Operando X-ray Absorption Study of the Redox Processes Involved

Feb 20, 2014 - Operando X-ray absorption spectroscopy investigations have been carried out to follow changes in the atomic and electronic local struct...
0 downloads 10 Views 3MB Size
Article pubs.acs.org/JPCC

Operando X‑ray Absorption Study of the Redox Processes Involved upon Cycling of the Li-Rich Layered Oxide Li1.20Mn0.54Co0.13Ni0.13O2 in Li Ion Batteries H. Koga,†,‡ L. Croguennec,*,† M. Ménétrier,† P. Mannessiez,† F. Weill,† C. Delmas,† and S. Belin*,§ †

CNRS, Université de Bordeaux, ICMCB, 87 avenue Schweitzer, Pessac, F-33600, France Toyota Motor Europe NV/SA, Hoge Wei 33, Zaventem, B-1930, Belgium § Synchrotron SOLEIL, L’orme des Merisiers Saint Aubin, Gif-sur-Yvette, F-91192, France ‡

S Supporting Information *

ABSTRACT: Operando X-ray absorption spectroscopy investigations have been carried out to follow changes in the atomic and electronic local structures of all three transition metals for the Li1.20Mn0.54Co0.13Ni0.13O2 layered oxide during the first and second charges and discharges of lithium batteries. The experiments were performed using a Quick-XAS monochromator on the SAMBA beamline at Synchrotron SOLEIL to record the three K-edges by edge-jumping between two energy ranges ([Mn, Co] and [Co, Ni]) every 3 min during the cycling of the battery. The results obtained especially at the Mn K-edge fully support the participation of oxygen in the reversible charge− discharge reaction of this Li- and Mn-rich layered material as a redox center and not only with oxygen loss, as was proposed previously.

I. INTRODUCTION Lithium ion batteries have been used for portable devices for 2 decades, and they are favorite candidates as power sources for future vehicles, such as hybrid, plug-in hybrid, and electric vehicles. LiCoO2 has been widely used in lithium ion batteries as positive electrode;1 however, it is difficult to maintain LiCoO2 in the batteries developed for vehicles due to its high price as a rare resource of Co, as well as due to safety concerns in the charged state. Recently, polyanionic materials exemplified by LiFePO4 were proposed as new positive electrodes for lithium ion batteries for transportation applications.2 LiFePO4 was found to be very attractive because iron as well as phosphate are abundant and thus low-priced raw materials; furthermore, it shows a very high thermal stability in the charged state of the battery with (for optimized material) very good reversibility even at high rates. Nevertheless, polyanionic materials have disadvantages as compared to LiCoO2, such as low electronic and ionic conductivities and low energy density3 The (1 − x)LiMO2·xLi2MnO3 (M = Ni, Co, Mn) system was reported as one of those delivering the largest capacity among layered oxides.4−6 Indeed, the charge and discharge mechanism was found to be different, with an irreversible “plateau” around 4.5 V vs Li+/Li during the first charge. It was thought to be at the origin of the overcapacity observed for that kind of Li- and Mn-rich system. During the plateau, Dahn and co-workers mentioned that oxide ions are oxidized and then lost from the structure,4,7−9 but the detailed mechanism is in fact still under debate. © 2014 American Chemical Society

We have shown in our group that irreversible structural reorganization occurs upon cycling of the lithium- and manganese-rich layered oxide Li1.20Mn0.54Co0.13Ni0.13O2 in lithium cells.10 A mixture of two phases is formed on the high voltage plateau and is then preserved upon long-range cycling. The impact of the particles’ size and cycling conditions (cycling rate, temperature, and number of cycles) on the distribution between the two phases shows that the surface and the bulk react differently. In both cases, oxygen participates in the reactions involved upon cycling of Li1.20Mn0.54Co0.13Ni0.13O2: for one phase, rather present in the bulk, through (reversible) oxygen oxidation and for the other phase, rather present at the surface, through oxidation of oxygen anion with departure of O2. In the latter, oxygen loss induces migration of the transition metal ions to the lithium vacancies from the surface to the bulk and a densification of the host structure. Chemical and redox titrations, magnetic measurements, and analysis of neutron diffraction data led us to propose an average chemical formula of Li1.11Mn0.58Co0.14Ni0.14O2 in the reduced state, with manganese ions mainly at the tetravalent state and the oxygen content close to that in the pristine material.11 We have shown that during the first charge about 0.40 electrons are exchanged on oxygen anions per formula unit. During the cell discharge and the following cycles, Ni and Co are involved in the redox processes in the entire particle (from the surface to the bulk) Received: December 13, 2013 Revised: February 15, 2014 Published: February 20, 2014 5700

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

while oxygen is electrochemically active only in the part of the particle where there are a significant number of lithium vacancies in interaction with MnO6 octahedra. In the other part, more at the surface, Mn participates in the redox processes. We report here XAS measurements to get more insight into changes occurring in Li1.20Mn0.54Co0.13Ni0.13O2, to the oxidation state of each transition metal ion, and also to their local environment during the charge and discharge. XAS measurements were performed operando with an in situ cell developed to prevent any degradation of the materials during their characterization but also to be closer to the materials really observed during operation of the battery.12

Figure 1. Energy range accessible by edge jumping with the Si(111) Quick-XAS monochromator for Li1.20Mn0.54Co0.13Ni0.13O2.

