1

Q. Chen, F. El Gabaly, F. Aksoy Akgul, Z. Liu, B.S. Mun, S. Yamaguchi, A. Braun, Observation of oxygen vacancy 1

filling under water vapor in ceramic proton conductors in-situ with ambient pressure XPS; Chem. Mater., 2013, 25 2

(23), pp 4690–4696 http://pubs.acs.org/doi/abs/10.1021/cm401977p 3

in-situ observation of oxygen vacancy filling under water vapor 4

in ceramic proton conductors with resonant 5

ambient pressure XPS and impedance spectroscopy 6

7

Qianli Chen1,2, Farid El Gabaly3, Funda Aksoy Akgul 4,5, Zhi Liu4, 8

Bongjin Simon Mun6, Shu Yamaguchi7, Artur Braun1,* 9

10

1 Laboratory for High Performance Ceramics, Empa. Swiss Federal Laboratories for 11

Materials Science and Technology, CH-8600 Dübendorf, Switzerland 12

13

2 Department of Physics, ETH Zürich, Swiss Federal Institute of Technology 14

CH-8057 Zürich, Switzerland 15

3 Sandia National Laboratories, Livermore, California 94551, USA 16

4 Advanced Light Source, Lawrence Berkeley National Laboratory, 17

Berkeley, CA 94720, USA 18

5 Physics Department, Nigde University, 51240 Nigde, Turkey 19

6 Department of Physics and Photon Science, School of Physics and Chemistry, Ertl Cen-20

ter for Electrochemistry and Catalysis, Gwangju Institute of Science and Technology, 21

Gwangju 500-712, Republic of Korea 22

7 Department of Materials Engineering, University of Tokyo, 113-8656 Tokyo, Japan 23

2

Abstract 24

25

The interaction of metal oxides with their ambient environment at elevated temperatures 26

is of significant relevance for the functionality and operation of ceramic fuel cells, elec-27

trolyzers and gas sensors. Proton conductivity in metal oxides is a subtle transport pro-28

cess which is based on formation of oxygen vacancies by cation doping and substitution, 29

and oxygen vacancy filling upon hydration in water vapor atmosphere. We have investi-30

gated the conductivity and electronic structure of the BaCeY-oxide proton conductor un-31

der realistic operation conditions from 373 K to 593 K and water vapor pressures up to 32

200 mTorr in-situ by combining ambient pressure X-ray photoelectron spectroscopy and 33

electrochemical impedance spectroscopy. We provide element specific spectroscopic evi-34

dence that oxygen vacancies are filled by oxygen upon water exposure and partly oxidize 35

Ce3+ and Y2+ towards Ce4+ and Y3+. Moreover, the resonant valence band spectra of dry 36

and hydrated samples show that oxygen ligand holes in the proximity of the Y dopant are 37

by around 0.5 eV closer to the Fermi level than the corresponding hole states from Ce. 38

Both hole states become substantially depleted upon hydration, while the proton conduc-39

tivity sets on and increases systematically. Charge redistribution between lattice oxygen, 40

Ce, and Y when BCY is exposed to water vapor at ambient and high temperature, give us 41

insight in the complex mechanism for proton incorporation in BCY. 42

43

44

45

3

Introduction 46

47

Protons can be structural elements in molecules and in hydrates, for example, and also 48

ionic and electric charge carriers. In the former case they are localized, in the latter case 49

they are delocalized. The proton is in both cases an elusive element as far as its interac-50

tion with the host lattice or molecular environment is concerned. 51

Ceramic proton conductors are prospective solid electrolytes for intermediate temperature 52

solid oxide fuel cells (IT-SOFC [1-3]). For its functionality as proton conductor, the dy-53

namics is very important with respect to lowering proton transport activation energies. 54

Impedance spectroscopy, inelastic and quasi-elastic neutron scattering and Raman and in-55

frared vibrational spectroscopy are the analytical tools to address the dynamics of the pro-56

ton conductivity. For improving and optimizing proton conducting materials, detailed 57

knowledge on the chemical interaction of the proton with the elements in the host lattice 58

is necessary. For example, exposure of metals to hydrogen atmosphere can cause embrit-59

tlement of the metal, i.e. structural disintegration. In metal oxides, exposure to hydrogen 60

atmosphere can form hydroxyl groups, which constitute a very rigid and polar species 61

that may affect integrity of the oxide and influence on charge transfer. In compounds with 62

oxygen vacancies, the oxygen from ambient water vapor molecules may fill the vacan-63

cies, whereby the protons from the water molecules are incorporated to the host lattice 64

[4]. 65

In the present work we have substituted BaCeO3 to 20% with Y2O3 (BaCe0.8Y0.2O3-δ) in 66

order to form oxygen vacancies to be filled with oxygen from water vapor molecules. The 67

chemical interactions of protons with the host metal oxide are investigated by ambient 68

