Thermal Behavior of Solid Electrolyte Interphase Films Deposited on Graphite Electrodes with Different States-of-Charge

The thermal degradation of surface ﬁlms (solid electrolyte interphase, SEI) and their repairing behaviors are examined on graphite electrodes that have different states-of-charge (SOC). In detail, a fully passivating SEI layer is generated in advance and then stored at 85 ◦ C for a prolonged time. The surface ﬁlms (SEI) are thermally degraded upon high-temperature exposure. The damaged ﬁlms are repaired by electrolyte decomposition and concomitant ﬁlm deposition, during which Li + ions/electrons are supplied from the graphite electrodes, which appears as an increase in the open-circuit voltage (OCV) of the graphite electrodes. The repaired surface ﬁlms show a comparable passivating ability along with a similar morphology and chemical composition to that of the initial SEI layers. The degradation/repairing continues until the graphite electrodes are fully de-lithiated (OCV = 3.0 V vs. Li/Li + ). Once the graphite electrodes are fully de-lithiated, the damaged surface ﬁlms cannot be repaired because the Li + ions/electrons are exhausted in the graphite electrodes. Because of the incomplete repairing, the graphite surface becomes poorly covered by surface ﬁlms, which leads to a loss of passivating ability.

At present, graphite is the most widely-used negative electrode for lithium-ion batteries (LIBs). At the working potential of the graphite electrode (0.2∼0.3 V vs. Li/Li + ), most of the conventional carbonatebased electrolytes are not stable against reductive decomposition, and the decomposed products deposit as a film, also known as the solid electrolyte interphase (SEI). [1][2][3][4][5][6] The formation of SEI is undesirable because Li + ions and the equivalent amount of electric charges are consumed in this process, which appears as an irreversible capacity. However, SEI formation is critically important for graphite electrodes to be successfully used as a negative electrode for LIBs. Namely, the SEI layers are electronically insulating. Hence, any electrochemical reactions at the electrode/electrolyte interface are hindered due to negligible electron tunneling through the SEI layer. As a result, additional electrolyte decomposition is prevented once it forms at a certain thickness. That is, the surface film passivates the graphite electrodes. Another beneficial feature of the SEI layer is that it is a Li + ion conductor, such that it does not induce concentration polarization in lithium reactions but enables reversible Li + intercalation/de-intercalation for graphite negative electrodes.
In order for the SEI layer to serve as an effective passivating film, however, it should meet other requirements in addition to its Li + ion conducting and electronically insulating properties. It should uniformly cover the electrode surface to passivate the whole graphite surface. A thinner SEI layer is generally favored to minimize the ohmic resistance as far as it plays the passivating role. Thermal stability is another requirement because SEI films are vulnerable to damage upon high-temperature exposure. The organic or inorganic ingredients in SEI layers may be either thermally decomposed or dissolved. [7][8][9][10] Once thermally degraded, the SEI layer loses its passivating ability to cause additional electrolyte decomposition. The damaged SEI is repaired by this electrolyte decomposition and concomitant film deposition. This repair process leads to several unfavorable features. For instance, the graphite electrodes are self-discharged because Li + ions/electrons in the graphite electrodes are consumed in the formation of the SEI. [7][8][9][11][12][13][14][15] Moreover, cell temperature increases due to heat evolution during the repair process, which can trigger a thermal runaway, in which the cell temperature increases uncontrollably because of continued exothermic reactions. 16,17 Changes in SEI morphology and chemical compositions upon high-temperature exposure have been reported in the literature. 7,9,12,15,18 In all these previous reports, however, little attention has been given to thermal stability and the passivating ability of SEI layers. This study investigates these issues. More specifically, in this work, SEI films were generated on graphite electrodes with different states-of-charge (SOC) and exposed to a high-temperature environment. It was found that the SEI films are degraded during the hightemperature exposure, but repaired by the forthcoming electrolyte decomposition and film deposition. Interestingly, the repairing behavior was critically affected by the degree of lithiation (states-of-charge, SOC) in the graphite electrodes. Thermal degradation/repairing of the SEI films were assessed by performing an open-circuit voltage (OCV) measurement at elevated temperatures. The change in surface films with respect to chemical composition and morphology was examined with X-ray photoelectron spectroscopy (XPS) and scanning electron microscope (SEM).

