Improved chemical and electrochemical stability …...Improved Chemical and Electrochemical...

1

Supplementary Information

Improved Chemical and Electrochemical Stability on Perovskite Oxides

by Less Reducible Cations at the Surface

Nikolai Tsvetkov1,2,†, Qiyang Lu1,3 †, Lixin Sun1,2, Ethan Crumlin4, and Bilge Yildiz1,2,3,*

1 Laboratory for Electrochemical Interfaces, 2 Department of Nuclear Science and Engineering, 3

Department of Material Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States.

4Advanced Light Source, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States

† These authors contributed equally to this work. * Corresponding author. E-mail: [email protected]

1. Parameters the chemical bath deposition of the additive cations at LSC surface

Additive cations were deposited from the aqueous solution of the metal chlorides at the

conditions shown in Table S1. All chlorides had purity of >99.0%, except for VCl3 and CoCl2,

(Table 1) and were purchased from Sigma-Aldrich. We examined the chemical composition at the

surface of the as-fabricated LSC and LSC-Me films using X-ray photoelectron spectroscopy. For

the LSC whose surfaces were not modified by the metal chloride solution, the excess Sr-rich layer

that forms during pulsed laser deposition (PLD) was removed by treating the films in dilute HCl

solution1-3. Because the metal chloride solution also etches the Sr-rich surface layer, removing

the Sr-segregated layer on LSC by dilute HCl provides a consistent comparison among these

samples.

The additive cation content (Me/(La+Sr+Co+Me) ratio) of 12-19% was chosen based on

the results of our previous study4. In Ref. 4, using Ti, as an example, we have shown that an

additive cation content lower than 10% did not efficiently prevent LSC degradation during

electrochemical testing at 420 °C and higher temperature, most possibly due to an insufficient

Improved chemical and electrochemical stability of perovskite oxides with less reducible cations at the surface

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT4659

NATURE MATERIALS | www.nature.com/naturematerials 1

© 2016 Macmillan Publishers Limited. All rights reserved.

http://dx.doi.org/10.1038/nmat4659

2

surface coverage with Ti. On the other hand, Ti content that is higher than 20% led to a decrease

of the degree of improvement at elevated temperatures.

Table S1. Deposition condition for chemical bath

Sample Reagent Molar concentration,

mM

Temperature,

°C

Time, s

LSC HCl 100 25 10

LSC-V12 VCl3, >97%,

Sigma-Aldrich

100 80 120

LSC-Co12 CoCl2, >97%,

Sigma-Aldrich

50 80 120

LSC-Ti15 TiCl4, >99.0%,

Sigma-Aldrich

3 25 20

LSC-Nb19 NbCl5,

>99.0%,

Sigma-Aldrich

5 80 20

LSC-Zr15 ZrCl4, >99.9%,

Sigma-Aldrich

20 80 120

LSC-Hf16 HfCl4,

>99.9%, Fluka

5 25 30

LSC-Al15 AlCl3, >99.9%,

Sigma-Aldrich

100 80 20

We evaluated and confirmed also that the LSC films that we report here are not

contaminated by chloride and thus their bulk or properties should not be affected by chloride that

might have been left on the surface during the surface modification procedure. Chloride precursors

are widely used in spray pyrolysis to deposit thin oxide films, and the presence and complicated

effects of chloride in such oxide films have been reported.5 However, our approach of making the

LSC films does not involve the pyrolysis step on a hot surface. The LSC films are made by PLD,

a route that yields highly clean and controllable compositions. Instead, we only process the very

top surface layer of the LSC films via a wet-chemical reaction at 25-80 oC. The LSC thin films


3

were dipped into dilute metal chloride solution, and during this process cation exchange/deposition

occurs, which results in a thin sub-monolayer coverage by these additive cations. The temperature

for this reaction is varied within 25-80°C just to tune the kinetics of the reaction to obtain the

targeted additive cation (Hf, Ti, Zr etc.) concentration at LSC surfaces. At these low temperatures,

chloride ions cannot diffuse and penetrate into the solid state LSC films. Then we use deionized

water to wash off the remaining chloride salt thoroughly from the sample surface. Since the

chlorides we use are all soluble in water, this step results in a clean surface free from chlorine.

Afterwards we clean the sample again using isopropanol. We also anneal the sample up to 260 °C,

which is meant for removing the remaining water and isopropanol on the sample surfaces before

the actual experiment. Therefore, our wet-chemical route to modify LSC surfaces at low

temperatures is clearly different from the spray pyrolysis to form the entire oxide film. This

difference guarantees that the LSC intrinsic properties were not affected by chlorine contamination.