4

pressure XPS under dry and hydrated conditions, meanwhile the proton conductivity is 69

studied using electrochemical impedance spectroscopy. 70

71

72

73

Experimental 74

BaCe0.8Y0.2O3-δ (BCY20) was prepared by solid state synthesis from precursors mixed in 75

stoichiometric amounts and fired at 1473 K for 12 hours in air, then ground and calcined 76

again at 1473 K for 12 hours. The obtained powder was pressed to pellets of 1 mm thick-77

ness and 18 mm diameter at 10 kbar and sintered for 24 hours at 1673 K in air [5, 6]. 78

Phase purity of the BCY20 was confirmed with powder X-ray diffraction on the finally 79

obtained pellets (Figure S1 in Supporting Information). The sintered pellet was subject to 80

a water vapor saturated N2 flow at 670 K for 16h [5]. Temperature dependent X-ray dif-81

fraction measurements were made on the sintered sample from room temperature up to 82

923 K in steps of 50 K. X-ray diffractograms were measured in a Bragg–Brentano geom-83

etry using a PANalytical X’Pert PRO θ-2θ scan system with the X-ray wavelength of 84

1.5406 Å (Cu-Kα1). 85

Au electrodes were deposited on the proton conducting electrolyte by evaporation in vac-86

uum from crucibles heated by an electron beam, with a mask to control the shape of the 87

electrode [7]. A sample holder with spring-loaded probes (Figure 1), specifically de-88

signed for the combination of XPS and impedance spectroscopy in-situ and operando un-89

der realistic high temperature electrochemical conditions, was used to provide good elec-90

trical contact with the electrodes [8]. The ionic conductivity for BCY20 was measured by 91

5

electrochemical impedance spectroscopy (EIS) using a PCI4/750 Potentiostat (Gamry). 92

The impedance spectra were analyzed with ZView (Scribner Associates). 93

X-ray photoelectron spectra were recorded at the ambient pressure photoemission spec-94

trometer chamber at Beamline 9.3.2 at the Advanced Light Source in Berkeley, California 95

[9, 10]. We exposed the material to water vapor (100 mTorr) at ambient temperature 96

while at the same time recording XPS core level spectra and valence band (VB) spectra 97

in-situ. The temperature was increased step-by-step to 773 K under the water pressure of 98

100mTorr. Again, XPS core level and VB spectra were recorded in-situ with impedance 99

spectra. This approach warrants that we can assess the chemical state of the proton con-100

ductor surface under reaction conditions and also proton conducting operation conditions. 101

Core level spectra for Ba 4d, Y 3d, Ce 4d, and O 1s were recorded for ambient tempera-102

ture, 373 K and 573 K in vacuum, and 573 K with 100 mTorr and 200 mTorr water vapor 103

pressure, respectively. The temperature was measured with a 2-color pyrometer (Mikron 104

M90-H1). Ce 4p1/2 resonant XPS spectra were recorded in vacuum, and with the water 105

vapor pressures mentioned above; Y resonant XPS spectra were recorded with respect to 106

the Y 3p1/2 and 3p3/2 from 299 eV to 310 eV, respectively. 107

6

108

Figure 1: BCY20 pellet with two sputtered Au current collector terminals, clamped in 109

sample holder with 3-electrode configuration for in-situ/operando high-temperature XPS 110

and impedance spectroscopy. 111

112

113

114

115 Results and Discussions 116

Before we turn to the systematic changes in the proton conductor during hydration, we 117

have to investigate the constitution of the proton conductor material (BCY20 pellet) be-118

cause it has been exposed to ambient conditions after synthesis. We have thus subjected 119

the pellet to a controlled drying protocol and monitored this process with thermogravime-120

try and XPS (Supporting Information). After this drying procedure, we exposed the 121

BCY20 pellet to 100 mTorr water vapor at around 573 K, while still recording XPS spec-122

tra. 123

7

124

Proton conducting ceramics develop their ionic conductivity at elevated temperatures T > 125

500 K. A typical variation of the impedance spectra of hydrated BCY20 with temperature 126

ranging from 320 K to 820 K is shown in ref. [Q. Chen, J. Banyte, X. Zhang, J. P. Embs, 127

A. Braun; Proton Diffusivity in Spark Plasma Sintered BaCe0.8Y0.2O3-δ: in-situ combina-128

tion of quasi-elastic neutron scattering and impedance spectroscopy; Solid State Ionics; in 129

press, http://dx.doi.org/10.1016/j.ssi.2013.05.009]. Figure 2-a shows a representative set 130

of impedance spectra of hydrated BCY20, recorded during heating in the UHV chamber 131

while XPS spectra were taken. 132

The impedance spectra in Figure 2-a show the typical semicircles, the radii of which have 133

been determined by deconvolution and least square fitting (ZView, Scribner Associates). 134

