Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

A2106

0013-4651/2004/151共12兲/A2106/7/$7.00 © The Electrochemical Society, Inc.

GraphiteÕElectrolyte Interface Formed in LiBOB-Based Electrolytes II. Potential Dependence of Surface Chemistry on Graphitic Anodes Kang Xu,*,z Unchul Lee, Sheng S. Zhang, and T. Richard Jow* Army Research Laboratory, Adelphi, Maryland 20783-1197, USA In an attempt to depict a dynamic picture of solid electrolyte interface 共SEI兲 formation on a graphitic anode surface during the initial forming cycle, we employed X-ray photoelectron spectroscopy in combination with a ‘‘pre-formation’’ technique to establish the dependence of the surface chemistry on the forming potential of the anode. A progressive transition in the 1s electron binding energies of the major elements was observed as the lithiation proceeded. However, the surface chemical species as well as their abundances seemed to stabilize around 0.55 V and remained constant during the subsequent delithiation process, indicating that a stable SEI exists thereafter. Integrating the information revealed by different analyses, we believe that the reductive decomposition of the BOB⫺ anion starts at ca. 1.00 V, while the effective protection of the graphene surface by SEI is available after the anode is lithiated below the potential of 0.55 V vs. Li. © 2004 The Electrochemical Society. 关DOI: 10.1149/1.1812732兴 All rights reserved. Manuscript submitted March 8, 2004; revised manuscript received May 12, 2004. Available electronically November 2, 2004.

The formation of a passivation film on the surface of metallic lithium or a carbonaceous anode constitutes the foundation for the rechargeable battery chemistry based on lithium. The film eventually enables these electrodes to operate reversibly in nonaqueous electrolytes at low potentials without the sustained decompositions of the electrolyte components.1,2 Due to the similar potentials of the fully lithiated carbon and lithium metal, the chemical composition of this protective film, universally known as a solid electrolyte interface 共SEI兲, has been suggested to be the same for either a lithium metal or carbonaceous electrode;2 however, a fundamental difference is believed to exist in their formation mechanisms. Because of the high electronegativity of lithium metal, free contact between it and the electrolyte components should never actually exist, according to Peled;1 thus, the reaction between the lithium electrode and electrolyte components occurs instantaneously, and the reduction by lithium should be indiscriminating to all components present in the solution. On the other hand, the intrinsic potentials of essentially all carbonaceous materials are much higher than the reduction potentials of most solvents and salts; therefore, the formation of the SEI on these materials should be stepwise, during which the reduction of certain components could be more favored than the others.3 It was this preferential reduction that presented the possibility of modifying SEI formation through the use of electrolyte additives, an approach now widely adopted in the lithium-ion battery industry.4 Once formed, the key chemical species in these SEI films, whether on lithium metal electrodes or carbonaceous electrodes polarized to low potentials, have been generally considered as alkylcarbonate 共or semicarbonate兲. The identification of these species, apparently the reduction products of carbonate solvents undergoing a one-electron process, was firmly established on the basis of spectroscopic analyses of various electrode surfaces, mainly conducted by Aurbach and co-workers,5-7 and have been one of the most important achievements concerning lithium-ion chemistry during the last decade. Compared with the extensive characterizations of SEI on various carbonaceous anodes through both spectroscopic and chemical means, there had been a limited number of investigations on the potential range where the SEI is formed. Early observations that carbonate-based electrolytes experienced a plateau near 0.80 Va,8 with concurrent gas evolution9,10 has led researchers to believe that this potential lies in the vicinity of 1.00 V, and in situ analysis using differential electrochemical mass spectroscopy confirmed this conclusion.11 However, a few other studies, employing the plasma spectrometer12 or electrochemical impedance spectroscopy 共EIS兲13

