Oxygen redox reactions in Li ion battery electrodes studied by resonant inelastic X-ray scattering

Accepted Manuscript Title: Oxygen redox reactions in Li ion battery electrodes studied by resonant inelastic x-ray scattering Author: L.-C. Duda K. Edstr¨om PII: DOI: Reference:

S0368-2048(16)30154-2 http://dx.doi.org/doi:10.1016/j.elspec.2017.06.003 ELSPEC 46682

To appear in:

Journal of Electron Spectroscopy and Related Phenomena

Received date: Revised date: Accepted date:

21-10-2016 10-5-2017 19-6-2017

Please cite this article as: L.-C. Duda, K. Edstrddotom, Oxygen redox reactions in Li ion battery electrodes studied by resonant inelastic x-ray scattering, (2017), http://dx.doi.org/10.1016/j.elspec.2017.06.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

L.-C. Dudaa , K. Edstr¨omb a Department

ip t

Oxygen redox reactions in Li ion battery electrodes studied by resonant inelastic x-ray scattering

us

cr

of Physics and Astronomy, Division of Molecular and Condensed Matter Physics, Uppsala University, Box 516, S-751 20 Uppsala, Sweden b Department of Chemistry - ˚ Angstr¨ om Laboratory, Uppsala University, Box 538, S-751 21 Uppsala, Sweden

an

Abstract

We present results using inelastic scattering x-ray spectroscopy (RIXS) combined with x-ray absorption spectroscopy on Li ion battery cathode and anode

M

materials, respectively. In particular, we discuss results obtained on the cathode materials Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 and Lix [Ni0.65 Co0.25 Mn0.1 ]O2 as well as in the composite anode material Ni0.5 TiOPO4 /C . We show that oxygen re-

d

dox reactions are an important aspect of many such systems and how one can

te

succesfully address them using RIXS. New insights on the formation of new oxygen species and on the details of cycling-induced structural disorder can be detected. We foresee a particular future focus on these issues considering

Ac ce p

the rapid development of new in operando RIXS techniques for Li ion battery research.

Key words: Li ion battery, anionic redox, resonant inelastic x-ray scattering

1. Introduction

Li ion batteries are found in a variety of applications ranging from large scale

grids, to energy sources for electrically powered vehicles, to consumer electronics [1, 2]. There is a strong desire to keep improving battery performance, in

5

particular to increase their capacity and the power density. Understanding the redox processes in Li ion batteries on an atomic level is crucial for the development of materials for the various battery components and for finding

Preprint submitted to Journal of LATEX Templates

May 10, 2017

Page 1 of 27

ways of improving the efficiency and energy storage capabilities of these devices [3, 4, 5]. Cathode materials typically consist of lithium 3d-transition-metal oxides,

ip t

10

i.e. LiTmO (Tm, transition metal), that operate by the deintercalation of Li+

cr

on charge and its reinsertion on discharge [6, 7]. The cathode material most commonly used in lithium ion batteries is LiCoO2 [8]. Other single-Tm oxides

15

us

such as LiNiO2 or LiMn2 O4 have been examined but compositions containing a combination of Ni, Co and Mn, i.e. Li(Ni,Mn,Co)O2 (Li-NMC) materials, can balance the drawbacks of the former ones, such as Ni-accumulation in the

an

Li-planes or phase changes [9, 10, 11] that limit capacity retention. Li(Ni1/3 Mn1/3 Co1/3 )O2 is an such example that has good combination of high capacity [12], good rate capability [13, 14] and the possibility to operate at high voltages. The redox capacity of the transition-metal ions presents a substantial bot-

M

20

tleneck for Li ion batteries by setting the ultimate limit to their entire capacity. Therefore, the improvement of cathodes and the understanding of the processes

d

involved on this side of Li ion batteries has received a lot of attention [17].

25

te

Specific capacities of 190 mAhg−1 have been demonstrated for Li[NiCoAl]O2 [15, 16]. Recently, Li-rich materials, such as Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 ex-

Ac ce p

hibit an even higher capacity of ∼230-250 mAhg−1 . However, this cathode material has been suspected to be prone to oxygen loss from its crystal lattice or to electrolyte decomposition, both leading to gas evolution such as O2 [20, 22] and/or CO2 [21, 22, 23, 24]. On the other hand, others have indicated that its

30

redox activity may not only be limited to the transition metals but may also be occuring at oxygen sited [18, 19]. Gaining extra capacity by oxygen redox reactions is an intriguing possibility that a couple of recent studies have addressed [25, 26]. Evidence from x-ray spectroscopy shows that in fact localized electron holes on the oxygen sites, O− , coordinated by Mn4+ and Li+ ions are formed

35

reversibly instead of the formation of true O2− peroxide with O–O bonding 2 lengths of ∼1.45 ˚ A. In this work, we will review how the results from Li rich Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 [25] compare to an Li-NMC material with high Ni content [27, 28], giving special attention to their oxygen redox reactions. 2

Page 2 of 27

On the anode side of Li ion batteries, graphite and other carbon-based com40

pounds are commonly used in commercial applications. However, alternative

ip t

anode materials are being seriously considered due to safety concerns around

graphite anodes when operating above the reductive decomposition of conven-

cr

tional electrolytes. Anodes based on titanium compounds show much promise

but are often limited by the amount of charge that can be reversibly stored in them. On the other hand, nickel titanium oxyphosphates could present an

us

45

alternative since it seems to access more of the redox-capable species in the material at potentials above the formation of the Solid Electrolyte Interphase (SEI)

an

lending them a potential to be used in Li ion battery technology. They have a good and stable specific capacity (270 mAhg−1 for Ni0.5 TiOPO4 cycled be50

tween 0.5 and 3.0 V vs Li+ /Li), a high rate performance, and the stability of the

M

covalently bonded PO4 group lessens safety concerns. However, low electronic conductivity as well as capacity loss and ageing issues are of concern and limit its industrial application. A significant increase of their electronic conductivity has

On the other hand, an extra irreversible capacity during the first lithiation pro-

te

55

d

been recently achieved by coating particles of this material with carbon [29, 30].

cess has puzzled and intrigued the scientific community [31, 29, 30, 32, 33].

Ac ce p

Spectroscopic studies have been recently undertaken to illuminate the involved processes [34] and it has been pointed out that solid electrolyte interphase formation may contribute to the problems with this material. In the second part of

60

the following, we will present new soft x-ray spectroscopy results that addresses this question.

