Chemical States of Overcharged LiCoO2 Particle Surfaces and

Jul 1, 2015 - The data suggest that Li-poor areas at surfaces develop into interior areas between charging states of 60% and 100%. ... Using the x val...
0 downloads 4 Views 6MB Size
Download PDF
Article pubs.acs.org/JPCC

Chemical States of Overcharged LiCoO2 Particle Surfaces and Interiors Observed Using Electron Energy-Loss Spectroscopy Jun Kikkawa,*,† Shohei Terada,‡ Akira Gunji,‡ Takuro Nagai,† Keiji Kurashima,† and Koji Kimoto† †

National Institute for Materials Science, 1-1 Namiki, Tsukuba 305-0044, Japan Hitachi Research Laboratory, Hitachi Ltd., 7-1-1 Omika, Hitachi 319-1292, Japan

‡

S Supporting Information *

ABSTRACT: Deterioration mechanisms of LiCoO2 electrode materials for lithium ion batteries remain unclear. Using electron energy-loss spectroscopy and transmission electron microscopy, this study investigated chemical states of LiCoO2 particles on first overcharging. We present a scheme for quantification of the Li/ Co atomic ratio. Using quantitative Li mapping and comprehensive probing of Li−K, Co−M2,3, Co−L3, and O−K edges, we observed that overcharging causes the progression of Co3+/Co2+ reduction with oxygen extraction from the particle surface to the interior. A gradual change in the chemical composition at and around the particle surfaces after charging of 60% revealed the presence of Co3O4-like and CoO-like phases at surface regions. We also observed nanocracks with deficient Li ions. These results are key factors affecting degradation on overcharging.

1. INTRODUCTION Rechargeable lithium (Li) ion batteries are extremely important energy storage devices underpinning the practical use of portable electronic devices, hybrid cars, electric vehicles, and many renewable sources of energy. Various electrode materials have been studied to improve battery energy density, power density, safety, and cycle life.1−3 Preventing irreversible reactions in electrode materials during charging and discharging is a crucially important issue for practical applications. The most widely used positive electrode material, LiCoO2,4,5 with theoretical capacity of 274 mA h/g, 0.5 Li per LiCoO2, is used in the electrochemical extraction and insertion of Li ions. It has practical capacity of ca. 140 mA h/g and electrochemical potential of 4.2 V vs Li/Li+ because overcharging, i.e., extraction of Li ions beyond x of ca. 0.5, causes LixCoO2 degradation with fading of capacity through subsequent electrochemical cycling. Capacity fading has been associated with structural instability, dissolution of cobalt ions, evolution of gases, and deterioration of LiCoO2 surfaces as a result of reactions with the electrolyte.6−9 Reactions occurring at electrode−electrolyte interfaces have attracted considerable attention in recent years. By coating surfaces of LiCoO2 particles with thin layers of oxides such as ZrO2, the cycle durability and high-potential durability are improved, but thick layers cause increased diffusion resistance for Li ions.10−14 Physical and chemical roles of coating films have been discussed. Recently, Takamatsu et al. demonstrated that reduction of Co3+ to Co2+ at the surfaces of LiCoO2 by contact with the electrolyte can cause initial degradation, probing a depth of ca. 3 nm of the uncoated and thin-film coated LiCoO2 using in situ X-ray absorption spectroscopy.15,16 Coating films that are electrically insulating © XXXX American Chemical Society

and stable against the reduction of electrolyte can function as a physical barrier preventing contact between LiCoO2 and the electrolyte. A thin coating layer can also prevent the increase of local distortion of the LiCoO2 from overcharging, leading to high-potential durability.15−17 However, the initial stage of deterioration of uncoated LiCoO2 itself when overcharging, especially on particle surfaces at the nanometer scale, is not fully understood despite its importance for clarifying details of the coating effects. To address this issue, investigation of the distribution of Li ion quantitatively at a high-spatial resolution is important, as is elucidation of the distribution of cobalt and oxide ions, valence states of cobalt ions, and local crystal structures. We use electron energy-loss spectroscopy (EELS) in combination with transmission electron microscopy (TEM) and scanning TEM (STEM). Probing Li−K edges using EELS spectrum imaging (SI) produced qualitative and quasi-quantitative Li mapping of particles from the interior to surface areas.18,19 Quantitative Li mapping in lithiated transition-metal oxides is not easy because the Li−K edge overlaps with transition-metal M2,3 edge on a background of plasmon tail in its energy range. For LiCoO2, the Li−K edge completely overlaps with Co−M2,3 edge,20 making it difficult to use a standard approach, i.e., the separation of Li−K edge from Co−M2,3 edge and subsequent calculations of their partial scattering cross sections.21 In this study, using a specific component of Li−K edges in EELS, we achieve quantitative Li mapping for charged Li xCoO2 particles to clarify the Received: March 9, 2015 Revised: June 11, 2015

A

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C distribution of Li ions at surfaces and interiors of the particles. We investigate local chemical composition with ca. 1 nm steps by comprehensive probing of EELS Li−K, Co−M2,3, Co−L3, and O−K edges.

