# Body

Structural changes in Li 2 MnO 3 cathode material for rechargeable Li-ion batteries are investigated during the fi rst and 33 rd cycles. It is found that both the participation of oxygen anions in redox processes and Li + -H + exchange play an important role in the electrochemistry of Li 2 MnO 3 . During activation, oxygen removal from the material along with Li gives rise to the formation of a layered MnO 2 -type structure, while the presence of protons in the interslab region, as a result of electrolyte oxidation and Li + -H + exchange, alters the stacking sequence of oxygen layers. Li re-insertion by exchanging already present protons reverts the stacking sequence of oxygen layers. The re-lithiated structure closely resembles the parent Li 2 MnO 3 , except that it contains less Li and O. Mn 4 + ions remain electrochemically inactive at all times. Irreversible oxygen release occurs only during activation of the material in the fi rst cycle. During subsequent cycles, electrochemical processes seem to involve unusual redox processes of oxygen anions of active material along with the repetitive, irreversible oxidation of electrolyte species. The deteriorating electrochemical performance of Li 2 MnO 3 upon cycling is attributed to the structural degradation caused by repetitive shearing of oxygen layers.

# 1 . Introduction

The exploration of electrochemical activation in Li 2 MnO 3 is important for two main reasons: fi rst, to enhance the fundamental understanding concerning the electrochemistry of Mn 4 + -containing cathode materials: and second, to obtain the knowledge necessary for designing a possible class of 'Li-rich' cathode materials, [ 1-9 ] in which Li 2 MnO 3 is an important component. Li 2 MnO 3 has an O3 structure where close-packed oxygen layers are stacked in an ABCABC sequence. In a layer notation, Li 2 MnO 3 can be written as Li[Li 1/3 Mn 2/3 ]O 2 where the interslab octahedral sites are occupied by Li only, while the octahedral sites within the [Li 1/3 Mn 2/3 ]O 2 slabs are occupied by both Li and Mn in a ratio of 1:2. [ 10 ] The fact that in Li 2 MnO 3 Mn 4 + ions are octahedrally coordinated by oxygen led to the initial belief that the material is electrochemically inactive. [ 11 ] However, this belief was disproven when a substantial charge could be extracted from the material during an activation process that occurs at 4.5 V vs. Li/Li + . [ 12 ] Over the past few years, several mechanisms concerning electrochemical activation of Li 2 MnO 3 during the fi rst charge have been postulated. It has been proposed that Li extraction from Li 2 MnO 3 occurs with the simultaneous release of oxygen. [ 11 ] Later, it was observed that Li extraction occurs via two competitive processes, namely oxygen removal and Li + -H + exchange. [ 13-15 ] The presence of structural defects such as oxygen vacancies also plays an important role in the electrochemical performance of Li 2 MnO 3 . [ 16,17 ] While most of these reports elucidate the structural changes upon Li extraction, [ 13-15,18 ] none provide information obtained by direct experimental evidence from electrochemically treated samples about what the structural modifi cations during Li re-insertion and upon prolonged cycling are. We study structural modifi cations in Li 2 MnO 3 cathode material during the fi rst and 33 rd cycles by X-ray absorption spectroscopy (XAS). The elemental selectivity of XAS provides a unique opportunity to probe chemical, electronic and structural changes at and around an absorbing atom. The near-edge region of the absorption spectra is called the X-ray absorption near edge structure (XANES) and provides qualitative information about the average valence state of absorbing atoms, their local symmetry, and electronic confi guration. On the other hand, the extended region of the absorption spectra is called the extended X-ray absorption fi ne structure (EXAFS) and provides quantitative information about geometrical changes in the vicinity of absorbing atoms. In the present study, the average valence state of Mn and its local symmetry in various Li 2 MnO 3 samples are qualitatively discussed in comparison with various manganese reference compounds, while structural changes such as the number of nearest neighbors, interatomic distances, and structural disorder are quantifi ed by fi tting a theoretical model to the EXAFS data.

# 2 . Results

As shown in Figure 1 a, the cell delivered a specifi c capacity of 388 mAh g -1 when charged to 5 V and of 200 mAh g -1 when discharged to 2 V with a fi rst cycle effi ciency of ≈ 50% (Figure 1 b). The long plateau above 4.5 V vs. Li/Li + during the fi rst charge that delivers most of the capacity is a characteristic of Li 2 MnO 3 [ 13-15,18,19 ] and Li 2 MnO 3 -containing cathode materials. [ 1-9 ] The capacity delivered during the fi rst charge (i.e., activation) depends on the rate at which the cell is charged. However, the capacity during subsequent cycles is independent of the rate. Therefore, the cell was fi rst-charged to 5 V at C/50, while subsequent cycling was carried out at C/10. As shown in Figure 1 b, the electrochemical performance of Li 2 MnO 3 degrades upon cycling. During the 33 rd cycle, the cell delivers a specifi c capacity of ≈ 110 mAh g -1 both during charge and discharge, which is just ≈ 55% of that of the fi rst discharge.