Resonant inelastic x-ray scattering (RIXS) has emerged as a powerful tool

to study redox processes in Li ion batteries [25, 26, 35, 36, 37]. RIXS can address element and site specific information of the occupied electronic states

65

thanks to resonant excitation at different absorption resonances. One of its most attractive aspects for in Li ion batteries is precisely its site selectivity because it allows one to separate contributions from the various atomic species in these complex systems. Here we focus on the soft x-ray regime that encompasses oxygen K-edge and the 3d-transition metals L-edges, i.e. 500-900 eV. 3

Page 3 of 27

70

RIXS is normally combined with x-ray absorption spectroscopy (XAS). XAS is a complementary technique, it can be used to study the electronic structure

ip t

changes of unoccupied states in the electrode materials in lithiation/delithiation.

Similarly to RIXS, XAS spectra reflect the structure and occupancy of the cor-

75

cr

responding unoccupied electronic states of the battery in an element and orbital

specific manner. Moreover, the area under a normalized spectrum can be con-

us

sidered to correspond to the number of electron holes in a specific orbital. It is well known that the spectral shape at the transition metal L-edge resonances can often be used as a fingerprint for different oxidation states of the metal

80

an

ions. The oxygen 2p orbitals bind and hybridize to all metal atomic species at different binding energies and thus the O K-XAS shows a number of separate peaks in complex transition metal oxide compounds. Thus, if a rigid band model

M

applies, the degree of lithiation in a specific orbital is reflected by the intensity of a certain XAS peak. XAS also offers depth sensitive information by using different detection modes. The total electron yield (TEY) mode is surface sensitive whereas the total florescence yield (TFY) mode is bulk sensitive. Spectral

d

85

te

differences between the two modes can help identify lithiation gradients as well as effects of the SEI formation, which is expected to occur in layers close to the

Ac ce p

surface.

2. Cathodes based on Ni-Co-Mn-oxides

90

2.1. Experimental Details

X-ray absorption spectroscopy (XAS) and resonant inelastic x-ray scattering

(RIXS) at the O K edge has been performed at beamline U41-PGM at BESSY II (Berlin) [38] and at ALS Beamline 8.0.1.4 (wet RIXS). The XAS spectra at the L-edges of Ti, Ni and P have been recorded at X1B at NSLS (Brookhaven)

95

NSLS [39]. For x-ray photoelectron spectroscopy (XPS) air exposure [40] must be strictly avoided because the outermost layers of cycled electrodes are almost immediately affected whereas XAS and RIXS are more bulk sensitive. Nevertheless,

4

Page 4 of 27

since the samples have some sensitivity to air exposure, a vacuum suitcase sys100

tem was available at Beamline 8.0.1.4 (ALS), a glove bag filled with Ar was

ip t

used for sample transfer at at X1B (NSLS) and at U41-PGM at (BESSY II) the sample transfer was performed quickly (∼1 min.) to the load lock chamber

cr

before immediately commencing with the pump down.

The cathode material Lix [Ni0.65 Co0.25 Mn0.1 ]O2 was synthesized by using a combustion method described by Saadoune et al. [41]. Details of the sample

us

105

preparation are described in [28].

an

2.2. O K XAS and RIXS Results

Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 is regarded as the archetypal example of a lithium-ion-battery cathode to exhibit extra capacity. The dominant mechanism of charge compensation has been shown [25] to involve the generation of localized

M

110

electron holes on the oxygens coordinated by Mn4+ and Li+ ions (the relatively ionic O–(Mn/Li) interactions serving to promote localization).

d

In the electrochemical charge curve there is a long, almost flat plateau (of

115

te

nearly constant voltage) that extends over most of the charge process. We therefore have chosen to investigate cathodes for potentials at the beginning of the plateau (BOP), the end of the plateau (EOP) and the end of the charge

Ac ce p

process (EOC), respectively. The cycling evolution of the O K XAS spectra are shown in Fig. 1. The black traces is the spectrum of the pristine material and the colored traces are spectra from cathodes left at various end potentials

120

in the charge process. A pronounced broadening of the features of the two resonances (528-532 eV) is observed. In this energy region, the oxygen states are heavily hybridized with the transition metals, where the bonding to Mn gives the dominating contribution due to its high content in the material. We find that the XAS spectra broaden and gain intensity along the charge curve, creating (at

125

least) two distinct new features, as seen in the difference spectra (gray traces). These have been described as additional hole states in the oxygen. The low energy feature that dominates in the beginning of the charge process is ascribed to be due to the Ni redox process and can be observed due to hybridization 5

Page 5 of 27

with oxygen states. Similarly, the second feature in the difference spectrum 130

corresponds to localized oxygen states that are formed on oxygen sites close to

ip t

Mn and Li ions in the crystal structure [25].

Presently, LiCoO2 is among the most common if not the most common

cr

cathode material in commercial batteries. For LiCoO2 the main drawbacks are

the high cost of cobalt, its toxicity, and itsminstability at high potentials. Pure LiNiO2 has also been considered, but here the serious drawbacks include safety

us

135

issues and the difficulty of synthesizing without unwanted extra nickel ending up in the lithium plane in the structure, inhibiting lithium diffusion. Therefore

an

attention has been directed toward LiNi1−y Coy O2 , in which cobalt supports the synthesis of the material. Moreover, with increased Co content in the solid 140

solution, the structure shows more of a two-dimensional character, which is

M

desirable. Additionally, the substitution of Mn for Ni and Co in LiNi1−y Coy O2 improves the electrical conductivity of the material and the stability of the structure.

plotted the spectra in a similar way but without an additional offset. The colors

te

145

d

The O K-spectra for Lix [Ni0.65 Co0.25 Mn0.1 ]O2 [28] are shown in Fig. 2. We

correspond to different lithiation levels x in the material, where higher charge

Ac ce p

means lower x. While the overall spectra structure and the charge behavior is similar to the Li rich material, here we see less of a broadening. Instead, there is a marked increase in intensity (number of hole states) under the peaks when

150

the lithium contents is below x∼0.75. Concurrently, there is a high energy state (∼526.5 eV) that evolves in the RIXS spectrum (excited at the main peak) as charge is extracted. Clearly, also cathode materials with high Ni content exhibit oxygen redox processes. In contrast to the Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 ,

a new state is found to form in the occupied O 2p band. With increasing

155

charge extraction, the number of states gradually increase, as well as the energy separation from the symmetric peak observed in the pristine material. In this section we have elucidated the role of oxygen in the charge compensation process of some important cathode materials. This has been sparked by novel discoveries that Li-rich NMC-materials appear to be prone to develop 6

Page 6 of 27

160

localized holes at certain oxygen sites. Here we have shown that oxygen redox processes are a more general phenomenon in lithiated NMC-oxides in the bat-

ip t

tery charge process whereas this can have different effects on the main oxygen band. We find that Li rich materials show a general broadening in the part of

165

cr

the O 2p band that hybridizes with Mn. On the other hand, for an ordinary

lithium NMC-oxide with high Ni content, additional states deeper in the O 2p

us

band are seen to be readily created. Future higher resolution x-ray spectroscopy, especially RIXS, will therefore is a promising path to gain insight to more details

an

of these intriguing processes.

