In Situ AFM Imaging of Surface Phenomena on Composite Graphite

Lithium Intercalation into Graphite. J. S. Gnanaraj , R. W. Thompson , J. F. DiCarlo , K. M. Abraham. Journal of The Electrochemical Society 2007 ...
2 downloads 0 Views 826KB Size
Download PDF
9000

Langmuir 2002, 18, 9000-9009

In Situ AFM Imaging of Surface Phenomena on Composite Graphite Electrodes during Lithium Insertion Doron Aurbach,* Maxim Koltypin, and Hanan Teller Department of Chemistry, Bar-Ilan University, Ramat-Gan 52900, Israel Received April 1, 2002. In Final Form: August 14, 2002 The behavior of composite graphite electrodes used as anodes in Li-ion batteries, in repeated lithium intercalation-deintercalation processes, largely depends on the solution’s composition. Failure mechanisms of graphite electrodes were studied in selected solutions using in situ imaging by atomic force microscopy. Of special interest is the stable behavior of lithiated graphite electrodes in ethylene carbonate solutions, while in solutions of a very similar solvent, propylene carbonate (PC), lithiated graphite electrodes fail. The morphological studies reported herein seem to confirm that the major failure mechanism of graphite electrodes in PC solutions involves cracking of the particles and electrical isolation of most of the active mass by surface films, rather than a massive exfoliation of the graphite particles due to cointercalation of solvent molecules. Cracking of the graphite particles during cathodic polarization depends on the crevices that originally exist in some types of graphite particles, especially in synthetic materials, and is promoted by gas formation within the crevices, due to the reduction of solvent molecules.

Introduction Development and commercialization of lithium ion batteries in recent years is one of the most important successes of modern electrochemistry. The major components in most of the commercial Li-ion batteries are graphite anodes, LiCoO2 cathodes, and electrolyte solutions based on alkyl carbonate solvents and LiPF6 as the salt.1 The electrodes for these batteries always have a composite structure that includes a metallic current collector (usually Cu or Al foil or grid for the anode and cathode, respectively), a polymeric binder, and the active mass in micrometric size particles. Graphite intercalates electrochemically and reversibly with lithium via phase transitions in four stages, up to a stoichiometry of LiC6 (corresponding to a capacity of 372 mAh/gr).2 However, it is generally known that the behavior of graphite electrodes in Li salt solutions is very sensitive to the solution composition. In most of the Li salt solutions based on solvents such as esters, ethers, and acyclic alkyl carbonates (e.g., dimethyl and diethyl carbonates), lithiation of graphite cannot be completed and its reversibility is poor.3 In contrast, when ethereal esters or acyclic carbonate solutions contain ethylene carbonate as a cosolvent or additives such as CO2 or SO2, lithiation of graphite is highly reversible and reaches its full capacity (LiC6 is formed).4 It was found that cathodic polarization of graphite electrodes in any polar aprotic Li salt solution leads to a massive reduction of solution species on the graphite surface, at potentials below 2 V versus Li/Li+.5 The products of reduction of solvents, salt anions, and common atmospheric contaminants in nonaqueous solutions in the presence of Li ions are always lithium salts, * To whom correspondence should be addressed. Fax: 972-35351250. Tel: 972-3-5318317. E-mail: [email protected]. (1) Li-ion Batteries, Fundamentals and Performance; Wakihara, M., Yamamoto, O., Eds.; Wiley-VCH: Weinheim, 1998. (2) Dahn, J. R.; Sleight, A. K.; Shi, H.; Reimers, J. N.; Zhong, Q.; Way, B. M. Electrochim. Acta 1993, 38, 1179. (3) Aurbach, D.; Ein-Eli, Y.; Markovsky, B.; Carmeli, Y.; Yamin, H.; Luski, S. Electrochim. Acta 1994, 39, 2559. (4) Aurbach, D.; Markovsky, B.; Gamolsky, K.; Levi, E.; Ein-Eli, Y. Electrochim. Acta 1999, 45, 67. (5) Winter, M.; Besenhard, J. O. In Handbook of Battery Materials; Besenhard, J. O., Ed.; Wiley-VCH: Weinheim, 1999; Chapter 5, p 383.