We prove this point by our XPS measurements, shown in Figure S1 below. We performed

XPS measurements on multiple LSC samples after treating their surfaces with metal chloride

solutions, searching for the possible existence of Cl remaining at the LSC surfaces. As an example,

data for the LSC samples after TiCl4 treatment and after dilute HCl treatment are shown in Figure

S1. The Cl 2p signal was not found in the expected binding energy range (197 eV~201 eV,

according to literature). This proves that the Cl is either not present on the sample surface, or that

its concentration is below the detection limit of XPS (~1% from our estimation based on the

ionization cross section of Cl 2p). Thus we do not think that the conclusions shown in the paper

were affected by possible Cl contamination from the deposition process.


4

Figure S1 X-ray photoelectron spectra showing the expected Cl 2p region for an as-deposited

LSC film, and two LSC films treated with dilute HCl and with TiCl4 solution. The Cl 2p

photoemission peak was not observed in any of these three samples shown as representative of

all the LSC films whose surfaces were modified by dilute HCl or by metal chloride solutions.

2. Surface morphology on the LSC and LSC-Me films

Surface morphologies of LSC, and LSC films threated with Co, V, Ti, Nb, Zr, Hf, and Al chloride

solutions were investigated by force microscopy (AFM) (Fig. S2). As can be seen from the AFM

images (Figures S2 a-k), these surfaces do not show any special or evident morphological features

associated with deposited cations within the resolution limits. Some difference can be observed

for the LSC-V12 sample which was treated with high VCl3 concentration at 80 °C to be able to

deposit the desired amount of V at the surface, and this resulted in partial etching of the surface

and increased roughness. Based on these AFM observations, we think that the metal additive is

deposited mainly in the form of a thin and smooth wetting layer of metal oxide layer, but not

necessarily a complete monolayer.


5

Figure S2. Surface topography characterized by atomic force microscopy on the (a) LSC, (b) LSC-

Co12, (c) LSC-V12, (d) LSC-Ti15, (e) LSC-Nb14, (f) LSC-Zr15, (j) LSC-Hf16, and (k) LSC-

Al15 after the deposition of the metals from the chloride solutions as described in Table S1.

Figure S3. Surface topography characterized by atomic force microscopy on the (a) LSC, (b) LSC-

Co12, (c) LSC-V12, (d) LSC-Ti15, (e) LSC-Nb14, (f) LSC-Zr15, (j) LSC-Hf16, and (k) LSC-

Al15 films that were electrochemically tested at 500 °C for 30 hours, as shown in Fig. 1a.

LSC

400 nm

LSC-Co12 LSC-V12 LSC-Ti15

LSC-Nb14 LSC-Zr15 LSC-Hf16

ba c d

e f j LSC-Al10k

LSC

1 µm

LSC-Co12 LSC-V12 LSC-Ti15

LSC-Nb14 LSC-Zr15 LSC-Hf16

ba c d

e f j LSC-Al15k


6

We note that, in theory, the chemistry of Zr and Hf should be very similar. However, it is

clear that LSC-Zr15 has been less stable both electrochemically (Fig. 1) and chemically (Fig. S3).

We believe that the following two reasons might be the origin of this non-trivial difference between

LSC-Zr and LSC-Hf:

1) One uncertainty is the exact structure of LSC surfaces modified by the less reducible

cations. It is possible that LSC-Zr and LSC-Hf possessed different distribution and structures at

the surface of LSC, for example, a uniform versus clustered distribution of the deposited cations

from the two different solutions onto the surface of LSC. If Hf is deposited very uniformly from

the chloride solution while Zr tends to cluster, then this could lead to LSC-Hf having better stability

because more of the surface area is affected. We tried to resolve this, but within AFM resolution

we cannot see any difference between the Zr-modified or Hf-modified surfaces (Fig. S2).

2) Another uncertainty is the relative stability of the Zr and Hf cations at the surface of

LSC. Zr and Hf cations are quite similar in terms of their oxidation state, ionic radii and oxygen

vacancy formation enthalpy of the corresponding binary oxides, but their atomic numbers (Z) are

quite different, i.e. Hf is much heavier than Zr. This can potentially lead to different diffusion

coefficients (in particular due to the different attempt frequencies) between Zr and Hf cations in

the perovskite matrix. According to our experimental observations, these less reducible cations are

effective in preventing Sr segregation only when they remain at near-surface region of LSC at a

high-enough fraction. It could be that Zr partially dissolves by diffusion into LSC relatively faster

than Hf, so that in LSC-Zr sample the concentration of additive cations at near-surface region

decreases with testing time which leads to more performance degradation compared with LSC-Hf

sample.