# 2.1 . XANES

As shown in Figure 2 a, the absorption edge is characterized by a variety of edge-features such as 1) an absorption threshold (or the fi rst peak in the derivative spectrum, not shown), 2) a 1s → 3d transition, corresponding to the pre-edge peak, 3) the main edge, corresponding to the vertically rising portion of the edge (or the second peak in the derivative spectrum, not shown), and 4) a 1s → 4p transition, corresponding to the maximum of the vertically rising portion of the edge. [ 20 ] A relative shift in the position of these features, often called a chemical shift, provides an empirical mean for the estimation of the average valence state of the central absorbing atom. The inset in Figure 2 a shows an enlarged pre-edge region of these spectra. In order to enhance the visibility of the pre-edge peaks of other reference compounds, the intense pre-edge peak of KMnO 4 is omitted from the inset. The pre-edge regions of MnO, MnO 2 and KMnO 4 are characterized by a single 1s → 3d peak, while for Mn 2 O 3 this peak is split into two poorly resolved t 2g and e g peaks. For LiMn 2 O 4 , this peak splitting is better resolved. The intensity of the pre-edge peak is minimum for MnO, intermediate for MnO 2 , and maximum for KMnO 4 . Inner d-orbitals are more tightly bound and, therefore, less sensitive to chemical changes than loosely bound outer p-orbitals. Thus, chemical shifts observed at the position of the 1s → 4p peak are more pronounced than those observed at any other features. [ 20,21 ] As a result, chemical shifts are interpreted in terms of the position of the 1s → 4p peak. Figure 2 b shows that with increasing formal valence state of Mn the position of this peak moves to higher energy and the average metal-ligand bond length decreases. The reported values of bond lengths are obtained by fi tting the EXAFS data of individual reference compounds (refer to Figure S1 and Table S1 in the Supporting Information). As shown in Figure 3 , the absorption spectra for various Li 2 MnO 3 samples appear more complex than those of manganese reference compounds due to the presence of several points of intersection along the main edge where some of the spectra cross. Closer examination of these spectra reveals that the trend in chemical shifts varies widely before and after the intersection points. As shown in the inset, the pre-edge region in the pristine state is characterized by a splitting of the Mn 3d orbitals into t 2g and e g orbitals. As the cathode is charged to 5 V, the intensity of the pre-edge peak increases and the peak splitting becomes less clear. When the cathode is discharged to 2 V, the intensity of the pre-edge peak decreases. For the cycled samples, changes in the pre-edge region are less obvious.

