SUPPLEMENTARY INFORMATION

Transcription

1 In the format provided by the authors and unedited. DOI: /NMAT4879 Real time impedance monitoring of oxygen reduction during surface modification of thin film cathodes Ghislain M. Rupp 1,*, Alexander K. Opitz 1, Andreas Nenning 1, Andreas Limbeck 1, Jürgen Fleig 1 1 Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9, Vienna, A 1060, Austria * Author for Correspondence, ghislain.rupp@tuwien.ac.at Decoration of LSC with La and Sr (sc) and detailed impedance interpretation Fig. S1 Representative Nyquist plots normalized to the surface area of the working electrode for asprepared thin films decorated with particles from (a) La and (b) Sr (sc) target measured in situ at 450 C and mbar p(o 2 ). The legends specify the total number of laser pulses after which the impedance spectra were recorded. (c) equivalent circuit used to fit impedance data shown Fig. 3 and Fig. S1 The high frequency intercept of impedance spectra (>100 Hz, ~ Ωcm 2 ), see Fig. 2 and S1a/b, is predominantly attributed to the oxide ion conduction in the YSZ single crystal electrolyte. Resistive contributions from wiring ( 3 Ω) could be neglected and the electronic sheet resistance was sufficiently small to guarantee a homogenous polarization of the entire working electrode (WE) despite very local contact by a Pt needle tip (see below). All impedance spectra obtained in this study were fitted to the equivalent circuit shown in Fig. S1c. The known temperature dependent ionic conductivity of YSZ 1 was used to determine the temperature from R YSZ and to adjust the temperature to 450 C throughout the entire study. Variations of R YSZ between the different impedance spectra thus reflect small temperature fluctuations (± 3 C). At lower frequencies ( Hz) the impedance spectra become semicircle like and can be fitted to a resistor in parallel with a constant phase element (exponential fitting parameter n > NATURE MATERIALS 1

2 0.98). The resistance is associated with the oxygen surface exchange reaction of the LSC working electrode; the ion transport resistance in the LSC electrode can be neglected under these and similar conditions 2. The capacitive element in parallel to R surf, exch has a value of > 20 mf/cm² and corresponds to the chemical capacitance of the LSC WE cf. interpretation in Ref. 2. The chemical capacitance describes the ability of the material to store chemical energy by undergoing stoichiometric changes and is therefore actually a bulk property. The bulk chemical capacitance of the LSC film remained constant at 1243 ± 64 F/cm³ throughout the study, irrespective of any surface modification. This is not surprising, since the surface of the 140 nm thick LSC film was only decorated by small amounts of oxides with a much lower non stoichiometry than LSC. It should be emphasized that the highly active porous counter electrode (CE) does not noticeably contribute to the impedance spectra. Even for the lowest electrode impedance (96 ± 17 Ωcm², after Co decoration) we do not have any indication of a CE effect since the shape of the impedance spectra remained unchanged and the chemical capacitance of the low frequency semicircle was not affected by Co decorations. The maximal impedance of the CE was deduced by another IPLD measurement series (Fig. S2), where up to 188 laser pulses of porous LSC were deposited on top of the dense LSC WE and the surface exchange resistance thus decreased by more than an order of magnitude to 16 Ωcm². This indicates that our much thicker porous CE has an even smaller polarization resistance. The electronic sheet resistance R eon, sheet of the dense LSC thin film WE was estimated by with an LSC film thickness t of ~140 nm, a needle tip diameter of ~5 µm (used to contact the WE), a sample size (5 x 5 mm) of ~ 5 mm diameter and an electronic conductivity σ of 1015 S/cm, which was measured on similar LSC thin films at 450 C by the Van der Pauw Method, see Ref. 2. An electronic sheet resistance of 69 Ω is thus obtained. Since, R eon, sheet is more than one order of magnitude lower than the sum of the ionic transport resistance of YSZ at 450 C (750 Ω) and the oxygen exchange resistance (~1000 Ω), it can be concluded that the entire electrode is electrochemically active, i.e. homogenously polarized. NATURE MATERIALS 2

