Interaction between Poly(vinylidene fluoride) Binder and Graphite in the Anode of

Mat. Res. Soc. Symp. Proc. Vol. 730 © 2002 Materials Research Society V5.17.1 Interaction between Poly(vinylidene fluoride) Binder and Graphite in the Anode of Lithium Ion Batteries: Rheological Properties and Surface ChemistryMikyong Yoo, a Curtis W. Frank, a,ba Department of Materials Science and Engineering andb Department of Chemical Engineering, Stanford University, Stanford, CA 94305ABSTRACTWe describe here the interaction of poly(vinylidene fluoride) (PVDF) with graphite basedon the rheological behavior of the slurries and the surface morphology of PVDF in the final composite anode. The rheological properties of slurries show the interaction between graphiteand PVDF with the viscosity varying over six orders of magnitude for different graphiteparticles. We correlated the suspension viscosity with the final film properties. The homogeneityof the PVDF distribution in the final composite film increases as the slurry viscosity increases.The interaction between graphite and PVDF is altered through the chemical properties ofpolymer such as molecular weight and functionality, leading to an improved morphology ofPVDF.INTRODUCTIONIn spite of the recent advances in lithium ion rechargeable batteries, there remainchallenging problems related to the fabrication of the anode and cathode, which are bothcomposite materials. The anodes consist predominantly of graphite particles (90-98% by weight) bound together by a polymeric binder such as poly(vinylidene fluoride) (PVDF). AlthoughPVDF is the most promising, other fluorinated binders have been used, includingpoly(tetrafluoro ethylene) (PTFE) [1,2] as well as non-fluorinated polymers such aspoly(ethylene propylene diene) (EPDM) [3,4]. The polymeric binder, which is necessary toprovide sufficient mechanical strength to the electrodes, can influence the electrochemicalreaction between lithium ions and electrolyte that takes place at the carbon surface, since it isknown that the details of the surface chemistry and the morphology of the carbon play importantroles in SEI formation [3,5], and that the binder influences heat generation in the lithium ion batteries [4,6].Moreover, it has been reported that up to 70% of the graphite surface was coveredby PVDF binder even though the concentration of binder was kept relatively low [7].Therefore,it is important to understand and control the binder/particle interactions and the relationshipbetween the processing parameters for the complex fluid suspension and the final composites.In this paper, we will examine binder interaction with graphite using different types ofgraphite mixed with PVDF binder. The PVDF binder/graphite interaction is manifested in the rheological properties in the high-solids-content slurries. Because such slurries are complexfluids with their rheological properties important for determining the stability of the suspension,their study is closely related to the processing of composite electrodes. And we will present the morphology of binders on the final composite anode and correlate the interaction betweengraphite and binder of the slurries with the surface properties of final composite films. This interaction and the surface distribution of PVDF are altered through chemical properties ofPVDF such as molecular weight and modified functionality.EXPERIMENTAL DETAILSWe used eight types of carbon materials, and three different PVDF, as shown in Table I, to investigate the effect of different surface area and nature of carbon and the chemical properties of PVDF on the interaction between carbon particles and PVDF. Anodes were prepared by mixing carbon slurries that contained the carbon particle, 1-methyl-2-pyrrolidinone (NMP,99.9+%, Mitsubishi Chemical Co.) as a carrier, and 10 wt% solution of PVDF binder in NMP solvent. The solid concentration of the slurries was 40 wt%. We spread the slurry using the doctor-blade method on a sheet of copper foil and dried it in an oven at 83 ºC in air for 1 hour to form 95/5 wt% of graphite/PVDF composite anode.Dynamic viscosities of the slurries were measured at 23°C by a stress-controlled rheometer (DSR, Rheometric Scientific) for MPG-V2 and MBC-N and a strain-controlled rheometer (Dynamic analyzer RDA II, Rheometric Scientific) for the other samples. Two rheometers were required because the viscosities of MPG-V2 and MBC-N slurries were too low for the range of the strain-controlled rheometer. We used a 50 mm parallel plate with a 1 mm gap, and samples were prepared just before the experiments. The viscosity was measured as a function of the frequency in the range of 0.1 and 100 rad·sec-1 at the maximum strain amplitude of the linear viscoelastic region.The spatial distribution of PVDF on the graphite particles was evaluated by mapping the fluorine of PVDF using electron probe X-ray microanalysis (EPMA) (JXA-733, JEOL) with 50 nA of current. In the case of different PVDF samples, energy dispersive spectroscopy (EDS, JSM-5600LV, JEOL) was