富锂层状锂电材料中氧空位的形成

Christopher R.Fell,†,∥Danna Qian,§Kyler J.Carroll,§Miaofang Chi,‡Jacob L.Jones,†

and Ying Shirley Meng*,†,§

†Department of Materials Science and Engineering,University of Florida,Gainesville,Florida32611,United States

‡Materials Science and Technology Division,Oak Ridge National Laboratory,Oak Ridge,Tennessee37831,United States

§Department of NanoEngineering,University of California San Diego,La Jolla,California92037,United States

*Supporting Information

during thefirst electrochemical cycle

exceeding200mAh/g in the Li-excess

vacancy,microstrain

Rechargeable lithium ion batteries are a potential candidate for the use as the primary onboard storage technology in plug-in hybrid electric vehicles(PHEVs)or electric vehicles(EVs). Specifically,the layered lithium-excess transition metal oxides, Li[Ni x Li1/3−2x/3Mn2/3−x/3]O2(0fied that the lithium chemical diffusion coefficient is significantly lower at the plateau region.11 Recently Li et al.reported that the capacity originating from the high voltage plateau is strongly dependent on the charging current indicating slow kinetics.12

Considerable research has been devoted to address the irreversible capacity losses as well as the poor rate capability by understanding the mechanisms associated with thefirst electrochemical charge/discharge cycle.Researchers have proposed simultaneous Li+and O2−diffusion creating oxygen vacancies and surface transition metal migration.7,8,13−15These structural rearrangements are believed to happen mainly during thefirst electrochemical cycle since the voltage plateau does not appear in subsequent cycles.For oxygen diffusion to occur, oxygen must overcome the activation energy barrier,a process that has been widely studied in ionic conductors such as solid oxide fuel cell electrolytes.16−20The diffusion mechanism can be explained by interstitialcy,interstitials,and vacancy Received:January2,2013

Revised:March20,2013

mechanisms depending on point defects in the system.The introduction of strain at or near interfaces has a powerful effect on the ionic diffusion properties of ceramics;21,22however,the relation of lattice strain and ionic transport in the lithium excess layered oxides is not well understood.Lattice strainsfirst elastically stretch the cation−oxygen bond,effectively weaken-ing the bond and decreasing the migration barrier leading to exponential increases in oxygen diffusivity.16,17,23,24During electrochemical cycling of the layered lithium-excess series of materials,significant cationic rearrangement occurs.It has also been shown that following electrochemical cycling the superlattice peaks disappear indicating a loss of Li/Mn honeycomb-type ordering.7,15,25During electrochemical cy-cling,the material shows volume expansion and significant shifts in the c/a lattice ratio.8,11,15,26At the surface of the material,changes in TM oxidation state and local environments as well as possible material loss have been reported.5,9,14,26 Recent research shows the formation of a defect-like spinel phase that may contribute to thefirst cycle irreversible capacity and poor rate capability.26,27We hypothesize that a combination of these changes during the electrochemical cycling provides the defect sites that enable oxygen mobility at ambient temperature.

It is crucial to understand how defects are generated and to quantify the microstrain associated with these defects.In this work,we carried out a detailed study using synchrotron X-ray diffraction(SXRD),aberration corrected scanning transmission electron microscopy(a/STEM),and electron energy loss spectroscopy(EELS)combined with electrochemical testing on Li[Li1/5Ni1/5Mn3/5]O2compounds during thefirst electro-chemical cycle at different states of charge to identify the structural changes.Our researchfindings provide new insights into the complex intercalation mechanisms and how they influence dynamic structural changes in the Li-excess layered oxide compound,Li[Li1/5Ni1/5Mn3/5]O2,during thefirst electrochemical charge and discharge cycle.

2.EXPERIMENTAL SECTION

Synthesis.A coprecipitation technique was used for the synthesis of the materials which was previously described.11Transition metal nitrates,Ni(NO3)2·6H2O(Fisher)and Mn(NO3)2·4H2O(Fisher), were titrated into a stoichiometric LiOH·H2O(Fisher)solution for a duration of two hours.The coprecipitated transition metal hydroxides were thenfiltered using a vacuumfilter and washed three times with deionized water.The collected transition metal hydroxides were dried in an oven at180°C for10h in air.The dried transition metal precursors were mixed with a stoichiometric amount of LiOH·H2O corresponding to the amount of M(OH)2from the coprecipitation step.This mixture was ground for30min to ensure adequate mixing and then placed into a furnace at480°C for12h.The precalcinated powders were prepared as a pellet for high temperature sintering. These samples were then calcinated at1000°C for12h in air.Samples were brought back to room temperature by furnace cooling. Electrochemistry.Electrochemical properties were measured on an Arbin battery cycler in galvanostatic mode between4.8and2.0V. Cathodes were prepared by mixing the active material Li-[Li1/5Ni1/5Mn3/5]O2with10wt%Super P carbon(TIMCAL)and 10wt%poly(vinylidenefluoride)(PVDF)in N-methylpyrrolidone (NMP)solution.The slurry was cast onto an Al foil using a doctor blade and dried in a vacuum oven overnight at80°C.The electrode discs were punched and dried again at80°C for6h before storing them in an argonfilled glovebox(H2O level<1ppm).2016type coin cells were used to electrochemically cycle Li[Li1/5Ni1/5Mn3/5]O2to different states of charge during thefirst electrochemical cycle seen in Supporting Information Figure S1.The batteries were prepared in the same Argonfilled glovebox using lithium metal ribbon as an anode and a1M LiPF6in a1:1ethylene carbonate/dimethyl carbonate (EC:DMC)electrolyte solution(Novolyte).Celgard model C480 separators(Celgard Inc.,USA)were used as the separator.