# 2.2 . EXAFS

The comparison between the χ ( k ) signals of various Li 2 MnO 3 samples in Figure 4 a reveals that the EXAFS oscillations are reduced as soon as the cathode is charged to 5 V. In the corresponding Fourier transforms (Figure 4 b), this is refl ected by a preferential reduction in the amplitude of the fi rst shell of O atoms in addition to an overall damping of other shells. Qualitatively, the χ ( k ) signal of the cycled-charged sample (i.e., the sample charged to 5 V during the 33 rd cycle) appears similar to that of the sample charged during the fi rst cycle, except that the signal is further damped. Along with the clear reduction in the peak amplitudes, the higher order shells of Mn atoms (at ≈ 4.7 Å) and O atoms (at ≈ 5.3 Å) largely disappear upon cycling. As shown in Figure 5 a, every feature of the χ ( k ) signal corresponding to the pristine sample is reproduced when the cathode is discharged to 2 V during the fi rst cycle, however, with reduced amplitudes. This is also indicated by similarities in the Fourier transforms (Figure 5 b). Again, the EXAFS signals corresponding to the cycled-discharged sample (i.e., the sample discharged to 2 V during the 33 rd cycle) appear similar to those of the sample discharged during the fi rst cycle, however, the higher order shells are damped. EXAFS data in the pristine state were fi tted by the monoclinic structure of Li 2 MnO 3 ( C2 / m ) [ 22 ] ( Figure 6 ). Best-fi t parameters are reported in Table S2 in the Supporting Information. A product of S 2 0 and the number of coordinating atoms ( N ) in a given shell determines its amplitude. As a result, these two parameters for a given shell cannot be varied independently in a fi t. This can be realized by the fact that when the number of oxygen nearest neighbors ( O N-N ) in the pristine Li 2 MnO 3 was constrained to that obtained from the theoretical model, the fi t refi ned S 2 0 to 0.76(4), which is consistent with S 2 0 values obtained by fi tting the EXAFS data of other manganese reference compounds. Conversely, when S 2 0 was constrained to 0.76, the fi t refi ned O N-N to 6.0(4), which is consistent with the fact that in Li 2 MnO 3 Mn 4 + ions are octahedrally coordinated by oxygen. Additionally, F 2 values, which represent disorder in the material also affect the amplitude and, therefore, the terms S 2 0 × N and F 2 for a given coordination shell, are highly correlated. Thus, determining N in highly disordered systems remains a challenge. The most signifi cant contribution to the EXAFS signal is due to backscattering of ejected photoelectrons from the nearest neighbors (i.e., O N-N in case of Li 2 MnO 3 ). The contribution from outer coordination shells decreases due to increased inelastic losses as photoelectrons propagate in the material. Since S 2 0 is the property of absorbing atoms, it is transferable between compounds where the absorbing atom in question is coordinated by the same type of nearest neighbors. On the other hand, F 2 cannot be estimated a priori. Thus, in situations where structural changes are to be followed as a result of any physical or chemical change that the sample has undergone, S 2 0 can be estimated from the starting material. From this aspect, S 2 0 determined from the pristine sample of Li 2 MnO 3 was regarded as the best estimate because it has a known oxygen coordination. The S 2 0 value thus determined was constrained when structural changes due to changes in O N-N in the material were to be estimated. It has been reported that Li extraction from Li 2 MnO 3 occurs with a simultaneous loss of oxygen, giving rise to the formation of a layered MnO 2 -type structure (trigonal, R 3 m ) with the average valence state of Mn remaining unchanged at + 4. [ 11 ] There are also reports claiming the formation of either a layered MnO 2 -type structure or an α -MnO 2 -type structure (tetragonal, I4/m ) upon chemical delithiation of Li 2 MnO 3 . [ 23,24 ] In order to assess such reports, the EXAFS data of the sample charged to 5 V during the fi rst cycle were fi tted using the crystal structures of α -MnO 2 and layered MnO 2 . The best fi t was obtained assuming the layered MnO 2 structure ( Figure 7 a). Best-fi t parameters are reported in Table S3 in the Supporting Information. The fi t constraining S 2 0 to 0.76 refi ned O N-N to 5.6(8), which is high considering that oxygen is released from the material. It has been proposed that Li re-insertion into layered MnO 2 occurs with the simultaneous reduction of Mn 4 + to Mn 3 + , thus forming LiMnO 2 . [ 6,25 ] Therefore, an attempt was made to fi t the EXAFS data of the sample discharged to 2 V during the fi rst cycle by a layered LiMnO 2 structure ( R 3 m ). However, there was no agreement between the data and theory. The data were rather explained by a Li 2 MnO 3 structure (Figure 7 b). Best-fi t parameters are reported in Table S4 in the Supporting Information. The fi t refi ned O N-N to 4.3(4) when S 2 0 was constrained to 0.76. It has been claimed that prolonged cycling of Li 2 MnO 3 gives rise to the formation of LiMn 2 O 4 -type spinel phase. [ 14 ] Therefore, an attempt was made to refi ne the fraction of such spinel phase in the cycled samples. However, there was no good agreement between the data and theory. For comparison, the EXAFS data for the LiMn 2 O 4 spinel are shown in Figure 8 . Qualitatively, the EXAFS data of the cycled samples appear similar to those of their counterparts during the fi rst cycle. As a result, a layered MnO 2 structure was assumed to fi t the data of the cycled-charged sample. When S 2 0 was constrained to 0.76, the fi t refi ned O N-N to 3.3(4), which is very low considering the weak pre-edge peak of the cycled samples (inset in Figure 3 ), suggesting octahedral coordination. This fi t result emphasizes the effect of the observed reduction in the EXAFS signals upon cycling. A gradually declining electrochemical performance of the cell during cycling (Figure 1 b) suggests a gradual loss in the intercalation ability of the material. Therefore, a parameter describing an effective concentration of electrochemically active material, X eff , was introduced to the model such that the amplitude of each coordination shell of the cycled samples was represented by the term S 2 0 × N × X eff . A fi t constraining S 2 0 to 0.76 and O N-N to 4.3, refi ned X eff to 0.73(6) (Table S5 in the