which in many cases are insoluble in the mother solutions. Thereby, these reduction products precipitate on the graphite surfaces, thus forming surface films.6 The stability of the graphite electrodes upon Li insertion and the amount of irreversible charge transfer involved in the above surface reactions depend on the passivation properties of these surface films, which are directly dependent on the cohesion of the surface species and their adhesion to the graphite particles.7 Naturally, the failure of graphite electrodes in Li insertion processes is attributed to cointercalation of solvent molecules together with the Li ions, which leads to an exfoliation of the graphite particles.8 We indeed found that exfoliation phenomena of graphite electrodes in ethereal Li salt solutions, during cathodic polarization of graphite electrodes, leads to their amorphization. Cointercalation of ether molecules with Li ions disconnects the graphene planes from each other and hence destroys the 3D structure of the graphite particles.4 In other cases (e.g., in propylene carbonate solutions), we found that graphite electrodes are deactivated, while the basic 3D structure of the active mass remains graphitic.9 Several research groups indeed attributed the failure of graphite electrodes in propylene carbonate (PC) solutions to exfoliation of the graphite, due to cointercalation of PC molecules with the Li ions.8,10 Some of these groups attributed the difference between ethylene carbonate (EC) and PC in this respect to the higher ability of PC molecules to solvate Li ions.11 Hence, in their opinion, cointercalation of PC molecules takes place because their desolvation from Li ions, which migrate from solution phase to intercalation sites in the graphite, is very difficult (compared with the case of other solvent molecules such as EC).8,10,11 (6) Peled, E.; Golodnitsky, D.; Penciner, J. In Handbook of Battery Materials; Besenhard, J. O., Ed.; Wiley-VCH: Weinheim, 1999; Chapter 6, p 419. (7) Ein-Eli, Y. Electrochem. Solid State Lett. 1999, 2, 212. (8) Winter, M.; Besenhard, J. O.; Spahr, M. E.; Novak, P. Adv. Mater. 1998, 10, 725. (9) Aurbach, D.; Levi, M. D.; Levi, E.; Schechter, A. J. Phys. Chem. B 1997, 101, 2195. (10) Chung, G. C.; Kim, H. J.; Jun, S. H.; Choi, J. W.; Kim, M. H. J. Electrochem. Soc. 2000, 147, 4398. (11) Wang, Y.; Balbuena, P. B. J. Phys. Chem. 2001, A105, 9972.

10.1021/la020306e CCC: $22.00 © 2002 American Chemical Society Published on Web 10/17/2002