* Electrochemical Society Active Member. z

E-mail: [email protected]

have placed the occurrence of SEI formation in much lower potential ranges, which overlap with the lithium intercalation process, i.e., below 0.25 V. These latter authors even suggested that, as the contribution to the building-up of this protective interface may not come from a singular reaction, the formation of SEI should progressively occur in a wide potential range, instead of being characterized by any particular electrochemical process at a certain potential. Nevertheless, in all those studies, the reduction of electrolyte solvents, especially cyclic carbonates such as propylene carbonate 共PC兲 and ethylene carbonate 共EC兲, was believed to be the predominant source of SEI building blocks, whereas the role of salt was considered inconsequential or even detrimental to the formation of such surface films. In this sense, our observation that the electrolytes based on lithium bis共oxalato兲borate 共LiBOB兲 stabilizes the graphitic anode materials even in the strongly exfoliating solvent, PC, probably served as the first evidence that the electrochemical reduction of salt anion may also actively participate in the building of the SEI.14 The emergence of this new salt, initially proposed by Lischka et al.15 and Xu and Angell,16 has undoubtedly generated a new SEI chemistry that is yet to be fully understood. A more recent report disclosed that the concentrated solution of lithium bis共perfluoroethylsulfonyl兲 imide in PC also behaves in a similar manner, further highlighting the participation of salt anion in SEI formation.17 Recognizing the potential importance of LiBOB for lithium-ion technology, this research group at the Army Research Laboratory has been engaged with an extensive characterization of this new salt regarding its electrochemical behavior in the environment of lithium-ion chemistry.14,18-22 Our efforts ranged from the fundamental ionics studies to the performance assessment of the electrolytes based on LiBOB at various temperature limits, while the major emphasis was placed on the understanding of the new SEI chemistry presented by LiBOB. Thus far we have learned that 共i兲 the reduction of BOB⫺ anion seems to occur below 1.0 V, very likely later than the reduction of carbonate solvents; (ii) the other reduction process that has been observed at 1.7 V in most LiBOB-based electrolytes should be attributed to the presence of impurities such as oxalate salts; (iii) the SEI formed in LiBOB-based electrolytes is generally characterized by the especially high abundance of semicarbonate species; and (i v ) the presence of EC in these electrolyte solutions is critical for high-temperature performance of the electrolytes in lithium-ion cells, suggesting that the co-presence of EC- and BOB⫺-originated species in the SEI layer are simultaneously responsible for the interface stability at elevated temperatures. Among the various analytic means we have adopted, X-ray photoelectron spectroscopy 共XPS兲 has been identified as an effective tool in revealing the chemical compositions of the surface film. Especially, due to the characteristic C 1s spectra feature associated

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

A2107

Figure 1. The potential dependence of the C 1s spectra for the anode surfaces that were partially formed in LiBOB/PC. The numeric labeling in the voltage profile 共central panel兲 indicates the PFPs at which XPS analysis was conducted and corresponds to the spectra in the left and right panels.

with the high abundances of semicarbonate species, XPS provides a facile way to detect the presence of the reduction products of the BOB⫺ anion.20 Here we intend to take advantage of the sensitivity of this technique to establish the potential dependence of the surface chemistry of graphitic anode, so that the onset potential for the reductive decomposition of BOB⫺ anion, determined by other means previously, can be confirmed. The knowledge obtained would be of special importance to the forming techniques of lithium-ion cells, the design and synthesis of new lithium salts, and the screening of electrolyte additives.

from the substrate 共carbon tab兲, was used as the internal reference to calibrate the energy position. The binding energy location of the elemental carbon 1s peak in graphite electrode remained unchanged 共⬍0.3 eV兲 during the entire lithiation/delithiation process and thus qualified as an internal reference. This constancy probably arises from the fact that the electrons in the 1s shell are less subjected to the change of electronic state caused by the presence of lithium ions in the graphene interstitial structure. To reveal the depth profile of the surface composition, continuous sputtering with Ar⫹ of 5 keV was conducted on an area of 4 ⫻ 4 mm at an interval of 4 min.