The intermittently cycled samples were recovered by disassembling cycled batteries in the same argon-filled glovebox.The cathode was washed by submerging in acetonitrile3times and then allowed to dry in argon atmosphere overnight.For TEM samples,the powders were suspended on a copper grid with lacey carbon.The approximate time of sample exposed to air(from a sealed environment to the microscope column)is less than5s.

Structural Characterization.Synchrotron X-ray Diffraction. Powder X-ray diffraction patterns were taken using synchrotron X-ray diffraction at the Advanced Photon Source(APS)at Argonne National Laboratory(ANL)on beamline11-BM(E=30keV,λ= 0.413225Å).All samples were hermetically sealed in1.0mm Kapton capillary to minimize air exposure.The beamline uses a sagittally focused X-ray beam with a high precision diffractometer circle and perfect Si(111)crystal analyzer detection for high sensitivity and resolution.Instrumental resolution at high Q is better thanΔQ/Q≈2×10−4,with a typical2θstep size of0.001°at30keV.XRD data analysis was carried out by Rietveld refinement methods using FullProf and GSAS/EXPGUI software.28−30Crystal structure refinement parameters included2θ,zero offset,intensity,scale factor,lattice parameters,isotropic atomic positions,atomic displacement parame-ters,and cation site occupancies.The refinements led to superior convergence factors when refined using the Thompson,Cox,and Hastings pseudo-Voigt function31with axial divergence symmetry described by the Finger,Cox,and Jephcoat model.32

a-STEM/EELS.Electron microscopy work was carried out on a Cs-corrected FEI Titan80/300-kV TEM/STEM microscope equipped with a Gatan Image Filter Quantum-865.All STEM images and EELS spectra were acquired at300KV and with a beam size of∼0.7Å.EELS spectra shown in this work were acquired from a square area of∼0.5×0.5nm with an acquisition time of3s and a collection angle of35 mrad.HAADF images were obtained with a convergence angle of30 mrad and a large inner collection angle of65mrad.Images acquired by an HAADF detector with a small convergence angle and a relatively large inner collection angle are also called“Z-contrast”images,where the contrast is proportional to Z1.7.33,34Atomic resolution Z-contrast images can be used to differentiate elements and provide atomic−structural information based on the contrast of the atomic columns. To minimize possible electron beam irradiation effects,EELS and HAADFfigures presented in this work were acquired from areas without prebeam irradiation.Mn L3to L2intensity ratio analysis was done by averaging over8to12spectra using the method described by Wang et al.35

3.RESULTS

Electrochemical measurements of Li[Li1/5Ni1/5Mn3/5]O2elec-trodes were performed in lithium half-cells.Figure S1 (Supporting Information)displays thefirst electrochemical charge/discharge curves between4.8and2.0V(at a rate of10 mA/g)for the Li[Li1/5Ni1/5Mn3/5]O2electrode.The lettered points along the curves indicate states of charge,which were prepared for the various characterization techniques.Support-ing Information Table S2indicates the voltages,capacities and corresponding Li remaining in the material assuming all of the electrochemical capacity corresponds to Li+removal from or insertion into the material.

3.1.Synchrotron X-ray Diffraction.High resolution synchrotron X-ray diffraction was analyzed for detailed crystal structure evolution during thefirst electrochemical cycle for the Li[Ni1/5Li1/5Mn3/5]O2electrode material(Figure1).The detailedfitting plots and refinement results are in Figure S3 and Table1.Table2compares the Rietveld reliability factors for different states of charge for respective models.

The X-ray di ffraction pattern of the pristine material (A in Figure S1)was reproduced from ref 27.26Rietveld re finement of the pristine material indicates the typical well-layered phase with 3%Li/Ni mixing.After charging the material to 4.40V,during the sloping region where Ni 2+is oxidized to Ni 4+during Li extraction,it is clear from XRD that the superlattice peaks remain visible and the transition metal ordering remains intact,as was previously shown from literature.8Rietveld re finement at this position (B in Figure S1)con firms that the material maintains the well-layered phase while the amount of Li/Ni mixing decreases slightly.The a lattice parameter decreases by 0.00Åwhile the c lattice parameter increases by 0.032Åfrom the pristine material (Table 2).This trend is consistent with the oxidation of Ni 2+to Ni 4+,during which the ionic radius decreases from 0.69Åto 0.48Åinducing a contraction in the overall lattice.36The re finement also con firms that the material remains a single layered phase with no obvious phase transformations occurring during the sloping region (compar-ison shown in Table 2).