Surface Phenomena on Graphite Electrodes

Our cumulative studies, based on scanning electron microscopy (SEM), X-ray diffraction (XRD), and standard electrochemical techniques, converged to the conclusion that in PC solutions, the failure mechanism of graphite electrodes is not exfoliation of the graphite particles but rather their cracking and electrical isolation by surface films.9,12 In the present work, in situ atomic force microscopy (AFM) imaging was applied for the study of several types of graphite electrodes in a few selected probe solutions. These included standard EC-DMC/LiPF6 or EC-DMC/LiAsF6 solutions in which graphite electrodes of all kinds behave highly reversibly, PC solutions in which graphite electrodes fail, and PC-based solutions in which graphite electrodes may behave reversibly,13 but the degree of reversibility and stability depends strongly on the morphology of the particles. We believe that this selection of working systems and the application of AFM may sharpen the conclusions regarding the most probable failure mechanisms of graphite electrodes in Li insertion processes. In situ AFM imaging of graphite electrodes has already been reported in the literature.14-18 Most of the work so far involved the use of highly oriented pyrolotic graphite (HOPG)14-16 and hence was carried out with relative ease. There are only a few publications on studies of composite graphite electrodes by AFM (which are much more difficult).17,18 The novelty of the present study lies in the use of in situ AFM imaging of a variety of graphite electrodes of different morphologies in several interesting selected solutions and the possibility of clarifying failure mechanisms of graphite electrodes in Li insertion processes. Experimental Section In this study, we used a commercial 1 M LiPF6 solution in EC-DMC 1:1 from Merck KGaA, Germany. In addition, a solution of 1 M LiClO4 (Tomiyama, Inc.) in EC-PC 2:3 (Merck KGaA) was prepared and tested. The water content of all the solutions was less than 15 ppm (monitored by Karl Fisher titration, a Metrohm Inc. 562 CF coulometer). We used synthetic graphite flakes, KS6 powder from Timrex Inc., Switzerland, carbon microbeads MCMB 25-28 from Osaka Gas Co. Inc., Japan, and natural graphite flakes (denoted as NG-15). The electrodes were made by mixing the powders (i.e., the active carbon material) and 10% PVdF binder, using a “wig-1-bug” amalgamator (5 min in air) and adding 1-methyl-2-pyrrolydone to obtain a homogeneous slurry. The slurry was then spread on 12 mm diameter copper disks that were prerubbed in order to achieve good adhesion. The electrodes’ mass was usually 4 mg. The electrodes were dried under vacuum for 12 h. The electrochemical/AFM cells used in these measurements were described previously.19 The cells were assembled in a glovebox (VAC Inc.) under a highly pure argon inert atmosphere. AFM measurements were carried out using a Topometrix Inc. AFM system, Discoverer model no. 2010, with Topometrix pyramidal silicon carbide tips, placed in a special homemade glovebox. This system was already described in detail.20 In addition, the experimental conditions and the way in which the in situ imaging of composite graphite electrodes by AFM was carried out were recently described in the literature.18 (12) Aurbach, D.; Teller, H.; Levi, E. J. Electrochem. Soc., in press. (13) Dahn, J. R.; Von Sacken, U.; Juzkow, M. W.; Al Janaby, H. J. Electrochem. Soc. 1991, 138, 22. Fong, R.; Von Sacken, U.; Dahn, J. R. J. Electrochem. Soc. 1990, 137, 2009. (14) Jeong, S. K.; Inaba, M.; Abe, T.; Ogumi, Z. J. Electrochem. Soc. 2001, 148 (9), A989. (15) Alliata, D.; Kotz, R.; Novak, P.; Siegenthaler, H. Electrochem. Commun. 2000, 2, 436. (16) Inaba, M.; Siroma, Z.; Funabiki, A.; Abe, T.; Mizutani, Y.; Asano, M.; Ogumi, Z. Langmuir 1996, 12, 1535. (17) Morigaki, K.; Fujii, T.; Ohta, A. Denki Kagaku 1998, 66, 1122. (18) Koltypin, M.; Cohen, Y.; Cohen, Y.; Markovsky, B.; Aurbach, D. Electrochem. Commun. 2002, 4, 17. (19) Aurbach, D.; Cohen, Y. J. Electrochem. Soc. 1996, 143, 3525. (20) Cohen, Y.; Aurbach, D. Rev. Sci. Instrum. 1999, 70, 4668.

Langmuir, Vol. 18, No. 23, 2002 9001

Figure 1. A typical potential-capacity curve in a first and a few subsequent cycles (E vs t translated to capacity in mAh/gr) of a composite graphite electrode comprised of synthetic flakes (KS6; Timrex, Inc.) in an EC-DMC/LiAsF6 1 M solution, cycled galvanostatically at C/10 h vs a lithium counter electrode.

Results Figures 1-3 describe the behavior of composite graphite electrodes comprising synthetic graphite flakes (KS-6, Timrex, Inc.) in an EC-DMC/LiAsF6 solution. Figure 1 shows a typical potential-capacity curve of these electrodes when cycled at constant current versus Li counter electrode (repeated lithium intercalation-deintercalation cycles). It reflects a reversible behavior: during the first cathodic polarization of this electrode, an irreversible charge is consumed in the formation of passivating surface films and generally appears as a voltage plateau in the 1.5-0.8 V range. As seen later in this paper, the potential of this plateau (related to the reduction of solution species) depends on both the salt and the solvent used.8 Then, upon cycling, the electrode’s behavior stabilizes, the Li intercalation stages are clearly seen at potentials below 0.3 V (Li/Li+), and a reversible capacity of >300 mAh/gr is usually recorded (and may reach the limit of 372 mAh/ gr, corresponding to the formation of LiC6). Figure 2 shows selected 3D AFM images of a similar electrode in the same solution, measured in situ during a first Li intercalation-deintercalation cycle at several constant potentials, as indicated. The images are focused on an area in which the border between two graphite particles (micrometric size) is clearly observed. Image a relates to a pristine electrode, and image b measured at 0.9 V (Li/Li+) shows the morphology of the surface films formed at this potential (see the plateau in Figure 1). Images c and d were measured at Li insertion potentials (stage III and stage I, respectively), and images e and f were measured during deintercalation. The AFM imaging of these electrodes in the EC-DMC-based solutions reflect stability. The changes observed are minor. Figure 3 schematically describes the processes of these electrodes in the standard solutions. During a first cathodic polarization, highly passivating surface films are formed, which stop the surface reactions, because they block electron transfer. They protect the fragile graphite structure from cointercalation of solvent molecules and exfoliation. As Li is inserted, there is a volume expansion of the graphite, and hence, the surface films on the edge planes (through which Li insertion takes place) are stretched. As Li is deintercalated, the graphite’s volume contracts back to its initial size. The passivation of the graphite particles depends on the degree of accommodation of the surface films to these volume changes and to what exent passivation is maintained (when the graphite is fully lithiated). As already discussed,18 we assume that upon the repeated volume changes of the graphite particles