Experimental

Results and Discussion

LiBOB was either prepared at the Army Research Laboratory through the procedures described in our previous publications18 or provided by Chemetal, Germany. The high purity of this salt from both sources was confirmed by various analytic means. PC from Ferro 共⬎99.9%兲 was dried over neutral alumina powder until the moisture level 共as determined by Karl Fischer titration兲 was below 10 ppm. The solution of 1.0 molality 共m兲 LiBOB in PC was prepared in a Vacuum Atmospheres glove box with oxygen and moisture levels below 10 and 5 ppm, respectively. A graphitic anode coated on Cu foil was provided gratis by Saft and cut into disks of 1.27 cm2 for coin cell assembly. Stainless steel coin cells 共size 2335兲 served as the testing vehicle for all the anode half-cells in galvanostatic cyclings. They were sealed by a Rayovac Multipress in a dry room with dew point below ⫺70°C. The cycling tests were conducted on a Maccor Battery Tester series 4000. Anode half-cells were assembled using a Saft anode and a lithium disk of the same area, which were then typically subjected to lithiation/delithiation procedures at a constant rate of C/10. Those cells were then terminated at various lithiation potentials before being opened in the glove box, so that the anode at different states of charge could be obtained for analysis. Before the XPS experiments, the graphite anodes were washed twice with the solvent mixture ␥-butyrolactone/dimethyl carbonate 共DMC兲, and then twice with neat DMC. The rationale behind this cleansing protocol was described in our previous paper.20 The anode samples thus cleansed were dried under vacuum 共⬍0.01 Torr兲 overnight at room temperature and loaded to a standard sample holder and covered by a stainless steel top with an aperture of ␾5 mm. As an external reference to interpret the spectra, the powdery LiBOB crystal was immobilized on a 12 mm carbon tab 共Structure Probe兲, which was then loaded into a similar sample holder without the stainless steel cover. A vacuum sample transporter with airtight adaptor was used to transport and load the dried samples from Ar atmosphere into the XPS target chamber. Surface analysis was then conducted on those electrodes with a PHI 5800 XPS system, where an Mg K␣ excitation source was used. The resolution of the measurement generally falls within 0.3 eV. The elemental carbon peak at 284.8 eV, which either arises from the active agent 共graphite兲 in the composite anode or

In the previous part of this paper, we discussed the effect of the solvent on the chemical nature of the SEI formed when LiBOB was the electrolyte solute, and showed that, by gradually removing the outlying surface species with Ar⫹-sputtering, it is possible to differentiate the contributions of solvent and salt to SEI.22 As a continued effort, in this work we attempt to study the change of surface chemistry with the cell potential in the initial formation cycle of a lithium-ion cell. To exclude any unwanted interference from EC reduction, we deliberately selected LiBOB/PC as the baseline electrolyte by taking advantage of the known fact that PC reduction could not form a protective SEI to enable the lithiation of the graphitic carbon. In this EC-free electrolyte solution, the chemical composition of the anode surface should predominantly reflect the consequence of BOB⫺ anion reduction, the products of which function alone as the SEI building blocks. Thus, the potential profile of the surface chemistry on the anode was obtained with the assistance of a ‘‘pre-formation technique,’’ as described in our earlier paper,19 in which a series of anode half-cells loaded with LiBOB/PC were charged 共lithiated兲 down to a series of potentials 共denoted the ‘‘pre-formation potential’’ or PFP兲, so that the graphitic anodes in those cells experienced only partial SEI buildup, as much as the PFP of each individual cell would allow. Because the lithiation of these cells is interrupted before a complete forming cycle 共i.e., lithiation followed by delithiation兲 is finished, the surface status of the graphite anodes should be preserved in the ‘‘snapshots’’ of the incomplete SEIs in the process of formation frozen at these specific PFPs. We then conducted XPS analysis on the individual graphitic anode samples recovered from these partially lithiated cells. To obtain such a potential profile in a complete forming cycle, similar procedures were also applied to the charging 共delithiation兲 process of those half-cells that had undergone full lithiation 共at 0.05 V兲. The potential dependence of the resultant XPS images thus established is shown in Fig. 1, which correlates the galvanostatic voltage profile of the anode half-cells 共central panel兲 to the C 1s spectra of these partially formed anode surfaces 共left panel兲 and the anode surfaces that have experienced the full lithiation but are ‘‘frozen’’ at various delithiation potentials 共right panel兲. Apparently, the characteristic peak corresponding to the semicarbonate species 共at ⬃289 eV兲 emerges only after 1.0 V, in good

A2108

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲 Table I. Comparison of the 1s X-ray photoelectron binding energy levels for graphitic surfaces that were formed in LiBOBÕPC at different states of charge.