The first semicircle at high frequencies originates from the bulk conductivity. Figure 2-b 135

shows hence the bulk conductivity. This conductivity increases with increasing tempera-136

ture homogeneously but not linear over the entire measured temperature range. Rather, 137

conductivity increases linear from 400 K to 450 K, increases then steep from 450 K to 138

550 K, and increases further with the similar slope to the temperatures region of 400 ~ 139

450 K. 140

141

8

142

Figure 2: 2-a: Representative set of impedance spectra of hydrated BCY20 pellet record-143

ed during heating in UHV. The inset shows magnified the high frequency semicircle from 144

the proton bulk conductivity. 2-b: Bulk proton conductivity of hydrated BCY20 as a 145

function of temperature, derived from impedance spectra set mentioned in 2-a. 2-c: Ex-146

cerpt of X-ray diffractograms from hydrated BCY20 recorded from 298 K to 973 K in air. 147

2-d: Variation of unit cell volume of hydrated BCY20 versus temperature, as derived 148

from X-ray diffractograms (2-c). 149

150

151

152

Naturally, the sample undergoes thermal expansion during annealing, which is reflected 153

by the change of the crystallographic unit cell volume (basically the thermal expansion) 154

as determined by high temperature X-ray diffraction in air (Figure 2-c). Note that the 155

9

BCY20 pellet has been saturated with water vapor by the same protocol as published in 156

[5]. The thermal expansion profile in Figure 2-d shows a similar behavior like the con-157

ductivity variation during annealing. We found recently [4] that a decrease of the thermal 158

expansion coefficient of BaZr0.9Y0.1O3-δ (BZY10) occurs at about the same temperature 159

of ~ 650 K where the quasi elastic neutron scattering shows an onset of lateral proton 160

mobility, revealing a correlation of proton conductivity and lattice spacing dynamics. 161

162

We begin with the systematic XPS study by heating the sample under exposure to water 163

vapor. Figure 3 shows three oxygen core level XPS spectra recorded when the sample 164

was dried in UHV, and then exposed to 100 mTorr and 200 mTorr water vapor pressure 165

at 592 K – 532 K. The spectrum recorded in dry condition in UHV (Figure 3-a) shows the 166

O-H peak at around 533 eV and the peak from structural oxygen near 529.5 eV. The two 167

peaks at 529 eV and 532.5 eV originate from structural oxygen Ox in the BCY20 perov-168

skite lattice and from hydroxyl groups (O-H) [12-15]. During exposure to 100 mTorr wa-169

ter vapor, the spectrum is shifted towards lower binding energy by about 0.3 eV, which 170

we have corrected for in Figure 3-b. This shift is likely due to increased electronic con-171

ductivity originating from the development of the space charge region through the grain 172

that influences strongly the inter-grain band-bending [8]. Concomitantly the O-H peak 173

height is increasing whereas the Ox peak intensity is decreasing. This observation sup-174

ports the suggestion that oxygen vacancies are filled and the concentration of protons in-175

creases and enhances the conductivity. During the injection of the water vapor into the 176

UHV chamber, the sample temperature decreased from 592 K to 545 K, an as of yet una-177

voidable technical side effect due to the heat capacity of water vapor. 178

10

179

In a further step, we increased the water pressure to 200 mTorr. The signature of water in 180

the gas phase comes up at around at 535.7 eV The spectrum shows two prominent and 181

well separated transitions at 528.5 eV and 532.5 eV, indicative to hydroxyl O-H and two 182

noticeably different structural oxygen ions. We believe these different oxygen ions could 183

be in proximity to Ce and to Y, respectively. An alternative interpretation could be that 184

these different structural oxygens are the well-known O1 and O2 oxygen ions in ortho-185

rhombic BCY [K.S. Knight, Solid State Ionics 2001, 145, 275–294.]. The exposure to 186

water vapor at this high temperature increases the electric conductivity of the BCY20 187

(Figure 2-b), which manifests in an additional shift of 0.4 eV to an overall shift of the 188

spectrum of 0.7 eV towards lower binding energies (Figure 3 shows the spectra (a) and (b) 189

after alignment on the energy axis). More noticeable is the redistribution of spectral 190

weight from the transition at 528.5 eV, originating from oxygen bound to Ce, toward the 191

corresponding transition originating from oxygen bound to Y, at 532.5 eV. This reveals 192

that the oxygen vacancies in BCY20 formed by substitution with Y are becoming filled, 193

and the corresponding states becoming more populated. 194

11

538 536 534 532 530 528 526 524

Ox (near Ce4+)OH(a) 592 K

UHV

O 1s

XPS

Inten

sity (b) 545 K

p(H2O)=100 mTorr

Ox (near Y3+ and Ce3+)