9002

Langmuir, Vol. 18, No. 23, 2002

Aurbach et al.

Figure 2. 3D AFM images measured in situ at constant potentials with a graphite electrode comprising synthetic flakes (KS6; Timrex, Inc.) during a Li insertion-deinsertion cycle in an EC-DMC/LiAsF6 1 M solution. Dimensions appear near the images: (a) OCV (≈3 V vs Li/Li+), a pristine electrode; (b) 0.9 V, after surface film formation; (c) 0.2 V during intercalation, formation of LiC36; (d) 0.06 V during intercalation, formation of LiC6; (e) 0.11 V during deintercalation, 2LiC6 f LiC12 +Li+ + e-;. (f) 0.24 V during deintercalation, 3LiC12 f LiC36 + 2Li+ + 2e-.

Figure 3. An illustration of morphology, surface processes, and changes during a Li insertion-deinsertion cycle of an electrode comprising synthetic graphite flakes in an EC-DMC/1 M LiAsF6 solution, in which the electrodes behave reversibly.

during the Li insertion-deinsertion cycling, the passivation by the surface films on the edge planes may be damaged, thus allowing a further small-scale reduction of solution species. These minor surface reactions that thicken the surface films during prolonged, repeated cycling explain the increase of impedance of these electrodes observed during prolonged charge-discharge cycling.21

Figure 4. Same as Figure 1: a potential vs capacity curve of a KS6 electrode (synthetic flakes) in an EC-PC 2:3/LiClO4 1 M solution. Note the pronounced irreversible capacity.

Figures 4 and 5 describe the behavior of composite graphite electrodes comprising synthetic graphite flakes (the same as the electrodes related to Figures 1-3) in an EC-PC 2:3/1 M LiClO4 solution. This electrolyte solution was chosen as the major probe solution, because graphite electrodes of different particle sources and morphologies behave very differently in EC-PC/LiClO4 solutions in (21) Mortinent, A.; Le Gorrec, B.; Montella, C.; Yazami, R. J. Power Sources 2001, 97-98, 83.

Surface Phenomena on Graphite Electrodes

Langmuir, Vol. 18, No. 23, 2002 9003

Figure 5. 3D AFM images measured in situ with a KS6 electrode during a first galvanostatic cathodic polarization from OCV (≈3 V vs Li/Li+) in an EC-PC 2:3/LiClO4 1 M solution. The images were obtained at constant potentials that the electrodes reached during the process, as indicated. Dimensions appear near each image.

terms of the irreversible capacity measured during a first cathodic polarization of the electrodes in these solutions and the potentials at which major reduction of solution species takes place (which are lower by more than 0.5 V compared with those in the EC-DMC/LiAsF6 solution). Nevertheless, after stabilization, the electrodes behave reversibly in these solutions, as seen in Figure 4. (After the first few cycles, the behavior of these systems stabilizes.) Figure 5 shows AFM images of a similar electrode measured in situ (at constant potentials) in an EC-PC 2:3/LiClO4 solution during the first cathodic polarization (related to the plateau starting around 1 V, seen in Figure 4). Polarization was conducted galvanostatically. As the electrodes reached certain potentials, they were held at these potentials and measured by AFM. As the electrodes are polarized from OCV (3 V vs Li/Li+; Figure 5, image a) to potentials below 1 V, pronounced morphological changes are observed. The surfaces of the particles at their boundaries are increasingly lifted upward. This is clearly seen in images b-d in Figure 5, focused on a gap between two particles. It is clear from these studies that the electrodes retain their integrity, and after the irreversible morphological changes observed by the AFM imaging, they can function electrochemically and reversibly (in repeated Li insertion-deinsertion processes). The fabrication of practical graphite anodes involves the application of some pressure, to obtain a compact