Experimental
To prepare the graphite composite electrodes, a mixture of mesocarbon microbeads (MCMB-10-28, average particle diameter = 10 μm, graphitization temperature = 2800 • C, Osaka Gas Co.), Super P (a conductive carbon), and poly(vinylidenefluoride) (PVdF, Kureha, KF-1300) binder (85:5:10 in wt. ratio) was dispersed in N-methyl pyrrolidone. The resulting slurry was coated onto a piece of copper foil (a current collector) and dried at 120 • C for 12 h under vacuum. Twoelectrode 2032-type coin cells were fabricated with an as-prepared MCMB composite electrode, Li foil (as a counter electrode) and a glass fiber filter (Advantec, GA-55, as a separator). The used electrolyte was 1.0 M LiPF 6 dissolved in a mixture of ethylene carbonate (EC)/diethyl carbonate (DEC) (1:1 in vol. ratio). The cells were assembled in an argon-filled dry box and cycled in a temperature-controlled oven.
The assembled coin-cells were cycled five times at 25 • C to generate a SEI layer on the MCMB electrode. To this end, the Li/graphite cell was cycled at a current density of 37.2 mA g −1 in a voltage range of 0.001∼2.0 V (vs. Li/Li + ). The OCV measurements were performed with the MCMB electrodes that had different SOCs. 19 For this experiment, the charge/discharge cycling was stopped at a predetermined SOC (SOC100 = 300 mA h g −1 ), and placed in a temperature-controlled oven to monitor the OCV change while the oven temperature was fixed at 85 • C.
For post-mortem field-emission SEM (FE-SEM, JSM-6700F, JEOL) and X-ray photoelectron spectroscopy (XPS) analyses, the cells were dismantled in an argon-filled dry box, and the cycled electrodes were collected and washed with DEC. A hermetic vessel was used to transfer the electrode samples from the dry box to the instrument chamber. The XPS data were collected in an ultra-high vacuum multipurpose surface analysis system (Sigma probe, Thermo, UK) that operates at a base pressure of < 10 −10 mbar. The photoelectrons were excited by Al Kα (1486.6 eV) radiation at a constant power of 150 W (15 kV and 10 mA); the X-ray spot size was 400 μm 2 . During data acquisition, a constant-analyzer-energy mode was used at a pass energy of 30 eV and a step of 0.1 eV. The atomic concentration, which is the ratio of the number of atoms of the element of interest to that of others, was calculated from the XPS spectra with the atomic sensitivity factor. 20 Fig. 1 shows the evolution of the open-circuit voltage (OCV) during storage at 85 • C, which was traced from three graphite electrodes with different SOCs: SOC0, SOC25, and SOC50. The initial OCV values before storage were 1.5, 0.15, and 0.12 V (vs. Li/Li + ), respectively. The OCV increased upon storage for the three samples. In the case of SOC0, the OCV increased rapidly to reach 3.0 V, which is the value for fully de-lithiated graphite (x = 0 in Li x C 6 ). When extending the time to reach 3.0 V for SOC25, it eventually reached the fully delithiated state (3.0 V). In the case of SOC50, however, the final OCV after 24 h of storage was still 0.6 V, showing that the graphite electrode is still partially lithiated (x > 0 in Li x C 6 ). Two features should be noted here. First, the OCV increase is a signature of a transfer of Li + ions and the equivalent amount of electrons from the graphite electrode to the electrolyte at the interface during the storage. Such a charge transfer is impossible if the SEI layer is perfectly passivating the graphite electrode. Hence, the OCV increase signifies a loss of passivating ability as a result of thermal degradation of the surface films. 19 This feature will be discussed in detail in a later section of the paper. Second, the graphite electrodes stored at high-temperature can be divided into two groups. In Fig. 1, the graphite electrode with SOC0 stored for 24 h is fully de-lithiated (OCV = 3.0 V). Similarly, the graphite electrodes with SOC0-12 h and SOC25-24 h are fully delithiated. These are named group A. A characteristic feature for group A is that the graphite electrodes are fully de-lithiated (Li + ions and electrons are exhausted, x = 0 in Li x C 6 ), such that Li + ions/electrons cannot be supplied from the graphite electrodes during the repair period of the thermally damaged SEI films. In contrast, the graphite electrodes with SOC25-12 h, SOC50-12 h and SOC50-24 h in Fig. 1,   Figure 2. Differential capacity (dQ/dV) plots obtained in the pre-cycling (before storage) and after the storage at 85 • C. The galvanostatic charge/discharge was performed at 25 • C, from which the dQ/dV profiles were derived: (a); in the initial two lithiation periods before storage, and (b)/(c); in the first cycle right after the high-temperature storage for group A and B.