Ox (near Ce4+)

Binding Energy [eV]

H2O (gas)

(c) 532 K

p(H2O)=200 mTorr

0.4 eV

195

Figure 3: O 1s core level XPS for BCY20 at (592 K to 532 K, temperature changes due 196

to heat capacity of injected water vapor), in (a) UHV, and in water vapor with (b) p(H2O) 197

= 100 mTorr and (c) p(H2O) = 200 mTorr. Photon energy = 700 eV. The spectra are nor-198

malized and aligned by the structural Ox oxygen peak (near Ce4+). 199

200

A shift of 0.4 eV has been observed by Higuchi et al. [19] on 10% Y-substituted barium 201

cerate depending on whether the sample was heated in air or in hydrogen. This is consid-202

ered a sign of the thermal activation of protons, preceding the onset of lateral proton dif-203

fusivity which constitutes the proton conductivity. Since the protons draw electrons away 204

from oxygen, when one proton is shared by two oxygen ions, each oxygen ion should be 205

more negatively charged, and thus the O 1s core level is chemically shifted to lower bind-206

ing energy. To maintain charge balance, we expect that the yttrium delivers electrons 207

when it is getting oxidized. As we will see later, this is indeed observed. 208

12

209

The temperature ranges that we are considering here with XPS show also characteristic 210

changes in the thermogravimetry analyses. The mass of BCY20, as shown in Figure 4-b 211

is noticeably increasing in an air filled TGA chamber at around 300 K. Comparison with 212

literature [11] (Figure 4-a) shows that at such temperature surface water is being released, 213

as shown by mass spectrometry, notwithstanding that the actual weight of the sample is 214

increasing, possibly by hydroxylation. For BCY20, the slope of the derivative of the ob-215

served mass change (lower curve in Figure 4-b) is positive between 300 K and 400 K, 216

virtually zero from 450 K to 650 K, and negative for T > 800 K. 217

218

Figure 4: 4-a: TGA and mass spectrometry data of BaCe0.85Yb0.15O3 reproduced from lit-219

erature [11]; 4-b: and TGA mass change and derivative from hydrated BCY20 in synthet-220

ic air. 221

222

13

It is expected that exposure to water fills oxygen vacancies and thus oxidizes the BCY20. 223

One interesting question is: which vacancies are filled first, i.e. which vacancies are filled 224

more easily. In the course of this in-situ XPS experiment, the BCY20 always showed Ce 225

in a mixed state of Ce3+ and Ce4+. A change in the relative spectral weight is found in the 226

peak ratio at higher binding energies (labeled W‴ and X‴ in Figure 5a, following the no-227

tation in [16]) and lower binding energies (labeled A-C). W‴ and X‴ are signature peaks 228

for Ce4+; these are absent in Ce3+ [15]. This change of ratio in the spectra reveals that 229

Ce3+ is partially oxidized to Ce4+ upon adding water vapor. This picture is paralleled by 230

the evolution of Y3d core level spectra (Figure 5-b) under the same conditions. The in-231

tensities of the structures at 157.5 eV and 159.5 eV are increasing during supply of the 232

water vapor, whereas the structures at 156.5 eV and 158.5 eV are decreasing. 233

130 125 120 115 110 105 100 95 164 162 160 158 156 154

ABCX'''

200 mTorr H2O

100mTorr

XPS

Inten

sity

Binding Energy [eV]

UHV

W'''

(a) Ce 4d

UHV

(b) Y 3d 42% Y3+ Y(3-x)+

Y(3-x)+64% Y3+

p(H2O)=100 mTorr

Y(3-x)+107% Y3+

p(H2O)=200 mTorr

Binding Energy [eV] 234 235

Figure 5: Ce 4d (a) and Y 3d (b) core level XPS for BCY20 at ~530 K, dry and hydrated 236

condition, Ce and Y become oxidized upon addition of water. hν = 700 eV. 237

238

14

As stated in the Introduction, proton conductivity is a subtle process. Weak spectral sig-239

natures are a manifestation of that. High resolution soft X-rayX-ray absorption and emis-240

sion spectra [18] show three O 2p states which have been termed hydrogen structures and 241

considered direct evidence of O-H bonds in the bulk of Y-substituted SrCeO3. These so-242

called hydrogen structures have very small intensity [18]. 243

In order to enhance a potential spectroscopic contrast between Ce and Y, we have applied 244

the valence band XPS experiment in the resonant mode with varied photon energy. Fig-245

ure 6 shows the resonant XPS spectra in the Ce 4p→4d energy region. The valence band 246

consists of a mixed state between 4d1L (A) and 4d0 (B) configurations, in analogy to [17]. 247