After the sloping region (4.45V,C in Figure S1),Rietveld re finement supports a single layered phase.The addition of tetrahedral Li +to the single phase model improved the Rietveld reliability factors R b and R wp to 10.73and 4.93.In addition,the amount of Li/Ni mixing continues to decrease while the a and c lattice parameters continue to decrease and increase,respectively.This indicates that slightly after the sloping region,Ni 2+is still being oxidized to Ni 4+,which is also con firmed by X-ray absorption spectroscopy (XAS)results (Figure S4).Rietveld re finement results of the XRD pattern cycled to this point provides the first evidence of tetrahedral Li +formation.This result is consistent with our previous findings that tetrahedral Li +ions are more energetically favorable to form in the middle of the first charge cycle.

26

Figure 1.Synchrotron X-ray di ffraction (XRD)patterns collected from di fferent states of charge during the first electrochemical charge/discharge cycle.

Table 1.Rietveld Re finement Results for

Li[Li 1/5Ni 1/5Mn 3/5]O 2at Di fferent States of Charge during the First Electrochemical Cycling Using One or Two Independent R 3m Phases

Charging

pristine

a =2.8608(2)c =14.2584(1)c /a =4.9z (O)=0.257(2)

N_Ni (in Li layer)=0.033(1)

R wp =11.18;R b =6.35

B

a =2.8519(1)c =14.2902(1)c /a =5.011z (O)=0.259(2)

n_Ni (in Li layer)=0.018(1)

R wp =9.67;R b =4.20

C

a =2.8516(3)c =14.2995(1)c /a =5.014z (O)=0.259(2)

n_Ni (in Li layer)=0.015(1)

n_Li (in tetrahedral site)=0.108(2)

R wp =10.73;R b =4.93

D

Phase 1:Phase 2:a =2.8540(1)a =2.8519(2)c =14.3456(2)c =14.3166(1)c /a =5.026c /a =5.020z (O)=0.259(1)z (O)=0.260n_Ni (in Li layer)=0.041(1)

n_Ni (in Li layer)=0.062(1)Phase Fraction:56%n_Li (in tetrahedral site)=

0.115(1)

R wp =7.61;R b =3.40

E

Phase 1:Phase 2:a =2.8595(2)a =2.8478(2)c =14.3620(1)c =14.3203(1)c /a =5.023c /a =5.028z (O)=0.260(1)z (O)=0.256(2)n_Ni (in Li layer)=0.050(1)

n_Ni (in Li layer)=0.071(1)Phase Fraction:57%n_Li (in tetrahedral site)=

0.099(1)

R wp =8.65;R b =4.19Discharging

G

Phase 1:Phase 2:a =2.8687(2)a =2.8655(1)c =14.3626(2)c =14.33(1)c /a =5.007c /a =5.003z (O)=0.261(1)z (O)=0.258(1)

n_Ni (in Li layer)=0.059(1)n_Ni (in Li layer)=0.070(2)Phase Fraction:53.3%n_Li (in tetrahedral site)=0.087(1)

R wp =7.34;R b =4.03

H

Phase 1:Phase 2:a =2.8676(1)a =2.86(2)c =14.3218(1)c =14.3059(1)c /a =4.994c /a =4.987z (O)=0.258(1)z (O)=0.258(1)

n_Ni (in Li layer)=0.021(1)n_Ni (in Li layer)=0.047(1)Phase Fraction:49.5%n_Li (in tetrahedral site)=0.0874(2)

R wp =7.27;R b =3.51

At the middle of the plateau region (4.60V,D in Figure S1)the re finement improved when two independent phases,R 3m space groups,were implemented.The addition of a second layered phase with tetrahedral Li +occupancy improved both Rietveld reliability factors R wp and R B to 7.61and 3.40,respectively.The use of a second phase increased the flexibility in the re finement to account for peak broadening as well as peak shoulders in the di ffraction pattern (Figure 1b).The re finement results continue to support the evidence of tetrahedral Li +ions forming in the second layered phase during the middle of the first charge cycle.

Lastly,at the end of the first complete charge (4.80V,E in Figure S1),the XRD pattern shows that the superlattice peaks remain visible (Figure 1a)indicating that long-range cation ordering remains present in the transition metal layer,even after complete delithiation.The addition of a second R 3m phase including tetrahedral Li +ion occupancy improves the pattern fitting and reduces the residual errors of the (003),(104),(110),and (108)peaks over a single layered R 3m phase or two independent R 3m layered phases without tetrahedral Li +occupancy (see Tables 1and 2).The amount of Li/Ni interlayer mixing increased 50%to 0.05at the end of the first charge.The increased amount of Li +vacancies increases the likelihood of Ni ion migration.Previous literature has identi fied oxygen loss during the plateau region in the Li-excess series of materials.7,8,14,37To explore the possibility of oxygen loss in the structure,oxygen occupancies were re fined without constraints.The Rietveld re finement results indicate that in the layered phase with tetrahedral Li +ion formation,22%oxygen loss is observed;while the oxygen content of the second layered phase remains constant at 2.0(see Table 3).Neutron di ffraction work is currently in progress to quantify the oxygen vacancy amount more accurately;the result will be reported elsewhere.