II. EXPERIMENTAL SECTION Li1.20Mn0.54Co0.13Ni0.13O2 was prepared using the sol−gel method, as described in detail in ref 13. The positive electrodes for ex situ XAS measurement consisted of 75 wt % of active material, 20 wt % of a carbon black/graphite (1:1) mixture, and 5 wt % of polytetrafluoroethylene (PTFE). They were cut into 14 mm diameter disks containing 11 mg/cm2 of active material and pressed at 40 MPa on an Al foil. Coin cells were assembled in an argon-filled glovebox with Li metal as the counter electrode and 1 M LiPF6 dissolved in a mixture of propylene carbonate (PC), ethylene carbonate (EC), and dimethyl carbonate (DMC) 1:1:3 by volume as the electrolyte. After cycling, positive electrodes were recovered from coin cells at different states of charge and discharge in order to compare their XAS spectra to those obtained operando during the cycling of the batteries (as described below). Prior to these ex situ experiments, the electrodes were rinsed in an excess of DMC in order to remove the residual electrolyte salt (LiPF6) and then dried under vacuum. The positive electrodes for operando XAS measurement were prepared with same processes as for ex situ XAS measurement, except that a 0.024 mm thick Al foil was used as current collector. The cell for operando XAS experiments was prepared in an argon-filled glovebox with the same counter electrode and electrolyte as those used for classical batteries, the specificity of the electrochemical cell developed for these kinds of experiments in transmission mode is the presence of two beryllium windows that allow X-rays passing through the overall stacking.12 The electrochemical measurements were performed between 2.5 and 4.8 V vs Li+/Li at C/10 cycling rate (1 C rate corresponding to the theoretical exchange of one electron in 1 h). We have taken advantage of the Quick-XAS monochromator14 on the SAMBA beamline (Synchrotron SOLEIL), to record the three K-edges by edge-jumping between two energy ranges, namely [Mn, Co] and [Co, Ni], every 3 min during operation of the cell (Figure 1). The Co Kedge was thus recorded twice, allowing us to check continuously that no significant changes occur within 3 min that correspond to the exchange of ∼0.00167 e−. The storage ring was operated in multibunch top-up at 2.75 GeV with a 400 mA current. The incident beam from the bending magnet was collimated by a first cylindrically bendable silicon Pd coated mirror and then was monochromatized using the channel-cut Si(111) QuickExafs monochromator,14 and the harmonic rejection was done thanks to the set of Pd-coated first and second mirrors tilted at 6 mrad. The beam size (7 mm × 0.6 mm, horizontal × vertical) and the position on the sample were kept constant during the acquisition except (for one or two chosen samples) when the cell was moved to check the

homogeneity within the electrode. The spectra were collected at the three K-edges (Mn, Co, and Ni) in transmission mode utilizing gas ionization chambers as detectors. As we have three detectors in series, we can measure during all the experiments a Co metal foil that can be used as reference to calibrate the energy. The measurement was carried out at 1 Hz with an amplitude of 3.9° around the Bragg angle. Analysis of X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) data was performed with the Athena and Artemis software packages.15

III. RESULTS AND DISCUSSION 1. Electrochemical Performance of the Cell Developed for Operando Study Using Synchrotron Radiation. XAS measurements with the cell developed for operando study of processes occurring in batteries during their charges and discharges have the advantage that unstable states of a positive electrode at high voltage can be measured in situ, without any additional sample preparation or manipulation. The electrochemical curves obtained for a Li//Li1.20Mn0.54Co0.13Ni0.13O2 battery using either a coin cell or the cell developed for these operando synchrotron studies are compared in Figure 2 to check reproducibility between these experiments. The charge and discharge curves obtained for the in situ cell do appear very similar to those obtained classically for a coin cell (although with a slightly larger polarization), showing that this setup can be used to get more insight into the redox and structural changes occurring in Li1.20Mn0.54Co0.13Ni0.13O2 upon cycling. To check the homogeneity of the electrochemical reaction within the electrode, XAS measurements were performed at different locations of the electrode, at the end of the relaxation process following the end of discharge, for instance. The spectra obtained were found to be identical, showing that, at a C/10 rate, the electrochemical reaction occurs homogeneously within the electrode in this cell running during the operando XAS study. 2. XANES Study during the Charge and Discharge of Li//Li1.20Mn0.54Co0.13Ni0.13O2 Batteries. Operando XAS measurements were performed during the first and second cycles to get more insight into the irreversible reaction occurring during the first cycle. In addition, the first cycle was performed in different potential windows, the charge being either stopped just before the plateau observed at high voltage around 4.5 V vs Li+/Li or at 4.8 V (i.e., after the plateau). XANES spectra obtained at the Ni, Co, and Mn K-edges during the first cycle including the plateau, the first cycle excluding the plateau, and the second cycle (including the plateau) are shown in Figures 3−8. Representative XANES spectra are given for each transition metal K-edge for various compositions, as 5701

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Figure 2. Charge and discharge curves obtained for Li// Li1.20Mn0.54Co0.13Ni0.13O2 lithium cells: (a) first cycle obtained using a coin cell and (b) first cycle including (black) or excluding (blue) the plateau during the first charge, as well as second cycle including the plateau (red); the last three were obtained using the cell developed for operando synchrotron studies.12