2. EXPERIMENTAL METHODS LiOH·H2O and Co2O3 (Wako Pure Chemical Industries Ltd.) were mixed using planetary ball milling. They were pelletized and fired at 950 °C for 12 h in air. The fired sample was crushed using planetary ball milling to obtain LiCoO2 particles of 0.5−1 μm size for positive electrode materials. Charged LixCoO2 particles were obtained after the first charge of 40% (112 mAh/g), 60% (168 mAh/g), and 100% (273 mAh/g) using coin-type cells with Li metal as a negative electrode and a 1 M solution of LiPF6 in ethylene carbonate (EC)/ethyl methyl carbonate (EMC)/diethyl carbonate (DEC) (2:4:4) as liquid electrolyte. The average Li/Co atomic ratios measured using inductively coupled plasma (ICP)−atomic emission spectrometry (AES) for the original LiCoO2 particles and 40%, 60%, and 100% charged particles were respectively 1.01, 0.60, 0.40, and 0.06. Specimens for TEM and EELS analyses were prepared using a focused ion beam (FB-2100; Hitachi Ltd.). Damage layers were removed using an argon ion milling instrument (Model 1040 NanoMill; Fischione Instruments Inc.) at initial and finishing voltages of 2 kV and 500 V. A monochromated electron microscope (Titan Cubed; FEI Co.) equipped with a spectroscope (Gatan imaging filter (GIF) Quantum ERS; Gatan Inc.) was operated at accelerating voltage of 80 kV. For EELS measurements, the energy dispersion was set as 0.05 eV/ ch. The full width at half-maximum of the zero-loss peak in vacuum was 0.15−0.20 eV. Convergence and collection semiangles in the STEM mode were respectively 21.4 mrad and 30−40 mrad.

Figure 1. Quantification of the Li/Co ratio. (a) EELS spectra including Li−K and Co−M2,3 edges acquired from middles of 40%, 60%, and 100% charged particles, with reference EELS spectra of LiCoO2 and Co3O4 particles. (b) Plots of the area ratio, SA/SB, in (a) as a function of the Li/Co ratio measured using ICP-AES.

3. RESULTS AND DISCUSSION 3.1. Quantification Method. Figure 1a displays EELS spectra including Li−K and Co−M2,3 edges for 40%, 60%, and 100% charged LixCoO2 particles acquired from ca. 60 nm φ areas in the center of particles, with reference EELS spectra of original LiCoO2 and Co3O4 particles, after background subtraction by fitting with first-order log-polynomials.21 The peak intensity of Li−K edges at 61.6 eV decreases as the charging percentage increases, and the spectrum feature approaches that for Co−M2,3 edges of Co3O4 after 100% charge. We define areas SA and SB between 60.3 and 62.5 eV for the intense peak of Li−K edge components at 61.6 eV characterized by Li 1s core−hole effects,20 as shown in Figure 1a. The boundary of areas of SA and SB is determined by the line that intersects the EELS spectrum at 60.3 and 62.5 eV. Areas SA and SB were measured for 5−8 particles for charged and reference particles with the relative thickness21 t/λ of 0.7− 1.4. In Figure 1b, ratios of the areas SA to SB, SA/SB, with standard deviation bars are shown as a function of the Li/Co ratio, i.e., x in LixCoO2 measured using ICP-AES. An empirical fitting function of the data is used for evaluating the Li/Co ratio with a given SA/SB value. In this study, the plot was fitted using a linear function: SA/SB = 0.813x + 0.00628. Before describing its application, we can explain some details, benefits, and cautionary points related to the SA/SB method. For this method, we assumed that the features of Co−M2,3 edges of LiCoO2 are not markedly different from those of Co3O4, having a gradual increase in EELS intensity with no

peaks between 60.3 and 62.5 eV (Figure 1a). These assumptions are reasonable because Co−M2,3 edges for LiCoO2 can locate at a higher energy side than Co3O4. Moreover, the intensity at 59.2 eV for Co3O4 (Figure 1a) can be weak, analogous to the differences of Co−L3 edges among LiCoO2, Co3O4, and CoO, as described later (Figure 3a). According to the assumption, SA and SB respectively include signals from Li−K edges and both Co−M2,3 and Li−K edges between 60.3 and 62.5 eV. Signals from Co−M2,3 edges in SB become dominant as the Li/Co ratio decreases. The SA/SB method presents two important benefits. First, plural inelastic scattering effects for Li−K edge are small, around 61.6 eV, which is different from the higher energy side, ca. 64−80 eV.22 Those small effects reduce the specimen thickness effects on measuring SA and SB, although the plural inelastic scattering effects can be removed using deconvolution techniques with low-loss EELS spectra including zero-loss peaks.21 Second, because of the approximately linear increase of the Co−M2,3 edge around the intense peak at 61.6 eV, we can effectively extract signals from Li−K edges as SA, especially for specimens with small Li/Co ratios. It is noteworthy that the SA/SB method does not separate Li−K edge signal from Co−M2,3 edge signal completely because SB includes a Li−K edge signal that decreases as the Li/Co ratio decreases. The complete separation of signals between Li−K and Co−M2,3 edges is probably extremely difficult even if one preferably selects a specific energy range, much more a wide energy range including B