Both of these possibilities could lead to the observed difference performance of LSC-Hf

and LSC-Zr. To resolve both of the uncertainties above, further work should be done, such as

resolving surface structure of the deposited cations, or use different deposition techniques, and

quantify diffusion coefficients of Hf and Zr in the same LSC lattice. However, the key point of our

paper being “decreasing the reducibility of the surface alone can improve the compositional and

electrochemical stability of perovskite catalysts” is not impacted by why Zr and Hf behave

differently.


7

3. Calculation of surface oxygen exchange coefficient

The half cells for electrochemical tests had the LSC or LSC-Me thin films as the working electrode

and the pasted porous Ag as the counter electrode. A typical impedance response of the cells with

the LSC-Hf16 thin film electrodes, and the equivalent circuit6,7 used for modeling the EIS data is

shown in Figure S4. The EIS data obtained on the cells were modeled with circuits consisting of

two R//CPE (a resistor in parallel to a constant phase element) (Fig. S4)6,8. The larger arc in the

lower frequency region is known to reflect the oxygen exchange reaction at the cathode surface6,9.

We obtained the surface polarization resistance, RS, values from the low frequency impedance

response, and used those values to calculate the surface oxygen exchange coefficient, kq, using the

following formula10:

𝑘𝑞 =𝑘𝐵𝑇

4𝑒2𝑐0𝑅𝑆,

where kB is the Boltzmann constant, T is temperature, e is the electronic charge, and co is the total

concentration of lattice oxygen determined according to data from Mizusaki et al.11.

The kq values on LSC and LSC-Me (Me concentration 12-16%) thin film cathodes as a function

of time at 530 °C in air.

Figure S4. (a) Representative electrochemical impedance spectrum of the cell with the LSC-Hf16

thin film electrode at 530 oC in air, and an equivalent circuit used to fit the experimental data

adopted from ref 3. Solid line was obtained by fitting the equivalent circuit parameters to the data.


8

4. Analysis of Sr 3d spectra

The analysis of the Sr 3d photoelectron core level peak allows one to quantify the atomic

concentration of Sr at the film surface, and to assess/differentiate the Sr binding environments in

the perovskite lattice and in non-lattice phase at the surface (Fig. S5)1. The contributions to the Sr

3d photoelectron spectrum were found to arise from the perovskite lattice-bound Sr ([Sr]Lattice) at

the lower binding energies and from the surface-bound Sr ([Sr]Non-lattice) at the higher binding

energies. [Sr]Lattice contribution was attributed to the Sr chemical environment of the perovskite

LSC1. The chemical environment of [Sr]Non-lattice could be attributed to the formation of species

such as SrCO3, SrO, or Sr(OH)21.

Figure S5. Representative Sr 3d x-ray photoelectron spectra recorded on the LSC film at 550 oC

and at oxygen partial pressure of 10-6 Torr.

5. Temporal stability of LSC chemical composition during AP-XPS measurements

Here we demonstrate that each data point in Figure 2 is not showing any temporal characteristics.

Rather, we show below that each data point represents the equilibrium state at each condition in

our experiments. This can be seen from the Figure S6 below, which shows the photoelectron

spectra of La, Sr and Co cations. There are two sets of data shown for each, one is denoted as “0

h” which we take as our initial condition (but in fact this data is taken after 30 min of equilibration

upon reaching that temperature or pressure), and another one is denoted as “1.5 h” which is taken

1.5 h after the data shown for “0 h”. The complete overlapping of these two data sets (at 0h and at

1.5 h) demonstrate the fact that the results shown in Figure 2 are not temporal effects but rather

dominated by the changes in the temperature and oxygen pressure.


9

Figure S6. Sr 3d, La 4d and Co 3p photoemissions (measured by APXPS) at (a-c) 450 °C, 10-6

Torr pO2 and (e-f) 550 °C, 10-6 Torr pO2, collected at 0 h and 1.5 h elapsed times. The overlapping

of two spectra for all three cations means that there is no detectable change occurring during the

1.5 h duration. Note: the data at “0 h” are actually collected after 30 minutes of equilibration at the

stated condition.