indicated by arrows during the charge and discharge. XANES spectra were recorded in 500 ms (two spectra at 1 Hz) and the spectra reported here are in fact the sums of 100 single spectra. 2.1. Ni K-Edge. XANES spectra obtained at the Ni K-edge for LiNiII1/3Co1/3Mn1/3O2 and LiNiIII0.80Co0.15Al0.05O2 are given as dotted lines in Figures 3 and 4, as references for the various oxidation states of Ni in a layered oxide structure. The oxidation state of Ni in the pristine material was determined as Ni2+, as its spectrum is similar to that of LiNiII1/3Co1/3Mn1/3O2 (Figure 3a). During the charge before the plateau, XANES spectra recorded at the Ni K-edge clearly shift continuously to higher energies by 3.1 eV, to a higher value than that observed for LiNiIII0.80Co0.15Al0.05O2, as highlighted by the arrow in Figure 3a. This shift of the Ni K-edge and comparison with results already reported for LixNiO2 indicate that Ni2+ ions are oxidized to the Ni4+ oxidation state during the charge before the plateau.16−19 When the lithium cell is then discharged (before the plateau) Ni K-edge XANES spectra shift reversibly to lower energy, reaching the same position as that observed for the pristine material (not reported here). This shows that Ni ions are reduced back to the Ni2+ oxidation state during discharge. During the charge through the plateau, Ni K-edge XANES spectra almost do not shift (i.e., by less than 0.2 eV), and only a very small change in their shape is observed: the maximum of the absorption edge moves in one direction, whereas the edge moves in the other (Figure 3b). As expected, no oxidation of Ni ions is observed, but local structural changes probably occur. For instance, as already reported by Yabuuchi et al., partial migration of nickel ions from the octahedral sites of the slab to the octahedral sites of the interslab space can already markedly modify the Ni K-edge spectra.20 Note that, contrary to the results recently reported by Oishi et al. for a material with similar composition (stated to be Li1.16Ni0.15Co0.19Mn0.50O2), we did not observe any reduction

Figure 3. Normalized representative XANES spectra at the Ni K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the first charge before the plateau (a), the first charge during the plateau up to 4.8 V vs Li+/Li (b), and the first discharge after the plateau (c). The spectra for LiNiII1/3Co1/3Mn1/3O2 and LiNiIII0.80Co0.15Al0.05O2 are given as references.

of Ni ions upon charging on the plateau.21 During the discharge after the plateau, Ni K-edge XANES spectra shift continuously to lower energies by 3.4 eV, down to the position observed initially for the pristine material (Figure 3c). This reveals that Ni ions are reduced back to the Ni2+ oxidation state during discharge. During the second cycle performed after a first cycle in the potential window 2.5−4.8 V vs Li+/Li, Ni K-edge XANES spectra show, as commonly observed for other layered oxides, the oxidation of Ni ions to the Ni4+ oxidation state during the charge and their reversible reduction to the Ni2+ oxidation state during the discharge (Figure 4a,b). 2.2. Co K-Edge. LiNi1/3CoIII1/3Mn1/3O2 and CoII,III3O4 were measured as references for various oxidation states for Co ions. The Co K-edge XANES spectrum for the pristine material is close to that observed for the layered oxide LiNi1/3CoIII1/3Mn1/3O2, as shown in Figure 5a, showing that the oxidation state of Co ions in the pristine material is Co3+. During the charge before the plateau, the half-height energy of the XANES spectra does not shift significantly, as shown in Figure 5a. However, their shapes change continuously, and especially, the position of the maximum of absorption shifts to higher energies (ΔE ∼ 2.4 eV). As reported for LiNi1/3Co1/3Mn1/3O2 and LiCoO2 in refs 19, and 23, respectively, such subtle changes reveal that Co3+ ions are 5702

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Figure 4. Normalized representative XANES spectra at the Ni K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the second charge (a), during the second discharge (b), and after a first cycle in the potential window 2.5−4.8 V. The spectra for LiNi II 1/3 Co 1/3 Mn 1/3 O 2 and LiNiIII0.80Co0.15Al0.05O2 are given as references.

oxidized to Co4+ during the first charge before the plateau. Then, during the discharge before the plateau, the Co K-edge XANES spectra change continuously and reversibly back to one similar to that of the pristine material, showing that Co ions are reduced back to the Co3+ oxidation state (not reported here). During the plateau, only slight modifications are observed in the Co K-edge XANES spectra (Figure 5b), showing that, as expected, Co ions are not oxidized further during the plateau. Again, we also did not observe any reduction of cobalt on the plateau, as Oishi et al. did,21 but only the maximum of the absorption edge that moves to slightly higher energies, whereas the edge itself moves to slightly lower energies. As shown in Figure 5c, during the first discharge after the plateau the Co Kedge XANES spectra shift to lower energies, showing that Co ions are reduced to Co3+ as in the pristine material. During the second cycle performed after a first cycle in the potential window 2.5−4.8 V vs Li+/Li, the XANES spectra continuously shift to higher energies during the charge and to lower energies during the discharge, as shown in Figure 6a−b, in good agreement with the reversible oxidation and reduction of Co ions as commonly observed in layered oxides. 2.3. Mn K-Edge. LiNi1/3Co1/3MnIV1/3O2, LiMnIII,IV2O4, and Li2MnIVO3 were measured as references for various oxidation states of Mn ions; the comparison of their XANES spectra with that of the pristine material in Figure 7a supports the presence of Mn4+ in Li1.20Mn0.54Co0.13Ni0.13O2 as in Li2MnIVO3. Whatever the state of charge or of discharge during the first cycle, the Mn K-edge XANES spectra do not change significantly; the maximum of the absorption edge shifts in one direction, by 1.2 eV only, whereas the edge moves in the other direction, suggesting that Mn ions do not participate in the redox processes (Figures 7a−c and 8a). This observation is very similar to that observed at the Mn Kedge for LiNi1/3Co1/3MnIV1/3O2 that showed no participation of Mn in the redox processes.22 It is interesting to mention also