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 2. Quantitative Li mapping for particles after the charge of 40% in (a−d), 60% in (e−h), and 100% in (i−l): (a), (e), and (i): ADF-STEM images; (b), (f), and (j): quantitative maps of the Li/Co ratio, x, in the rectangle areas with dotted lines (a), (e), and (i), respectively; (c), (g), and (k): histograms of x for (b), (f), and (j) with Gaussian fits; (d), (h), and (l): plots of x in the rectangle areas in (b), (f), and (j) from particle edges to interiors.

materials that have a sharp peak in Li−K edge feature without abrupt variation in transition metal−M2,3 edge feature at and around the sharp peak. 3.2. Quantitative Li Mapping. Applying the SA/SB method to EELS SI, quantitative Li mapping was achieved for 40%, 60%, and 100% charged particles, as shown in Figure 2. Constant scanning steps of electron probes were 9.2−10.4 nm, with a typical probe current of ca. 16 pA and acquisition time of 0.5 s/pix. Figures 2b, 2f, and 2j respectively display quantitative maps of the Li/Co ratio: x in LixCoO2 in the rectangle areas with dotted lines in annular dark-field (ADF)STEM images of Figures 2a, 2e, and 2i. Corresponding histograms of x for Figures 2b, 2f, and 2j are displayed respectively in Figures 2c, 2g, and 2k. Average x from the particle edge sides to centers in the rectangle areas of Figures 2b, 2f, and 2j are shown respectively in Figures 2d, 2h, and 2l, where the edge positions are defined as zero. For the 40%

fully Li−K and Co−M2,3 edges, as in the case of LiFePO4 described by Moreau et al.23 However, the Li/Co ratio quantification does not necessarily require the complete separation of signals between Li−K and Co−M2,3 edges as long as the dependence of a variable parameter, i.e., SA/SB in the present case, on the Li/Co ratio (Figure 1b) is prepared in combination with other analytical schemes. Strictly speaking, it is necessary to investigate the difference of Co−M2,3 edge features among Co3O4, LiCoO2, and LixCoO2 (0 ≤ x < 1), variation of the Li−K edge feature in LixCoO2 (0 < x ≤ 1), and the dependences of Li−K and Co−M2,3 edge features on scattering angles and specimen orientation for anisotropic crystals, LixCoO2. Such effects are included in the fitting function in this study. Caution is called for when applying the SA/SB and similar methods to other Li-containing transitionmetal oxides where the Li−K edges overlap with transitionmetal M2,3 edges. The methods are probably useful for C

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C charged particle, Li ions were certainly extracted without causing distinct separation of the Li-poor phase from Li-rich phase at particle interior areas, although some Li-poor areas exist at the particle edges (Figure 2b). Gaussian fits in Figure 2c indicate the presence of a main component at x = 0.54 and a minor component at x = 0.37. The main component is close to the average Li/Co ratio, 0.60 measured using ICP-AES. The x profile for area α in Figure 2b shows an abrupt change from 0 to 0.55 at the left side of the particle edge (Figure 2d), whereas that for area β shows a short plateau-like feature around x = 0.35 at the right side of the particle edge, corresponding to the minor phase of x = 0.37 in Figure 2c. For the 60% charged particle, Figure 2f revealed that Li-poor areas are extended along the particle edges. Gaussian fits in Figure 2g indicate the presence of two components at x = 0.36 and 0.13. Actually, x = 0.36 is close to the average Li/Co ratio measured using ICPAES. The x = 0.13 component exists near the particle edge as Li-poor phase, whereas the x = 0.36 component exists inside the particle. The value of x increases with an approximate gradient of 4.6 × 10−3 [1/nm] from the particle edge to the interior and reaches ca. 0.42 as shown in Figure 2h, with a short plateau-like profile around x = 0.1. The gradient is smaller than that for the 40% charged particle. For the 100% charged particle, more Li ions were extracted from the whole region of the particle (Figure 2j), leaving inhomogeneous distribution of Li/Co ratio at the several tens of nanometer scale. Gaussian fits in Figure 2k indicate the main component at x = 0.05 and the minor component at x = 0.23. The x = 0.05 component corresponds approximately to the average Li/Co ratio of 0.06 measured using ICP-AES. The fluctuation of Li/Co ratio between x = 0.05 and 0.23 in Figure 2I reflects the presence of two phases at the several tens of nanometers scale. Results of quantitative Li mapping show that different electrochemical reactions occur between particle interior and surface areas, especially for a charging state of around 60%. The data suggest that Li-poor areas at surfaces develop into interior areas between charging states of 60% and 100%. Detailed investigations of interior and surface areas are described below. 3.3. Chemical States of the Particle Interiors. We specifically examined the particle interior areas, which are major areas of the particles, reflecting physical properties characterized by macroscopic analyses such as ICP-AES. Figure 3 portrays representative EELS spectra of Co−L3 and O−K edges acquired from middles of the 40%, 60%, and 100% charged particle (shown as circles in Figures 2a, 2e, and 2i), with the reference EELS Co−L3 and O−K edges of LiCoO2, Co3O4, and CoO particles. The origins in Figure 3 were calibrated by the simultaneous acquisition of zero-loss peaks in dual-EELS modes.24 Peak broadening of Co−L3 edge and the first peak of O−K edges occurred when x decreased to ca. 0.4, as described in reports of previous studies.20 Increase in EELS intensity at the lower energy side of the first peak of O−K edge (denoted as ▼ in Figure 3b) also occurred.20,25 It is noteworthy that features of Co−L3 and O−K edges of the 100% charged particle are analogous to those of Co3O4 particle. The main peak of the Co−L3 edge and the first peak of O−K edges respectively shift to lower and higher energy-loss sides. The results imply that reduction in Co valence states and extraction of oxygen occurred on charging states between 60% and 100% in the particle interior areas. Consequently, the average chemical compositions of the particle interior after 100% charge can be described as LixCoO2−δ (δ of ca. 0.67). Both Co−L3 edges for Co3O4 and 100% charged particles