LiBOB salt Graphite surface 共before lithiation, 2.00 V兲 Graphite surface 共fully lithiated, 0.05 V兲 Graphite surface 共fully delithiated, 2.00 V兲 a

Figure 2. The change of C 1s spectra collected from the graphitic anodes with the lithiation process. The potentials of these anodes 共PFPs兲 are indicated in the graph, and the bulk salt LiBOB is also shown as an external reference to observe the shift in the binding energy of semicarbonate species on graphitic anodes.

agreement with the previous conclusions drawn on the results of either electrochemical or EIS experiments.19 More significantly, because the same cleansing procedure was applied to all the anode samples, the complete absence of this peak in the initial two anode surfaces that were formed down to 1.0 and 2.0 V actually confirmed that the semicarbonate species observed in the other anode surfaces with lower PFPs were the result of the electrochemical reduction of BOB⫺ anion, instead of the simple precipitation or encapsulation of the unreacted BOB⫺ anion. Furthermore, the effectiveness of the cleansing procedure toward LiBOB, already proven previously in similar XPS experiments, is again verified, in addition to the ‘‘blank test.’’20 The fact that the signature peak at 289 eV remains independent from the further lithiation and delithiation steps indicates the characteristic of a stable SEI, as shown by the constant presence of this peak during the remaining potential range below 1.0 V and also the entire process of delithiation. This same trend is repeatedly seen in this work when the photoelectron spectra of other elements of interest 共O, Li, and B兲 are analyzed. It is expected that in the extended cycles following the formation, these semicarbonate species as the key ingredient of the protective interface would remain inert against electrochemical reduction or dissolution and simultaneously serve as the electron-insulating layer, while the cell chemistry proceeds at the electrolyte/electrode interfaces. For a closer examination of the chemical changes during the initial formation stage, we plotted in Fig. 2 the C 1s spectra of bulk salt LiBOB together with those collected on the surface of anodes that were only partially lithiated, as shown in the left panel of Fig. 1. The carbon tab substrate, on which the powdery LiBOB is immobi-

C 1sa 共semicarbonate兲

O 1s

Li 1s

B 1s

289.72 NA

533.47 532.57

56.80 56.01

194.02 192.75

289.02

532.47

55.45

192.50

289.05

532.47

55.50

192.59

For C 1s only the peak corresponding to semicarbonate 共⬃289 eV兲 was listed, and the bulk LiBOB salt was listed as an external reference.

lized, generates an elemental carbon signal, which is barely sufficient in abundance to be used as reference for the calibration of the energy scale because of the conspicuous signal at ⬃289.72 eV corresponding to oxalate carbonyl. The difference between oxalate carbonyl and the semicarbonate species observed on lithiated anode surfaces 共289.02-289.05 eV, see Table I兲 is ca. 0.70 eV, which lies well above the resolution limit of the current analytic technique and therefore should be considered as the consequence of electrochemical reduction. In combination with the results of Fourier transform infrared 共FT IR兲 analysis,21 we believe that the lower binding energy for the carbonyl on the anode surfaces should be attributed to the ring-opening process of the BOB⫺ anion reduction. This process produces a three-coordinated borate that is less electronegative, therefore it would be easier for the photoelectrons in the 1s shell of the carbonyl carbon atom to be excited and subsequently escape 共see Scheme 1兲.

Scheme 1. The possible single electron mechanism for the electrochemical reduction of LiBOB on the surface of graphitic anode.