Figure 5. Normalized representative XANES spectra at the Co K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the first charge before the plateau (a), the first charge during the plateau up to 4.8 V vs Li+/Li (b), and the first discharge after the plateau (c). The spectra for LiNi1/3CoIII1/3Mn1/3O2 and CoII,III3O4 are given as references.

that Mn K-edge XANES spectra were reported to shift significantly upon cycling for other Mn-rich materials showing changes in Mn oxidation state; in addition, significant changes were also observed in the shape of their edges depending on the presence (or not) of Jahn−Teller distorted Mn3+ ions.24,25 Note that the changes observed at the Mn K-edge during the second cycle are very similar to those observed during the first cycle, i.e., no obvious shift in energy but only a continuous modification of the shape (Figure 8b−c). Such small changes are expected despite nonparticipation of Mn in the redox processes, since oxidation of Ni to the Ni4+ state induces an increase in the covalency of the Ni−O bonds and consequently changes in the distribution of the electrons between Mn and O in antagonistic bonds. The pre-edge of Mn K-edge XANES spectra is also expected to be very sensitive to changes in the oxidation state of Mn ions,26 with the peak at lower energy corresponding to the transitions from 1s to 3deg states and that at higher energy to the transitions from 1s to 3dt2g states. In our experiments, the pre-edge of XANES spectra recorded at the Mn K-edge during the first cycle does not shift continuously, as shown in Figure 9, confirming that the manganese oxidation state does not change. The continuous change in intensity of the global pre-edge during charge and discharge (Figure 9) reveals that the distortion of the octahedral sites increases during the first charge and decreases during the next discharge,27 but without recovering that of the pristine material. 5703

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Figure 6. Normalized representative XANES spectra at the Co K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the second charge (a) and the second discharge (b), after a first cycle in the potential window 2.5− 4.8 V. The spectra for LiNi1/3CoIII1/3Mn1/3O2 and CoII,III3O4 are given as references.

This is a further reason to conclude that Mn ions were not oxidized over the tetravalent state. In the case of the formation of supervalent Mn ions (i.e., Mnx+ with x > 4), significant changes of intensity should be observed for the pre-edge due to their migration from octahedral sites to tetrahedral sites for their stabilization.28 As shown in Figure 10, no difference is observed between the spectra obtained ex situ or operando, whatever the metal Kedge. Such reproducibility between operando and ex situ experiments is not surprising, as the studies were performed operando at C/10, which is a rather slow rate. As already alluded to, the Ni K-edge spectra recorded for the pristine material and the material obtained after the first cycle including the plateau are very similar in position and in shape, whereas the Co and Mn spectra are different in shape for the same materials. This difference between Ni and (Co, Mn) is expected considering the cation ordering observed in the slabs.13 Ni is surrounded by oxygen as first neighbors and then by Co and Mn as second neighbors. Its local environment in the slabs thus appears not to change significantly between the beginning and the end of the first cycle. On the contrary, Mn and Co are surrounded by oxygen as first neighbors and by [2Li, 1Ni, 3(Co, Mn)] as second neighbors; the difference in the Co and Mn spectra between the beginning and the end of the first cycle thus suggests that an irreversible modification occurs at the local scale, which could be attributed first to irreversible lithium deintercalation from the slabs but also to cation reorganization within the lithium vacancies created.11 A comparison of these XANES spectra at the Ni, Co, and Mn K-edges with those obtained after the 10th cycle is also given in Figure 10. All of these spectra are found to be very similar to those obtained for the material after the first cycle, showing that the oxidation state and environment of each transition metal ion does not change significantly between the first and the 10th cycles. In parallel, an irreversible plateau is observed only in the

Figure 7. Normalized representative XANES spectra at the Mn K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the first charge before the plateau (a), the first charge during the plateau up to 4.8 V vs Li+/Li (b), and the first discharge after the plateau (c). The spectra for LiNi1/3Co1/3MnIV1/3O2, LiMnIII,IV2O4, and Li2MnIVO3 are given as references.

first charge, and the following charge and discharge curves continuously change upon cycling.10 From the XANES analysis only we cannot therefore state that this change be caused by a change in the redox processes. In order to get more insight into changes in the local structure, EXAFS spectra were analyzed in detail in the following at the Ni, Co, and Mn K-edges. 3. Analysis of the EXAFS Spectra Recorded Operando upon Cycling. The metal−oxygen and metal−metal distances can be estimated from EXAFS spectra to follow changes in the oxidation state of each transition metal and in their local environment during the charge and discharge. A fit of the EXAFS data was performed using Artemis software. Better fitting results could be obtained by considering the description of the structure in the C2/m space group, i.e., with a cation local ordering as in Li2MnO3: Ni ions are surrounded by six transition metals, whereas Co and Mn ions are surrounded by four transition metals and two Li+ ions (see the Supporting Information, Figure S1). For comparison, a description of the structure considering the R3̅m space group as for LiNiO2 would lead to all transition metal ions being surrounded by 4.8 transition metal ions (as 0.2Li and 0.8M are present in the slabs). This result is another confirmation of the cation local ordering and is in good agreement with the results previously obtained by electron diffraction,7Li MAS NMR, and Raman spectroscopy.13 5704

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Figure 8. Normalized representative XANES spectra at the Mn K-edge for Li1.20Mn0.54Co0.13Ni0.13O2 during the first discharge before the plateau (a) and the second charge and the second discharge after a first cycle in the potential window 2.5−4.8 V (b, c). The spectra for LiNi1/3Co1/3MnIV1/3O2, LiMnIII,IV2O4, and Li2MnIVO3 are given as references. Figure 9. Pre-edge of normalized XANES spectra at the Mn K-edge of Li1.20Mn0.54Co0.13Ni0.13O2 during the first charge before the plateau (a), during the plateau up to 4.8 V vs Li+/Li (b), the first discharge after the plateau (c), and the first discharge before the plateau (d).