6. Evolution of the Co L3-edge position at different temperature and oxygen partial pressures

As shown in Figure 3b in main text and also in Figure S7, a 0.4 eV difference was observed

comparing LSC and LSC-Hf15 at the first measurement condition (300 °C, 0.76 Torr). According

to Hu et al.12, this means the difference in Co oxidation state between these two samples is ~0.4.

However, the Co oxidation state can also be affected by the concentration of Sr in LSC lattice.

Here we argue that this 0.4 eV difference cannot be from the difference in [Sr]lattice. From Figure

2(c) in the main text, the difference in [Sr]lattice/[Co] ratio is only ~0.05, which corresponds to a

change in Co oxidation state of only 0.05, which is much smaller than what we observed from in

situ XAS measurements (i.e. 0.4). Therefore, we can conclude that this difference in Co oxidation

state is from the effect of additive cations, most likely the suppressed formation of oxygen

vacancies, rather than the difference in [Sr]lattice among the samples.


10

It was possible to detect the instability of the bare LSC surface chemistry already at as low

temperatures as 300 °C, characterized by a significant increase of the non-lattice Sr at the film

surface (Fig. 2b). This corresponds to the initial phase of LSC surface degradation. So, for

suppressing the segregation of non-lattice Sr, this initial phase must be stopped or delayed to much

higher temperatures. This is why we mainly compared the signatures of Co L2,3 edge XAS, O K

edge XAS and the VB structure at 300 °C in Figs. 3, 4 among the unmodified and modified LSC

samples. As shown in Fig. 2c, the [Sr]Lattice increased significantly upon going to higher

temperatures on LSC and LSC-Ti3 due to much larger driving force and/or faster kinetics for Sr

segregation on those samples. More Sr cations in LSC lattice means more p-type doping, which

leads to a more oxidized surface, as seen in Fig. S7 and in Figs. 3, 4 for LSC and LSC-Ti3 samples

at temperatures higher than 300 °C. Since this evolution of lattice Sr concentration complicates

the interpretation, we limited our discussion on the effect of Ti and Hf to the first measurement

condition (300 °C, 0.76 Torr). At this condition, the differences in [Sr]Lattice among the four samples

were rather small (within only ~0.05). Therefore, the Co L-edge, O K-edge and VB structure

differences really arise from the effect of the additive cations on the oxygen vacancy concentration.

Figure S7. The Co L3-edge peak position measured at different conditions for LSC, LSC-Ti3,

LSC-Ti15 and LSC-Hf15.

7. Reducibility of the Co cation on LSC and LSC-Me films

To support XAS measurements, we compared reducibility of the LSC and LSC-Me films by

measuring the Co 2p photoelectron spectra at 450 and 550 oC in ultra-high vacuum (~10-9 Torr)

(Fig. S8) using a conventional laboratory X-ray source.


11

At 550 °C, the Co 2p main peak satellite peak at ~ 786 eV indicates the presence of reduced Co2+.13

This satellite peak was clearly seen on LSC but not on LSC-Ti15, LSC-Nb14, LSC-Zr15, LSC-

Al15 or LSC-Hf16. The surface modification with V, on the other hand, facilitated the Co

reduction as apparent with a clear satellite peak (Fig. 3d). On LSC-V12, the Co2+ satellite peak

becomes apparent already at 450 °C, while no satellite peak was observed for the bare LSC film

at the same temperature. These observations mean that adding the less reducible cations of Ti, Nb,

Zr, Al or Hf, made Co in LSC also less reducible (LSC surface being more oxidized), while V

enhanced the reducibility of Co on LSC.

Figure S8. (a) Co 2p photoelectron spectra on LSC and LSC-Me films recorded at (a) 550 oC and

(b) 450 oC in ultra-high vacuum.

8. Valence band spectra at the bare and modified LSC surfaces

We next show that the evolution of the valence band (VB) supports also a more oxidized

surface when LSC is modified by Hf and Ti, consistent with the Co L2,3-edge XAS above. The VB

spectra of LSC, LSC-Ti3, LSC-Ti15, and LSC-Hf16 films at different conditions are shown in Fig

S9. These four samples can be divided into two groups based on the similarity of the VB spectra.

For the LSC and LSC-Ti3, the intense peak located at around 1.5 eV at 300 °C arises from the Co

792 788 784 780 776

LSC

LSC-Ti15

LSC-Nb19

LSC-Zr15

LSC-Hf16

LSC-Al15

Binding Energy, eV

No

rmalized

In

ten

sit

y, a.u

.

ba Co 2p3/2 at 550 oC; UHV Co 2p3/2 at 450 oC; UHV

792 788 784 780 776

Binding Energy, eV

LSC-V12

LSC-Co12

LSC

No

rmalized

In

ten

sit

y, a.u

.2p3/2

Co2+

sat.