3. Anodes based on titanium oxyphosphates 3.1. Material synthesis and sample preparation

M

170

Lix Ni0.5 TiOPO4 /C composite samples were prepared for analysis with soft x-ray spectroscopy [42] by assembling a batch of eight identical batteries and cy-

d

cle them as shown in Fig. 3. Thus the electrodes from each sample are retrieved by disassembling the batteries at different end potentials before mounting on a sample holder, packaging in an inert atmosphere, and transportation to the

te

175

synchrotron radiation facilities.

Ac ce p

The anode material Ni0.5 TiOPO4 /C preparation is described in [33]. The

sample synthesis, battery assembly and cycling was performed at the department of chemistry at the ˚ Angstr¨om Laboratory. For assembling batteries, the

180

electrodes were prepared by coating an aluminum foil current collector with a slurry composed of Ni0.5 TiOPO4 /C composite with conductive carbon black and poly(vinylidene fluoride) (PVdF) binder in N-methyl-2-pyrrolidone. The composition Ni0.5 TiOPO4 /C : PVdF: carbon black of the electrodes

was 75:10:15 (weight %). Cell assembly was performed in an argon dry box

185

with metallic lithium sheet as a counter electrode, the electrolyte consisted of 1 M LiPF6 in a nonaqueous solution of ethylene carbonate (EC) and dimethyl carbonate (DEC) with a volume ratio of 2:1. Charge-discharge measurements were performed at room temperature with a Digatron BTS 600 battery test

7

Page 7 of 27

system (Fig. 3). The cells were cycled between 0.5 and 3.0 V vs Li/Li+ at C/10 190

rate.

by XAS and RIXS are listed in Table 1. active mass

Current

Voltage

Capacity

[mg]

[mA]

[V]

[mAh] -

cr

Li content

x=0

-

-

-

B1

x=0.5

5.74

0.081831

1.290

0.41

B2

x=1

6.46

0.092086

1.189

0.92

B3

x=2

6.69

0.095398

1.053

1.90

B4

x=2.5

9.20

0.131079

0.829

3.275

B5

EoD

6.73

0.095932

0.50

2.853

B6

x=1

6.53

0.093048

1.955

0.930

B7

EoC

6.81

0.097107

3.00

1.766

an

us

B0

M

Cells

ip t

The Li content x in the Lix Ni0.5 TiOPO4 electrodes that were investigated

d

Table 1: Parameters for the Ni0.5 TiOPO4 /C battery cells that have been cycled to a

certain potential (i.e. degree of lithiation) for measuring with x-ray spectroscopy. EoD

te

stands for “End of Discharge” and EoC stands for “End of Charge”, designating the

Ac ce p

end points of the respective cycling branches.

3.2. XRD characterization of Ni0.5 TiOPO4 /C In situ XRD investigations were performed on the Ni0.5 TiOPO4 /C compos-

195

ite [43]. Pristine Ni0.5 TiOPO4 crystallizes as a monoclinic system with a P21 /c space group, shown in Fig. 4, and that the coating layer of C had an amorphous structure. During the first discharge, when Li is added to the Ni0.5 TiOPO4 /C

, in situ XRD spectra were collected, Fig. 5A at degrees of lithiation given in Fig. 5B.

200

The XRD pattern groups into three regions, as follows. D0 to D4 : little change is detected. D5 to D13 : a new peak appears at 2θ∼15.8◦ (see arrow between D5 and D7) and the most intense peaks decline. In D15 to D23 : all peaks slowly broaden and become less intense. Thus, in the initial lithiation of 8

Page 8 of 27

the material, we observe a loss of long range crystalline order in Ni0.5 TiOPO4 that could be due to amorphization.

ip t

205

3.3. RIXS and XAS results

cr

In order to learn more about the processes that govern the amorphization that occurs during the charge cycle and the redox processes that take place

during lithiation and delithiation of Ni0.5 TiOPO4 /C we turn to soft x-ray spectroscopy.

us

210

Figure 6 shows O K-edge absorption data of Lix Ni0.25 TiOPO4 . In the bulk

an

sensitive TFY-mode XAS (right panel) we observe four distinct peaks, marked A–D, which are stationary in energy position and vary in intensity depending on lithiation degree. These spectra resemble those of the Ni0.25 TiOPO4 /C compound with a lower Ni-concentration [39]. The TEY-mode spectrum of the

M

215

pristine sample (black trace in left panel) shows similar structures to those of Ni0.25 TiOPO4 /C whereas the spectral appearance changes dramatically with

d

cycling (B1-B7). In particular, peaks A-C lose most of their intensity early in

220

te

the cycle (due to filling of the states). However, note the transient recurrence (x=1 and x=2) of intensity close to peak C and the development of a shoulder (marked S) on the low energy slope of peak D that is more pronounced than in

Ac ce p

the TFY mode.

The observed differences in spectral shape between TFY- and TEY-mode

suggest that there is a significant depth dependent of the electronic structure.

225

True for both modes is that the electronic structure shows irreversible changes. This is clearly illustrated by the superposition of spectra from the pristine sample and the spectrum after a full cycle when the battery is recharged (dashed grey trace). Also the comparison between the latter and the spectrum of the fully discharged battery (solid grey trace) is interesting as it illustrates that very

230

little change occurs in the recharging half-cycle. To faciliate a detailed view of the evolution at different atomic sites we have integrated the intensities under the four peaks A-D of the TFY and the TEY mode XAS, respectively. This is shown in the panels 1 and 2 of Fig. 9. Panel 3 of Fig. 9 shows the integrated 9

Page 9 of 27

intensity over the entire range of the O K edge. In the latter, the trends at 235

specific sites cannot be well discerned as they might cancel each others effects.

ip t

Instead, the trends are more pronounced when only integrated over the individual peaks A-D. While the TFY-peaks show an overall downward trend until

cr

EOD, the TEY-peaks appear to undergo a maximum after x=0.5. Therefore in this lithiation range there is a depth dependence of the oxygen occupancies.