k3-Weighted Fourier transforms of EXAFS spectra recorded at the Ni, Co, and Mn K-edges are compared in Figure S2 (Supporting Information) for the pristine material, the material obtained in charge just before the plateau, the material obtained after a charge before the plateau followed by a first discharge, the material obtained after a full charge up to 4.8 V vs Li+/Li, and, finally, the material after the first full cycle. Changes in the intensity and position, especially for the first peak, which is related to the M−O distance, are observed during the first cycle. These changes are considered by some authors to be caused by oxygen loss during the first charge of Li-rich layered oxides.26 However, similar changes in intensity were observed by other authors in normal (non overlithiated) layered oxides17−19 and in materials with other structures.24 Furthermore, according to our results, the intensity of the first peak in EXAFS data decreases already during the charge before the plateau, i.e., at a step at which oxygen loss is not expected to happen. In our opinion, the change in the peak intensity is thus not directly linked to oxygen loss (which would indeed lead to a decrease in the number of oxygen neighbors) but most probably to a larger disorder (i.e., distribution of distances) as suggested by a larger Debye−Waller factor (see Table S1 in the Supporting Information). We have, however, also considered possible changes in the number of oxygen neighbors around each transition metal ion;

indeed, according to the mechanism of oxygen loss proposed to explain the plateau, one oxygen first neighbor out of six would be lost. No significant change from six was in fact observed in our case, even taking into account the possibly strong correlation between the Debye−Waller factor and the number of neighbors deduced from the fit. This result supports again that almost no oxygen is lost during the plateau.10,11 Metal−first oxygen neighbor distances and metal−first metal neighbor distances obtained from the refinement of k3-weighted Fourier transforms of EXAFS spectra are given in Figures 11−13, and the absolute error on each distance is ±0.01. During the first cycle, changes observed in the Ni−O distances including or not the plateau are compared in Figure 11a to those of Ni II −O and Ni III −O distances in the LiNiII1/3Co1/3Mn1/3O2 and LiNiIII0.80Co0.15Al0.05O2 references, respectively. The Ni−O distance decreases during the charge before the plateau to values smaller than that observed for LiNiIII0.80Co0.15Al0.05O2. This is in good agreement with a decreasing ionic radius for the Ni ions and their oxidation to NiIV upon Li deintercalation. Then, in discharge after a charge 5705

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Figure 11. Distances determined between Ni and its first oxygen neighbors (a) and between Ni and its first metal neighbors (b) from the analysis of the EXAFS spectra recorded operando during the first cycle with (full symbols) and without (open symbols) the plateau. Ni− O and Ni−M distances determined from analysis of the EXAFS spectra during the second cycle of Li//Li1.20Mn0.54Co0.13Ni0.13O2 (c, d). LiNi1/3Co1/3Mn1/3O2 and LiNi0.80Co0.15Al0.05O2 are given as reference for NiII−O(M) and NiIII−O(M), respectively.

Figure 10. Normalized XANES spectra of Li1.20Mn0.54Co0.13Ni0.13O2 at the Ni K-edge (a), Co K-edge (b), and Mn K-edge (c) for the pristine material, the pristine material within the cell developed for operando studies, the material recovered ex situ after the first cycle of a lithium cell, the material after the first cycle within the operando cell, and the material recovered ex situ from a lithium cell after 10 cycles.

just before the plateau, the Ni−O distance increases and recovers its initial value, i.e., that observed for the pristine material. On the other hand, the Ni−O distance remains stable during the plateau, again supporting that Ni ions are not oxidized during the plateau. In discharge after a charge including the plateau (up to 4.8 V vs Li+/Li), the Ni−O distance increases and reaches that observed for the pristine material, again suggesting a reduction of Ni. Such redox reactions involving Ni ions during the first cycle agree with the analysis above considering the shift of XANES spectra. The Ni−M distance changes similarly, in parallel to the Ni−O distance, as shown in Figure 11b, decreasing during the charge before the plateau, being stable during the plateau, and increasing in discharge. In the second cycle, the Ni−O and Ni−M distances change continuously and reversibly as classically observed for normal layered oxides, as shown in Figure 11a−b. Changes in the Co−O distance during the first cycle are shown in Figure 12a and compared with that of CoIII−O in LiNi1/3CoIII1/3Mn1/3O2. These changes follow exactly the same trends as those for the Ni−O distance during the first and second cycles, although to a smaller extent since Co ions are oxidized from Co3+ to Co4+, whereas Ni ions are oxidized from

Figure 12. Distances determined between Co and its first oxygen neighbors (a) and between Co and its first metal neighbors (b) from the analysis of the EXAFS spectra recorded operando during the first cycle with (full symbols) and without (open symbols) the plateau. Co−O and Co−M distances determined from the analysis of the EXAFS spectra during the second cycle of Li// Li1.20Mn0.54Co0.13Ni0.13O2 (c, d). LiNi1/3Co1/3Mn1/3O2 is given as reference for CoIII−O(M).