Figure 6. Same as Figures 1 and 4: a potential vs capacity curve of a first and subsequent Li insertion-deinsertion cycles of an MCMB electrode; EC-PC 2:3/LiClO4 1 M solution (galvanostatic mode).

electrode structure. However, when the active mass is comprised of graphite flakes (e.g., like the electrodes dealt with above, composed of KS-6 Timrex, Inc., particles), application of pressure orients the particles with their wide dimensions (the basal planes) parallel to the current collector. Since lithium is inserted through the edge planes, such a compact, oriented electrode’s structure is very bad for the electrodes’ kinetics, because a problem of contact between solution species and the entire active

9004

Langmuir, Vol. 18, No. 23, 2002

Aurbach et al.

Figure 7. (a,b) Same as Figure 5: 3D AFM images of an MCMB electrode in an EC-PC 2:3/LiClO4 1 M solution, during a first cathodic polarization. The relevant potentials are indicated, and dimensions appear near each image. Illustrations of the morphological changes observed are also present.

Surface Phenomena on Graphite Electrodes

mass is created.22 This problem is solved by the use of round graphite particles such as mesocarbon microbeads (MCMB).23 Electrodes comprised of MCMB graphite particles were also tested in the present study. Figure 6, which shows a typical potential-capacity curve of an MCMB electrode charged and discharged galvanostatically in an EC-PC 2:3/1 M LiClO4 solution, demonstrates a similar behavior pattern to that of electrodes comprised of synthetic graphite flakes (see Figure 4). The irreversible capacity measured is much higher than the ideal reversible capacity of this material (≈372 mAh/gr). After the first few Li insertion-deinsertion cycles, the behavior of these systems stabilizes. Figure 7a,b shows AFM images measured in situ during a first galvanostatic polarization of an MCMB electrode, as well as illustrations that explain the phenomena observed. The images measured at OCV (not seen in this figure) reflect initially a rough surface. The first image presented relates to the carbon surface covered by surface films, whose morphology differs considerably from that of the pristine material. (The surface films cover all the surface and, in fact, hide the initial rough structure.) As seen in Figure 7a, as polarization proceeds, the imaging shows the clear formation of cracks that deepen and widen during polarization, at potentials corresponding to the plateau around 0.8 V (irreversible surface processes) in Figure 6. Upon further polarization, the cracks are filled, probably by solution reduction products, which are formed within the cracks. Imaging MCMB electrodes enables us to follow cracking of the graphite particles due to the surface reactions. Here, the advantage over electrodes comprised of graphite flakes is that with the MCMB particles, many edge planes, along which the particles are cracked, face upward, and hence cracking can be easily imaged. In the case of the graphite flake electrodes, mostly the basal planes (the wide dimension of the particles) face upward, and hence cracking can be followed only indirectly, appearing as a lifting up of the particles’ edges, as seen in Figure 5. The morphology of the surface films on MCMB differs from that of the surface films formed on the flakes. This is probably due to the fact that the surface films imaged in Figure 5 are mostly on the basal planes of the flakes, whose initial morphology is relatively smooth. The surface films imaged in Figure 7a are those formed on edge planes, whose initial morphology is rough. Hence, the difference in the morphologies of the surface films in Figures 5 and 7a reflects the difference in the initial morphologies of the graphite surfaces, which influence the morphologies of the surface films formed on them. Figure 8 shows a typical potential versus capacity curve of a graphite electrode comprised of natural graphite flakes in an EC-PC 2:3/1 M LiClO4 solution. As was found by scanning electron microscopy,12 the edge planes of these particles (i.e., the facets perpendicular to the basal planes, through which Li insertion takes place) are much smoother than the edge planes of the synthetic graphite particles (either flakes or MCMB). As seen in Figure 8, the irreversible capacity measured with these electrodes, even in solutions containing 60% PC, is relatively low and similar to that measured in the standard solutions: ECDMC/LiAsF6 or EC-DMC/LiPF6. After the first few Li insertion-deinsertion cycles, the behavior of these systems stabilizes. (22) Gnanaraj, J. S.; Cohen, Y. S.; Levi, M. D.; Aurbach, D. J. Electroanal. Chem. 2001, 516, 89. (23) Lampe-Onnerud, C.; Shi, J.; Onnerud, P.; Chamberlain, R.; Barnett, B. J. Power Sources 2001, 97-98, 133.

