2110 Journal of The Electrochemical Society, 147 (6) 2110-2115 (2000) S0013-4651(99)08-066-0 CCC: $7.00 © The Electrochemical Society, Inc. The Li-ion cell, which was first commercialized in the early 1990s by Sony Corporation, contains a metal oxide (LiCoO 2 ) posi- tive electrode and a carbon negative electrode. 1-3 A variety of car- bons, which are classified as “hard” (nongraphitizable) carbons and “soft” (i.e., graphitizable) carbons have been evaluated as potential- ly negative electrode materials. Broadly speaking, the hard carbons exhibit both a high reversible capacity and higher irreversible capac- ity loss (ICL) than the soft carbons. The ICL originates from the de- composition of the electrolyte to form both a solid electrolyte inter- face (SEI) layer and gaseous products on the carbon electrode dur- ing the initial charge/discharge cycles. Carbon particles consist of hexagonal arrays of carbon atoms ar- ranged in benzene-type structures. The hexagonal arrays can form ordered layer structures, and in the case of graphite, the layer planes are separated by 3.354 Å. The surface of the hexagonal arrays forms the basal plane of graphite and the termination of the hexagonal array form the edge sites. The chemical reactivity of the basal plane and edge sites was recognized many years ago to be vastly different. For example, the oxidation rate and oxygen chemisorption at the edge sites are much higher than that of the basal plane. 4-6 The fac- tors responsible for the ICL are still a subject of intense research and debate. Both the catalytic properties and physical structure of carbon are believed to play important roles in the ICL. 7-16 The evidence 7-11 suggests that the ICL is strongly affected by the relative amounts of basal plane and edge sites. These studies con- cluded that the edge sites are the more active (catalytic) sites for elec- trolyte decomposition. Winter et al. 9 concluded that both the total surface area and the average ratio of the basal plane and edge thick- ness dimension influenced the ICL. Bar-Tow et al. 10 reported that the mechanism of electrolyte decomposition was different on the basal and edge sites in an electrolyte consisting of LiAsF 6 in ethylene car- bonate (EC)-diethyl carbonate (DEC). The SEI layer formed on the edge sites is rich in inorganic compounds, whereas the SEI layer formed on the basal plane is rich in organic compounds. On the other hand, Yamamoto et al. 11 reported that no electrochemical reaction involving electrolyte decomposition occurred on the basal plane. The ICL (mAh/g) increases linearly with an increase in the surface area of the carbon electrode in Li-ion cells. 7,9,17 This relationship does not provide insight into the role of basal and edge sites on the ICL. The study by Kinoshita and co-workers 8 related the significance of the crystallographic parameters L a and L c on the magnitude of the ICL. However, they were not able to ascertain the change in the ICL with a change in the relative amount of basal and edge sites. The pur- pose of the present paper is to extend the earlier study to assess the relationship between the relative fraction of edge and basal plane sites and the ICL. To achieve this goal, the ICL on natural graphite powder of different particle sizes was measured. The relative fraction of edge and basal plane sites was estimated from a theoretical analysis of the change in surface sites with particle size of a prismatic structure that resembled the graphite particles. To the best of our knowledge, a detailed analysis of the fraction of edge and basal plane sites, and their influence on the irreversible capacity loss has not been presented. Chung and co-workers 7 pre- sented an interesting approach to correlate the ICL and the edge sur- face area of graphite carbons by experiments involving the use of pro- pylene carbonate (PC) as an electrolyte additive. They reported that the ICL increased with an increase in the relative amount of PC in the electrolyte, but their study was not able to quantitatively show the relationship between the ICL and edge-site concentration. However, they presented a simple model of a circular disk to resemble a graphite particle and showed that the fraction of edge sites for a typical parti- cle size of 10 m is very low (i.e., 1%). The results obtained by Chung and co-workers are very relevant to our analysis and provided useful information that prompted the study presented in this paper. Experimental Six samples of natural graphite powders with average particle sizes of 2, 7, 12, 20, 30, and 40 m were obtained from commercial sources. These samples represent a wide range of particle sizes, and a particle morphology which is similar. The Brunauer-Emmett- Teller (BET) surface area was measured with a Quantachrome Autosorb automated gas sorption system using N 2 gas. X-ray dif- fraction (XRD) analysis (Siemens D500 diffractometer) was used to determine the d 002 spacing and the crystalline size, L c . Scanning electron microscopy (SEM, Hitachi) and particle size analysis (Microtrac X100 particle analyzer) were employed to determine the morphology and dimensions of the edge and basal plane dimensions, respectively. The average particle size is calculated from duplicate measurements, and the instrumental error is 0.1 m. The thickness of the edge planes was obtained from magnified images from SEM micrographs. The thickness is an average of four measurements of the edge planes in the samples. The error is 0.01 m. Raman spectroscopy measurements were carried out at room temperature in ambient atmosphere using an argon-ion laser (Coher- ent Inc. model Innova 70) tuned to 514.5 nm. The resolution of this instrument is approximately 1.7 cm -1 . The crystallite dimension, L a , is obtained by using the method of Tuinstra and Koenig, 18 i.e., L a = 44/(I 1372 /I 1576 ), where I 1372 and I 1576 are the integrated intensities of the Raman peaks at 1372 and 1576 cm -1 , respectively. The working electrode was fabricated from a mixture of the nat- ural graphite and poly(vinylidene fluoride (PVDF) dissolved in 1- methyl-2-pyrrolidinone (NMP). The slurry was spread onto a copper grid and dried under vacuum at 95°C for 24 h. The working elec- Effect of Graphite Particle Size on Irreversible Capacity Loss Karim Zaghib, a, * ,z Gabrielle Nadeau, a and Kimio Kinoshita b, * a Institute de Recherche d’Hydro-Québec, Varennes, Québec, Canada J3X 1S1 b Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA Electrolyte decomposition and irreversible capacity loss (ICL) occur on carbon electrodes in Li-ion cells. The nature of the surface sites and their role in the amount of electrolyte decomposition on carbon electrodes is not fully understood. Therefore, a study was undertaken to analyze the relationship between the ICL and the active sites on natural graphite of prismatic structure. The ICL was measured on natural graphite of predominantly two-dimensional platelets of average particle size varying from 2 to 40 m. The fraction of edge and basal plane sites was determined for ideal prismatic structures of different particle size and used as a model for the natural graphite particles. This analysis permitted an analysis of the relationship between the electrolyte decomposition and the distribution of surface sites. From this analysis we conclude that these sites play an important role in the magnitude of the ir- reversible capacity loss on natural graphite. © 2000 The Electrochemical Society. S0013-4651(99)08-066-0. All rights reserved. Manuscript submitted August 19, 1999; revised manuscript received February 5, 2000. * Electrochemical Society Active Member. z E-mail: karimz@ineq.ca