3 Decoration of dense LSC by porous LSC to estimate CE influence Fig. S2 (a) Representative Nyquist plots normalized to the surface area of the working electrode for as prepared thin films decorated with particles from an LSC target and measured in situ at 450 C and mbar p(o 2). The legend specifies the total number of laser pulses after which the impedance spectra were recorded. (b) Decrease of the surface exchange resistance with increasing amount of laser pulses and independency of the chemical capacitance. In order to estimate the influence of the porous LSC counter electrode on the low frequency semicircle, porous LSC was continuously deposited on an as prepared dense LSC top electrode at 450 C in mbar p(o 2 ). Three impedance spectra, each taking 10 minutes, were measured after the deposition of LSC using 1, 2, 5, 10, 20, 50 and 100 laser pulses at a repetition rate of 1 Hz. Representative Nyquist plots normalized to the electrode surface area are shown in Fig. S2a and the legend shows the total amount of laser pulses after which the impedance spectra were recorded. The high frequency intercept is not influenced by the LSC deposition and was again used to determine and adjust the temperature of the sample. Already after the deposition of 1 single laser pulse the diameter of the low frequency semicircle decreases significantly. This means that the catalytic activity for oxygen surface exchange of the top electrode is enhanced by partly covering it with fresh LSC particles. This is in agreement with earlier results 3, where we showed that dense LSC electrodes deposited at higher temperature (600 C) exhibit a Sr rich termination layer after deposition, which decreases the catalytic activity, while low deposition temperatures lead to highly active surfaces 4. Deposition of additional LSC material on top of the dense electrode further decreases the surface exchange resistance, see also Fig. S2b. The impressive enhancement of the oxygen surface NATURE MATERIALS 3

4 exchange activity, however, is not solely caused by a chemical improvement of the surface but also by enlarging the participating electrode surface area, since under these deposition conditions porous LSC is grown on top of dense LSC 5. The chemical capacitance remains constant during the entire experiment, despite an increase of the electrode volume by 2 3% by the end of the experiment. However, such a small increase might easily be missed due to data scattering. Since neither the shape of the impedance spectra nor the chemical capacitance showed any significant change after deposition of 188 laser pulses it can be safely assumed that the counter electrode does not contribute to the low frequency semicircle and shows superior oxygen exchange kinetics under the given conditions. This conclusion is further supported by the fact that 188 laser pulses were already enough to improve the surface exchange resistance of the top electrode from 276 to 16 Ωcm², while for the fabrication of the counter electrode 9000 laser pulses were used leading to a porous electrode with much larger area. Stability of the LSC electrochemical performance Fig. S3 (a) Representative Nyquist plots normalized to the surface area of the working electrode for asprepared thin films annealed for 6 h at 450 C in mbar p(o 2) and measured during annealing. The legend specifies the times after which the impedance spectra were recorded. (b) Progression of the relevant resistances with annealing time. The temperature of the sample calculated from the high frequency intercept (ionic transport resistance of YSZ) varies around 452 ± 3 C during the 6 hours. The surface exchange resistance follows the temperature fluctuations and does not show any indication of degradation. It can be concluded that the LSC WE is stable in the setup under these conditions and excellently suitable for the study. NATURE MATERIALS 4

5 Quantification by ICP MS Fig. S4 Cation amount detected by ICP MS plotted versus the total number of laser pulses used to fabricate the calibration samples. The slope of each curve represents the deposition rate of the respective targets: 6.3 ± 0.2 pmol Co/pulse, 18.5 ± 0.5 pmol La/pulse, for Sr (sc) 2.3 ± 0.1 pmol Sr/pulse and for Sr 12.7 ± 0.7 pmol Sr/pulse. LSC surface chemistry measured by XPS before and after decoration Prior to mounting, the samples were heated up to ~360 C for 5 minutes in air to reduce (but not fully remove) the contamination with adventitious carbon. At the chosen temperature, no cation segregation was observed with LEIS even after 16 hours 3. However, a significant amount of carbonates was left on the samples, especially on those with Sr surface decoration, indicating the formation of SrCO 3. These carbonates complicate the chemical analysis and it remained an open question whether binding energies measured in XPS are representative for the PLD conditions, which is why a main focus was laid on the compositional quantification. For all quantifications, a Shirley background and Schofield cross sections 6 were used. Correction for photoelectron inelastic mean free paths were not necessary, since the photoelectron peaks Sr3d, La4d, Co3p were chosen and the photoelectron energy therefore varies in the small range between 1350 and 1420 ev. By integrating all angular channels (photoelectrons collected from 20 to 80 relative to the surface normal) of the ~0.25 mm² spot size, 68% of the total signal came from the top 1.5 nm, according to the TPP 2m equation 7. NATURE MATERIALS 5