2p3/2

Co2+

sat.


12

t2g states hybridized with the O 2p states14,15. On the other hand, on LSC-Ti15 and LSC-Hf16, this

peak was absent at the same condition. The intensity of this peak is tied to the number of electrons

at the Co t2g orbital15,16, and provides information on the Co oxidation state. Therefore, this intense

t2g peak on LSC and LSC-Ti3 suggests more t2g electrons and a lower Co oxidation state. The

absence of this peak on LSC-Ti15 and LSC-Hf16 indicates that Co is more oxidized on these

samples than on LSC, in line with the Co L-edge XAS results.

Figure S9. Evolution of the valence band structure from X-ray photoelectron spectra

measured in situ on (a) LSC, (b) LSC-Ti3, (c) LSC-Ti15, and (d) LSC-Hf16. The arrow indicates

the low energy peak which reflects the hybridization of Co t2g states with the O 2p orbital. The

greater the intensity of this peak, the more electrons in the t2g states of Co.

6 4 2 0 -2

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

Binding Energy, eV

No

rma

lize

d I

nte

ns

ity

, a.u

.

6 4 2 0 -2

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

Binding Energy, eV

No

rmalized

In

ten

sit

y, a.u

.

6 4 2 0 -2

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

No

rma

lize

d I

nte

ns

ity

, a.u

.

Binding Energy, eV

6 4 2 0 -2

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

Binding Energy, eV

No

rma

lize

d I

nte

ns

ity

, a.u

.

a

d

Co t2g-O 2p

c

Co t2g-O 2p

bCo t2g-O 2p

Co t2g-O 2p

LSC LSC-Ti3

LSC-Ti15 LSC-Hf16


13

At 450-550 °C, the VB structure of LSC, LSC-Ti3, LSC-Ti15 and LSC-Hf16 become

similar, with the disappearance of the Co t2g-O 2p peak on LSC and LSC-Ti3. The variations of

the [Sr]Lattice as a function of temperature can explain this behavior. Raising the temperature up to

450-550 °C substantially increases the [Sr]Lattice/[Co] (Fig. 2c) on LSC and LSC-Ti3. The larger

Sr doping level in the near-surface region is charge compensated by Co becoming more oxidized.

This decreases the intensity of Co t2g-O 2p peak on the LSC and LSC-Ti3 at 450-550 °C. As a

result, the difference in the intensity of the Co t2g-O 2p peak among the samples at 300 oC is mainly

due to a difference in oxygen vacancy concentrations. On the other hand, at 450-550 °C the

enrichment of [Sr]Lattice near the surface of LSC and LSC-Ti3 governs the shape of the VB near the

Co t2g-O 2p peak.

9. O K-edge spectra at the bare and modified LSC surfaces

It is well know that for LSC, there is significant hybridization of the Co 3d and O 2p orbitals

due to the covalent characteristics of this material17. Therefore, the O K-edge XAS also provides

insights into the electronic structure of LSC surfaces. Fig. S10 summarizes the evolution of the O

K-edge spectra taken in situ on LSC, LSC-Ti3, LSC-Ti15, and LSC-Hf16 films as a function of

measurement conditions. In line with the unchanged valence band structure (Figs. S9 c, d), the O

K-edge spectra on LSC-Ti15 and LSC-Hf16 remained unaltered throughout the measurements. A

sharp pre-edge peak at around 528 eV (shown by the arrows in Fig. S9) indicates the existence of

O 2p ligand holes18. The O 2p ligand hole peak is clearly evident on LSC-Ti15 and LSC-Hf16

surfaces. In contrast, this peak was absent on LSC and LSC-Ti3 at 300°C and pO2 of 0.76 Torr.

The presence of this peak indicates increased p-type doping and a more oxidized Co valence19 on

LSC-Ti15 and LSC-Hf16 compared to that on LSC and LSC-Ti3.

The difference in the O K-edge XAS spectra among the four samples disappears by 550

°C and high pO2 of 0.76 Torr. We attribute this evolution of the O K-edge spectra to the increase

of the [Sr]lattice concentration on LSC and LSC-Ti3, using the same reasoning as we had for the

evolution of the VB spectra above.