used with a gun voltage of 10 kV. We determined the degree of homogeneity of PVDF by quantifying the image of fluorine dots using a spatial autocorrelation function. From the characteristics of the autocorrelation function, we were able to calculate the area of a fluorine dot cluster and the spacing between clusters. From these values, we derived the degree of homogeneity.Table I. Information on carbon particles and PVDF used for the experiments.Manufacturer Type of carbon Averageparticlesize(µm) BET surface area (m2/g)MPG-V2 Mitsubishi Chemical Co. Synthetic graphite 18.4 2.8 MBC-N Mitsubishi Chemical Co. Amorphous carbon 18.0 5.0 MCMB Oosaka Gas Chem. Synthetic graphite 7.8 3.3 SFG44 Timcal Co. Ltd. Synthetic graphite 22.0 5.0 SFG75 Timcal Co. Ltd. Synthetic graphite 27.0 3.5 KS15 Timcal Co. Ltd. Synthetic graphite 7.7 12.0 SFG15 Timcal Co. Ltd. Synthetic graphite 8.1 8.8 KS6 Timcal Co. Ltd. Synthetic graphite 3.3 20.0Manufacturer Functionality Molecularweight H-H defects(%) (19F NMR)KF1300 Kureha N/A ~ 350,000 4.22 (± 0.53) MKB212A Atofina -OH(a little – COOH)~ 350,000 5.54 (± 0.67) Kynar301F Atofina N/A ~ 500,000 6.55 (± 1.24)1.0E-011.0E+011.0E+031.0E+051.0E+071.0E-01 1.0E+00 1.0E+01 1.0E+02Frequency (rad/s, log scale)R e a l V i s c o s i t y (N -s /m 2, l o g s c a l e )RESULTS AND DISCUSSIONRheological properties of slurriesTo investigate the interaction between carbon particles and PVDF before doctor-blading, we measured the viscosity of the slurries. Figure 1 shows the frequency dependence of the real viscosity, which shows over six orders of magnitude difference. We used the real viscosity here (loss modulus divided by frequency) instead of the dynamic viscosity because the storage moduli of MPG-V2 and MBC-N showed negative values in the high frequency region, which does not have any physical meaning. For MPG-V2 and MBC-N, the real viscosities are nearlyindependent of frequency on a log-log scale, which indicates that these behave as Newtonian interfaces. On the other hand, the viscosities of the others decrease linearly as frequency increases. This shear thinning is typical non-Newtonian behavior. In concentratedparticle/polymer suspensions, the shear thinning occurs due to structural breakdown. When a polymer adsorbs at a solid-liquid interface, a portion of the polymer loop or tail may be extended from one particle to another particle. Therefore, flocculation of suspensions may be modeled by a bridging process [8]. In a developed three-dimensional network of flocs, structural breakdown is induced with increasing shear rate, leading to shear thinning behavior.High BET surface area of graphite particles generally causes the high interaction between graphite and PVDF through more adsorption area. The influence of reactive sites on theinteraction can be examined by using a modified PVDF. When we used a PVDF modified by the hydroxyl functional group, which has the same molecular weight as that of KF1300, the dynamic viscosity increases in terms of the same graphite as shown in Fig. 2. This indicates that thealtered interaction between graphite and PVDF through the hydroxyl functional group enhances the adsorption of PVDF onto the graphite particles. The higher molecular weight PVDF,Kynar301F, causes the viscosities to be increased because high molecular weight polymer can bridge more particles together with longer chains, leading to a more flocculated particlestructure. PVDF is adsorbed onto the surfaces of graphite through hydrogen bonding [9] and physisorption, and loops and tails can bridge more than one particle together, leading toaggregated particles with high viscosity.Figure 1. Dependence of real viscosity on frequency for slurries at room temperature. Viscosities are significantly different over six orders of magnitude. Whereas MPG-V2 and MBC-N showcharacteristics of a Newtonian fluid, other samples show non-Newtonian shear thinning.Figure 2. Viscosity influenced by the chemical properties of PVDF in terms of graphite particles. The high molecular weight PVDF (Kynar301F) and the modified one (MKB212A) show a higher viscosity.Surface distribution of PVDFThe location and morphology of PVDF on the carbon can be observed by detecting fluorine with EPMA because PVDF has a repeating unit of –(CH 2-CF 2)–. Since the depth of field of EPMA is about 1 µm, we cannot detect the exact surface morphology of PVDF, but rather we determine a near -surface morphology. Figure 3 (a)-(f) show fluorine dot mappings of the composite films with different carbon materials. These images are shown in the order ofincreasing slurry viscosity (see Fig. 1), illustrating that the distribution of PVDF in the final composite film becomes more homogeneous with increasing viscosity. We used different magnifications to minimize the influence of carbon particle size on the images of apparentpolymer distribution. We also used the normalized value considering the magnification and the average particle size. We quantified the degree of homogeneity by using the autocorrelation function, C(R), which describes the spatial correlation of the density of dots in an image and isFigure 3. Fluorine dot mapping of (a) MBC-N, (b) MCMB, (c) SFG44, (d) KS15, (e) SFG15, and (f) KS6. These images are shown in the order of increasing viscosity. 