Langmuir, Vol. 18, No. 23, 2002 9005

Figure 8. Same as Figures 1, 4, and 7: a typical potential vs capacity curve of a first and subsequent cycles of an electrode comprised of natural graphite flakes. EC-PC 2:3/LiClO4 1 M solution (galvanostatic mode).

Figure 9 shows AFM images of an electrode comprised of the natural graphite flakes, measured in situ in the EC-PC/LiClO4 solution during a first galvanostatic polarization from OCV (≈3 V vs Li/Li+). When the electrode reached some selected potentials (indicated near each image), it was measured by AFM, while the potentials were held constant. The images of these electrodes obtained in situ during polarization definitely reflect a relatively stable morphology with only minor changes, which should be attributed to the formation of surface films on top of the particles. We did not observe cracking of particles with these electrodes, as was observed with the synthetic flakes and the MCMB electrodes. We suggest that the natural graphite flake electrodes behave differently in the PC-containing solutions because their edge planes are much smoother and contain fewer crevices as compared with synthetic flakes or MCMB particles (see discussion section). Figure 10 shows a typical chronopotentiogram (E vs t curve) of an electrode comprised of natural graphite flakes in a PC/LiClO4 solution. The voltage profile presented in this figure reflects pronounced surface processes in which solution components are intensively reduced and then deactivation of the electrode, so that Li insertion does not take place at all. Figure 11 shows AFM images measured from a similar electrode in a PC/LiClO4 solution during polarization at constant potential (0.9 V, corresponding to the plateau in Figure 10 and the irreversible process). The images of Figure 11 reflect pronounced morphological changes in the electrode, which can be understood in terms of remarkable changes in the orientation of the particles due to the electrochemical processes, as illustrated in the figure in the attached drawing. The surface processes of these systems were accompanied by a pronounced formation of gas bubbles that could be observed by the microscope’s CCD camera, as demonstrated in Figure 12. Note that in this figure, the CCD camera traces pronounced visible changes on the electrode’s surface due to its polarization (compared with the OCV picture which is also presented in Figure 12). Hence, the AFM imaging and the CCD camera show continuous morphological changes that correlate well with the electrode’s deactivation demonstrated in Figure 10. Discussion The major goal of the studies reported herein was to resolve the questions regarding the failure mechanisms of graphite electrodes during lithium insertion in solutions based on propylene carbonate.

9006

Langmuir, Vol. 18, No. 23, 2002

Aurbach et al.

Figure 9. Same as Figures 5 and 8: AFM images of an electrode comprised of natural graphite flakes measured in situ at constant potentials (indicated) during a first galvanostatic, cathodic polarization in an EC-PC 2:3/LiClO4 solution. Illustrations of the morphological changes observed are also present.

Figure 10. A typical chronopotentiogram (E vs t curve) of a graphite electrode (natural graphite flakes) in a PC/LiClO4 1 M solution.

As explained in the Introduction, a major suggestion for the failure mechanism of Li-graphite electrodes in PC solutions involves cointercalation of solvent molecules and exfoliation of the graphite.5,8,10 This suggestion suffers from the following weak points: 1. If graphite electrodes exfoliate due to cointercalation of PC molecules with lithium, they should cointercalate EC molecules as well and hence should fail in EC solutions as well. 2. Previous studies of Li insertion into graphite in PC solutions by in situ and ex situ XRD showed clearly that

deactivated graphite electrodes after cathodic polarization in PC solutions retained their graphitic structure.4,9 If the failure mechanism of graphite electrodes in PC solutions was exfoliation, the graphitic structure would be destroyed, which should be clearly indicated by XRD measurements. 3. Finally and most important, the studies presented herein by in situ AFM imaging showed clearly that the graphite particles undergo upon cathodic polarization in PC solutions morphological changes but not exfoliation. Also, the different behavior of different graphite particles is striking. As clearly seen in Figures 4-9, the irreversible capacity of synthetic flakes and MCMB is high and the morphological changes are pronounced in EC-PC/LiClO4 solutions while for natural graphite flakes, both the irreversible capacity and the morphological changes are relatively small. If the failure mechanism of graphite electrodes were due to cointercalation of solvent molecules and exfoliation, the particle morphology would not have such a pronounced impact on the electrodes’ behavior. We thereby suggest the following scenario, described in Figure 13, which explains the failure of graphite electrodes in PC solutions, based on both the electrochemical and the morphological studies: At potentials below 1.5 V, PC is reduced on any metallic or carbon electrode, in the presence of Li ions.24 The major (24) Aurbach, D.; Moshkovich, M.; Gofer, Y. J. Electrochem. Soc. 2001, 148, E155.