Figure 3. Electronic states of middles of charged particles. EELS spectra of Co−L3 edges in (a) and O−K edges in (b) acquired from the circled area in the 40%, 60%, and 100% charged particles in Figures 2a, 2e, and 2i, respectively, with reference EELS spectra of LiCoO2, Co3O4, and CoO particles.

comprise two predominant peaks. The intensity ratio of the peak at lower energy-loss side (denoted as ■) to that at higher energy-loss side (denoted as ●) for the 100% charge particle is larger than that for Co3O4, which probably indicates that the fraction ratio Co2+/Co3+ for the 100% charged particle is larger than that for Co3O4, as expected from the feature of divalent Co−L3 edges for CoO (Figure 3b), consistent with the Li+ presence in the 100% charged particle, although the determination of valence states of Co ions using L3 edge of reference materials with Co ions in different environments is not simple. Using the x value from the SA/SB method (Figure 2k), we can estimate the chemical composition of the major phase in Figure 2i, approximately as Li 0.05CoO1.33 (= Li0.15Co3O4) using EELS. Results for particle interior areas in Figure 3 revealed that Li ions extraction occurred with regular charge compensation, i.e., extraction of electrons from the hybridization states of Co 3d and O 2p orbitals up to 60% charging at least, although Li ions extraction occurred with different charge compensation, i.e., extraction of oxygen between charging of 60% and 100%. Preferential extraction of electrons from O 2p orbitals is likely to induce the formation of neutral O or O2 species, which desorb from the particle and which can form CO2 and H2O by reaction with the electrolyte.6,26 Some electrons from the O 2p orbitals were deposited to Co 3d orbitals, reducing valence states. The results for 100% charged particles (Figure 3) show that Li-poor areas along particle surfaces (Figure 2f) are related with those irreversible reactions. 3.4. Chemical States of the Particle Surfaces. Next, we specifically examine the surface areas of the charged particles to investigate their chemical compositions and electronic structures. Figure 4a shows bright-field STEM images of the edge region of 60% charged particle in Figures 2e and 2f. The EELS SI data for Li−K edges with Co−M2,3, Co−L3, and O−K edges were acquired separately with scanning steps of 0.95−1.2 nm in the rectangular area of Figure 4a from the particle edge A to interior A″ sides through the A′ position. The corresponding D

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 4. Chemical and electronic states at the surface region of 60% charged particle: (a) bright-field STEM image of the 60% charged particle in Figure 2e. (b) Plot of x from A to A″ in (a). [(c)−(e)] EELS spectra acquired from A to A′ in (a), with the energy range of Li−K and Co−M2,3 edges in (c), Co−L3 edges in (d), and O−K edges in (e). Reference EELS spectra of CoO, Co3O4, and LiCoO2 are also presented in (c). (f) Schematic drawing of the variation of chemical composition from the particle edge, A, to the interior.

