Enhanced Cycle Life of Si Anode for Li-Ion Batteries by Using Modified ...

Electrochemical and Solid-State Letters, 8 共2兲 A100-A103 共2005兲

A100

1099-0062/2004/8共2兲/A100/4/$7.00 © The Electrochemical Society, Inc.

Enhanced Cycle Life of Si Anode for Li-Ion Batteries by Using Modified Elastomeric Binder Wei-Ren Liu,a Mo-Hua Yang,b Hung-Chun Wu,b S. M. Chiao,c and Nae-Lih Wua,*,z a

Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan 106 Materials Research Laboratories, Industrial Technology Research Institute Chutung, Hsin-Chu, Taiwan 310 c Department of Chemical Engineering, Tung Hai University, Taichung, Taiwan 407 b

Cycle-life of the particulate electrode of Si, either with or without carbon coating, for Li-ion battery has significantly been improved by using a modified elastomeric binder containing styrene-butadiene-rubber 共SBR兲 and sodium-carboxyl-methylcellulose 共SCMC兲. Compared with poly-vinylidene-fluoride 共PVdF兲, the (SBR ⫹ SCMC) mixture binder shows smaller moduli, a larger maximum elongation, a stronger adhesion strength on Cu current collector, and much smaller solvent-absorption in organic carbonate. There were demonstrated cycle lives of ⬎50 cycles for bare Si at 600 mAh/g or carbon-coated Si at 1000 mAh/g, as contrast to ⭐8 cycles for PVdF-bound electrode in all cases. © 2004 The Electrochemical Society. 关DOI: 10.1149/1.1847685兴 All rights reserved. Manuscript submitted July 28, 2004; revised manuscript received September 27, 2004. Available electronically December 16, 2004.

Being an anode for Li-ion batteries, Si possesses a maximum capacity exceeding 3000 mAh/g.1,2 However, this material exhibits dramatic volume expansion and shrinkage during lithiation and delithiation, respectively, and this volumetric variation has resulted in very poor cycle life. Several approaches have so far been known for tackling this crucial problem. One approach, is to use the thin-film type of Si anode, prepared mainly by vacuum sputtering.3,4 By way of either tailing the microstructures of the film and film-substrate interfaces3 or using films of amorphous Si-containing alloys,4 good cycling stability at high charge/discharge 共C/D兲 capacities has been demonstrated. However, vacuum sputtering is costly, and this thinfilm approach may not be commercially viable for large-scale applications. On the other hand, by adopting the conventional thick-film electrode made of particulate materials, several studies5-9 have coated the Si particles with various conducting materials, which not only serve to enhance the conductivity of the electrode but also could act as a buffer to partially accommodate the volumetric variations during C/D cycling. However, except for using graphitic coating,5,6 which has demonstrated some beneficial effect on cycle life, most of these attempts have shown very limited success. We have systematically looked into the effects of the typical constituents of a thick-film electrode on cycle-life of Si anode. We believe that optimizing the properties of these constituents will remain fundamental and crucial to the performance of Si anode, even with or without introduction of coating material. Among others, the binder material has been found to play a very important role in stabilizing the electrode. As reported here, very pronounced improvement in cycle life of thick-film electrode of Si, either with or without carbon-coating, was observed by using a modified elastomeric binder, namely, a mixture of styrene butadiene rubber 共SBR兲 and sodium carboxyl methyl cellulose 共SCMC兲, which exhibited very different mechanical properties and swelling behavior in electrolyte solvent from poly-vinylidene fluoride 共PVdF兲.

The Si electrodes constitutes, on the dry basis, 62 wt % active material 共i.e., Si or C-Si兲, 30 wt % conductive additives and 8 wt % binder. The elastomeric binder is a mixture of SBR 共L1571, Asahi Chemicals兲 and SCMC 共WS-C, Cellogen, DKS International, Inc.兲 with 1:1 weight ratio. SBR was used as purchased in the form of an aqueous emulsion solution, and SCMC was added to act as the thixotropic agent. The mixture will hereafter be referred as the SBR mixture binder, and the electrodes bound by this mixture as the SBR-electrodes, for brevity. The PVdF-bound electrodes 共referred as the PVdF-electrodes兲, on the other hand, use solely PVdF as the binder material, and N-Methyl Pyrrolidone 共NMP, Mitsubishi Chemical兲 was used as the solvent. The conductive additives consist of graphitic flakes 共KS6, 3 ␮m, Timcal兲 and nanosized carbon black 共Super P, 40 nm, Timcal兲 with a weight ratio of 5:1. The electrodes had a thickness of 45-50 ␮m and contained 3 mg of the active material (Si or C-Si) per cm2 of electrode. CR2032 coin cells were fabricated from the Si-containing electrodes, Li foils as the counter electrode and a commercial electrolyte of 1 M LiPF6 in ethylene carbonate 共EC兲: ethyl methyl carbonate 共EMC兲共vol % of 1:2兲. All the electric potential values reported herein are referenced to Li/Li⫹. A typical C/D cycle consists of the following steps: discharge 共lithiation兲 at a constant-current 共CC兲 of 0.3 mA/mg until either the designated specific capacity 共mAh/g兲 or the zero-potential bound is reached, and charge 共delithiation兲 at 0.3 mA/mg until the cutoff voltage 共1.2 V兲 is reached. An equilibrating period of 5 h was adopted at the end of the first discharging. The