14

Figure S10. O K-edge spectra of (a) LSC, (b) LSC-Ti3, (c) LSC-Ti15, and (d) LSC-Hf16

films at different temperatures and oxygen partial pressures. The dashed lines in each plot mark

the position of the O 2p ligand hole peak. The presence of this peak indicates p-type doping and

therefore an increased Co oxidation state, as seen on LSC-Ti15 and LSC-Hf16.

10. Bonding Environment of additive cations

The presence as dopants, such as Hf, Zr, Nb or Al, into the Co-site of LSC alters the oxygen

vacancy formation energy at LSC surface because the bonding between the added metal cations

and oxygen is much stronger than the Co-O bonds, and thus, more difficult to break. The oxygen

vacancy formation energy in the binary oxides of Zr, Nb, Ti, Hf, and Al (e.g. TiO2 and HfO2) is

524 528 532 536 540 544 548

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

No

rma

lize

d I

nte

ns

ity

, a

.u.

Photon Energy, eV

524 528 532 536 540 544 548

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

No

rma

lize

d I

nte

ns

ity

, a.u

.

Photon Energy, eV

524 528 532 536 540 544 548

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

No

rmalized

In

ten

sit

y, a.u

.

Photon Energy, eV

524 528 532 536 540 544 548

Photon Energy, eV

300 C 0.76 Torr

450 C 10-6

Torr

550 C 10-6

Torr

550 C 0.76 Torr

No

rmalized

In

ten

sit

y, a.u

.a b

O 2p hole

c d

O 2p hole

O 2p holeO 2p hole

LSC LSC-Ti3

LSC-Ti15 LSC-Hf16


15

higher compared to that in LSC. Therefore, it is expected that the bonding between added metal

cations (Ti, Hf, etc.) and oxygen is much stronger than the Co-O bonds, and thus, more difficult

to break. It is reasonable to expect that such stronger bonding increase the oxygen vacancy

formation energy (and decrease the oxygen vacancy concentration) on the LSC surfaces modified

by these cations.

Even though Ti was found to be incorporated into the perovskite lattice at elevated

temperatures, the exact composition of the very surface layer of the modified LSC still needs

further investigation. We also consider the possibility that a thin layer of the SrTiO3 can be formed

at the top surface of LSC. However, it is highly unlikely that there was no intermixing between

the added (Ti) and host (Co) cations, especially at elevated

temperatures. Therefore, we believe that the most likely Ti

bonding environment is Ti substituting the Co site in LSC

perovskite lattice, forming Ti-doped LSC on the top surface

layer.

Figure S11. Ti L2,3 XAS of (a) SrTiO3, (b) rutile TiO2, and (c)

anatase TiO2. The solid lines give the estimated line strength of

the dipole transition probability 3d0-2p53d1 in (a) Oh and (b) D2h

symmetry. Reprinted from Ref.20 with the permission of

American Physical Society.

11. Discussion on influence of additive cations on surface polarity

We also considered the possible effect of the added cations on the polarity of LSC surface.

As it was shown by Harrison, the polar surface of (100)-terminated (La,Sr)MnO3 (LSM) can be in

part contributing to the Sr-enrichment at the surface21. The surface charges can be eliminated by

depleting La and enriching Sr near and at the surface of LSM, and the same reasoning should be

able to be applicable to LSC. However, oxygen non-stoichiometry is left out from that picture.

Considering the (100) CoO2 termination on La1-xSrxCoO3 in the absence of oxygen vacancies, the


16

surface takes a charge of –(1–x)e- per B-site, thus introducing instability due to the charged surface

termination. The surface charges can be relieved when positively charged oxygen vacancies are

formed in this BO2 layer. Consistent with this, it is also known that oxygen vacancies are enriched

at perovskite surfaces22. In our case, one may argue that the positively charged 𝑀𝑒𝐶𝑜∙ (such as the

added Hf, Ti, Zr cations) in CoO2 layer can reduce the charge and thus stabilize the surface.

However, follow the same logic, since the oxygen vacancy is doubly negatively charged (𝑉𝑂∙∙) while

𝑀𝑒𝐶𝑜∙ has only single negative charge, one should expect that oxygen vacancies have stronger

effect of relieving polar surface than added cations. That is to say, an increasing in surface oxygen

vacancy concentration should lead to a decreased charge in CoO2 layers and consequently more

stable surface. However, this is opposite to our experimental observations, which explicitly

showed that LSC surfaces that are modified by the less reducible cations, i.e. smaller concentration

of oxygen vacancies on LSC, had much better stable surface chemistry. Therefore, we believe that

the enhanced surface stability cannot be explained by the effect of modifying polar surface of LSC

with additive cations, or the effect is rather small.