Results and Discussion
which are named group B, still carry Li + ions and electrons (x > 0 in Li x C 6 ). The SEI layers on group B exposed to a high-temperature environment are thermally degraded, yet the graphite electrodes are still partially lithiated. It is thus expected that the Li + ions and electrons in the graphite electrodes can participate in the repair process.
In order to confirm the thermal degradation of the SEI layers and loss of passivating ability during high-temperature storage, the graphite electrodes belonging to group A and B were placed back into the temperature-controlled oven (25 • C) and cycled galvanostatically. Fig. 2a shows the lithiation dQ/dV profiles obtained in the pre-cycling period (before high-temperature storage), in which a reduction peak appears at 0.7 V (vs. Li/Li + ) in the first cycle, but disappears in the second cycle. Obviously, this feature is associated with the reductive decomposition of carbonate-based electrolytes and concomitant SEI deposition on the graphite electrodes. After storage, the reduction peak at 0.7 V appears again in group A (Fig. 2b), illustrating that electrolyte decomposition is severe because the SEI layer has been degraded during storage. In contrast, the reduction peak at 0.7 V is absent in the group B electrodes (Fig. 2c), illustrating that the SEI layers still possess a good passivating ability even after high-temperature storage. Fig. 3 shows a detailed analysis on the loss of passivating ability for the SEI layers on the group A and B electrodes. Fig. 3a shows the evolution of Coulombic efficiency before and after storage. Be- fore storage (pre-cycling period), the Coulombic efficiency is as low as 85∼87% in the first cycle for five samples, but it increases up to >99% within a few forthcoming cycles. This ensures a full passivation of the graphite electrodes as a result of SEI deposition in the pre-cycling stage. 19,21 After the high-temperature storage, however, the passivating ability of the SEI layers is sharply different between the two groups. As shown in Fig. 3a, group A shows a Coulombic efficiency of ∼85% right after storage, but the value steadily increases in the next few cycles. In contrast, the Coulombic efficiency of group B is >95% from the first cycle. The first irreversible capacity and the Coulombic efficiency obtained after storage are compared in Fig.  3b. A higher irreversible capacity and lower Coulombic efficiency is observed for group A. Generally, the irreversible capacity in graphite electrodes originates from irreversible charge consumption for electrolyte decomposition and SEI formation. [22][23][24][25] It is thus clear that the SEI layer in group A has lost its passivating ability during the high-temperature storage. Hence, electrolyte decomposition and film deposition are severe when they are cycled after storage. In contrast, the SEI layer on group B is rather robust maintaining its passivating ability even when exposed to a high-temperature environment. Fig. 4 shows the SEM images taken before and after storage, from which the morphological change of the SEI films upon hightemperature exposure is addressed. Fig. 4a shows the SEM image of a pristine electrode surface, in which the MCMB (larger particles) and Super P (smaller particles) are bound by PVdF binders. After pre-cycling, foreign materials are deposited on the electrode surface, which must be the surface film (SEI) derived from the electrolyte decomposition (Fig. 4b). After storage, the surface film on SOC0-24 h (Fig. 4c), which represents group A (surface film A, hereafter), crumbles and loses its initial morphology. In contrast, the surface film on SOC50-24 h, which represents group B (surface film B, hereafter), still uniformly covers the electrode surface (Fig. 4d). Evidently, surface film A is severely damaged, whereas surface film B appears intact.  To gain an insight into what happens on the surface films during high-temperature storage, the film compositions are analyzed with XPS. Fig. 5 shows the O 1s and F 1s spectra obtained from three different electrodes; before storage, one from group A and one from group B. All the XPS spectra are fitted according to the reported binding energy in Table I. [26][27][28] The spectra taken before storage show the presence of O-and F-containing chemical species in the SEI layer, which are derived from electrolyte decomposition (Fig. 5a). The changes in chemical compositions upon high-temperature storage, however, differ between the two groups. Namely, a notable spectral change appears on surface film A (Fig. 5b) compared to the spectra taken before storage (Fig. 5a), whereas the change is marginal for surface film B (Fig. 5c). In detail, when the O 1s spectra for surface film A are compared with those taken before storage (Fig. 5a), the peak at 533.5 eV increases at the expense of the peak at 532 eV. In addition, the intensity of the F 1s peak at 686.6 eV greatly increases on surface film A. In contrast, surface film B (Fig. 5c) is quite similar in chemical composition to that observed before storage (Fig. 5a). The strongest O 1s peak at 532 eV and F 1s peak at 685 eV are still observed in surface film B.