Surface Phenomena on Graphite Electrodes

Langmuir, Vol. 18, No. 23, 2002 9007

Figure 11. AFM images measured in situ from a natural graphite electrode during a first cathodic polarization in a PC/LiClO4 1 M solution. All the images except the one related to OCV (indicated) were measured at 0.9 V (Li/Li+). Dimension scales appear near the images. The morphological changes of the electrode, as reflected by the AFM imaging, are illustrated.

Figure 12. CCD camera’s pictures of a pristine, natural graphite electrode and the electrode during cathodic polarization in a PC-LiClO4 1 M solution (top and bottom, respectively). The pictures show the device to which the AFM tip is mounted (top view) and the electrode surface below. Note the gas bubble, which accumulates at the upper part of the cell, as indicated (bottom picture).

products are CH3CH(OCO2Li)CH2OCO2Li and propylene gas.25 The former solid product is insoluble in PC and hence precipitates on the active mass of the electrode as surface films.

Figure 13. An illustration of the processes of synthetic graphite flakes during polarization of the electrodes in the PC-based solution. Reduction of solvent molecules in the crevices leads to formation of gas and a buildup of internal pressure, which cracks the particles.

The synthetic graphite particles have many crevices in their edge planes, as was analyzed by scanning electron microscopy.12 As PC is reduced inside the crevices and the passivation by the reduction products is not sufficiently (25) Aurbach, D.; Gottlieb, H. Electrochim. Acta 1989, 34, 141.

9008

Langmuir, Vol. 18, No. 23, 2002

efficient, the coproduct, propylene gas, may accumulate and form internal pressure inside the particles, which should lead to their cracking. This is illustrated in Figure 13. Cracking by a buildup of internal pressure lifts the particles upward, as is clearly observed by the in situ imaging by AFM. Cracking also increases the active surface, which further reacts with solution species. This explains the very high irreversible capacity measured in these systems, as demonstrated in Figure 4. Cracking of the graphite particles due to the formation of internal pressure was clearly observed in electrodes comprising MCMB particles (Figure 7a,b) because their round shape allows imaging of edge planes facing upward (see the illustrations in Figure 7a,b). The images of Figure 7b also show how the cracks thus formed are filled with the reduction products of solution species on the newly formed active surfaces. The natural flakes have smoother edge planes than the synthetic particles (both flakes or MCMB), as was evident by SEM measurements.12 Hence, reduction of solvent molecules inside deep crevices and formation of internal pressure that cracks the particles is much less possible in this case (as is indeed demonstrated in Figures 9 and 10). Hence, the major failure mechanism of graphite electrodes in PC solutions involves cracking which allows intensive surface reactions on the cracked particles. These surface reactions form surface films that surround the particles and isolate them electrically from the current collector, which deactivates the electrode. The buildup of pressure inside the graphite particles depends critically on passivation phenomena. In cases where the solid reduction products of the solvent molecules form highly cohesive and adhesive surface films, the surface reactions are blocked before massive reduction processes, which form the gas molecules, take place. When passivation of the graphite is not reached quickly enough, intensive surface reactions build up the internal pressure that cracks the particles and leads to their deactivation. The adhesion of the ROCO2Li species, formed by reduction of the alkyl carbonate solvents,25,26 to the carbon surface, as well as the cohesion of the ROCO2Li species, goes through the carbonate groups and the Li ions that bridge between the negatively charged oxygen atoms and between the surface species and the negatively charged carbon atoms. EC and PC reduction forms (CH2OCO2Li)2 and CH3CH(OCO2Li)CH2OCO2Li, respectively, which differ from each other in the methyl group. We suggest that the (CH2OCO2Li)2 species formed by EC reduction precipitate as adhesive and cohesive surface films more quickly and efficiently than the PC reduction products which have an extra methyl group. We assume that such a methyl group should interfere badly with adhesion and cohesion of the Li dicarbonate species, due to aspects of steric hindrance [which do not exist with (CH2OCO2Li)2]. From the above discussion, it is clear that the stabilization or failure of graphite electrodes depends on a delicate balance between passivation phenomena (due to formation of highly cohesive and adhesive surface films) and a buildup of internal pressure due to reduction of solution species inside crevices in the graphite particles. This delicate balance can be attenuated, as demonstrated in this study, by both solution composition (solutions of ECDMC, EC-PC, PC, etc.) and the morphology of the (26) Aurbach, D.; Gofer, Y.; Ben-Zion, M.; Aped, P. J. Electroanal. Chem. 1992, 339, 451.