It is also interesting to note in Fig. 2 that, as the lithiation proceeds, the transition of the carbonyl signals from 289.70 eV 共oxalate兲 to 289.02-289.05 eV 共semicarbonate species兲 actually occurs with the PFPs being lower than 0.55 V, while the first appearance of the semicarbonate species on the graphitic anode surface 共PFP 1.0 V兲 still bears close resemblance to the original BOB⫺ anion in terms of the binding energy scale. We suggest that this shift in binding energy location may be the indicator of the emergence of a protective surface layer consisting of the reduction products of BOB⫺ anion, which comes into shape around 0.55 V, again in good accordance with our previous observations through the pre-formation technique19 and EIS studies. The binding energy of the semicarbonate species that were generated by the electrochemical reduction of BOB⫺ anion remains between 289.02-289.05 eV throughout the rest

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

A2109

Figure 3. The change of O 1s photoelectron binding energy on the surface of graphitic anode during 共left兲 lithiation and 共right兲 delithiation processes. The scan labeled OCV in lithiation process 共left兲 corresponds to the anode sample that underwent electrolyte soaking and routine washing but no lithiation.

of the forming cycle, including the delithiation process, as shown in Table I. Assisted by the XPS sputtering results obtained in the previous part of this paper,22 it is logical to infer that the reduction of BOB⫺ anion occurs after the reduction of EC if the latter is present in the electrolyte solution. However, in the current work, because PC reduction 共occurring at ⬃0.80 V兲 is unable to produce any products that could effectively adhere well to the anode surface, the SEI composition as detected by XPS should reflect the changes that arise predominantly from the chemistry of the BOB⫺ anion. Similarly, Fig. 3 compares the O 1s spectra collected on the anode surfaces that either were partially lithiated 共Fig. 3, left兲 or were fully lithiated at 0.05 V but were interrupted during delithiation 共Fig. 3, right兲. Although the anode surface that underwent no lithiation 共labeled OCV兲 generates a peak near 533.47 eV that is typical of carbonate oxygen, a smooth shift toward a lower binding energy accompanies the lithiation process. The binding energy location of the final carbonate oxygen eventually stabilizes at 532.47 eV after 0.55 V and remains constant even during the subsequent delithiation 共Table I兲. This trend would obviously lead to the same conclusion that was drawn on the C 1s spectra results. One notable observation of the O 1s spectra collected on the graphitic anode surfaces is the absence of the signal corresponding to Li2 O, which should be located at 528 eV and has been observed on anode surfaces that were formed in LiPF6 - or LiAsF6 -based electrolytes by numerous authors, including Kanamura et al.23 and BarTow et al.24 Considering that Li2 O is the fully reduced form of the SEI species and Kanamura et al. once proposed that such simple inorganic anions should be present only in the inner layer of SEI,23 we tentatively attributed the absence of Li2 O in our XPS results to either the insufficient exposure of the electrolyte to lithiated carbon during the initial forming cycle or the resultant thick SEI depositions on the anode surface from the reduction products of BOB⫺ anion. However, similar tests conducted on the well-cycled anode surfaces in various LiBOB-based electrolytes along with the prolonged sputtering in these tests still fail to detect this chemical species. Further investigations are underway to gain more insight into this interesting discrepancy. Relative to the C 1s and O 1s spectra, the Li 1s spectra were much less informative due to the narrow energy band of all Li atoms in various chemical environments with the given resolution of instrumentation. Still, a steady transition was observed during the