Features of Co−M2,3 edges differ between Co3O4 and CoO because of the chemical shift and difference in partial density of states for Co 3d orbitals. The intensity maximum locates at 62.2 eV for CoO, although a linear increase in intensity exists between 60.5 and 64.0 eV for Co3O4. It is noteworthy that the feature of Co−M2,3 edge for CoO does not fatally disturb quantitative Li mapping based on the SA/SB method in Figures 2b, 2f, and 2j because of rough scanning steps, ca. 10 nm, although x was set to zero less than 6 nm in Figure 4b. In Figure 4d, the intensity of the shoulder of the main Co−L3 peak at the lower energy-loss side, i.e., Co2+ component increases by about 10−20 nm, while visible variations do not occur in the range of 20 and 30 nm. At the distance of ca. 8 nm, the features of Co−L3 edges resemble those for Co3O4, implying that the Co2+/Co3+ ratio is approximately 0.5. The Co2+/Co3+ ratio increases rapidly, as approaching from 8 nm to the particle edges, A. The features of Co−L3 edges at 6−8 nm are similar to those for a 100% charged particle in Figure 3a. The feature of Co−L3 edges at the distance of ca. 4−5 nm is similar to that for CoO, implying that all Co ions can be

rectangle area is also shown in the inset Li map of Figure 4a. Figure 4b displays a plot of Li/Co ratio, x, from A to A″, using the SA/SB method. The Li/Co ratio decreases from the interior to the surface. A plateau exists around x = 0.13 in the region of approximately 40−70 nm distance from the particle edge, corresponding to the phase of x = 0.13 in Figure 2g. EELS spectra of Li−K and Co−M2,3, Co−L3, and O−K edges from A to A′ are displayed respectively in Figures 4c, 4d, and 4e, with reference EELS spectra of CoO, Co3O4, and LiCoO2. Although the positions of origin are not exact in Figures 4d and 4e, different from Figure 3 acquired by dual EELS, it is possible to discuss relative differences in the energy positions. In Figure 4c, the peak intensity for Li−K edges at 61.6 eV (denoted as ▼) decreases from the particle interior to the edge in the range of 12 and 34 nm. From 12 nm to the particle edge, 0 nm, features of two types of Co−M2,3 edges become much clearer. The EELS spectra of 6−12 nm resemble the EELS Co−M2,3 edges of Co3O4 or much more to EELS spectra from the particle interior after 100% charge (Figure 2a), whereas EELS spectra of 0−6 nm are analogous to EELS Co−M2,3 edges of CoO. E

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. 60% charged particles with nanocracks: (a) ADF-STEM image; (b) quantitative map of the Li/Co ratio, x, for the rectangle area with the dotted line in (a); (c) EELS spectra for Co−L3 edge acquired along A to A′ in (b); (d) enlarged ADF-STEM image of the squared area with the solid line in (a), showing nanocracks; (e, f) a pair of ADF-STEM and quantitative Li/Co ratio maps from the rectangle area in (d); (g) EELS spectra for areas α and β in (f).

at narrow areas of particle surfaces after 40% charge (Supporting Information), indicating that the extraction of oxygen and reduction in Co valence state can occur partly at surfaces on charging up to 40%. Consequently, the Co3+/Co2+ reduction progress at the particle surfaces even on charging states between 40% and 60% (Figures 2b and 2f). Takamatsu et al. reported that the Co3+/Co2+ reduction of uncoated LiCoO2 surfaces of ca. 3 nm immediately after soaking in an electrolyte and the Co2+/Co3+ oxidation at the surfaces on subsequent charging up to ca. 55% .15,16 Such Co2+/Co3+ oxidation at the particle surfaces was not confirmed, although our result does not contradict their result. Our results demonstrate that overcharging causes the progression of Co3+/Co2+ reduction with oxygen extraction from surface areas to interior areas. The reactions, which are probably irreversible, are likely to cause capacity fading during subsequent discharge−charge cycling. In addition to the reactions described above, another reaction was found for degradation. We observed 60% charged particles with nanometer-sized cracks in all areas (Figure 5a), which existed at the rate of about 40%. The square area with the solid line in Figure 5a is shown as enlarged in Figure 5d. Using electron diffraction, we confirmed that the nanocracks are extended to the direction normal to stacking layers of LixCoO2 (Supporting Information), which can be correlated with the abrupt change in lattice parameter along the [001] axis of the hexagonal unit cell at x of ca. 0.3.29 Figure 5b displays quantitative maps of Li/Co ratios, x, in the rectangle area, with dotted lines of Figure 5a. More Li ions are extracted from the interior of particle in Figure 5b than from the particle in Figure 2f. The major phase of the histogram of Figure 5b is located at x = 0.28 by Gaussian fits (Supporting Information). Figure 5c presents EELS Co−L3 edges acquired in the rectangular area with a solid line from the particle edge, A to A′, as presented in Figure 5a, demonstrating the existence of CoO-like phases at 0−20 nm, Co3O4-like phases at 45−60 nm, and their transitional phases at 20−45 nm. Figures 5e and 5f respectively display ADF-STEM image and quantitative map of Li/Co ratio