5706

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

Ni2+ to Ni4+. The changes observed during the second cycle are expected and observed also for normal (non overlithiated) layered oxides. As mentioned above, changes in the oxidation state of Mn ions during the first cycle were not clear from the analysis of XANES spectra at the Mn K-edge, and therefore, a complementary analysis of the Mn−O distances is required to support further our conclusions. The Mn−O distances determined during the first cycle are compared in Figure 13a

larger one, showing again that Mn is not reduced below the Mn4+ oxidation state. We note that the Mn−M distance, in the reduced state, is slightly larger than in the pristine material (Figure 13b). This effect is not observed for Ni, whereas it is also observed for Co (Figure 12b). It supports again that the local second neighbors’ environment of Mn and Co, sitting in the same crystallographic site in the pristine material, is significantly altered after the first charge. In contrast, as expected, this alteration does not occur for Ni. 4. Discussion. Electrochemical experiments have revealed that during the first charge 1.03 mol Li is deintercalated from the layered structure out of which 0.39 mol are compensated for by oxidation of Ni2+ and Co3+ to Ni4+ and Co4+, respectively.10,11 These oxidations are confirmed by shifts to higher energy for XANES spectra recorded at the Ni and Co Kedges and by changes observed in the Ni−O and Co−O distances from EXAFS spectra recorded at the Ni and Co Kedges. It was proposed that the next 0.64 mol Li deintercalated during the first charge are compensated for by oxidation of oxide ions before their release as oxygen gas and then that Mn participates, with Ni and Co, as an additional redox center with its reduction to Mn3+ in discharge and its oxidation to Mn4+ in charge.9 In fact, changes in the shape of the XANES spectra recorded at the Co and Mn K-edges during the plateau in the first charge indicate that alteration of the local structure around Co and Mn ions occurs, but do not indicate at all their further oxidation. Then, as expected, reduction of Ni and Co from Ni4+ and Co4+ to Ni2+ and Co3+ is confirmed in discharge by the shift to lower energy of XANES spectra recorded at Ni and Co K-edges, but no reduction of Mn ions could be observed in discharge from the XANES spectra recorded at the Mn K-edge and from the Mn−O distances deduced from EXAFS spectra. The contribution of Mn to the redox processes appears, in fact, negligible during the first cycles. All the results deduced from these XAS data, from the local structures to the transition metals’ oxidation states, support thus again a participation of oxygen to the redox processes, such that oxide ions are oxidized during the plateau during the first charge and, for most of them, reversibly reduced in the next discharge.10,11,26

Figure 13. Distances determined between Mn and its first oxygen neighbors (a) and between Mn and its first metal neighbors (b) from the analysis of the EXAFS spectra recorded operando during the first cycle including (full symbols) and excluding (open symbols) the plateau. Mn−O and Mn−M distances determined from the analysis of the EXAFS spectra during the second cycle of Li// Li1.20Mn0.54Co0.13Ni0.13O2 (c, d). LiNi1/3Co1/3Mn1/3O2 as well as Li2MnO3 and LiMn2O4 are given as reference for MnIV−O(M) and MnIII,IV−O(M), respectively.

IV. CONCLUSIONS Redox reactions of Li1.20Mn0.54Co0.13Ni0.13O2 during the charge and discharge were characterized by operando XAS measurements. Changes in the oxidation state of each transition metal ion were followed from the shift of XANES spectra recorded at the Ni, Co, and Mn K-edges and from modifications in the metal−oxygen distances from EXAFS data. Ni and Co are oxidized in the first charge before the plateau and maintained at the Ni4+ and Co4+ oxidation states during the plateau. Mn is stable as Mn4+ throughout the first charge. This entails that oxide ions are oxidized on the plateau with Li deintercalation. In discharge, Ni and Co are reduced and recover their initial states observed in the pristine material, whereas Mn is not reduced. Therefore, these results support again, as those reported in refs 10 and 11, that the oxygen loss model is not sufficient to explain the reactions occurring upon cycling of Li1.20Mn0.54Co0.13Ni0.13O2, and thus, the involvement of oxygen in the reversible redox reactions, in addition to that of Ni and Co, is occurring. Getting direct evidence of the participation of oxygen in the redox processes is not obvious, as probing the electronic state of oxygen is not straightforward by X-ray absorption, especially when the sample is recovered from a battery with electrolyte

with MnIV−O ones in LiNi1/3Co1/3MnIV1/3O2 and Li2MnO3 and with Mn”3.5+”−O ones in LiMnIII,IV2O4. The Mn−O distance decreases continuously from the beginning of the charge to high voltage and then reversibly increases in discharge, recovering its initial value. The total variation of the Mn−O distance over the potential window is small (0.01 Å). As shown in Figure 13a, similar changes occur during the second cycle, but again smaller than those observed for Ni and Co ions. As similar changes in the Mn−O distance were observed in LiNi1/3Co1/3Mn1/3O2, for which Mn maintains at the tetravalent state all along the charge and discharge processes,29 they could suggest modification in the local environment around Mn. Indeed, in a layered structure made of edge-sharing octahedra, the size of a given octahedron is strongly related to (or constrained by) the size of the neighboring ones. In oxidation when cobalt and nickel ions are oxidized, there is a general lattice contraction (the ahex. parameter decreases)10 leading to a small decrease in the Mn− O distances. The partial oxidation of oxygen at the end of the charge acts in the same way. The opposite effects occur in discharge, where the Mn−O recovers its initial value, but not a 5707