lithiation process from 56.80 eV of LiBOB to a lower binding energy value at 55.45 eV, indicating a less electronegative environment for the 1s electrons of lithium 共Fig. 4, left兲. Again, this lower binding energy level remained almost constant during the subsequent delithiation process 共Fig. 4, right兲. Attempts to resolve the issue about Li2 O using Li 1s spectra failed to render more conclusive results than the O 1s spectra, and further complications arose from the inconsistency in the data of Li 1s binding energy for Li2 O reported in literature. According to Kanamura, the photoelectron in the 1s shell of Li in Li2 O should generate a peak at ca. 53 eV,23 which would be rather distinguishable from the peak at 55-56 eV of Li in LiOH or Li2 CO3 . However, the Handbook of X-Ray Photoelectron Spectroscopy, published by Physics Electronics, Inc. and widely used as the bench reference, placed the Li 1s binding energy value of Li2 O between 55-56 eV,25 i.e., in the same range with the other lithium-containing species such as LiOH, Li2 CO3 , lithium semicarbonates and lithium oxalates. Those species are expected to be detected on the anode surfaces, and their presence would likely camouflage any Li2 O signals. Nevertheless, the Li 1s spectra in the current work still fail to provide an answer to the question about the absence or presence of Li2 O on the anode surfaces formed in LiBOB electrolytes, no matter which source of binding energy data is more accurate. No discernible signals below 54 eV were detected, and deconvolution of the main signals centered between 54-56 eV did not yield any meaningful interpretation. As in the O 1s spectra, prolonged sputtering of the anode surfaces did not detect any new signals in the region either, leaving the issue of Li2 O unsettled. However, the information derived from the B 1s spectra seems to be more definitive in clarifying the surface chemistry that occurs during the formation of the SEI. Repeating the trend as observed in the C 1s, O 1s, and Li 1s spectra, the binding energy of the boron 1s electrons exhibited a gradual transition from 193.86 eV before lithiation to 192.50 eV at the full lithiated state, and remained relatively constant thereafter in the delithiation processes 共Fig. 5, left and right, and Table I兲. Considering that the binding energy of the B 1s electron in the unreacted LiBOB salt, which has a boron center coordinated by four electron-withdrawing carbonyls, is located at 194.02 共Fig. 5, left兲, it is apparent that the shift toward a lower binding energy is caused by a configuration change of the boron center, where the electrons in the B 1s shell become less bound,

A2110

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

Figure 4. The change of Li 1s photoelectron binding energy on the surface of graphitic anode during 共left兲 lithiation and 共right兲 delithiation processes. The scan labeled OCV in lithiation 共left兲 corresponds to the anode sample that underwent electrolyte soaking and routine washing but no lithiation. Bulk LiBOB was shown as external reference.

probably due to the decreased number of electron-withdrawing coordination units. Because the tri-coordinated borate species, such as B2 O3 or H3 BO3 , all exhibited the B 1s signal in the vicinity of 192-193 eV, it is a reasonable assumption that this configuration change involves a ring-opening process that reduces the symmetry of boron from the tetragonal of the BOB⫺ anion to the trigonal of an orthoborate. A similar conclusion was drawn earlier based on the interpretation of FTIR analysis, which was conducted on the anode surface formed in LiBOB-based electrolyte. Thus, it seems likely that during the electrochemical reduction of BOB⫺ anion the tetrahedral symmetry of the anion is broken through a ring-opening mechanism, probably as shown in Scheme 1, and the resultant species that eventually constitute the main composition of the SEI layer consist of carbonyl moieties and tri-coordinated boron centers.

To reveal the correlation between the PFP of the anodes and the surface chemistry thereon in a more apparent and probably also more quantitative manner, the shifts in binding energy (⌬B.E.), as referenced against the bulk salt LiBOB, for various spectra were plotted against the PF P values at which the lithiation is terminated 共Fig. 6兲. Taking the external reference LiBOB as the origin where electrochemical reduction has not occurred to the BOB⫺ anion, one can clearly see that for all these spectra the major shifts unanimously occurred at around 0.55 V and stabilized thereafter, regardless of the vast energy range 共⬎500 eV兲 in which these electrons in different elements were excited. We strongly believe that all the plots in Fig. 6 unequivocally indicate a major electrochemical process that may start at potentials as high as 1.00 V, as suggested by

Figure 5. The change of B 1s photoelectron binding energy on the surface of graphitic anode during 共left兲 lithiation and 共right兲 delithiation processes. The scan labeled OCV in lithiation 共left兲 corresponds to the anode sample that underwent electrolyte soaking and routine washing but no lithiation. Bulk LiBOB is shown as external reference.

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

Figure 6. The potential dependences of the binding energy shift (⌬B.E.) for C 1s, O 1s, Li 1s, and B 1s spectra, respectively. The shift was referenced against the corresponding signals of the bulk salt LiBOB. The entire lithiation/delithiation cycle is represented; the left and right halves correspond to the discharge 共lithiation兲 and charge 共delithiation兲 process for the anode half-cell, respectively.