characterized as divalent with a local structure analogous to CoO. The feature of Co−L3 edges for 0−3 nm exhibits a predominantly divalent nature, although its features differ slightly from those for standard CoO, probably because of worse signal-to-noise ratios. The existence of a CoO-like layer covering a Co3O4-like layer is similar to reduced nature of particle edges of Co3O4.27 In Figure 4e, the first and second peaks of O−K edges (denoted respectively as ● and ■) shifts slightly to the higher energy-loss side with decrease in the integral intensity of the first peak relative to that of the third peaks from the distance of 10−14 nm, although no visible variation of spectrum features is observed for 14−34 nm. From distances of 10 to 4 nm, the first peak shifts to the higher energy-loss side with decrease in the width of the first peak and in the integral intensity of the first peak (●) relative to that of the third peaks (denoted as ▲), whereas the third peak shifts to the lower energy-loss side. The feature of O−K edges in the region of 0−3 nm, which is considerably different from that for the interior, is analogous to that for CoO with a worse signalto-noise ratio. Based on results presented in Figure 4, chemical compositions near the particle edge can be categorized as (i) LixCoO2 (x ≥ 0.05), (ii) LixCoO2−δ (0 < x ≤ 0.05, 0 < δ ≤ 0.67), and (iii) CoO2−δ (0.67 < δ ≤ 1), with increasing δ and decreasing x from the particle interior to the edge, as shown in Figure 4f. 3.5. Degradation Mechanism. The existence of Co3O4like (i.e., LixCoO1.33) and CoO-like (i.e., CoO2−δ) phases at particle edges after 60% charge (Figure 4) and the LixCoO1.33 (x = 0.05) phase at particle interior after 100% charge (Figure 3) implies that extraction of oxygen and reduction in Co valence states progress substantially from the particle surface areas to interior areas around the charging state of 60%. The results are consistent with those of previous studies which report the formation of Co3O4 and CoO phases from Li0.5CoO2 by reaction with electrolytes,26,28 although distribution of these extra phases within particles was not investigated in their studies. The presence of CoO-like phases was observed partly F

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

■

in the rectangular area of Figure 5d: Both data were acquired simultaneously. The central part of Figure 5e, showing a nanocrack, is dark because it has less thickness. Figure 5f shows that the Li/Co ratio at the nanocrack is small. In Figure 5g, EELS spectra at area α are characterized as LixCoO2 with average x = 0.41, whereas EELS spectra at area β can be attributed to LixCoO2−δ (x of ca. 0, 0 < δ ≤ 0.67) as in Figure 4. This attribution is particularly interesting because the formation of nanocracks is related not only with mechanical properties such as lattice strains induced by extraction of Li ions and oxygen and change in crystal structure; the formation is also related to local deficiency in Li ions and around the nanocracks. Finally, we briefly comment on two points related with our experimentally obtained results. One is the similarities and differences of degradation processes in LixCoO2 between overcharging with ζ ≤ x ≤ 1 (0 ≤ ζ < 0.5) and repetitive charge−discharge cycling with 0.5 ≤ x ≤ 1. We can speculate on them from the perspective of two factors: (i) reaction at the electrolyte/LiCoO2 interfaces and (ii) irreversible structural change caused by the removal of oxygen and nanocracks. For overcharging, the factor (i) is dominant at the first half of charge and both factors contribute to degradation in the second half of the charging. For repetitive charge−discharge cycling, the factor (i) probably contributes to degradation throughout cycling.8,15 In addition, an irreversible structural change caused by the cation disorder rather than the removal of oxygen also contributes to degradation.30 The other point is the influence of liquid electrolyte on chemical states of LiCoO2. We used 1 M solution of LiPF6 in EC/EMC/DEC as the liquid electrolyte in this study. A recent study demonstrated that EC is preferentially absorbed on LiCoO2 surfaces.31 In addition, liquid electrolytes containing EC react with Li0.5CoO2 and assist the reduction of Co ions forming Co3O4 and CoO, whereas an increase in the LiPF6 concentration can slow its reaction at the electrolyte/LiCoO2 interface.26,28 These reports suggest that the results and degradation mechanisms of overcharging process presented in this paper are probably common to LiCO2 in other EC-containing electrolytes. Investigating differences in deterioration using liquid electrolytes of several kinds with and without EC is necessary.

Article

ASSOCIATED CONTENT

S Supporting Information *

Figures S1−S3. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.jpcc.5b02303.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]; phone +81 29 860 4756 (J.K.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS J.K. appreciates fruitful comments from T. Aizawa, T. Hara (NIMS), and K. Ishizuka (HREM Research Inc.) and the use of the NanoMill, as offered by M. Mitome and W. Zhang.