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

[NixLi1/3−2x/3Mn2/3−x/3]O2. J. Electrochem. Soc. 2002, 149, A778− A791. (8) Jiang, M.; Key, B.; Meng, Y. S.; Grey, C. P. Electrochemical and Structural Study of the Layered, “Li-Excess” Lithium-Ion Battery Electrode Material Li[Li1/9Ni1/3Mn5/9]O2. Chem. Mater. 2009, 21, 2733−2745. (9) Armstrong, R.; Holzapfel, M.; Novak, P.; Johnson, C. S.; Kang, S.H.; Thackeray, M. M.; Bruce, P. G. Demonstrating Oxygen Loss and Associated Structural Reorganization in the Lithium Battery Cathode Li[Ni0.2Li0.2Mn0.6]O2. J. Am. Chem. Soc. 2006, 128, 8694−8698. (10) Koga, H.; Croguennec, L.; Ménétrier, M.; Mannessiez, P.; Weill, F.; Delmas, C. Different Oxygen Redox Participation for Bulk and Surface: A Possible Global Explanation for the Cycling Mechanism of Li1.20Mn0.54Co0.13Ni0.13O2. J. Power Sources 2013, 236, 250−258. (11) Koga, H.; Croguennec, L.; Ménétrier, M.; Douhil, K.; Belin, S.; Bourgeois, L.; Suard, E.; Weill, F.; Delmas, C. Reversible Oxygen Participation to the Redox Processes Revealed for Li1.20Mn0.54Co0.13Ni0.13O2. J. Electrochem. Soc. 2013, 160, A786−A792. (12) Leriche, J. B.; Hamelet, S.; Shu, J.; Morcrette, M.; Masquelier, C.; Ouvrard, G.; Zerrouki, M.; Soudan, P.; Belin, S.; Elkaïm, E.; et al. An Electrochemical Cell for Operando Study of Lithium Batteries Using Synchrotron Radiation. J. Electrochem. Soc. 2010, 157, A606− A610. (13) Koga, H.; Croguennec, L.; Mannessiez, P.; Ménétrier, M.; Weill, F.; Bourgeois, L.; Duttine, M.; Suard, E.; Delmas, C. Li1.20Mn0.54Co0.13Ni0.13O2 with Different Particle Sizes as Attractive Positive Electrode Materials for Lithium-Ion Batteries: Insights into Their Structure. J. Phys. Chem. C 2012, 116, 13497−13506. (14) Fonda, E.; Rochet, A.; Ribbens, M.; Belin, S.; Briois, V. The SAMBA Quick-EXAFS Monochromator: XAS with Edge Jumping. J. Synchrotron Radiat. 2012, 19, 417−424. (15) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: Data Analysis for X-ray Absorption Spectroscopy Using IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537−541. (16) Kim, Ni_M. G.; Sung, N. E.; Shin, H. J.; Shin, N. S.; Ryu, K. S.; Yo, C. H. Ni and Oxygen K-edge XAS Investigation into the Chemical Bonding for Lithiation of LiyNi1−xAlxO2 Cathode Material. Electrochem. Acta 2004, 50, 501−504. (17) Balasubramanian, M.; Sun, X.; Yang, X. Q.; McBreen, J. In Situ X-ray Diffraction and X-ray Absorption Studies of High-rate LithiumIon Batteries. J. Power Sources 2001, 92, 1−8. (18) Mansour, A. N.; Yang, X. Q.; Sun, X.; McBreen, J.; Croguennec, L.; Delmas, C. In Situ X-Ray Absorption Spectroscopy Study of Li(1‑z)Ni(1+z)O2 (z ≤ 0.02) Cathode Material. J. Electrochem. Soc. 2000, 147, 2104−2109. (19) Yoon, W.-S.; Chung, K. Y.; McBreen, J.; Fischer, D. A.; Yang, X.Q. Electronic Structural Changes of the Electrochemically Li-Ion Deintercalated LiNi0.8Co0.15Al0.05O2 Cathode Material Investigated by X-ray Absorption Spectroscopy. J. Power Sources 2007, 174, 1015− 1020. (20) Yabuuchi, N.; Yoshii, K.; Myung, S.-T.; Nakai, I.; Komaba, S. Detailed Studies of a High-Capacity Electrode Material for Rechargeable Batteries, Li2MnO3−LiCo1/3Ni1/3Mn1/3O2. J. Am. Chem. Soc. 2011, 133, 4404−4419. (21) Oishi, M.; Fujimoto, T.; Takanashi, Y.; Orikasa, Y.; Kawamura, A.; Ina, T.; Yamashige, H.; Takamatsu, D.; Sato, K.; Murayama, H.; et al. Charge Compensation Mechanisms in Li1.16Ni0.15Co0.19Mn0.50O2 Positive Electrode Material for Li-Ion Batteries Analyzed by a Combination of Hard and Soft X-ray Absorption Near Edge Structure. J. Power Sources 2013, 222, 45−51. (22) Kim, J.-M.; Chung, H.-T. Role of Transition Metals in Layered Li[Ni,Co,Mn]O2 under Electrochemical Operation. Electrochim. Acta 2004, 49, 3573−3578. (23) Nakai, I.; Takahashi, K.; Shiraishi, Y.; Nakagome, T.; Izumi, F.; Ishii, Y.; Nishikawa, F.; Konishi, T. X-ray Absorption Fne Structure and Neutron Diffraction Analyses of De-Intercalation Behavior in the LiCoO2 and LiNiO2 Systems. J. Power Sources 1997, 68, 536−539.

degradation products (containing also oxygen) at its surface. Very recently, Tarascon and co-workers have studied the Li2Ru1‑ySnyO3 system as a lithium-rich layered oxide showing also an irreversible high voltage plateau during the first charge and an overcapacity versus the number of electrons that can be exchanged considering the transition metal ions only.30 Interestingly, using X-band EPR measurements at room temperature and 4 K, they got direct evidence for the formation of oxidized oxygen species during the first charge and came, as we did, to the conclusion that overcapacity in the lithium-rich layered oxides is associated with reversible oxygen participation in the redox processes.