The XRD patterns collected during the first electrochemical discharge indicate that the superlattice peaks remain evident until 3.30V (position G)as seen in Figure 1a.The superlattice peaks start to fade during the remainder of the discharge cycle,as seen from the XRD pattern following the first full electrochemical cycle,consistent with previous literature.15,25Rietveld re finement of patterns collected during the discharge using a single layered phase no longer leads to pattern convergence and con firms that the material retains the two independent phases with tetrahedral Li +occupancy remaining constant during discharge.Within the states during discharge,the c lattice parameter begins to contract while the a lattice parameter expands,which corresponds to the reduction of Ni 4+to Ni 2+.

After the first electrochemical cycle,for the original layered phase,the c lattice parameter increases from the pristine value of 14.2584Åto 14.3218Åfollowing discharge to 2.0V (H in

Figure S1).Moreover,the a lattice parameter expands from 2.8608Åto 2.8676Åfollowing complete discharge.The expanded lattice parameters following discharge are consistent with previous research.8

3.2.Strain E ffects During Electrochemical Cycling.Microstrain e ffects were extracted by examining the line broadening observed in the XRD patterns obtained during the first electrochemical cycle.From our SEM images,the average particle size does not change much during the first electrochemical cycle,that is,it remains to be above 100nm.XRD peak broadening contributed from size e ffect can be excluded.Figure 2shows microstrain values obtained at di fferent states of charge during the first electrochemical cycle.Williamson −Hall type microstrain can be explained by nonuniform strain e ffects originating from systematic shifts of atoms from their ideal positions resulting from defects such as point defects,site-disorder,and vacancies as well as plastic deformation (see S5for more details).38−40The plot shows that the microstrain remains constant from the pristine material through the sloping region to

4.45V.At the end of the first charge (E)the microstrain increases by 0.1%,which

doubles

Table 2.Comparison of Rietveld Reliability Factors from Single Layered R 3m Phase,Single Layered R 3m Phase with Tetrahedral Li +Occupancy,Two Independent R 3m Layered Phases with and without Tetrahedral Li +Occupancy

C 10.73 4.9510.70 4.9310.17 5.2110.21 5.21

D 10.73 4.938.78 3.7.68 3.577.61 3.40

E 10.82 5.3710.71 4.829.30 4.828.65 4.19G 9.54 6.449.27 5.437.59 5.007.34 4.03H

8.49

4.74

8.50

4.6

7.51

3.93

7.27

3.51

Table 3.Comparison of Fitting Models for Position E Using a Single Phase Model and Two Independent Layered Phase with and without Tetrahedral Li +Occupancy

position E:single phase position E:two layered phases position E:two layered phases and tetrahedral Li position E:two layered phases and tetrahedral Li with oxygen vacancies Phase 1:Phase 1:Phase 1:a =2.8485a =2.8575a =2.8595a =2.8595c =14.3263c =14.3446c =14.3620c =14.3620c /a =5.029c /a =5.020c /a =5.023c /a =5.023z (O)=0.260z (O)=0.260z (O)=0.260z (O)=0.260

n_Ni (in Li layer)=0.063

n_Ni (in Li layer)=0.079

n_Ni (in Li layer)=0.050n_Ni (in Li layer)=0.050R wp =10.71;R b =4.82

Phase Fraction:57%Phase Fraction:57%n_Oxygen =2.00Phase Fraction:57%Phase 2:Phase 2:Phase 2:a =2.8476a =2.8478a =2.8478c =14.3217c =14.3203c =14.3203c /a =5.029c /a =5.028c /a =5.028z (O)=0.260z (O)=0.259z (O)=0.260

n_Ni (in Li layer)=0.053n_Ni (in Li layer)=0.071n_Ni (in Li layer)=0.075

R wp =9.30;R b =4.82

n_Li (in tetrahedral site)=0.099n_Li (in tetrahedral site)=0.102R wp =8.65;R b =4.19

n_Oxygen =1.56R wp =8.52;R b =3.80

comparing to the pristine value.However,the errors associated with the microstrain values with states of charge at 4.60V,4.80V,and 3.30V have increased signi ficantly.TEM images taken at these states of charge show a high degree of stacking faults and the formation of dislocations within the material.An improve-ment to the fitting can be applied using the modi fied Williamson −Hall plot,40−42which will be explored in future studies.The microstrain increase correlates well with the amount of Li +ion vacancies in the delithiated sample as well as site-disordering from the possibility of tetrahedral Li +site formation.Microstrain e ffects may also originate from cation migration accompanied by oxygen vacancy,Li/Ni site mixing,and a second layered phase formation,which are con firmed from Rietveld re finement of the XRD patterns.The microstrain values continue to increase during the first half of the discharge cycle.At this point,Li +ions are re-entering the structure,and the microstrain decreases as additional Li +are intercalated into

the material at the end of the discharge cycle.We propose that the decrease in microstrain is a result of oxygen ions re-entering the structure,which is evidenced by changes in both intensity and position of our EELS oxygen pre-edge and K-edge data.The microstrain generated during the first electrochemical cycle does not completely recover to the pristine material value,indicating irreversible changes during the first cycle,which may contribute to the irreversible capacity loss.