In order to obtain information about changes of the occupied states close to

us

240

the Fermi level, we performed RIXS measurements. Fig. 8 shows the result for excitation at peaks A-D in the XAS spectra of Fig. 6. The RIXS spectra have

an

been normalized to unity so that the evolution of the spectral shape throughout the lithiation/delithiation cycle can be seen more easily. The various incident 245

energies (A–D) preferentially project the occupied oxygen band at different sites.

M

At the lowest energy A in the O K XAS we expect that hybridized metal states contribute strongly with their empty 3d-states. Thus the O K RIXS at this energy reflects oxygen sites that are strongly bound to Ti that should have the

a small peak about 5 eV above the main peak of non-bonding oxygen states.

te

250

d

lowest energy empty 3d-states. The corresponding hybridized states appear as

Lithiation leads to a slight increase in of spectral intensity of the hybridized

Ac ce p

states (located close to the incident energy ∼529 eV) whereas the nonbonding oxygen peak (∼524 eV) shows a some band boadening. By contrast, at excitation energies B–D, additional antibonding states appear at 520 eV and 523.5

255

eV that are strongly affected by lithiation. The spectral changes in the O K RIXS at A are found to be relatively small

whereas spectra excited at B-D show substantial narrowing (during lithiation) as well as intensity changes of the low energy structures below the main peak. We point out that renewed delithiation in the re-charge process (dashed lines

260

in Fig. 8) does not reverse this effect. It is also interesting to compare these results to Ni0.25 TiOPO4 /C where the the O K RIXS changes much less during the cycling process [36]. Since the O K RIXS was recorded in second order diffraction of the x-ray spectrometer grating its position is located to an energy correpsonding to 263 eV 10

Page 10 of 27

265

(i.e. 525/2 eV) in first diffraction order. The detector is wide enough it is possible to simultaneously record first order (non-resonantly excited) C K-emission

ip t

(∼283 eV) from the carbon coating and carbon black. The left panel of Fig. 7

shows the spectra normalized to the integrated intensity of the carbon K emis-

270

cr

sion spectrum that the sample emits when excited at the energy corresponding to the peak A in Fig. 6. The right panel of Fig. 7 then displays the integrated

us

intensity of the O K RIXS for each of the four excitation energies normalized in this way and thus reflecting its partial RIXS yield (PRIXSY). The PRIXSY for the different incidence energies can give us additional information about

275

an

the evolution of charge uptake related to the various other metal ions. In the PRIXSY evolution curves, one observes a strong initial dip for a lithiation of, which is to be expected in a rigid band model because of the loss of available

M

empty states. However, from x = 0.5 to x = 1 there is a sharp rise in intensity. Both of these behaviors (initial dip and sharp rise) find their parallels in the TFY and the TEY spectra (not displayed). For the sample with x=2.5 (second last point on the discharge branch) we find the behavior of the PRIXSY excited

d

280

at C and D.

te

at the first two energies, A and B, show a different behavior than when excited

Ac ce p

Before we turn to an explanation of this unexpected non-monotonous behavior we will look at another non-metal x-ray absorption edges to gather additional

285

clues. In Fig. 10 the XAS spectra of the P L-edge of Ni0.5 TiOPO4 /C is shown as recorded in the bulk sensitive TFY-mode. The dependence of the intensities as a function of lithiation degree are displayed in the inset. The behaviors are similar at both energies, marked with vertical dashed lines and labelled with A and B. The (more surface sensitive) TEY-mode XAS spectra of the P L-edges

290

have also been measured (not shown) but they do not show a declining trend as the TFY. 3.4. Discussion We conclude that the XAS and RIXS results suggest that both mass transport and redox reactions take place simulataneously in the battery cathode 11

Page 11 of 27

295

during its first discharge. A tentative model of what this could look like is given in Fig. 11. The pristine material starts out as crystalline particles cov-

ip t

ered with amorphous carbon. The XRD shows (Fig. 5) that a deterioration of the crystalline structure into a more amorphous state takes place. It seems

300

cr

reasonable to assume that the amorphization starts with the outer layers and

then progresses inward. Furthermore, the XRD suggest that first a secondary

us

crystalline phase, containing oxides of both Ni and Ti but without P, forms and also becomes increasingly amorphous. Generally, lithiation above x=0.5 initiates amorphization in the outer particle layers (the blue layer depicted in Fig.

305

an

11) and there is a concurrent volume expansion in the cathode material due to the incorporation of Li+ -ions. An SEI forms as the outermost layer on top of the carbon coating, although the non-crystalline phases may also increasingly mix

M

with these layers. For instance, P appears to be promoted to the outer amorphous layers, which is supported by the comparison of the P L-edge TFY and TEY data: in contrast to the TFY signal strength, which drops monotonously, the variation of the TEY-signal is low. Thus, while P is expelled from the inner

d

310

te

part, leaving behind an amorphous layer, a new layer of crystalline Ti and Ni oxides is formed further out (in the red and purple layers depicted in Fig. 11).

Ac ce p

This newly formed oxide crystal structure is probably the one belonging to the peak that emerges at 2θ∼15.8◦ in the the XRD for a lithiation level of about

315

x=0.5 (Fig. 5).

If one assumes that the C K-emission intensity is independent of the lithi-

ation process (which is suggested by the fact there exist different trends for various excitation energies) the ratios between the oxygen RIXS integrated intensity and the integrated intensity of the carbon x-ray emission (excited non-

320

resonantly simultaneously) can be used as a measure of the partial fluorescence yield for the oxygen at specific excitation energies. The PRIXSY can serve as an indicator for the filling of the initially empty states of the pristine material. In particular, it is striking that there is a deviation from the expected monotonous filling of empty states, which would be expected due to electron donation by Li

325

during the discharge process. The ordinary O K XAS spectra (Fig. 6) should 12

Page 12 of 27

reflect the relative occupation during the lithation process in the O 2p-orbitals hybridized with the Ni, Ti, and P-orbitals. However, there is a large background

ip t

from other atomic species in the TFY-signal (mostly carbon K-, phosphorous K-, and Ti L-emission) and therefore the PRIXSY-signal gives valuable addi-

tional information of the filling of the empty oxygen states at different energies

cr

330

above the Fermi level in the deeper parts of the particles where we expect a

us

more crystalline structure.

