.J. Jeong, K.K. Kim, S.Y. Jeong, M.H. Park, C.W. Yang, Y.H. Lee, J. Phys. Chem. B 108 (2004) 46. [12] J. Sun, L. Gao, Carbon 41 (2003) 1063. [13] R. Dominko, M. Gabersˇcˇek, J. Drofenik, M. Bele, J. Jamnik, Electrochim. Acta 48 (2003) 3709. [14] J. Fan, P.S. Fedkiw, J. Power Sources 72 (1998) 165. [15] R.S. Rubino, H. Gan, E.S. Takeuchi, J. Electrochem. Soc. 148 (2001) A10. [16] M. Aizawa, M.S.P. Shaffer, Chem. Phys. Lett. 368 (2003) 121. [17] Y. Liang, H. Zhang, B. Yi, Z. Zhang, Z. Tan, Carbon 43 (2005) 3144. [18] S.C. Ray, T.I.T. Okpalugo, P. Papakonstantinou, C.W. Bao, H.M. Tsai, J.W. Chiou, J.C. Jan, W.F. Pong, J.A. McLaughlin, W.J. Wang, Thin Solid Films 482 (2005) 242. [19] Y.M. Choi, S.I. Pyun, Solid State Ionics 99 (1997) 173. [20] D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi, H.J. Kim, Electrochim. Acta 47 (2002) 1899. [21] M.D. Levi, G. Salitra, B. Markovsky, D. Aurbach, U. Heider, L. Heider, J. Electrochem. Soc. 146 (1999) 1279.