L denotes the hole in the valence band, which is mainly composed of the O 2p state. The 248

valence band spectra show a remarkable difference between dry and hydrated state, 249

whereas at first glance a difference between on-resonance and off-resonance cannot be 250

made out. 251

10 8 6 4 2 0 -2 -4

UHV p(H2O)=100 mTorr

off-resonancehn = 223 eV

Inten

sity

Binding Energy [eV]

on-resonance (Ce 4p)hn = 232 eV

O 2p AB

252

Figure 6: On- and off-resonance XPS spectra for dry and hydrated BaCe0.8Y0.2O3-δ 253

measured at ~537 K with hν = 232 eV (Ce 4p 1/2) and 223 eV, respectively. The spectra 254

measured in water vapor are aligned to the spectrum obtained under UHV by shifting 0.6 255

15

eV to higher binding energy. The broad bump (blue) at 0 eV when hν = 232 eV is a sec-256

ond harmonic effect. 257

258

The leading peak A at around 5 eV in Figure 5 therefore represents the Ce3+ state and 259

peak B represents the Ce4+ state. The BCY20 dried in UHV shows in the resonant and 260

off-resonant mode two distinct peaks for Ce3+ and Ce4+ of equal heights, revealing that a 261

substantial amount of the Ce is in the Ce3+ valence at the probed region, i.e. the BCY20 262

surface in the dry condition. When water vapor is supplied at around 573 K, the spectral 263

weight of the leading Ce3+ state decreases by 50%. The spectral weight of the Ce4+ state 264

increases accordingly. A remarkable shoulder from Ce3+ spectral weight remains upon 265

hydration at this level, but the spectral differences are striking. The Ce3+ ions are likely 266

located on a sample surface or in an oxygen defect site. Water fills the oxygen defects 267

and consequently the spectral weight of Ce3+ decreases while the spectral weight of Ce4+ 268

increases. Moreover, for the hydrated BaCe0.8Y0.2O3-δ, the valence band shifts by 0.6 eV 269

to lower binding energy, suggesting higher conductivity due to hole doping induced by 270

hydration (Figure 6 shows the spectra after the aligning of the binding energy B.E. with 271

peak A and B). Slight shifts of the spectra on the energy axis have been observed for ex-272

ample in the VB and oxygen core level spectra on BaCe0.9Y0.1O3-δ (0.4 eV) depending on 273

whether the samples had been annealed in air or in hydrogen [19]. 274

A broad, low intensity peak with the FWHM of ~1 eV appears at 0 eV binding energy 275

when using hν = 232 eV (Ce resonant) as excitation energy, as shown in Figure 6. This 276

broad peak is a spectroscopic artifact and originates from the second order effect out of 277

16

the beamline grating, likewise in the Y resonance spectra (Figure 6). At the Ce resonance 278

energy, this second order effect can be significant. 279

The observation that we make on the cerium resonant VB spectra is paralleled by the yt-280

trium resonant VB spectra at 299 eV and 311 eV. Figure S6 shows a survey spectrum 281

from 200 eV to 350 eV, showing where Y and Ce resonance energies are located. 282

Figure 7 shows the Y-resonant VB spectra of the BCY20 recorded at the 3p3/2 and 3p1/2 283

resonant energies under dry and hydrated conditions at 585 K and 548 K, respectively. 284

We recall for the reader that the heat capacity of the water vapor had effect on the tem-285

perature at the sample when the water vapor was injected in the UHV chamber. Here, too, 286

the BCY20 has a double peak in the dry UHV condition. While we could assign the dou-287

ble peak to Ce3+ and Ce4+ in Figure 5, the origin of the double peak for the Y resonant 288

spectra is not immediately clear, because we anticipate no Y4+ state. We recall that the 289

valence band with respect to the Ce consists of a mixed state between 4d1L (A) and 4d0 290

(B) configurations, in analogy to [17]. Therefore, the leading peak in the Y-resonant 291

spectra at around 4.5 eV binding energy in Figure 6, should be assigned to a reduced Y 292

species such as Y(3-x)+, with x<1. Interesting to note is also that this peak has slightly 293

higher spectral weight than the neighboring conjugated Y3+ peak at around 6 eV. Upon 294

supplying water vapor, we again notice the shift of roughly 0.6 eV towards the Fermi en-295

ergy, suggesting hole doping. Hence, Ce and Y show qualitatively the identical electronic 296

response towards hydration, this is, a slight oxidation with hole doping from the O 2p 297

states. Moreover, it appears that we are unable to discriminate between Ce and Y, despite 298

the resonant excitation that we want to take advantage of. 299

17

14 12 10 8 6 4 2 0 14 12 10 8 6 4 2 0p(H2O)=100mTorr

XPS

Inten

sity

Binding Energy [eV]