■

REFERENCES

(1) Tarascon, J. M.; Armand, M. Issues and Challenges Facing Rechargeable Lithium Batteries. Nature 2001, 414, 359−367. (2) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable Li Batteries. Chem. Mater. 2010, 22, 587−603. (3) Bruce, P. G.; Scrosati, B.; Tarascon, J. M. Nanomaterials for Rechargeable Lithium Batteries. Angew. Chem., Int. Ed. 2008, 47, 2930−2946. (4) Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. LixCoO2 (0 < x ≤ 1): A New Cathode Material for Batteries of HighEnergy Density. Mater. Res. Bull. 1980, 15, 783−789. (5) Ohzuku, T.; Ueda, A.; Nagayama, M.; Iwakoshi, Y.; Komori, H. Comparative-Study of LiCoO2, LiNi1/2Co1/2O2 and LiNiO2 for 4-V Secondary Lithium Cells. Electrochim. Acta 1993, 38, 1159−1167. (6) Ohsaki, T.; Kishi, T.; Kuboki, T.; Takami, N.; Shimura, N.; Sato, Y.; Sekino, M.; Satoh, A. Overcharge Reaction of Lithium-Ion Batteries. J. Power Sources 2005, 146, 97−100. (7) Amatucci, G. G.; Tarascon, J. M.; Klein, L. C. Cobalt Dissolution in LiCoO2-Based Non-Aqueous Rechargeable Batteries. Solid State Ionics 1996, 83, 167−173. (8) Aurbach, D.; Markovsky, B.; Rodkin, A.; Levi, E.; Cohen, Y. S.; Kim, H. J.; Schmidt, M. On the Capacity Fading of LiCoO2 Intercalation Electrodes: The Effect of Cycling, Storage, Temperature, and Surface Film Forming Additives. Electrochim. Acta 2002, 47, 4291−4306. (9) Vetter, J.; Novak, P.; Wagner, M. R.; Veit, C.; Möller, K. C.; Besenhard, J. O.; Winter, M.; Wohlfahrt-Mehrens, M.; Vogler, C.; Hammouche, A. Ageing Mechanisms in Lithium-Ion Batteries. J. Power Sources 2005, 147, 269−281. (10) Cho, J.; Kim, Y. J.; Kim, T. J.; Park, B. Zero-Strain Intercalation Cathode for Rechargeable Li-Ion Cell. Angew. Chem., Int. Ed. 2001, 40, 3367−3369. (11) Kim, Y. J.; Kim, T. J.; Shin, J. W.; Park, B.; Cho, J. P. The Effect of Al2O3 Coating on the Cycle Life Performance in Thin-Film LiCoO2 Cathodes. J. Electrochem. Soc. 2002, 149, A1337−A1341. (12) Chen, Z. H.; Dahn, J. R. Effect of a ZrO2 Coating on the Structure and Electrochemistry of LixCoO2 When Cycled to 4.5 V. Electrochem. Solid State Lett. 202, 05, A213−A216. (13) Chen, Z. H.; Dahn, J. R. Studies of LiCoO2 Coated with Metal Oxides. Electrochem. Solid State Lett. 2003, 6, A221−A224. (14) Miyashiro, H.; Yamanaka, A.; Tabuchi, M.; Seki, S.; Nakayama, M.; Ohno, Y.; Kobayashi, Y.; Mita, Y.; Usami, A.; Wakihara, M. Improvement of Degradation at Elevated Temperature and at High State-of-Charge Storage by ZrO2 Coating on LiCoO2. J. Electrochem. Soc. 2006, 153, A348−A353. (15) Takamatsu, D.; Koyama, Y.; Orikasa, Y.; Mori, S.; Nakatsutsumi, T.; Hirano, T.; Tanida, H.; Arai, H.; Uchimoto, Y.; Ogumi, Z. First In Situ Observation of the LiCoO2 Electrode/Electrolyte Interface by

4. CONCLUSION In summary, we conducted quantitative Li mapping and Co− M2,3, Co−L3, and O−K EELS analyses for LixCoO2 particles on the first overcharge process. Overcharging causes the progression of Co3+/Co2+ reduction with oxygen extraction from the surface to interior areas. The chemical composition at and around the surface of 60% charged particle is categorized as (i) LixCoO2 (x ≥ 0.05), (ii) LixCoO2−δ (0 < x ≤ 0.05, 0 < δ ≤ 0.67), and (iii) CoO2−δ (0.67 < δ ≤ 1), with increased δ and decreased x from the particle interior to the edge. The Co3O4 and Li-inserted Co3O4 phases develop into the particle interior, with CoO phase at the topmost surface. Moreover, results show that nanocrack formation is related not only with mechanical properties but also with local deficiency in Li ions. These reactions, which are probably irreversible, are crucially important factors affecting degradation caused by overcharging. Results from this study of uncoated LiCoO2 are expected to be important for optimizing coating effects on LiCoO2 and related electrode materials. G