3.3.TEM/STEM.Figure 2depicts images at three di fferent points along the first electrochemical cycle (C,D,and G points).Multiple grains were selected for study,and the results are consistent,therefore only representative data are shown here.It has been found by previous studies that the pristine material shows well faceted surfaces,a high degree of crystallinity,and well layered properties in the bulk that extend to the surface.5The TEM image following charging to the end of the sloping region at

4.45V (C)illustrates the formation

of

Figure 2.Change in the percent microstrain of Li[Li 1/5Ni 1/5Mn 3/5]O 2plotted corresponding to the state of charge during the first electrochemical cycle.Above are TEM images of the highlighted points illustrating increased strain.Error bars shown correspond to σ

.

Figure 3.High resolution STEM images of (a)pristine and (b)after 10cycles.

nanocracking extending through the first 20nm of the material.The figure shows that the faceted surfaces in the pristine material begin to become less clearly de fined which may be an indication of material loss.Attempts to capture HAADF-STEM images were unsuccessful from the beginning of the voltage plateau region throughout the remainder of the first electro-chemical charge/discharge cycle because the evolution of strains and defects prevented visualization of the atomic columns.Blue-framed low magni fication TEM images taken at the middle of the plateau region (4.60V,D)correspond to a charging capacity of 200mAh/g.The figure shows a signi ficant increase in the formation of nanocracks that span into the bulk of the material.

Upon discharge to 3.30V (G),approximately 0.5Li +ions are reintercalated into the structure.The TEM image in the maroon frame suggests that the amount of stacking faults and defects continue to increase in the material.The electron di ffraction pattern of the particle reveals that there are no distinct spots associated with a well layered structure,but streaking indicates changes in the long-range ordering and layeredness of the material (Figure S6).This streaking is consistent with the large amount of microstrain observed by XRD.Following discharge to 2.00V (H),the nanocracks as well as the defects seen at previous points during the electrochemical cycle are not commonly observed anymore in most of the particles,which enables the HAADF-STEM imaging of the material.The image shows that the bulk of the material still maintains the well-layered structure;however,the contrast within the first 2nm of the surface changes.Along the (001)direction,the dark columns become much brighter on the surface.43The contrast matches that of the neighboring TM columns.This suggests that,following the first electro-chemical cycle,a second phase has formed,consistent with previous findings.26Figure 3compares the HAADF-STEM images from the pristine and after 10electrochemical cycles recorded along the [110]zone axis.The stacking sequence of the layers in the pristine material completely changed after 10electrochemical cycles.In the pristine material,there is evidence

of stacking faults;however,following electrochemical cycling,the material adopts a more uniform structure.This may suggest that the oxygen framework tends to adopt a di fferent stacking after the oxygen vacancy formation and cation migration,and such stacking changes may be responsible for the disappearance of the superstructure peaks in the XRD pattern.

3.4.EELS.Figure 4a compares the EELS spectra of the oxygen K-edge and manganese L-edges from the bulk of the structure taken at di fferent states during electrochemical cycling.The structural evolution during the first electro-chemical cycle can be interpreted using changes in the onset energy and the fine structures in the spectra.The intensities of all the spectra are normalized to the highest intensity peak.The onset energy of O K-edge prepeak is aligned to 532eV.Therefore,our analysis of the O K-edge is limited to the fine structures and not the chemical shift of O K-edge.The L 3and L 2of transition metals are due to the transition from 2p 3/2to 3d 3/2and 3d 5/2and from 2p 1/2to 3d 3/2,respectively.Their intensities are correlated to the unoccupied bands in 3d orbitals.Previous studies have shown that the L 3/L 2ratio is sensitive to the valence state of Mn.44,45

Examining the Mn L-edge onset energy during the sloping region of the first charging cycle shows that the peak shifts to a lower energy loss beginning at position C (Figure 4b)which indicates a lower oxidation state.Analysis of the L 3/L 2edge ratio (shown in Figure 4d)further supports the change in the Mn oxidation state to a value at position C.During the discharge,the Mn valence state returns back to 4+following the first electrochemical cycle.The onset energy for the Mn L 3peak remaining shifted to lower energies following discharge may signify a changing local atomic arrangement.

Figure 4c shows the oxygen K-edge spectra.The splitting into two peaks for the O K-edge is a characteristic of the layered material.In the layered material,oxygen forms an O3framework and the TM resides in octahedral sites.The crystal field of the TM splits into three t 2g orbitals at a lower energy and two e g orbitals at a higher energy level.The K-edge of oxygen is the consequence of the transition of 1s electrons

to

Figure 4.Representative EELS spectra of Mn L-edge and O K-edge from the bulk of the Li[Ni 1/5Li 1/5Mn 3/5]O 2during the first electrochemical charging and discharging cycles.(a)Overall EELS spectrum;(b)Mn L-edge;(c)O K-edge;(d)Mn L 3/L 2ratio;(e)O K-edge comparison of pristine and after one cycle.

the unoccupied2p orbitals,which hybridized with the TM3d orbitals.The splitting of the O K-edge corresponds to the splitting of the TM3d orbitals.