Experimental Si powder 共99.9%, Aldrich兲 was ballmilled in a PE jar containing two different Al2 O3 balls 共5 and 2 mm diam, respectively兲 at rotation speed of 300 rpm for 24 h. The milled powder has a bimodal particle size distribution 共Fig. 1兲 having an volume average size of ⬃3 ␮m and contains less than 15% of the particles larger than 5 ␮m. Carbon-coated Si 共C-Si兲 powder was synthesized by passing benzene vapor through the milled Si powder at 900°C. The coating has a d002 of 0.347 nm and is 27% by weight.

* Electrochemical Society Active Member. z

E-mail: [email protected]

Figure 1. Particle size distribution of the silicon powders employed in the present study.

Electrochemical and Solid-State Letters, 8 共2兲 A100-A103 共2005兲 designated capacity 共mAh/g兲 is calculated based on the weight of the active material, either Si or C-Si, in the electrode. However, the graphitic conductive additive K26 could also participate the lithiation process, and the true lithiation capacity of the active material will be smaller than the designated value. Based on a lithiation capacity of 370 mAh/g for graphite, the maximum contribution from the graphitic component is 150 mAh/g of the active material. Solution-cast samples were prepared for mechanical and swelling tests of the binder materials. The SBR 共containing SCMC兲 emulsion was cast under 80°C in vacuum to remove humidity. The synthesized SBR sheets were further hot-pressed under 110°C and 30 kgf/cm2 for 30 min. NMP was used to dissolve PVdF, and was subsequently removed under 110°C in vacuum. PVdF samples were then hot pressed under 120°C and 50 kgf/cm2 for 30 min. The dynamic mechanical spectra were recorded in the three-point bending mode using a DMA7e dynamic mechanical analyzer 共Perkin Elmer兲 at 1Hz. The storage moduli, loss moduli, and loss tangent were recorded as functions of temperature, which ramped from ⫺50°C to ⫹50°C in 2°C/min. The tensile test followed ASTM D412 using Instron 4467. The crosshead speed was 500 mm/min. The compatibility of the binder systems with electrolyte solvent was examined by the swelling test. Binder sheets prepared as described above were placed in propylene carbonate 共PC兲 at either room temperature or 80°C. PC is used because it is available to us. However, because it has the same major functional group (⬎O2 C ⫽ O) as the electrolyte solvent (EC ⫹ EMC) used in the coin-cell test, we believe the results obtained from the former can be extended to the latter. Weight measurements were made by blotting the samples dry and immediately weighting them. The swelling solvent was then removed by heating the sample at 70°C under vacuum until the equilibrium weight was achieved. The swelling ratio was defined as the weight ratio of the amount of solvent absorbed to the dry weight of the tested binder sheet. The adhesion strength between the Si-containing overlay and Cu current collector of the electrode was investigated by a peeling test. In this test, a 3M tape was pasted onto the surface of the overlay with a covering area of 1.8 ⫻ 1.8 cm, and pulled at a speed of 4.5 cm/min. The adhesion strength was recorded 共Digital Force Gauge, DPX-1TR, IMADA Corp.兲 as the maximum force occurring during the course of peeling. It was confirmed that de-bonding takes place between the overlay and Cu in all the peeling tests. The resistance across the electrode (overlay ⫹ Cu) was also measured by using a four-point method. It is clear that the resistance is contributed mainly by the overlay.

A101

Figure 3. The potential curves for 共a兲 the first-fifth cycles of the PVdFelectrode, and 共b兲 the third-twentieth cycles, SBR-electrode, at the designated discharge capacity of 600 mAh/g.

Figure 2 summarizes the cycling capacity data versus cycle number with a designated discharge capacity of 600 mAh/g, while Fig.

3a and b show the potential curves for selected cycles for the PVdFand SBR-electrodes, respectively. For the PVdF-electrode, the designated discharge capacity could not be reached and the discharge process was cut-off at zero-potential. Upon charge, there exist large irreversible capacities, which in general increase with increasing cycle number until the eighth cycle when the electrode became completely un-chargeable 共Fig. 2兲. The potential curves of the PVdFelectrode 共Fig. 3a兲 showed increasing polarization with cycling. This is an indication that the electrode structure was getting loose and electric contact was increasingly poor between particles.