UHV

(a)

(b)

p(H2O)=100mTorr

UHV

Binding Energy [eV] 300

Figure 7: Y 3p1/2 resonant XPS (photon energy = 311 eV) at 585 K and 548 K(a); and Y 301

3p3/2 resonant XPS (photon energy = 299 eV) at 585 K and at 544 K (b). The spectra 302

measured in water vapor are aligned to the spectrum obtained under UHV by shifting 0.6 303

eV to higher binding energy. 304

305

When we subtract the VB XPS spectra recorded under wet condition from those recorded 306

under dry condition, we obtain a difference spectrum which should contain the spectral 307

signature of the oxygen defects. Figure 8 shows these difference spectra in resonant con-308

dition for Y and Ce. The maxima of the pronounced difference peak at around 5 eV bind-309

ing energy are shifted by about 0.5 eV, revealing that the gap state of oxygen vacancies 310

next to Y is by 0.5 eV closer to the Fermi level than a gap state from an oxygen vacancy 311

next to Ce. 312

18

313

Figure 8: Difference spectra from Y and Ce resonant XPS measurements under dry and 314

wet condition show a chemical shift of 0.5 eV near the Fermi energy. 315

316

The temperature at which we study the BCY20 with VB XPS is the temperature where 317

we expect the water molecules to be split and their oxygen ions to fill oxygen vacancies 318

that were formed by the substitution with yttrium, which we have sketched in Figure 8. 319

This scenario is indeed confirmed by the change of our spectra upon hydration. Ce3+ is on 320

a sample surface or in an oxygen defect site. The oxygen from the water molecules fills 321

the oxygen defects and consequently the spectral weight of Ce3+ decreases while the Ce4+ 322

spectral weight increases. Note while we here discuss the behavior of cerium, the oxygen 323

vacancies should be actually concentrated at around the yttrium ions. Apparently, this 324

makes spectroscopically little difference when water vapor is supplied. This suggestion is 325

19

corroborated by the observation that the Y-resonant and Ce-resonant VB XPS show iden-326

tical behavior. 327

328

Figure 9: Sketch of the BCY20 structure showing Y, Ce, O, and protons from the water 329

or from the hydroxyl groups. 330

331

The oxygen core level spectra in Figure 3 show that the temperature range (600 K to 500 332

K) where the VB XPS spectra are recorded contains still significant spectral weight for 333

the O-H groups (592 K and 545 K), notwithstanding that at 532 K the spectral weight 334

from O-H groups is clearly dominating the spectrum. The hydrogen in the O-H groups 335

are confirmed at say 545 K (Figure 3) in the VB XPS spectra recorded at 537 K (on- and 336

off- Ce resonant), and at 548 K and 544 K (on- and off- Y resonant). It appears therefore 337

that we must include in the interpretation of the VB XPS spectra and in the formulation 338

of the electronic structure not only the B-site metal ions and the structural oxygen, the 339

oxygen vacancies and the oxygen from the water molecules, but also the protons in the 340

20

hydroxyl groups, which may be structural (localized) protons at low temperatures, and 341

“free” or mobile (delocalized) or polaron protons at higher temperatures when the hydro-342

gen bonds with the structural oxygen (including the oxygen ions that have become struc-343

tural by filling the oxygen vacancies) “melt”. Early studies which suggest a critical role 344

of protons on the electronic structure are based on optical spectroscopy. Sata et al. [20] 345

observed that the optical absorption edge shifted and the band gap increased depending 346

on the Yb-doping concentration in SrZrO3, suggesting that holes are formed at the top of 347

the VB due to Yb doping. In analogy to observations made on dry and wet CaZrSc-oxide 348

with optical spectroscopy, doped protons from moist environment will exchange with 349

doped holes and oxygen vacancies that have been formed by B-site cation doping [21]. 350

The absorption in CaZr0.95Sc0.05O3-δ is lower upon annealing in moist atmosphere, indi-351

cating that the doped proton exchanges with a hole or an oxygen vacancy [21]. An X-ray 352

spectroscopy study conducted on In-doped CaZrO3 suggested that proton states exist in 353

the bulk, and maybe also surface states, proton induced level at the top of the VB [22]. 354

We believe that this experimental in-situ study on the chemistry and the changes of the 355

electronic structure of proton conductors during hydration and annealing will be helpful 356

for understanding the conditions when the proton changes from a localized state to a de-357

localized state where proton conductivity actually sets on. 358

359

Conclusion 360

We investigated the chemical interactions of water with the BaCe0.8Y0.2O3-δ proton con-361

ductor under realistic working condition at elevated temperature and high water pressure 362

in-situ combining ambient pressure X-ray photoelectron spectroscopy and electrochemi-363

cal impedance spectroscopy. We observe three different temperature regimes for the pro-364