The initial lithiation step from x=0 to x=0.5 leads to a drop in intensity in as seen in Fig. 7 (as well as in the ordinary TEY and TFY modes). This is in accord with the expectation that the density of unoccupied states is reduced

an

335

when filled with the electrons donated by the inserted Li atoms. However, at a lithiation corresponding to about 1 electron (x=1) per unit cell there is

M

a sudden unexpected rise in the intensity. Continued lithiation (x>1) then leads to a further anomalous increase in the TEY-signal up to the point of 340

maximum lithiation, whereas the TFY-signal gradually decreases again, showing

d

that there is a clear depth inhomogeneity of the lithiation process. Moreover,

te

since there is no change in the spectral appearance of the O K XAS spectra (Fig. 6) that suggests a new crystalline phase, this is interpreted as a loss of charge

Ac ce p

in these hybridized orbitals as well as a loss in the amount of such orbitals due 345

to amorphization. Recharging shows a weak reversal of the trend in general but the original values are by far not re-attained. Instead, a state corresponding to about x=0.5 is reached, indicating that irreversible restructuring has occured. In summary, the first lithiation/delithiation cycle of the Li-ion battery elec-

trode composite material Lix NiTiOPO4 /C was studied using soft x-ray absorp-

350

tion and resonant inelastic x-ray scattering spectroscopy. The micron-sized particles undergo a phase separation due to a partial amorphization of outer layers and thus there is a depth segregation. The charge compensation behavior of these phases is found differ strongly, from analyzing the evolution of the partial RIXS yield at different excitation energies. After an expected initial drop,

355

we observe an anomalous intensity increase of excitation energy dependent O K-emission until the end of the first charge half-cycle (lithiation), whereas this 13

Page 13 of 27

trend is reversed in the discharge half-cycle (delithiation). This is contrasted by the behavior of the integrated intensity of the O K-absorption spectra, which

360

ip t

shows the expected overall decrease of intensity during the charge half-cycle. Therefore we conclude that charge uptake mainly takes place in the amorphous

cr

phase whereas the crystalline phase donates additional charge to the amorphous

4. Concluding Remarks and Outlook

us

phase at values x > 0.5.

We have presented XAS and RIXS results of cathode materials (Li-NMC) and the anode material (Ni0.5 TiOPO4 /C ) from cycled Li ion batteries. We find

an

365

that oxidation of oxygen is ubiquitous in these materials and in favorable cases

M

this may be a reversible oxygen redox process that can add to the cationic charge compensation during battery cycling. For instance, Li-rich NMC materials show such anionic redox reactions that account for their so-called extra capacity. We have shown that RIXS complements XAS measurements with information

d

370

about specific sites and species and thereby contains valuable information about

te

the complex processes during in the first charge/discharge cycle. Finally, our results on Ni0.5 TiOPO4 /C strongly suggest that irreversible charge capacities

Ac ce p

in oxyphosphate anodes are due to mass transport within the carbon-coated

375

particles, leading to phase segregation and amorphization during the lithiation process.

In future, we may expect to see more RIXS studies that address oxidation

and redox reactions on oxygen sites as well as other non-metallic ions in battery electrode materials. This will benefit from the advent of facilities that offer

380

high resolution (viz. >10000) RIXS, allowing investigate and distinguish different types of new binding species that are created during battery cycles. Also collective excitations such as phonons and magnons will be resolvable and may have some bearing on the understanding of some aspects of battery materials, seeing that they are transitions metal oxides. Finally, a subject that deserves

385

great attention in its own right is the further development and application of in

14

Page 14 of 27

operando RIXS spectroscopy for the study of battery cathode and anode materials. The study of these materials while they are in operation is not an easy

ip t

task but will undoubtedly lead to new insights that complement post mortem

390

cr

material studies that have been presented in this work.

5. Acknowledgements

us

We are indebted to H. Hollmark and P. Kristiansen for performing the synchrotron experiments for their respective PhD theses. T. Gustafsson is thanked for valuable discussions in the early stages of this work. K. Lasri, M. Dahbi, and

395

an

I. Saadoune are thanked for preparation of the Lix NiTiOPO4 /C material as well as for helpful discussions. The Li-rich materials, Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 , have been measured in collaboration with M. Roberts and co-workers from the

M

University of Oxford. K.E. Smith is gratefully acknowledged for access to the Boston University x-ray spectrometer and the endstation equipment at beamline X1B of the National Synchrotron Light Source (NSLS), Brookhaven. J.H. Guo is gratefully acknowledged for assistance at the wet RIXS station of beam-

d

400

te

line 8 at the Advanced Light Source (ALS), Berkeley. The authors gratefully

Ac ce p

acknowledge financial support from the Swedish Research Council (VR).

References

[1] M. Armand, J.-M. Tarascon, Building better batteries, Nature 451 (2008)

405

652-657.

[2] C. Daniel, Materials and processing for lithium-ion batteries, JOM 60 (9) (2008) 43-48.

[3] M.S.Whittingham, MRS Bull. 33 (4) (2008) 411-419. [4] A. Patil, V. Patil, D.W. Shin, J.-W. Choi, D.-S. Paik, S.-J. Yoon, Mater.

410

Res. Bull. 43 (2008) 19131942. [5] J. Hassoun, P. Reale, B. Scrosati, Recent advances in liquid and polymer lithium-ion batteries, J. Mater. Chem. 17 (2007) 36683677. 15

Page 15 of 27

[6] D. Guyomard, Advanced cathode materials for lithium batteries, in: Energy Storage Systems for Electronics, in: T. Osaka, M. Datta (Eds.), New Trends in Electochemical Technology, vol. 1, Gordon and Breach, Amster-

ip t

415

dam, 2000, pp. 253350.

cr

[7] M.S. Whittingham, Lithium batteries and cathode materials, Chem. Rev. 104 (2004) 4271-4301.

420

us

[8] E.Antolini,LiCoO2:Formation,structure, lithium and oxygen nonstoichiometry, electrochemical behavior and transport properties, Solid State Ionics

an

170, 159-171 (2004).

[9] Q. Liu, D. Mao, C. Chang, F. Huang, Phase conversion and morphology evolu- tion during hydrothermal preparation of orthorhombic LiMnO2

425

M

nanorods for lithium ion battery application, J. Power Sources 173 (2007) 538544.

d

[10] J.Molenda, M.Ziemnick, J.Marzec, W.Zajac, M.Molenda, M.Bucko, Electrochemical and high temperature physicochemical properties of or-

te

thorhombic LiMnO2 , J. Power Sources 173 (2007) 707-711.

Ac ce p

[11] M.M. Thackeray, Spinel electrodes for lithium batteries, J. Am. Ceram. 430

Soc. 82 (12) (1999) 33473354.