12. Repeatability of electrochemical measurements

For selected surface compositions, 2-3 samples were tested electrochemically (Fig. S12). Variation

of the measured kq values ranged from +/-10% for bare LSC to +/-40% for LSC-Hf, in part because

of the variability in the PLD-prepared base LSC films and, in part, because of the different

concentrations of additives that were put at the surface. The results for LSC, LSC-Ti15, and LSC-

Hf16 shown in Fig. 2 represent a batch of samples that were deposited at the same time to ensure

consistency of the base film. Regardless of the batch of samples, the general trend shown in Fig.

2 reveals that the significantly better stability and higher kq are consistent among all samples with

the less reducible additive cations at the surface.


17

Figure S12. The oxygen surface exchange coefficient, kq, quantified from electrochemical

impedance spectroscopy measurements over time at 530 °C in air, for the LSC and LSC-Me films

deposited at different PLD sessions and with different additive Me concentrations.

13. Effect of concentration of the less reducible cation at the LSC surface

In our earlier work, we performed a detailed electrochemical study as a function of

concentration of surface additives, using specifically Ti as an example: see Fig. S13b below from

Ref. 4. We found in fact a similar “volcano” behavior (see Fig. S13c) with increasing concentration

of Ti at the surface, i.e. there was an optimum concentration of Ti on LSC which yields the most

stable LSC surfaces. A low Ti or a too high Ti concentration does not improve much the stability

of electrochemical performance. The current paper can shed light into that behavior based on the

same explanation (and spectroscopic proof we provide here). As a result, now, the good/stable

behavior of LSC-Ti15 compared to the bad/unstable behavior of LSC-Ti3 electrochemically tested

in our previous study can be explained also based on the Co oxidation states reported in Figure 3b

of the current paper. (Note the LSC-Ti2 and LSC-Ti10 in Ref. 4 corresponds to the same

compositions as on LSC-Ti3 and LSC-Ti15, respectively, in this paper when considering the same

emission angle in XPS as in this paper). In fact, the reason that we had for choosing on average a

15% of additive cation at the surface in this study was based on our previous finding that LSC-Ti

had optimal stability with 15% Ti. Now, the only variable is the oxygen vacancy formation

enthalpy (modulated by the surface additives, as indicated in the x-axis of Figure 6) while the %

is nearly fixed around 15%.

0 10 20 30 40 5010

-10

10-9

10-8

10-7

LSC-Hf9, set 1

LSC-Hf16, set 2

Time, h

LSC-Nb19, set 3

LSC-Nb22, set 4

LSC-Al15, set 4

LSC-Al17, set 5

LSC, set 1

LSC, set 2

LSC, set 3

kq, cm

/s


18

Figure S13 (Adapted from Ref. 4). (a) Representative electrochemical impedance spectra of the

cells with the LSC and LSC-T10 thin film electrodes at 530 oC in air, and an equivalent circuit

used to fit the experimental. Solid lines were obtained by fitting the equivalent circuit parameters

to the data. (b) The oxygen surface exchange coefficient, kq, over time at 420-530 oC, for LSC and

LSC-T films with 2-13% Ti at the surface. The y-axis shows the kq values measured after 90 hours

of testing at 530°C in air. Figures S13 a,b are reproduced with permission from The Royal Society

of Chemistry. (c) The dependence of oxygen exchange coefficient, kq, of LSC-Ti on Ti content

measured by XPS at the surface. Lines connecting that data in (b and c) are a guide to the eye.

Please note that the 2, 6, 10, 13% of Ti (in Ref. 4) corresponds to 3, 9, 15 and 19% of Ti by using

the same emission angle (20o) as used in the current paper.

0 3 6 9 12 1510

-10

10-9

10-8

kq,

cm

/s

Ti Content, %

c


19

References

1 Cai, Z., Kubicek, M., Fleig, J. & Yildiz, B. Chemical Heterogeneities on La0.6Sr0.4CoO3−δ Thin Films—Correlations to Cathode Surface Activity and Stability. Chemistry of Materials 24, 1116-1127.

2 Jung, W. & Tuller, H. L. Investigation of surface Sr segregation in model thin film solid oxide fuel cell perovskite electrodes. Energy Environ. Sci. 5, 5370-5378 (2012).

3 Kubicek, M., Limbeck, A., Frömling, T., Hutter, H. & Fleig, J. Relationship between Cation Segregation and the Electrochemical Oxygen Reduction Kinetics of La0.6Sr0.4CoO3−δ Thin Film Electrodes. J. Electrochem. Soc. 158, B727-B734 (2011).