21

ton transport along with the structural change of BCY. Applying water pressure at inter-365

mediate temperature affects the oxygen core level, Ce and Y core level as well as the va-366

lence band spectra, revealing the filling of oxygen vacancies in BCY. The corresponding 367

increase in electric conductivity is paralleled by chemical shifts in the oxygen core level 368

spectra. Changes in the oxygen core level spectra, particularly emerging new spectral 369

weight, suggests that oxygen ions near Y3+ and Ce3+ can be distinguished from oxygen 370

ions near Ce4+. The filling of oxygen vacancies with oxygen from water vapor is impres-371

sively reflected by the substantial decrease of the leading peak in the valence band spec-372

tra. Difference spectra of Ce4p1/2 and Y3p1/2 resonant VB spectra show a shift of the lead-373

ing peak by 0.5 eV, which we interpret as that gap states of oxygen vacancies next to Y is 374

0.5 eV closer to the Fermi energy than the corresponding gap state of an oxygen vacancy 375

next to Ce. 376

377

22

Acknowledgement 378

The research leading to these results received funding from the European Community's 379

Sixth Framework Marie Curie International Reintegration Programme grant n.° 042095 380

(HiTempEchem - X-ray and Electrochemical Studies on Solid Oxide Fuel Cells and Re-381

lated Materials), Swiss National Science Foundation project # 200021-124812 (Effect of 382

lattice volume and imperfections on the proton-phonon coupling in proton conducting 383

lanthanide transition metal oxides: High pressure and high temperature neutron and im-384

pedance studies) and by the Korean-Swiss Cooperative Program in Science and Technol-385

ogy project “Spectroscopy on Photoelectrochemical Electrode Materials (SOPEM)”(Call 386

2010). We are grateful to Selma Erat (Empa, ETHZ) for assistance at the beamline, and 387

Songhak Yoon (Empa) for the high temperature XRD measurements. This research was 388

supported by the U. S. Department of Energy through the Sandia Laboratory Directed 389

Research and Development program under Contract DE-AC04-94AL85000. The ALS is 390

supported by the Director, Office of Science/BES, of the U.S. DoE, No. DE-AC02-391

05CH11231. 392

393

23

References 394 395 [1] Norby, T. Solid State Ionics 1999, 125, 1-11. 396

[2] Kreuer, K. D. Annu. Rev. Mater. Res. 2003, 33, 333-359. 397

[3] Malavasi, L.; Fisher, C. A. J.; Islam, M. S. Chem. Soc. Rev. 2010, 39, 4370-4387. 398

[4] Braun, A.; Ovalle, A.; Pomjakushin, V.; Cervellino, A.; Erat, S.; Stolte, W. C.; 399

Graule, T. Appl. Phys. Lett. 2009, 95, 224103. 400

[5] Chen, Q.; Braun, A.; Yoon, S.; Bagdassarov, N.; Graule, T. J. Eur. Ceram. Soc. 2011, 401

31, 2657-2661. 402

[6] Chen, Q.; Huang, T.-W.; Baldini, M.; Hushur, A.; Pomjakushin, V.; Clark, S.; Mao, 403

W. L.; Manghnani, M. H.; Braun, A.; Graule, T. J. Phys. Chem. C 2011, 115, 24021-404

24027. 405

[7] El Gabaly, F.; Grass, M.; McDaniel, A. H.; Farrow, R. L.; Linne, M. A.; Hussain, Z.; 406

Bluhm, H.; Liu, Z.; McCarty, K. F. Phys. Chem. Chem. Phys. 2010, 12, 12138-12145. 407

[8] Whaley, J. A.; McDaniel, A. H.; El Gabaly, F.; Farrow, R. L.; Grass, M. E.; Hussain, 408

Z.; Liu, Z.; Linne, M. A.; Bluhm, H.; McCarty, K. F. Review of Scientific Instruments 409

2010, 81, 086104. 410

[9] Grass, M. E.; Karlsson, P. G.; Aksoy, F.; Lundqvist, M.; Wannberg, B.; Mun, B. S.; 411

Hussain, Z.; Liu, Z. Rev. Sci. Instr. 2010, 81, 053106-053107. 412

[10] Hussain, Z.; Huff, W. R. A.; Kellar, S. A.; Moler, E. J.; Heimann, P. A.; McKinney, 413

W.; Padmore, H. A.; Fadley, C. S.; Shirley, D. A. J. Electron. Spectrosc. Relat. Phenom. 414