In positions C and D,the ratio of thefirst peak to the second peak increases,which may result from a larger amount of unoccupied t2g orbitals and a change in local environment of oxygen,such as bond length,oxygen vacancy formation,etc. The increasing ratio can be from the oxidation of TM,e.g.,Ni2+ to Ni4+.The energy difference in the two peaks can also be an indication of the oxygen local environment change.From position C through the end of thefirst cycle,the difference in the two oxygen peaks is the same,which is larger than positions A and B.This may be a result from oxygen vacancy formation as well as stronger bonding between oxygen and TM,which is not reversible at the end of thefirst cycle.These results show that,in the bulk structure beginning at the voltage plateau region,the local environments of Ni4+,Mn4+,and oxygen are simultaneously changing and may all participate in charge compensation during the voltage plateau region,some of which is not reversible.

4.DISCUSSION

4.1.Source of Anomalous Capacity−Oxygen Activa-tion.Previous studies of the Li-excess series of layered materials have identified the oxygen activation mechanism as the source for the anomalous capacity.7,8,14,46The present research uses the combined results obtained from SXRD,TEM, and EELS at various states of charge to further explain the source of anomalous capacity identified within this series of materials.Rietveld refinement of SXRD patterns following the first charging cycle(E in Figure1C)identified the loss of approximately20%oxygen from the bulk structure within one of the two layered phases refined.The amount of oxygen loss to accommodate the removal of the additional0.6mol Li+ions corresponds to0.3mol(15%),assuming the contribution of2 electrons.Recent reports have used refinement to identify the loss of structural oxygen;however,only less than half of the theoretically proposed oxygen loss was found.7A possible explanation for the smaller than expected oxygen loss is that the transition metal ions are also contributing to the anomalous capacity.Analysis of the Mn oxidation state from the EELS reveals a change from4+to nearly3+then returns to4+during the discharge cycle.It is thefirst identification of the changing Mn oxidation state within this series of materials during thefirst electrochemical cycle.The changing Mn oxidation state leads to a shift in the oxygen octahedral positions,which can cause local distortions in the lattice causing increased peak broadening as evidenced in the elevated microstrain values from the Williamson−Hall plots at position E.Mn3+may also go through disproportionation reaction,which may contribute to the irreversible capacity during thefirst cycle.Further evidence of changing oxygen local environment is evidenced through our EELS data.

4.2.Structural Changes Caused by Oxygen Vacancies. During the initial electrochemical charging before the voltage plateau region,structural changes observed through Rietveld refinement of XRD patterns and the TEM images are minimal. The amount of Li/Ni interlayer mixing decreases,indicating that the layered properties of the material are improving. Rietveld refinement indicates that neither the tetrahedral Li+ ions nor the second layered phase forms within this region. When the electrochemical charging cycle reaches the voltage plateau,at position D,there is some evidence of a second phase along with tetrahedral Li+ion formation.EELS spectra indicate a possibility of oxygen vacancy formation at the beginning of the plateau region,and Rietveld refinement confirms the oxygen loss following the plateau region.We propose that these oxygen vacancies and the distortion in the oxygen octahedral create stacking faults within the lattice seen in the TEM micrographs(Figure2).It has been previously reported that ordered and disordered vacancies,such as vacant Li+and O2−sites,as well as the creation of stacking faults are driven by transition metal or lithium layer gliding to reduce the structural free energy.47,48The layer gliding provides a mechanism for the formation of domain structure and second phase formation, which result in different stacking sequences of the crystal structure(Figure3).This second phase is created during the plateau region,and Rietveld refinement confirms the presence throughout the remainder of thefirst electrochemical cycle. The TEM micrographs show that at the beginning of the discharge,a large amount of stacking faults are evident.

With the introduction of the stacking faults and the corresponding increase in microstrain within the material,the cations within Li[Li1/5Ni1/5Mn3/5]O2migrate and undergo significant electronic state changes.At the beginning of the plateau region the Ni and Mn ions are completely oxidized to the4+states and the amount of Li/Ni interlayer mixing remains below0.02.Within the plateau region,thefirst evidence of tetrahedral Li+is identified,indicating favorable cation migration environments.The vacancies created in the oxygen layers not only induce the stacking faults but also enable Li/Ni cation migration by hopping through nearby vacancies. During states of charge within the voltage plateau,the Li/Ni interlayer mixing increases to a maximum mixing of0.071at the end of thefirst charge cycle and remains elevated at the initial stages of the discharge cycle.Furthermore,electronic state changes occur in the Mn ions.The EELS spectra reveal a change in the energy loss to a lower position and detailed analysis of the L3/L2peak intensity ratio shows evidence of Mn3+formation.Both of these changes show that not only does the oxygen participate in the electrochemical compensation during the voltage plateau,but that TM also contributes.The stacking faults generated during electrochemical cycling contributed to the microstrain seen during thefirst electro-chemical cycle.These factors prevented high resolution structural imaging at the high state of charge(point E); however,following discharging to 2.0V the structure has reversibly reconfigured to a model closer to the pristine material where the Mn oxidation in the bulk is almost fully back to4+and the Ni ion has reduced to2+.