[12] S.G.Stewart,V.Srinivasan,J.Newman,Modeling the performance of lithiumion batteries and capacitors during hybrid-electric-vehicle operation, J. Electrochem. Soc. 155 (9) (2008) A664A671.

[13] S.K. Martha, E. Markevich, V. Burgel, G. Salitra, E. Zinigrad, B.

435

Markovsky, H. Sclar, Z. Pramovich, O. Heik, D. Aurbach, I. Exnar, H. Buqa, T. Drezen, G. Semrau, M. Schmidt, D. Kovacheva, N. Saliyski, A short review on surface chemical aspects of Li batteries: A key for a good performance, J. Power Sources 189 (2009) 288296.

[14] S.K. Martha, H. Sclar, Z.S. Framowitz, D. Kovacheva, N. Saliyski, Y. Gofer, 440

P. Sharon, E. Golik, B. Markovsky, D. Aurbach, A comparative study 16

Page 16 of 27

of electrodes comprising nanometric and submicron particles of LiNi0.50

O2 layered compounds, J. Power Sources 189 (2009) 248-255.

ip t

Mn0.50 O2 , LiNi0.33 Mn0.33 Co0.33 O2 , and LiNi0.40 Mn0.40 Co0.20

[15] Robert, R., Villevieille, C. Novak, P. Enhancement of the high potential specific charge in layered electrode materials for lithium-ion batteries. J.

cr

445

Mater. Chem. A 2, 8589-8598 (2014).

us

[16] Weaving, J. S. et al. Development of high energy density Li-ion batteries based on LiNi1-x-yCoxAlyO2. J. Power Sources 97-98, 733-735 (2001).

[18] Koga, H. et al. Different oxygen redox participation for bulk and surface:

a possible global explanation for the cycling mechanism of

M

450

an

[17] J. Fergus, Journal of Power Sources 195 (2010) 939-954.

Li1.20Mn0.54Co0.13Ni0.13O2. J. Power Sources 236, 250-258 (2013). [19] Koga,

H. et al. Operando X-ray absorption study of the re-

Li1.20Mn0.54Co0.13Ni0.13O2 in Li ion batteries. J. Phys. Chem. C 118,

te

455

d

dox processes involved upon cycling of the Li-rich layered oxide

5700-5709 (2014).

Ac ce p

[20] Armstrong, A. R. et al. Demonstrating oxygen loss and associated structural reorganization in the lithium battery cathode Li[Ni0.2Li0.2Mn0.6]O2. J. Am. Chem. Soc. 128, 8694-8698 (2006).

460

[21] 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. 133, 4404-4419 (2011).

[22] Castel, E., Berg, E. J., El Kazzi, M., Novak, P. Villevieille, C. Differential 465

electrochemical mass spectrometry study of the interface of Li2MnO3·(1-x) LiMO2 (M = Ni, Co, and Mn) material as a positive electrode in Li-ion batteries. Chem. Mater. 26, 5051-5057 (2014).

17

Page 17 of 27

[23] Hong, J. et al. Critical role of oxygen evolved from layered Li-excess metal oxides in lithium rechargeable batteries. Chem. Mater. 24, 2692-2697 (2012).

ip t

470

[24] Hy, S., Felix, F., Rick, J., Su, W. N. Hwang, B. J. Direct in situ observation

cr

of Li2O evolution on Li-rich high-capacity cathode material, Li[NixLi(12x)/3 Mn(2-x)/3]O2 (0≤x≤0.5). J. Am. Chem. Soc. 136, 999-1007 (2014).

[26] Luo, K. et al. JACS 2016.

an

475

us

[25] Luo, K. et al. Nature Chem. 8, 684-691 (2016).

[27] H˚ akan Hollmark, PhD thesis, (2011) unpublished.

[28] H. M. Hollmark, L.-C. Duda, M. Dahbi, I. Saadoune, T. Gustafsson and K.

M

Edstr¨ om Journal of The Electrochemical Society, 157, A962-A966 (2010). [29] Maher, K.; Edstrom, K.; Saadoune, I.; Gustafsson, T.; Mansori, M. The Electrochemical Behaviour of the Carbon-Coated Ni0.5TiOPO4 Electrode

d

480

te

Material. J. Power Sources 2011, 196, 2819-2825. [30] Zhang, X. J.; Zhang, Y.; Zhou, Z.; Wei, J. P.; Essehli, R.; El Bali, B. Core-

Ac ce p

Shell Ni0.5TiOPO4/C Composites as Anode Materials in Li Ion Batteries. Electrochim. Acta 2011, 56, 2290-2294.

485

[31] Belharouak, I.; Amine, K. New Active Titanium Oxyphosphate Material for Lithium Batteries. Electrochem. Commun. 2005, 7, 648-651.

[32] Godbole, V. A.; Villevieille, C.; Sommer, H. H.; Colin, J. F.; Novak, P. A Structural and Electrochemical Study of Ni0.5TiOPO4 Synthesized Via Modified Solution Route. Electrochim. Acta 2012, 77, 244-249.

490

[33] Lasri, K.; Dahbi, M.; Liivat, A.; Brandell, D.; Edstrom, K.; Saadoune, I. Intercalation and Conversion Reactions in Ni0.5TiOPO4 Li-Ion Battery Anode Materials. J. Power Sources 2013, 229, 265-271.

18

Page 18 of 27

[34] Rickard Eriksson, Karima Lasri, Mihaela Gorgo, Torbjrn Gustafsson,

495

J. Phys. Chem. C , 119, 9692-9704 (2015).

ip t

Kristina Edstr¨ om, Daniel Brandell, Ismael Saadoune, and Maria Hahlin

[35] Hollmark, H. M.; Gustafsson, T.; Edstrom, K.; et al. PHYSICAL CHEM-

cr

ISTRY CHEMICAL PHYSICS, 13, Pages: 20215-20222 (2011)

[36] Hollmark, H. M.; Maher, K.; Saadoune, I.; et al. PHYSICAL CHEMISTRY

500

us

CHEMICAL PHYSICS, 13, Pages: 6544-6551 (2011)

[37] P.T. Kristiansen, M. Dahbi, T. Gustafsson, PHYSICAL CHEMISTRY

an

CHEMICAL PHYSICS, 16 Pages: 3846-3852 (2014).