1996, 80, 401-404. 415

[11] Wu, J. Defect chemistry and proton conductivity in Ba-based perovskites. Disserta-416

tion, Caltech, 2004. 417

24

[12] Koel, B. E.; Praline, G.; Lee, H. I.; White, J. M.; Hance, R. L. J. Electron. Spectrosc. 418

Relat. Phenom. 1980, 21, 31-46. 419

[13] Dupin, J.-C.; Gonbeau, D.; Vinatier, P.; Levasseur, A. Phys. Chem. Chem. Phys. 420

2000, 2, 1319-1324. 421

[14] Deng, X.; Lee, J.; Wang, C.; Matranga, C.; Aksoy, F.; Liu, Z. Langmuir 2011, 27, 422

2146-2149. 423

[15] Mullins, D. R.; Overbury, S. H.; Huntley, D. R. Surf. Sci. 1998, 409, 307-319. 424

[16] Burroughs, P.; Hamnett, A.; Orchard, A. F.; Thornton, G. J. Chem. Soc., Dalton 425

Trans. 1976, 1686-1698. 426

[17] Higuchi, T.; Tsukamoto, T.; Matsumoto, H.; Shimura, T.; Yashiro, K.; Kawada, T.; 427

Mizusaki, J.; Shin, S.; Hattori, T. Solid State Ionics 2005, 176, 2967-2970. 428

[18] Higuchi, T.; Yamaguchi, S.; Sata, N.; Shin, S.; Tsukamoto, T. Jpn. J. Appl. Phys. 429

2003, 42, L1265-L1267. 430

[19] Higuchi, T.; Matsumoto, H.; Shimura, T.; Yashiro, K.; Kawada, T.; Mizusaki, J.; 431

Shin, S.; Tsukamoto, T. Jpn. J. Appl. Phys. 2004, 43, L731-L734. 432

[20] N. Sata, M. Ishigame and S. Shin: Solid State Ionics 86–88 (1996) 629. 433

[21] T. Higuchi, T. Tsukamoto, N. Sata, M. Ishigame, S. Yamaguchi, S. Shin, 434

Optical Absorption Spectra of Protonic Conductor CaZr0:95Sc0:05O3, Jpn. J. Appl. 435

Phys. 2003, 42(3), 1331-1332. 436

[22] T. Higuchi, S. Yamaguchi, K. Kobayashi, T. Takeuchi, S. Shin, T. Tsukamoto, Elec-437

tronic Structure in the Bulk State of Protonic Conductor CaZrO3 by Resonant Soft-X-ray 438

Emission Spectroscopy, Jpn. J. Appl. Phys. 2002, 41(2), L938-L940. 439

440

25

441

442

443 Supporting Information 444 445 446

447

448

Figure S1: X-ray diffractogram of powder from the BCY20 pellet recorded with Cu K-a 449

radiation, confirming phase purity (which phase is it?). 450

451

452

453

454 X-ray 455

26

456

457

We have monitored the evolution of the spectral weight at 593 K of either peak in UHV 458

condition in excess of 7 hours, collecting a total of 485 spectra. The heating profile is 459

shown in Figure S4 a. The transient of the relative peak heights of Ox and O-H and their 460

ratio O-H/Ox is displayed in Figure S4 b. The spectral weight of the hydroxyl groups is 461

rapidly increasing, and the spectral weight of the structural oxygen is decreasing to the 462

same extent. Overall, the ratio of the spectral weights increases sharply within 1 hour of 463

keeping the temperature around 593 K, and then remains virtually constant for the fol-464

lowing 6 hours. 465

We believe thus that the O-H groups are formed during annealing on the sample surface 466

from residual water from exposure to ambient conditions, thus causing a relative decrease 467

of the spectral weight from Ox. Comparison with thermogravimetry data (Figure S5 a, 468

adopted from BaCe0.85Yb0.15O3 in Ref. [11]) shows that surface water is released at 373 K 469

and structural water at 593 K, suggesting that structural oxygen Ox in BCY20 indeed 470

could be released from the sample at 593 K in UHV. 471

472

27

473

Figure S4: Heating profile (left) and evolution of spectral weight (right) of the O1s core 474

level XPS peaks for O-H, Ox, and their ratio O-H/Ox. 475

476

Figure S5: Ion current from water release (solid line from 350 K to 600 K) and XPS peak 477

height ratio for O-H/Ox. 478

479

28

480

Figure S6: Survey spectrum for the range of Ce 4p and Y 3p core levels. 481

360 320 280 240 2001x105

2x105

3x105

Ba 4sCe 4p1/2

Y 3p3/2

XP

S In

tensit

y [a

.u.]

Binding Energy [eV]

Y 3p1/2