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C Total-Reflection X-ray Absorption Spectroscopy. Angew. Chem., Int. Ed. 2012, 51, 11597−11601. (16) Takamatsu, D.; Mori, S.; Orikasa, Y.; Nakatsutsumi, T.; Koyama, Y.; Tanida, H.; Arai, H.; Uchimoto, Y.; Ogumi, Z. Effects of ZrO2 Coating on LiCoO2 Thin-Film Electrode Studied by In Situ X-ray Absorption Spectroscopy. J. Electrochem. Soc. 2013, 160, A3054− A3060. (17) Takamatsu, D.; Nakatsutsumi, T.; Mori, S.; Orikasa, Y.; Mogi, M.; Yamashige, H.; Sato, K.; Fujimoto, T.; Takanashi, Y.; Murayama, H.; Oishi, M.; Tanida, H.; Uruga, T.; Arai, H.; Uchimoto, Y.; Ogumi, Z. Nanoscale Observation of the Electronic and Local Structures of LiCoO2 Thin Film Electrode by Depth-Resolved X-ray Absorption Spectroscopy. J. Phys. Chem. Lett. 2011, 2, 2511−2514. (18) Kikkawa, J.; Akita, T.; Tabuchi, M.; Shikano, M.; Tatsumi, K.; Kohyama, M. Real-Space Observation of Li Extraction/Insertion in Li1.2Mn0.4Fe0.4O2 Positive Electrode Material for Li-Ion Batteries. Electrochem. Solid-State Lett. 2008, 11, A183−A186. (19) Taguchi, N.; Akita, T.; Sakaebe, H.; Tatsumi, K.; Ogumi, Z. Characterization of the Surface of LiCoO2 Particles Modified by Al and Si Oxide Using Analytical TEM. J. Electrochem. Soc. 2013, 160, A2293−A2298. (20) Kikkawa, J.; Terada, S.; Gunji, A.; Haruta, M.; Nagai, T.; Kurashima, K.; Kimoto, K. Understanding Li−K Edge Structure and Interband Transitions in LixCoO2 by Electron Energy-Loss Spectroscopy. Appl. Phys. Lett. 2014, 104, 114105. (21) Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron Microscope, 3rd ed.; Springer: Berlin, 2011. (22) Ahn, C. C. Transmission Electron Energy Loss Spectrometry in Materials Science and the EELS Atlas; John Wiley & Sons: New York, 2006. (23) Moreau, P.; Boucher, F. Revisiting Lithium K and Iron M2,3 Edge Superimposition: The Case of Lithium Battery Material LiFePO4. Micron 2012, 43, 16−21. (24) Gubbens, A.; Barfels, M.; Trevor, C.; Twesten, R.; Mooney, P.; Thomas, P.; Menon, N.; Kraus, B.; Mao, C.; McGinn, B. The GIF Quantum, A Next Generation Post-Column Imaging Energy Filter. Ultramicroscopy 2010, 110, 962−970. (25) Mizokawa, T.; Wakisaka, Y.; Sudayama, T.; Iwai, C.; Miyoshi, K.; Takeuchi, J.; Wadati, H.; Hawthorn, D. G.; Regier, T. Z.; Sawatzky, G. A. Role of Oxygen Holes in LixCoO2 Revealed by Soft X-Ray Spectroscopy. Phys. Rev. Lett. 2013, 111, 056404. (26) MacNeil, D. D.; Dahn, J. R. The Reaction of Charged Cathodes with Nonaqueous Solvents and Electrolytes - I. Li0.5CoO2. J. Electrochem. Soc. 2001, 148, A1205−A1210. (27) Zhang, Z. L. Surface Effects in the Energy Loss Near Edge Structure of Different Cobalt Oxides. Ultramicroscopy 2007, 107, 598− 603. (28) MacNeil, D. D.; Dahn, J. R. The Reactions of Li0.5CoO2 with Nonaqueous Solvents at Elevated Temperatures. J. Electrochem. Soc. 2002, 149, A912−A919. (29) Ohzuku, T.; Ueda, A. Solid-State Redox Reactions of LiCoO2 (R3m ̅ ) for 4 V Secondary Lithium Cells. J. Electrochem. Soc. 1994, 141, 2972−2977. (30) Wang, H. F.; Jang, Y. I.; Huang, B. Y.; Sadoway, D. R.; Chiang, Y. T. TEM Study of Electrochemical Cycling-Induced Damage and Disorder in LiCoO2 Cathodes for Rechargeable Lithium Batteries. J. Electrochem. Soc. 1999, 146, 473−480. (31) Yu, L.; Liu, H. J.; Wang, Y.; Kuwata, N.; Osawa, M.; Kawamura, J.; Ye, S. Preferential Adsorption of Solvents on the Cathode Surface of Lithium Ion Batteries. Angew. Chem., Int. Ed. 2013, 52, 5753−5756.

H

DOI: 10.1021/acs.jpcc.5b02303 J. Phys. Chem. C XXXX, XXX, XXX−XXX