4 Tsvetkov, N., Lu, Q. & Yildiz, B. Improved electrochemical stability at the surface of La0.8Sr0.2CoO3 achieved by surface chemical modification. Faraday Discussions (2015).

5 Rupp, J. L. M., Drobek, T., Rossi, A. & Gauckler, L. J. Chemical Analysis of Spray Pyrolysis Gadolinia-Doped Ceria Electrolyte Thin Films for Solid Oxide Fuel Cells. Chemistry of Materials 19, 1134-1142.

6 Adler, S. B. Factors Governing Oxygen Reduction in Solid Oxide Fuel Cell Cathodes. Chemical Reviews 104, 4791-4844 (2004).

7 Jiang, S. P. A comparison of O2 reduction reactions on porous (La,Sr)MnO3 and (La,Sr)(Co,Fe)O3 electrodes. Solid State Ionics 146, 1-22 (2002).

8 Yamada, A. et al. Ruddlesden-Popper-Type Epitaxial Film as Oxygen Electrode for Solid-Oxide Fuel Cells. Advanced Materials 20, 4124-4128 (2008).

9 Jamnik, J. & Maier, J. Generalised equivalent circuits for mass and charge transport: chemical capacitance and its implications. Physical Chemistry Chemical Physics 3, 1668-1678 (2001).

10 Baumann, F. S., Fleig, J., Habermeier, H.-U. & Maier, J. Impedance spectroscopic study on well-defined (La,Sr)(Co,Fe)O3−δ model electrodes. Solid State Ionics 177, 1071-1081 (2006).

11 Mizusaki, J., Mima, Y., Yamauchi, S., Fueki, K. & Tagawa, H. Nonstoichiometry of the perovskite-type oxides La1−xSrxCoO3−δ. Journal of Solid State Chemistry 80, 102-111 (1989).

12 Hu, Z. et al. Difference in spin state and covalence between La1-xSrxCoO3 and La2-xSrxLi0.5Co0.5O4. Journal of Alloys and Compounds 343, 5-13 (2002).

13 Vaz, C. A. F., Prabhakaran, D., Altman, E. I. & Henrich, V. E. Experimental study of the interfacial cobalt oxide in Co3O4-Al2O3(0001) epitaxial films. Physical Review B 80, 155457 (2009).

14 Abbate, M. et al. Electronic structure and spin-state transition of LaCoO3. Physical Review B 47, 16124-16130 (1993).

15 Mizokawa, T. et al. Photoemission and x-ray-absorption study of misfit-layered (Bi,Pb)-Sr-Co-O compounds: Electronic structure of a hole-doped Co-O triangular lattice. Physical Review B 64, 115104 (2001).

16 Tsukada, I. et al. Ferromagnetism and Large Negative Magnetoresistance in Pb Doped Bi–Sr–Co–O Misfit-Layer Compound. Journal of the Physical Society of Japan 70, 834-840 (2001).

17 Suntivich, J. et al. Estimating Hybridization of Transition Metal and Oxygen States in Perovskites from O K-edge X-ray Absorption Spectroscopy. The Journal of Physical Chemistry C 118, 1856-1863 (2014).

18 Moodenbaugh, a. et al. Hole-state density of La1-xSrxCoO3-δ (0~x~0.5) across the insulator/metal phase boundary. Physical Review B 61, 5666-5671 (2000).

19 Abbate, M. et al. Controlled-valence properties of La1-xSrxFe03 and La1-xSrxMnO3 studied by soft-x-ray absorption spectroscopy. Physical Review B 46, 4511-4519 (1992).

20 van der Laan, G. Polaronic satellites in x-ray-absorption spectra. Physical Review B 41, 12366-12368 (1990).

21 Harrison, W. A. Origin of Sr segregation at La1-xSrxMnO3 surfaces. Physical Review B 83, 155437 (2011).


20

22 Chueh, W. C. & Haile, S. M. Electrochemistry of Mixed Oxygen Ion and Electron Conducting Electrodes in Solid Electrolyte Cells. Annual Review of Chemical and Biomolecular Engineering 3, 313-341, doi:10.1146/annurev-chembioeng-073009-101000 (2012).


Date post:	27-May-2020
Category:	Documents
Upload:	others
View:	9 times
Download:	0 times

Improved chemical and electrochemical stability …...Improved Chemical and Electrochemical...

Documents