B 1s and Li 1s spectra, but completes at the threshold potential of 0.55 V. An independent semiquantitative analysis was also carried out based on the abundances of these related elements detected on the anode surfaces, and the results led to basically the same conclusion. Thus, the concentrations of all four elements are plotted against the PFP, as shown in Fig. 7. As expected, during the initial lithiation stage, the elemental carbon 共arising from the graphene structure of

A2111

the anode and represented by the signal at 284.5 eV in the C 1s spectra兲 was the dominant species on the surface. A drastic drop in its abundance occurs between 3.0-1.0 V, accompanied by concomitant increases in the abundances of O, B, and Li, and indicates the gradual coverage of the bare graphene surface by the reduction products from either solvent 共PC兲 or salt anion (BOB⫺). Note that the simultaneous increase in the surface concentrations of B and Li during this time may be caused by the decrease of carbon on the surface rather than the actual increase of these two elements; however, the increase of oxygen, most of which is in the form of carbonate-like species, unambiguously marks the appearance of new chemical species on the surface. The decreasing and increasing abundances for carbon and oxygen both stabilize in the vicinity of 0.50 V and remain relatively constant thereafter considering the experimental errors. This potential dependence is markedly consistent with the variations of either the emergence of the semicarbonate signal or the binding energy shift with the lithiation/delithiation state of the graphitic anode. Because C 1s spectra are particularly sensitive to the presence of BOB⫺ anion reduction products that are characterized by carbonylrich species, a further quantitative analysis was carried out with distinctions made between the total carbon content on the surface and that of the semicarbonate species represented by the signal at ⬃289.0 eV. The abundance of this latter species, which is generally believed to be the key ingredient for an effective SEI and hence of high significance to the effectiveness of the resultant SEI, was then plotted in a similar manner against the PFP of the anode, as shown in the inset of Fig. 7. Note that, although the total abundance of carbon decreases drastically during the initial lithiation process 共3.00-1.0 V兲, the surface concentration of carbon contained in the semicarbonate species exhibited an increase, matching that of the oxygen-containing species in the same potential range. A logical rationale would be that the new species deposited on the anode surface at this stage consists predominantly of semicarbonate species that are rich in carbonyls, which is the source of the O 1s signal at ⬃532 eV and the C 1s signal at ⬃289 eV. Chemically interpreted, the plots in Fig. 7 paint a dynamic picture of how the bare graphene surface is gradually covered by these BOB⫺-originated products in the potential range of 1.0-0.55 V. The stability of the resultant SEI against solvent dissolution is reflected by the relatively constant concentrations of all these chemical species on the anode surface. These observations are in good agreement with the conclusion we drew earlier: that the electrochemical reduction of BOB⫺ anion occurred after 1.0 V, and that a stable surface layer with complete protection of the anode surface is achieved in the vicinity of 0.55 V. Conclusion

Figure 7. The relative abundances of various elements on the graphitic anode surface and their potential dependences during a full forming cycle 共left: lithiation; right: delithiation兲. Inset: Potential dependence of the relative abundance of semicarbonate species located at ⬃289 eV in C 1s spectra. Its relative abundance was normalized against that of the total carbonaceous species.

The surface chemistry of graphitic anodes that were cycled in LiBOB/PC to various cell potentials was studied by XPS, and its dependence on the electrode potential was established, mainly based on the analysis of the signature C 1s spectra originated from the decomposition of BOB⫺ anion. During the initial lithiation between 3.00-1.00 V, the binding energy locations of the photoelectrons excited from the 1s shells of all the elements of interest 共C, O, B, and Li兲 consistently showed a gradual shift from those of the bulk salt 共LiBOB兲 to lower values, with the characteristic signal of semicarbonate species emerging at 1.00 V. The potential dependences of both the binding energy shifts and the surface abundances of the related elements unanimously showed an apparent deflection at ⬃0.55 V, which should mark a major surface event change. In combination with previous results, we believe that this event should correspond to the completion of SEI formation, for which the electrochemical reduction of BOB⫺ anion should be responsible. The fact that these parameters 共binding energy shift and surface abundance兲 stabilize after this threshold potential betokens the stability of the SEI formed during the cell operation. From the matching potential dependences of both C and O concentrations on the anode