[38] Ch. Jung, F. Eggenstein, S. Hartlaub, R. Follath, J.S. Schmidt, F. Senf,

M

M.R. Weiss, Th. Zeschke and W. Gudat, First results of the soft X-ray microfocus beamline U41-PGM, Nuclear Instruments and Methods in Physics 505

Research Section A467 - 468, (2001).

d

[39] H. M. Hollmark, K. Maher, I. Saadoune, T. Gustafsson, K. Edstrom, and

te

L.-C- Duda, Resonant inelastic X-ray scattering and X-ray absorption spectroscopy on the negative electrode material Li0.5 Ni0.25 TiOPO4 in a Li-ion

Ac ce p

battery, Phys. Chem. Chem. Phys. 13, 6544-6551 (2011). 510

[40] B. Philippe at al., Journal of The Electrochemical Society, 163 (2) A178A191 (2016)

[41] I. Saadoune, M. Dahbi, M. Wikberg, T. Gustafsson, P. Svedlindh, and K. Edstrm, Solid State Ionics, 178, 1668 (2008).

[42] Paw Kristiansen, PhD thesis, (2013) unpublished.

515

[43] Rickard Eriksson, Kenza Maher, Ismael Saadoune, Mohammed Mansori, Torbjrn Gustafsson, Kristina Edstrm Solid State Ionics 225, 547-550 (2012). [44] Karima Lasri, Mohammed Dahbi, Anti Liivat, Daniel Brandell, Kristina Edstr¨ om and Ismael Saadoune Intercalation and conversion reactions in

19

Page 19 of 27

Ni0.5 TiOPO4 Li-ion battery anode materials, Journal of Power Sources, 229. 265 - 271 (2013).

Ac ce p

te

d

M

an

us

cr

ip t

520

20

Page 20 of 27

ip t

O K RIXS and XAS spectra

cr

raw and difference to pristine

us an

4.5 V - EOP

M

Intensity [arb. units]

4.8 V - EOC

te

d

4.4 V - BOP

Ac ce p

520

Figure

1:

Oxygen

K

525

530

535

Photon Energy/eV

RIXS

and

XAS

spectra

(TFY

mode)

of

Li1.2 [Ni0.13 Co0.133 Mn0.544 ]O2 excited at the peak of the absorption resonance

(marked with an arrow). The cycle potentials are indicated as well as the corresponding location on the charge curve (BOP: beginning of plateau, EOP: end of plateau, and EOC: end of charge). The XAS traces in gray color are difference spectra between the spectra from the respective cycled material (colored) and the spectrum from the pristine material (black).

21

Page 21 of 27

Figure

2:

cr us Oxygen

525

an

520

530 535 Energy / eV

540

M

Intensity / a.u.

x=1.0 x=0.9 x=0.75 x=0.6 x=0.4

x=1.0 x=0.9 x=0.75 x=0.6 x=0.4 x=0.2

ip t

XAS O K-edge

RIXS O K-edge

K

RIXS

and

XAS

545

spectra

550

(TFY

mode)

of

Lix [Ni0.65 Co0.25 Mn0.1 ]O2 excited at the peak of the absorption resonance (indi-

Ac ce p

te

indicated.

d

cated by the arrow) and color-coded corresponding to the degree of lithiation x as

Figure 3: Discharge-charge profile under a galvanostatic mode during the first cycle of

Ni0.5 TiOPO4 /C in the 3.0 to 0.5 V range vs Li/Li+ (at C/10 rate). B1–B7 refers to the sampling points of the investigation.

22

Page 22 of 27

ip t cr us an M

Figure 4: Structure of Ni0.5 TiOPO4 . The marked tunnel constitutes one of the favor-

te

d

able paths for lithium insertion.

Ac ce p

A#

B4#

B#

B3#

B2#

B1#

B0#

Figure 5: (A) In situ XRD of a Ni0.5 TiOPO4 /C composite anode. (B) Charge curve

for the sample. The letters and dots denote the potentials at which in situ XRD has been performed. The vertical and horizontal arrows depict the potentials at which some (B0-B4) of the batteriy anodes used for x-ray spectroscopy were stopped.

23

Page 23 of 27

ip t cr B C

A

B6# B5# B4# B3# B2#

B5# B7#

B1#

us

B

D

S

C

an

B7#

O K XAS (TFY)

D S

Intensity (arb. units)

pristine x=0.5 x=1 x=2 x=2.5 x=3.1 x=2.2 x=1.2

A

B7#

B6# B5# B4#

M

Intensity (arb. units)

O K XAS (TEY)

B3#

B2#

d

B0#

te

530 535 540 incident photon energy/eV

B1# B0#

B7# B5# pristine x=0.5 x=1 x=2 x=2.5 x=3.1 x=2.2 x=1.2

530 535 540 incident photon energy/eV

Ac ce p

Figure 6: The left (right) panel shows the cycling evolution of O 1s XAS spectra in the

TEY-mode (TFY-mode). The spectra are color-coded corresponding to the degree of lithiation x as indicated (and belong to the points in the cycle B0-B7 of Fig. 3). For convenience, the last spectra from the discharging, respectively, the charging branch is also superimposed in grey onto the spectrum of the pristine sample (black trace).

24

Page 24 of 27

B

0.7

0.6

D

1.6

A

1.5

S# 1.0

C

0.5

B

0.0

ip t

1.7

TEY intensities

0.9

0.8

1.5

1.8

A B C C' D

D

cr

C

Intensity @ D [a.u.]

Intensity @ A, B and C [a.u.]

1.0

2)#TEY#peaks#

1.9

A B C D

A

us

XAS max intensities 1)#TFY#peaks#

1.1

pristine x=0.5 B5# x=1 B6# EoC B2# x=2 B7# B0# B1# x=1 B3# x=2.5 B4# EoD

x=2 B7# B3# x=2.5 B4# EoD B5# x=1 B6# EoC 0###0.5###1####2###2.5#EOD#1###EOC#

0###0.5####1####2####2.5#EOD##1##EOC#

pristine x=1 B0# x=0.5 B1# B2#

3)#Integrated#O#K5edge# 25.5

an TEY##

24.5

24.0

21

20

TEY##

23.5

23.0

Int TEY 526.9 - 550.6

Int TFY/I0 527.6 - 551

22 25.0

19

B0# B1# B2# B3# B4# B5# B6# B7# 0###0.5##1####2##2.5#EOD##1##EOC# x=0.5

x=1

x=2

x=2.5

EoD

x=1

EoC

M

pristine

Figure 7: Panel 1: Evolution of the TFY intensities for excitation energies A-D. Panel

d

2: Evolution of the TEY intensities for excitation energies A-D. Panel 3: Evolution of

Intensity [arb. units]

pristine x=0.5 x=1 x=2 antibonding x=2.5 O 2p-states EoD x=1 EoC

515

520

nonbonding O 2p-states

hybridized with TM 3d states

C nonbonding O 2p-states

525

Intensity [arb. units]

A

Intensity [arb. units]

te

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

Ac ce p

Intensity [arb. units]

the integrated TFY and TEY intensities, respectively for excitation energies A-D.