120 pixel corresponds to 20 µm for (a) and 10 µm for (b), (c), (d), and (f). For (e), 116 pixel corresponds to 10 µm. (a) (b) (c)(d) (e) (f) 1.0E+001.0E+011.0E+021.0E+031.0E+041.0E+051.0E-01 1.0E+00 1.0E+011.0E+02Frequency (rad/s, log scale)V i s c o s i t y (N -s /m 2, l o g s c a l e )defined as the average normalized by C(0) [10],)0(C ])R r (][)r ([)R (C o o >−−−<=ρρρρ (1) where ρ is a dot density distribution, and the angular brackets indicate averaging over allcoordinates r. When r is equal to 0, there is a strong self-correlation peak corresponding to each image correlating with itself. When r is equal to the mean displacement, there is a second peak called a positive displacement peak, which corresponds to the first image correlating with the second image. From this characteristic of the autocorrelation function, we can evaluate the area of a cluster and the spacing between clusters. The two-dimensional image was transformed to the one-dimensional spatial peak as shown in Fig. 4(a). The area under the first peak indicates the size of a cluster and the distance to the second peak corresponds to the radial average spacing between two clusters. We plotted the spacing versus the average size of clusters of each sample in Fig. 4(b); small cluster size and reduced distance between clusters of dots indicatehomogeneous distribution. The data approaches a small cluster size and spacing as the viscosity increases. From these data, we relate the processing condition such as viscosity to the surface properties of final films.Figure 5 shows the effect of the molecular weight and functionality of PVDF on the surface distribution of PVDF. We plotted the values for each graphite and PVDF and connected points in terms of PVDF. MKB212A shows an improved homogeneity compared to KF1300 in all carbon particles through enhanced interaction with graphite particles. In the case of Kynar301F, better homogeneity of PVDF is detected for some carbon samples, which can be caused by long chains of high molecular weight PVDF that can be more adsorbed onto graphite particles. Thehomogeneity range of Kynar301F PVDF is small compared to other PVDF materials because of the aggregation of long chains of PVDF after the adsorption of PVDF onto the surface of101010102103104105(N -s /m 2, a t 1 r a d /s , l o g s c a l e )R a d c i n g b e t w e e n C l s t e r s (µm )r e a C l u s t e r (a) (b) Figure 4. (a) 1-D radial averaged autocorrelation normalized by C(0). The area of the first peak indicates the size of a cluster and the distance to the second peak corresponds to the radial average spacing between two clusters. (b) 3-D graph showing the relationship between the degree of homogeneity and real viscosity. As the viscosity increases, the homogeneity becomes improved.Figure 5. Radial spacing between clusters of fluorinedots versus normalized areaof cluster in terms of PVDF. Kynar301F and MKB212A show an improved homogeneity compared to KF1300.graphite. The Kynar301F and MKB212A show higher interaction with graphite particles,determined by the rheological properties, leading to a more homogeneous distribution of PVDF on the surface of graphite particles.CONCLUSIONSIn this paper, we have demonstrated that the interaction between graphite and PVDF in slurries depends on the nature of graphite particles and the chemical properties of PVDF and can be correlated with the surface properties of the final composite film. This study of particle/PVDF interaction is important in understanding and controlling the final film properties, which could contribute to the improvement of the electrochemical efficiency in lithium ion batteries.ACKNOWLEDGMENTS This work was supported by Mitsubishi Chemical Corporation.REFERENCES1. G. B. Li, R. J. Xue, and L. Q. Chen, Solid State Ionics 90, 221 (1996).2. W. F. Liu, X. J. Huang, G. B. Li, Z. X. Wang, H. Huang, Z. H. Lu, R. J. Xue, and L. Q. Chen,J. Power Sources 68, 344 (1997).3. O. Chusid, Y. E. Ely, D. Aurbach, M. Babai, and Y. Carmeli, J. Power Sources 43, 47(1993).4. M. N. Richard and J. R. Dahn, J. Power Sources 83, 71 (1999).5. E. Peled, C. Menachem, D. BarTow, and A. Melman, J. Electrochem. Soc. 143, L4 (1996).6. H. Maleki, G. P. Deng, I. KerzhnerHaller, A. Anani, and J. N. Howard, J. Electrochem. Soc.147, 4470 (2000).7. K. A. Hirasawa, K. Nishioka, T. Sato, S. Yamaguchi, and S. Mori, J. Power Sources 69, 97(1997).8. T. W. Healy, J. Colloid Sci. 16, 609 (1961).9. A. Biswas, R. Gupta, N. Kumar, D. K. Avasthi, J. P. Singh, S. Lotha, D. Fink, S. N. Paul,and S. K. Bose, App. Phys. Lett. 78, 4136 (2001).10. G. R. Strobl and M. Schneider, J. Polym. Sci. B 18, 1343 (1980). 0510152025303500.020.040.060.080.1Normalized Area of Cluster R a d i a l S p a c i n g b e t w e e n C l u s t e r s (µm )。