Aurbach et al.

graphite particles (i.e., the structure of the edge planes and the presence of crevices). Conclusion Understanding the failure mechanisms of graphite electrodes in Li insertion-deinsertion processes is an important issue in the field of Li batteries. In this respect, the failure of Li-graphite electrodes in PC solutions poses an interesting question, in light of the fact that in ECbased solutions, lithiated graphite electrodes behave highly reversibly, even though PC and EC are so similar in their structure and properties. In this study, we tested three types of graphite electrodes, synthetic flakes, natural flakes, and MCMB, in three solutions: EC-DMC/LiAsF6, in which all graphite electrodes behave highly reversibly; PC/LiClO4, in which all the graphite electrodes fail and cannot be lithiated; and EC-PC/LiClO4, in which there is a great difference in the behavior of the three types of electrodes in terms of irreversible capacity and stability. This difference is connected to the different morphologies of the graphite particles used. In situ imaging of this matrix of systems (three types of electrodes and three types of solutions) by AFM was found to correlate very well with the electrochemical response of the various systems and, in fact, helped to explain it. With the EC-DMC/LiAsF6 solutions, the morphology of the graphite electrodes is generally stable during repeated Li insertion-deinsertion cycling. With PC/LiClO4 solutions, the in situ imaging showed a remarkable change in the electrodes’ morphology. The particles crack and change orientation. With PC-EC/ LiClO4 solutions, the morphology of the electrodes comprised of natural flakes remains stable during polarization and surface film formation, while the morphology of the synthetic flakes and MCMB changes considerably during the surface film formation. We were able to follow the formation of cracks in the particles, in situ. The behavior of these systems can be explained in light of their surface chemistry and the fact that in addition to surface species of the ROCO2Li (Li alkylene dicarbonates) type, reduction of both EC and PC forms alkylene gas. When EC reduction dominates the surface chemistry, the (CH2OCO2Li)2 thus formed can precipitate with good adhesion and cohesion, to form highly passivating surface films. When the surface chemistry is dominated by PC reduction, the surface species that contain the methyl group are less cohesive and adhesive; thus, formation of passivating surface films requires an intensive reduction of solution species, and thus it is accompanied by a massive propylene gas formation. Graphite particles always contain crevices in their edge planes. Reduction of PC in the crevices in the absence of a rapid formation of passivating surface films leads to a buildup of pressure inside, which leads to the cracking and splitting of the particles and an increase of the surface area, and hence, massive surface reactions take place and form surface films that electrically isolate most of the active mass. In cases of EC-PC mixtures, where EC reduction products also contribute to the formation of passivating surface films, electrodes comprised of natural graphite flakes whose edge planes are relatively “smooth” can behave reversibly with relatively low irreversible capacity losses. However, when the electrodes are comprised of synthetic flakes, or MCMB whose edge planes have a high concentration of crevices, the passivation in an EC-PC-based solution is not efficient enough to avoid the buildup of internal pressure, which cracks the particles. However, cracking of these particles in EC-PC-based solutions is not massive enough to

Surface Phenomena on Graphite Electrodes

interfere adversely with the electrode’s integrity, nor to change the particles’ orientation and lead to an electrical isolation of the active mass, as happens in pure PC solutions. Therefore, the electrodes can reach passivation in PC-EC solutions after pronounced consumption of irreversible charge and then lithium can be inserted reversibly.

Langmuir, Vol. 18, No. 23, 2002 9009

Acknowledgment. Partial support for this work was obtained from the BMBF, the German Ministry of Science, in the framework of the DIP Program, and also from the EC, under Contract Number ENK6-CT99-00006. LA020306E