From the similarity in chemical composition between the initial film and surface film B, one can assume that the film formation mechanism is similar for the two processes; the initial SEI formation during the pre-cycling period and the repairing period for the damaged SEI layer. Namely, the initial surface film is generated by decomposition of electrolyte ingredient (carbonate-based organic solvent and LiPF 6 ), for which Li + ions/electrons are supplied from the graphite electrode. In the case of the group B electrodes, Li + ions and electrons can be supplied from the graphite electrodes even under an open-circuit condition because they are still partially lithiated. In other words, the SEI layer on the group B electrodes is damaged upon high-temperature storage either by dissolution or decomposition, but the surface film is repaired by a reaction between the electrolyte component and the Li + ions/electrons supplied by the partially lithiated graphite electrodes. If this is the case, the chemical composition of surface film B should be similar to that of the initial SEI layer because the films are derived from the electrolyte ingredients for both cases. Direct evidence for the supply of Li + ions/electrons from the graphite electrodes is the steady increase of OCV during storage in Fig. 1.
The absence of a self-remedying action on the group A electrodes can be assumed based on the absence of Li + ions/electrons in the fully de-lithiated graphites. The SEI layer is damaged during storage, but the surface film cannot be regenerated because Li + ions/electrons are exhausted in the graphite electrodes. Even in this case, however, thermally decomposed products could be deposited if any electrolyte components are thermally unstable. This possibility is confirmed by the XPS data shown in Fig. 5, in which a high population of Li x PF y O z is found in surface film A, which seems to be derived from LiPF 6 that is thermally unstable. 29,30 The results presented so far can be summarized by tracing what happens in the SOC25 sample during high-temperature storage under an open-circuit condition (Fig. 1). This graphite electrode is exposed to a high temperature while being partially lithiated. Upon high-temperature exposure, the SEI film is degraded but repaired with the aid of Li + ions/electrons inside the graphite. The OCV of graphite electrodes increases because of the consumption of Li + ions/electrons in this process. This degradation and repair are continued with an increase in OCV until the Li + ions/electrons are exhausted. The graphite electrodes belonging to group B, in which the SEI is repaired, still possess the passivating ability (Fig. 2c). In contrast, the graphite electrodes in group A cannot be repaired. Instead, the thermally unstable component (for instance, LiPF 6 ) is decomposed without the aid of the Li + ions/electrons from the graphite, and the resulting products (Li x PF y O z ) are deposited. The passivating ability of the latter electrodes is poor, such that an appreciable amount of electrolyte decomposition is observed when cycled at room temperature after the high-temperature storage (Fig. 2b).

Conclusions
The thermal behavior of surface films that are deposited on graphite electrodes with different SOCs is investigated. Specifically, the degradation/repairing behavior of the surface films is analyzed as a function of the SOC. The major findings are summarized below.
(i) The OCV measurement can effectively be used to examine the thermal degradation of surface films. During storage at 85 • C, the OCV of the MCMB electrodes continuously increases to reach 3.0 V (vs. Li/Li + ). During this period, the SEI film is degraded and repaired until the Li + ions/electrons in the graphite are exhausted. The time to reach 3.0 V (fully de-lithiated state) is proportional to the degree of lithiation (SOC) for the graphite electrodes. (ii) When the SEI films on partially lithiated graphite electrodes (group B in this study) are exposed to a high-temperature environment, the damaged layers are repaired by the reaction between the electrolyte and Li + ions/electrons supplied from the graphite. Hence, the passivating ability of the surface films is largely maintained. In addition, these newly formed SEI films have a similar chemical composition to the initially formed ones. (iii) Once the surface films are thermally damaged and the graphite electrodes are fully de-lithiated, the damaged films are not repaired. Instead, thermally decomposed products can be deposited. The passivating ability of these films is poor. (iv) One practically important message of this work is that the full discharging of LIBs is not desirable because of the graphite negative electrode in such a fully discharged cell. If this cell is occasionally exposed to high-temperature environment, the surface films on the de-lithiated graphite electrode are damaged without repairing. Due to the poor passivating ability of surface films, the cell performances are deteriorated.