A2112

Journal of The Electrochemical Society, 151 共12兲 A2106-A2112 共2004兲

surface, it is also apparent that the predominant chemical species deposited as the main building material of SEI are likely the semicarbonates. Acknowledgment The authors acknowledge Dr. Kamen Nechev and Dr. Guy Chagnon of Saft America for the electrode materials and Dr. JanChristoph Panitz of Chemetal for the LiBOB salt of high purity. The U.S. Army Research Laboratory assisted in meeting the publication costs of this article.

References E. Peled, J. Electrochem. Soc., 126, 2047 共1979兲. R. Fong, U. von Sacken, and J. R. Dahn, J. Electrochem. Soc., 137, 2009 共1990兲. Y. Ein-Eli, Electrochem. Solid-State Lett., 2, 212 共1999兲. G. E. Blomgren, J. Power Sources, 119Õ121, 326 共2003兲. D. Aurbach, M. L. Daroux, P. W. Faguy, and E. Yeager, J. Electrochem. Soc., 134, 1611 共1987兲. 6. D. Aurbach, Y. Gofer, M. Ben-Zion, and P. Aped, J. Electroanal. Chem., 339, 451 共1992兲. 7. O. Chusid, Y. Ein-Ely, D. Aurbach, M. Babai, and Y. Carmeli, J. Power Sources, 43Õ44, 47 共1990兲. 8. A. N. Dey and B. P. Sullivan, J. Electrochem. Soc., 117, 222 共1970兲. 1. 2. 3. 4. 5.

9. M. Arakawa and J. Yamaki, J. Electroanal. Chem. Interfacial Electrochem., 219, 273 共1987兲. 10. Z. X. Shu, R. S. McMillan, and J. J. Murray, J. Electrochem. Soc., 140, 922 共1993兲. 11. R. Imhof and P. Nova´k, J. Electrochem. Soc., 145, 1081 共1998兲. 12. Y. Matsumura, S. Wang, and J. Mondori, J. Electrochem. Soc., 142, 2914 共1995兲. 13. S. Zhang, M. Ding, K. Xu, J. Allen, and T. R. Jow, Electrochem. Solid-State Lett., 4, A206 共2001兲. 14. K. Xu, S. Zhang, B. A. Poese, and T. R. Jow, Electrochem. Solid-State Lett., 5, A259 共2003兲. 15. U. Lischka, U. Wietelmann, and M. Wagner, Ger. DE, Pat. 19,829,030 C1 共1999兲. 16. W. Xu and C. A. Angell, Electrochem. Solid-State Lett., 4, E1 共2001兲. 17. S. Jeong, M. Inaba, Y. Iriyama, T. Abe, and Z. Ogumi, Electrochem. Solid-State Lett., 6, A13 共2003兲. 18. K. Xu, S. Zhang, T. R. Jow, W. Xu, and C. A. Angell, Electrochem. Solid-State Lett., 5, A26 共2002兲. 19. K. Xu, S. Zhang, and T. R. Jow, Electrochem. Solid-State Lett., 6, A117 共2003兲. 20. K. Xu, U. Lee, S. Zhang, M. Wood, and T. R. Jow, Electrochem. Solid-State Lett., 6, A144 共2003兲. 21. G. V. Zhuang, K. Xu, T. R. Jow, and P. N. Ross, Jr., Electrochem. Solid-State Lett., 7, A224 共2004兲. 22. K. Xu, U. Lee, S. Zhang, J. L. Allen, and T. R. Jow, Electrochem. Solid-State Lett., 7, A273 共2004兲. 23. K. Kanamura, H. Tamura, S. Shiraishi, and Z. Takehara, J. Electrochem. Soc., 142, 340 共1995兲. 24. D. Bar-Tow, E. Peled, and L. Burstein, J. Electrochem. Soc., 146, 824 共1999兲. 25. Handbook of X-Ray Photoelectron Spectroscopy, Physical Electronics, Inc., Eden Prairie, MN 共1995兲.