ASSOCIATED CONTENT

S Supporting Information *

Fourier transforms of k3-weighted EXAFS spectra of pristine material Li1.20Mn0.54Co0.13Ni0.13O2 at the Ni, Co, and Mn Kedge and calculation estimated with Artemis software considering the C2/m unit cell (Figure S1); Fourier transforms of k3-weighted EXAFS spectra of Li1.20Mn0.54Co0.13Ni0.13O2 at the Ni, Co, and Mn K-edge for the pristine material, the material obtained in charge just before the plateau, the material obtained at the end of the charge up to 4.8 V vs Li+/Li, the material obtained after the first cycle with the plateau, and the material obtained after the first cycle without the plateau (Figure S2); EXAFS parameters of Li1.20Mn0.54Co0.13Ni0.13O2 at the Ni, Co, and Mn K-edges (Table S1). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Authors

*L.C.: Tel: +33 (0) 5 4000 2234 (or 2647). E-mail: crog@ icmcb-bordeaux.cnrs.fr . *S.B.: +33 (0) 1 6935 9645 (or 9723). E-mail: stephanie. [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank Région Aquitaine and Toyota for their financial support.



REFERENCES

(1) Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. LixCoO2 (0 < x < −1): A New Cathode Material for Batteries of High Energy Density. Mater. Res. Bull. 1980, 15, 783−789. (2) Masquelier, C.; Croguennec, L. Polyanionic (Phosphates, Silicates, Sulfates) Frameworks as Electrode Materials for Rechargeable Li (or Na) Batteries. Chem. Rev. 2013, 113, 6552−6591. (3) Ellis, B. L.; Lee, K. T.; Nazar, L. F. Positive Electrode Materials for Li-Ion and Li-Batteries. Chem. Mater. 2010, 22, 691−714. (4) Lu, Z.; MacNeil, D. D.; Dahn, J. R. Layered Cathode Materials Li[NixLi(1/3−2x/3)Mn(2/3−x/3)]O2 for Lithium-Ion Batteries. Electrochem. Solid State Lett. 2001, 4, A191−A194. (5) Park, Y. J.; Hong, Y. S.; Wu, X.; Ryu, K. S.; Chang, S. H. Structural Investigation and Electrochemical Behaviour of Li[NixLi(1/3−2x/3)Mn(2/3−x/3)]O2 Compounds by a Simple Combustion Method. J. Power Sources 2004, 129, 288−295. (6) Wu, Y.; Manthiram, A. High Capacity, Surface-Modified Layered Li[Li(1−x)/3Mn(2−x)/3Nix/3Cox/3]O2 Cathodes with Low Irreversible Capacity Loss. Electrochem. Solid-State Lett. 2006, 9, A221−A224. (7) Lu, Z.; Beaulieu, L. Y.; Donaberger, R. A.; Thomas, C. L.; Dahn, J. R. Synthesis, Structure, and Electrochemical Behavior of Li5708

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709

The Journal of Physical Chemistry C

Article

(24) Chan, H.-W.; Duh, J.-G.; Lee, J.-F. Valence Change by In Situ XAS in Surface Modified LiMn2O4 for Li-Ion Battery. Electrochem. Commun. 2006, B, 1731−1736. (25) Kim, G.-Y.; Yi, S.-B.; Park, Y. J.; Kim, H.-G. Electrochemical Behaviors of Li[Li(1‑x)/3Mn(2‑x)/3Nix/3Cox/3]O2 Cathode Series (0 < x < 1) Synthesized by Sucrose Combustion Process for High Capacity Lithium Ion Batteries. Mater. Res. Bull. 2008, 43, 3543−3552. (26) Ito, A.; Sato, Y.; Sanada, T.; Hatano, M.; Horie, H.; Ohsawa, Y. In situ X-ray Absorption Spectroscopic Study of Li-Rich Layered Cathode Material Li[Ni0.17Li0.2Co0.07Mn0.56]O2. J. Power Source 2011, 196, 6828−6834. (27) Simonin, L.; Colin, J.-F.; Ranier, V.; Canevet, E.; Martin, J. -F.; Bourbon, C.; Baehtz, C.; Strobel, P.; Daniel, L.; Patoux, S. In Situ Investigations of a Li-Rich Mn−Ni Layered Oxide for Li-Ion Batteries. J. Mater. Chem. 2012, 22, 11316−11322. (28) Balasubramanian, M.; McBreen, J.; Davidson, I. J.; Whitfield, P. S.; Kargina, I. In Situ X-Ray Absorption Study of a Layered Manganese−Chromium Oxide-Based Cathode Material. J. Electrochem. Soc. 2002, 149, A176−A184. (29) Tsai, Y. W.; Hwang, B. J.; Ceder, G.; Sheu, H. S.; Liu, D. G.; Lee, J. F. In-Situ X-ray Absorption Spectroscopic Study on Variation of Electronic Transitions and Local Structure of LiNi1/3Co1/3Mn1/3O2 Cathode Material during Electrochemical Cycling. Chem. Mater. 2005, 17, 3191−3199. (30) Sathiya, M.; Rousse, G.; Ramesha, K.; Laisa, C. P.; Vezin, H.; Sougrati, M. T.; Doublet, M.-L.; Foix, D.; Gonbeau, D.; Walker, W.; et al. Reversible Anionic Redox Chemistry in High-Capacity LayeredOxide Electrodes. Nat. Mater. 2013, 12, 827−83.

5709

dx.doi.org/10.1021/jp412197z | J. Phys. Chem. C 2014, 118, 5700−5709