It is clear that permanent structural rearrangements have occurred.There is permanent lattice expansion in both the a and c lattice parameters as well as the formation of tetrahedral Li+ions and a second layered phase,as well as the change in the layer stacking in the original layered phase.These permanent structural modifications become more evident and the phase percentages increase as the material continues electrochemically cycling as we have shown previously.26Moreover,following the first electrochemical cycle,the STEM image clearly shows the well-layered structure in the bulk;however,the surface structure of the cycled grain has changed to a defect spinel phase.Identifying how this material continues to evolve during long-term cycling is a critical piece necessary to identify the source of capacity fading as well as the depression in discharge voltage in this class of compounds.

5.CONCLUSION

Based on the experimental results presented,we have shown that the mechanisms providing charge compensation during the first electrochemical cycle of the Li-excess layered oxide materials are complex.We have presented direct evidence to describe the lithium deintercalation mechanisms at different states of charge.Oxygen vacancies forming and possible oxygen activation within the bulk structure are identified through EELS and Rietveld refinement of SXRD.The evidence indicates that the oxygen activation may be responsible for the formation of stacking fault defects as well as for facilitating cation migration, including Ni migration to the Li layer.These defects created within the structure lead to the increase of significant microstrains observed within the bulk structure.The micro-strain continually increases during the voltage plateau and into the beginning of the discharge cycle where the largest amount of cation mixing,oxygen vacancies,and lattice parameter expansion are observed.

Our results presented here suggest that a combination of these mechanisms during thefirst electrochemical cycle, particularly in the voltage plateau region,impede lithium diffusion,which may contribute to the low intrinsic rate capabilities and large irreversible capacity losses of this material. Identification of these mechanisms forming during thefirst electrochemical cycle raises questions about how surface coatings improve irreversible capacity fading and whether they remove or change these dynamic processes during thefirst electrochemical cycle.The present research also raises questions about the continually changing nature of these electrodes upon cycling.Only through continued research using sophisticated instrumentations such as atomic resolution STEM and EELS as well as high resolution synchrotron X-ray diffraction and neutron diffraction will the complex nature of the interactions in the Li-excess layered transition metal oxides

be uncovered.

■ASSOCIATED CONTENT

*Supporting Information

Electrochemical voltage profile offirst cycle(S1).State of charge,voltages,and capacities summary of different points (S2).Profilefits for Rietveld refinement of different points (S3).X-ray absorption spectroscopy(S4).The Williamson-Hall method(S5).Electron diffraction pattern(S6).This material is

available free of charge via the Internet at http://pubs.acs.org ■AUTHOR INFORMATION

Corresponding Author

*E-mail:shirleymeng@ucsd.edu.

Present Address

∥Global Technology&Innovation,Power Solutions,Johnson Controls,Milwaukee,WI53209,USA.

Notes

The authors declare no competingfinancial interest.■ACKNOWLEDGMENTS

UCSD work was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy,Office of Vehicle Technologies of the U.S.Department of Energy under Contract No.DE-AC02-05CH11231,Subcontract No. 70512under the Batteries for Advanced Transportation Technologies(BATT)Program. C.R.F.acknowledges the financial support from Florida Energy System Consortium through University of Florida under Award No.80859.D.Q. would like to acknowledge Dr.Huolin Xin for providing the scripts for L3/L2analysis.The synchrotron X-ray diffraction patterns were collected at Argonne National Laboratory on beamline11-BM through the general user proposal mail-in program(GUP-13210).a-S/TEM and EELS analysis is carried out at the ORNL Shared Research Equipment(SHaRE)User Facility,which is sponsored by the Office of Basic Energy Sciences,U.S.Department of Energy.The authors acknowledge Dr.Gabriel Veith and Dr.Nancy Dudney for assistance and Dr.

Bo Xu and Dr.M.Yang for their valuable discussions.■REFERENCES

(1)http://www.mineralprices.com/(2012).

(2)Boulineau,A.;Simonin,L.;Colin,J.-F.;Cane v et,E.;Daniel,L.; Patoux,S.Chem.Mater.2012,24,3558.

(3)Thackeray,M.M.;Kang,S.-H.;Johnson,C.S.;Vaughey,J.T.; Benedek,R.;Hackney,S.A.J.Mater.Chem.2007,17,3112.

(4)Meng,Y.S.;Ceder,G.;Grey,C.P.;Yoon,W.S.;Jiang,M.; Breger,J.;Shao-Horn,Y.Chem.Mater.2005,17,2386.

(5)Lei,C.H.;Bareno,J.;Wen,J.G.;Petrov,I.;Kang,S.H.; Abraham,D.P.J.Power Sources2008,178,422.

(6)Yoon,W.S.;Paik,Y.;Yang,X.Q.;Balasubramanian,M.; McBreen,J.;Grey,C.P.Electrochem.Solid-State Lett.2002,5,A263.

(7)Yabuuchi,N.;Yoshii,K.;Myung,S.-T.;Nakai,I.;Komaba,S.J. Am.Chem.Soc.2011,133,4404.

(8)Lu,Z.;Dahn,J.R.J.Electrochem.Soc.2002,149,A815.

(9)Robertson,A.D.;Bruce,P.G.Electrochem.Solid-State Lett.2004, 7,A294.