530

pristine x=0.5 (x=1) x=2 antibonding x=2.5 O 2p-states EoD x=1 EoC

pristine x=0.5 x=1 antibonding x=2 O 2p-states x=2.5 EoD x=1 (EoC)

515

photon energy/eV

B nonbonding O 2p-states

520

D nonbonding O 2p-states

525

530

photon energy/eV

Figure 8: O K RIXS spectra normalized to unity for all excitation energies (A–D).

The variation of the spectral shape of the antibonding states as well as of the weak metal-hybridization states at the top of the valence band is evident. The spectra are color-coded corresponding to the degree of lithiation x as indicated.

25

Page 25 of 27

PFY intensities PRIXSY'

1

10

4 8

cr

2

Intensity @ A, B and C [a.u.]

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

3

12

A B C D

5

3

6

1.2 TFY

2

us

510

520

530 540 550 Photon Energy [eV]

560

x=2 x=2.5 EoD x=1 EoC 0'''0.5''''1''''2''''2.5'EOD''1''EOC'

pristine x=0.5 x=1

570

(a)

1.0

(b)

0.80 0.6

0.70

0.4

0.65

0.3

0.60 0.55 p

FIG. 6: (a) RIXS obtained from A in figure 5 normed to

0.5 1 2 2.5 EoD1 EoC

B, 138.9

0.8

Figure 9: (a): Evolution of theorder RIXS for Max excitation energies A when normalthe the second C intensity feature. (b) RIXS intensities Intensity [a.u.]

an

0.75

0.5

Int @ 138.9 eV

4

0

Int @ 137.5 eV

O K-edge RIXS @ A, 529.75 eV

Intensity @ D

Intensity normed to C K-edge

4

tensities at A and B on lithiation degree in the inset. The TEY-mode XAS spectra (b) offset vertically to each other, where th progresses from bottom to top, except for of the last sample at EoC that is superp spectrum of the pristine sample at the b one observes a continuous drop in spectra the TFY-mode XAS P L-edges there is no the TEY-mode XAS.

ip t

were normalized is a similar fashion and the maximum intensities of all of them are shown in Fig. 6(b).

A, 137.5

when the spectra are normalized to unity on the second ized to the integrated C K emission intensity. (b): Integrated O K RIXS for excitation 0.6 order carbon feature.

M

energies A-D normalized to the C K emission, i.e. partial RIXS yield (PRIXSY). 0.4

1.0

1.0

O K-edge RIXS @ B, 532.15 eV

520

525 Photon Energy [eV]

1.0

0.6

0.31

0.4

0.30

521.6

Intensity [a.u.]

Intensity [a.u.]

0.32

521.8

Intensity [a.u.]

0.75 0.70 0.65

0.55

510

pristine 137.5 x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

0.2

520 525 Photon energy [eV]

521.2

0.0

520 Photon Energy [eV]

525

TEY

1.0 O K-edge RIXS @ D, 538.00 eV

8

B, 138.9

138.0

A, 137.5

0.34 0.33 0.32

0.6

0.31 0.30 522.0

0.4

522.2

522.4

522.6

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

13

B, 138.9

510

(c)

515

520 525 530 x 1/8[eV] Photon energy

p 0

6

535

(d)

0.0

2.0 1.5 1.0 0.5

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1

0.0 535

138.5 139.0 Photon energy [eV]

0.8

0.2

530

137.5

530

(b)

522.0

0.4

515

515

0.5 1 2 2.5 EoD1 EoC

0.2

0.0

521.0

137.0

0.60

0.6

520.8

0.80

0.3

A,

520.6

0.2

510

0.4

p

520.4

0.0

0.5

O K-edge RIXS @ C, 533,73 0.8eV 0.33

0.26

530

0.6

Ac ce p

1.0

0.2

0.4

Int @ 138.9 eV

(a)

Int @ 137.5 eV

1.2 TFY

0.27

Int @ 137.5

515

0.6

Intensity [a.u.]

0.0

0.8

0.28

te

0.2

0.30 0.29

d

Intensity [a.u.]

0.6

0.4

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

0.31

0.8

pristine x=0.5 x=1 x=2 x=2.5 EoD x=1 EoC

0.8

137.5 139.04 /C 139.5samples 140.0 FIG. 7: RIXS137.0 from the138.0 Ni0.5138.5 TiOPO normed to

Intensity [a.u.]

O K-edge RIXS @ A, 529.75 eV

4

Photon energy [eV]

unity. (a) at 529.75 eV, (b) at 532.15 eV, (c) at 533.73 eV and (d) 538.00 eV. The inserts in (b)–(d) are of the position 2 and intensity of the low-energy shoulder/peak generally Figure 10: The cycling evolution of O K XAS spectra (TFY) for Ni0.5 TiOPO4 /C . The around 521.5 eV. spectra are color-coded corresponding to the degree of lithiation x as indicated.

In Fig. 7 the RIXS spectra obtained using the four incident energies (A–D) are displayed when the oxygen feature is normalized to unity. At incident energy A (529.75 eV) the spectrum displays little sensitivity with respect to the degree of lithiation, whereas the three higher excitation energies (B–D) all spectra have low en26 ergy features that vary in intensity and position as the degree of lithiation is changed. This is plotted in the insets. In Fig. 8 the XAS spectra of the P L-edge of Lix NiTiOPO4 is shown. The TFY-mode XAS spectra in (a) are plotted superposed and the dependence of the in-

0 137

138 139 Photon energy [eV]

FIG. 8: XAS of the P L-edge, (a) is TFY a

In Fig. 9 TEY-mode XAS spectra at the shown. The Ni 2p3/2 main resonance pea with the letters D and E. The ratio of of D and E is characteristic of the ioniz the Ni atom and thus offers a way of deter valence[4]. The features in incident energy Page 26 of 27 been attributed to the fact that higher or

ip t cr us an

M

SEI Carbon

d

Amorphous

P

te

Crystal

Ac ce p

Figure 11: Structural model and restructuring process of the Lix NiTiOPO4 /C battery

cathode material particles. See text for details.

27

Page 27 of 27