Figure 2. Cycling performance at the discharge capacity of 600 mAh/g. 共Open symbols: discharge capacity; solid symbols: charge capacity.兲

Figure 4. The differential capacity plots (⫺dQ/dV) as a function of potential (V) for the PVdF- and SBR-electrodes.

Results and Discussion


A102

Figure 6. Storage modulus vs. temperature for PVdF and SBR. Figure 5. Charge capacities at the designated discharge capacities of 800 mAh/g for PVdF- and SBR-Si anodes; 1000 mAh/g for SBR-bound C-Si anode.

For the SBR-electrode, it in some cases took a couple of induction cycles to reach the full designated discharge capacity 共Fig. 2兲, and the columbic efficiency reaches 98% at the fifth cycle. The total irreversible capacities during the ‘‘induction’’ phase are less than 10 mAh/g in all cases. The SBR-electrode retained 95% capacity up to 70 cycles. Its potential curves 共Fig. 3b兲 showed smaller polarization than the PVdF-electrode even for initial cycles and exhibited good reversibility. The enhancement in cycle stability of the Si anode by replacing PVdF with SBR as the binder is evident. The differential capacity plots 共Fig. 4兲 showed essentially the same redox events for both electrodes. They include a broad reduction 共lithiation兲 peak within 0.1 ⬃ 0.3 V and two major oxidation 共de-lithation兲 peaks within 0.15-0.6 V. 共An oxidation shoulder near 0.2 V observed for both electrodes corresponds to de-lithiation from the graphitic component of the conductive additives.兲 The differences in intensity of the redox peaks between the two electrodes are associated with the kinetics of the redox processes, which are in turn affected by several factors, including the type and amount of the binders, the Si particle size, film thickness, and so on. Detailed discussion on the characteristics of these redox events will be published, along with results from other electrochemical analyses, in a future publication. It nevertheless suffices to say here that the SBR mixture binder employed in the present study is electrochemically stable and does not cause any electrochemical reaction other than lithiation and de-lithiation of Si. Figure 5 shows the C/D capacity data with designated discharging capacity of 800 mAh/g for Si electrode and 1000 mAh/g for C-Si, respectively. For bare Si, the cycle life decreases slightly when the discharge capacity increased from 600 to 800 mAh/g. Coating with a carbon surface layer, nevertheless, further improves the stability, and a cycle life (⬎97% retention兲 ⬎50 cycles was obtained at 1000 mAh/g. A cycle life ⬍8 cycles was for the PVdF electrode in either case. A coating material would affect the interfacial properties between the binder and particles, and these results indicate that the SBR mixture binder also works well with carbon-coated Si particles. PVdF is a thermoplastic material while SBR is a typical elastomer. Their mechanical properties are very different. Storage

moduli obtained from the DMA are shown in Fig. 6. Temperature ranges from ⫺50°C to ⫹50°C. The SBR mixture binder has much smaller moduli, and hence is much softer than PVdF. The difference in storage modulus between these two binder materials reaches as large as by two orders at room temperature. A significant drop on the storage modulus was found for the SBR mixture binder at about 0°C 共Fig. 6兲, which can be attributed to a first-order phase transition.11 On the other hand, a slight depression in the storage modulus was for PVdF at around ⫺40°C 共Fig. 6兲, which corresponds to a second order phase transition.12 Results of the tensile tests are summarized in Table I. Both the Young’s modulus and tensile strength of PVdF are greater than those of the SBR mixture binder, as expected, but the elongation of PVdF at break is smaller. The effect of SCMC on SBR is also noticed. That is, the Young’s modulus measured here is much greater than the one obtained from the dynamic tests 共Fig. 6兲, and is caused by the domination of SCMC in the tensile mode. In addition, the elongation of the SBR ⫹ SCMC mixture binder shown here is smaller than that of pure SBR. Finally, peeling tests showed that the SBRbound overlay exhibited an adhesion-strength on the Cu substrate almost 30 times that of the PVdF-bound overlay 共Table I兲. The swelling ratios of the SBR mixture binder were found to be all below 1% 共Fig. 7兲, which indicates that there is basically no solvent absorbed by SBR either at room temperature or 80°C. On the other hand, although we did not see dissolution of PVdF, it took up the solvent 共PC兲 up to 60% of its own weight. When the solvent swells into the binder matrix, it can easily migrate into the interface region between the binder and Si particles. The adhesion between the binder and silicon may thus be weakened, or even destroyed. Similar adverse effect would also occur at the interface between the binder and Cu current collector. This might contribute in part to the very large difference in adhesion behavior observed between the two sets of electrodes 共Table I兲. Although the microstructures of the electrodes have yet been studied in detail, the very different cycling stabilities exhibited by the SBR- and PVdF-electrodes may be understood in part in terms of the differences in the macroscopic mechanical properties of the binder materials. For instance, because the SBR mixture binder has a smaller modulus than PVdF, it allows the electrode to expand and shrinkage more easily and hence reduces the forces exerted between