(10)Park,Y.J.;Hong,Y.-S.;Wu,X.;Ryu,K.S.;Chang,S.H.J.Power Sources2004,129,288.

(11)Fell,C.R.;Carroll,K.J.;Chi,M.;Meng,Y.S.J.Electrochem.Soc. 2010,157,A1202.

(12)Li,J.;Klo p sch,R.;Stan,M.C.;Nowak,S.;Kunze,M.;Winter, M.;Passerini,S.J.Power Sources2011,196,4821.

(13)Kang,S.H.;Thackeray,M.M.J.Electrochem.Soc.2008,155, A269.

(14)Armstrong,A.R.;Holzapfel,M.;Novak,P.;Johnson,C.S.; Kang,S.H.;Thackeray,M.M.;Bruce,P.G.J.Am.Chem.Soc.2006, 128,8694.

(15)Tran,N.;Croguennec,L.;Menetrier,M.;Weill,F.;Biensan,P.; Jordy,C.;Delmas,C.Chem.Mater.2008,20,4815.

(16)De Souza,R.A.;Ramadan,A.;Horner,S.Energy Environ.Sci. 2012,5,5445.

(17)Kushima,A.;Yildiz,B.J.Mater.Chem.2010,20,4809.

(18)Araki,W.;Imai,Y.;Adachi,T.J.Eur.Ceram.Soc.2009,29,2275.

(19)Kilner,J.A.Nat.Mater.2008,7,838.

(20)Mavrikakis,M.;Hammer,B.;Nørskov,J.K.Phys.Rev.Lett. 1998,81,2819.

(21)Maier,J.Nat.Mater.2005,4,805.

(22)Blamire,M.G.;MacManus-Driscoll,J.L.;Mathur,N.D.; Barber,Z.H.Adv.Mater.2009,21,3827.

(23)Schichtel,N.;Korte,C.;Hesse,D.;Janek,J.Phys.Chem.Chem. Phys.2009,11,3043.

(24)Garcia-Barriocanal,J.;Rivera-Calzada,A.;Varela,M.;Sefrioui, Z.;Iborra,E.;Leon,C.;Pennycook,S.J.;Santamaria,J.Science2008, 321,676.

(25)Jiang,M.;Key,B.;Meng,Y.S.;Grey,C.P.Chem.Mater.2009, 21,2733.

(26)Xu,B.;Fell,C.R.;Chi,M.;Meng,Y.S.Energy Environ.Sci. 2011,4,2223.

(27)Ito,A.;Shoda,K.;Sato,Y.;Hatano,M.;Horie,H.;Ohsawa,Y.J. Power Sources2011,196,4785.

(28)Rodríguez-Carvajal,J.In Satellite Meeting on Powder Diffraction of the XV Congress of the IUCr Toulouse,France,1990;p127. (29)Larson,A.C.;Dreele,R.B.V.General Structure Analysis System (GSAS);Los Alamos National Laboratory Report(LAUR)86-748; Los Alamos National Laboratory:2004.

(30)Toby,B.J.Appl.Crystallogr.2001,34,210.

(31)Thompson,P.;Cox,D.E.;Hastings,J.B.J.Appl.Crystallogr. 1987,20,79.

(32)Finger,L.W.;Cox,D.E.;Jephcoat,A.P.J.Appl.Crystallogr. 1994,27,2.

(33)Pennycook,S.J.Ultramicroscopy19,30,58.

(34)James,E.M.;Browning,N.D.Ultramicroscopy1999,78,125.

(35)Wang,Z.L.;Yin,J.S.;Jiang,Y.D.Micron2000,31,571.

(36)Shannon,R.D.;Prewitt,C.T.Acta Crystallogr.1969,B25,925.

(37)Wu,Y.;Murugan,A.V.;Manthiram,A.J.Electrochem.Soc.2008, 155,A635.

(38)Stephens,P.J.Appl.Crystallogr.1999,32,281.

(39)Williamson,G.K.;Hall,W.H.Acta Metallurg.1953,1,22.

(40)Groma,I.;Ungar,T.;Wilkens,M.J.Appl.Crystallogr.1988,21, 47.

(41)Ungar,T.;Borbely,A.Appl.Phys.Lett.1996,69,3173.

(42)Krivoglaz,M.A.;Ryboshapka,K.P.Phys.Met.Metallogr.1963, 15,18.

(43)Carroll,K.J.;Qian,D.;Fell,C.R.;Veith,G.M.;Chi,M.; Dudney,N.J.;Meng,Y.S.Energy Environ.Sci.2012,submitted.

(44)Kurata,H.;Colliex,C.Phys.Rev.B1993,48,2102.

(45)Riedl,T.;Gemming,T.;Wetzig,K.Ultramicroscopy2006,106, 284.

(46)Yu, D.Y.W.;Yanagida,K.;Kato,Y.;Nakamura,H.J. Electrochem.Soc.2009,156,A417.

(47)Paulsen,J.M.;Thomas,C.L.;Dahn,J.R.J.Electrochem.Soc. 1999,146,3560.

(48)Robertson,A.D.;Armstrong,A.R.;Bruce,P.G.Chem.Mater. 2001,13,2380.