Table I. Mechanical and electrical properties of Si electrodes. Electrode type SBR-Si PVdF-Si

Binder SBR/SCMC ⫽ 1:1 PVdF

Tensile strength 共Mpa兲

Young’s modulus 共Mpa兲

Maximum elongation 共%兲

Adhesion strength 共Kgw/1.8 cm兲

Electrode resistance 共m⍀兲

24.82

137.6

14.40

0.117

0.0427

40.08

441.4

10.15

0.004

0.261


Figure 7. Swelling ratio vs. swelling time in PC for PVdF and SBR at two different temperatures: 25 and 80°C.s: 25 and 80°C.

particles. These results are in accordance with one earlier suggestion made by Chen et al.10 that cycling stability of the electrodes, such as Si and Sn, showing large volume changes might benefit from increasing the elasticity of the binder material. Because the SBR mixture binder show a greater adhesion strength and larger breakage elongation, it can stick to the particles and current collector to withstand a greater stress before de-bonding and to tolerate a greater extent of volume expansion with reversibility. All of these considerations point to improved cycle life. In addition, the binders also require different solvents for preparing the coating pastes, which have different rheological properties. The water-based paste in the case of the SBR mixture binder might have provided favorable rheological properties that result in an electrode architecture of lower resistance, by, for instance, enabling a more uniformly distributed conductive additives between Si particles. This may explain the observation that the SBR-electrode has a lower resistance than the PVdF-electrode even in the fresh state 共Table I兲. The uniformity in conductivity within the electrode to start with could also help to enhance the stability of the electrode upon cycling. While the enhancement in the cycle life of Si anode by replacing PVdF with a SBR-SCMC mixture binder has clearly been demon-

A103

strated in the present work, it is also understood that neither the binder nor the electrode architecture has yet been optimized. For instance, it was found in some cases among the SBR-electrodes that it takes a couple of initial ‘‘induction’’ cycles before the electrodes begin to show large charging capacity. The cycle number of the ‘‘induction’’ phase was found to decrease with decreasing C/D rate. In addition, it can be avoided by either using less binder 共e.g., 4 wt %兲 or reduce the film thickness, say to 30 ␮m. The possible explanation for the induction cycle is that the surfaces of Si particles in the 8% binder electrode were largely covered with the binder, which hindered the access of electrolyte to the Si surfaces. Pores might have been created during initial cycles. Trade-off may be needed between high cycling stability and avoidance of the induction phenomenon. More studies are currently underway to understand the effects due to the composition 共SBR/SCMC ratio兲 and amount of the binder, along with other nonbinder factors, such as Si particle size and conductive additive content, and to understand the interplay between the cycle life performance and the various mechanical properties of the binder material. National Taiwan University assisted in meeting the publication costs of this article.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

C. J. Wen and R. A. Huggins, J. Solid State Chem., 37, 271 共1981兲. M. N. Obrovac and L. Christensen, Electrochem. Solid-State Lett., 7, A93 共2004兲. H. Yagi and H. Tarui, US Pat. 6,649,033 B2 共2003兲. L. Y. Beaulieu, T. D. Hatchard, A. Bonakdarpour, M. D. Fleischauer, and J. R. Dahn, J. Electrochem. Soc., 150, A1457 共2003兲. T. Umeno and K. Fukuda, Eur. Pat. 1024544A2 共2000兲. M. Yoshio, H. Wang, K. Fukuda, T. Umeno, N. Dimov, and Z. Ogumi, J. Electrochem. Soc., 149, A1598 共2002兲. II-seok Kim, G. E. Blomgren, and P. N. Kumta, Electrochem. Solid-State Lett., 6, A157 共2003兲. G. A. Robert, E. J. Cairns, and J. A. Reimer, J. Power Sources, 110, 424 共2002兲. G. X. Wang, L. Sun, D. H. Bradhurst, S. Zhong, S. X. Dou, and H. K. Liu, J. Am. Ceram. Soc., 306, 249 共2000兲. Z. Chen, L. Christensen, and J. R. Dahn, J. Electrochem. Soc., 150, A1073 共2003兲. D. R. St. CYR, Encyclopedia of Polymer Science and Engineering, 2nd ed., Vol. 14, J. I. Kroschwitz, Editor, John Wiley & Sons, New York 共1985兲. J. E. Dohany and J. S. Humphery, Encyclopedia of Polymer Science and Engineering, 2nd ed., Vol. 17, J. I. Kroschwitz, Editor, John Wiley & Sons, New York 共1985兲.