6 Semi quantitative results on the elemental composition of as deposited and modified LSC samples are shown in Fig. S5. The observed A site and Sr enrichment for as prepared LSC thin films is in accordance with results obtained by XPS for similarly prepared LSC thin films in Ref. 8 and Low Energy Ion Scattering (LEIS) in Ref. 3. The latter showed a Sr rich termination (81 %) and a Sr enrichment for the first 3 nm of asprepared LSC thin films. In our study, surface decoration by La and Sr led to the strongest change because of the much higher deposition rates. Clear trends in cation composition of layers with no modification, after 88, and 888 applied laser pulses can be seen for all cation decorations (see Fig. S5) and are in accordance with ICP MS measurements. Fig. S5 XPS results of the cation composition, determined from Sr3d, La4d, and Co3p peak areas, for an as prepared sample (dotted line) and samples after deposition of 88 and 888 laser pulses of each target ( La, Sr (sc), Sr, Co ). Approximately 68% of the information is derived from the first 1.5 nm. A qualitative comparison of the Sr3d, Co2p and O1s peaks for different LSC modifications is shown in Fig. S6. Co2p and Co3p binding energies were constant within the measurement error for all samples and peak shapes were identical except for a slightly sharper Co2p peak on the Co 888 sample. The absence of strong changes in cobalt peak positions and shapes does not imply equal chemical conditions, because Co 2+ and Co 3+ have very similar binding energies. Recent X ray absorption experiments also revealed that O anions, rather than Co 4+ counterbalance the (relative negative) charge of the Sr 2+ ions 9. Anyway, the observed Co oxidation states are not necessarily representative for operating conditions because of the strongly different po 2, temperature and carbonate surface coverage. A qualitative comparison of the Co2p region of Co 88 and Co 888 (Fig. S6a) to the reference materials Co 3O 4 and CoO, shown in Ref. 10 and 11, suggests that large amounts of Co deposits retained the Co 3 O 4 target stoichiometry. NATURE MATERIALS 6

7 The Sr3d spectra shown in Fig. S6b were fitted by two spin orbit doublets arising from lattice bound Sr (Sr bulk) and surface bound Sr (Sr surface), in analogy to Ref. 8. Components of each doublet had a separation of 1.7 ev, equal full width at half maximum and an area ration of 2:3. Angle resolved measurements confirmed an increase of the Sr surface /Sr bulk ratio at shallow photoelectron emission angles. With increasing amount of deposited SrO (as deposited Sr (sc) 88 Sr (sc) 888 Sr 888 ) an enhanced presence of Sr surface is observed. The chemical state of the Sr surface, however, remains unclear. Very similar Sr3d spectra, including peak shifts of Sr surface, were found for annealed LSC thin films after Sr surface segregation 8,12. Thus, the same Sr species seem to be involved in our decoration experiments and in degradation studies. This will be detailed in a forthcoming paper based on near ambient pressure XPS measurements in combination with EIS to monitor the degradation process of LSC thin films in situ 12. The O1s peaks (Fig. S6c) were fitted with three Gaussian Lorentzian peaks and no constraints. The low binding energy component at 528 ev is strongest for the as deposited and barely decorated Sr (sc) 88 sample. Depth profiling confirms that this component is part of the perovskite bulk. The intense intermediate peak at ev for Co 888 was also found for Co 3O 4 in Ref. 13. The similar O1s profile of the La 888 sample and La 2 O 3 investigated in Ref. 14 also suggests a successful deposition of La 2 O 3 on top of LSC. The high binding energy peak at ~531 ev, which is predominant for Sr (sc) or Sr decorated samples is most likely correlated to SrCO 3. We are convinced that this carbonate coverage is an artefact caused by air exposure after the IPLD experiment, since earlier near ambient pressure XPS measurements showed that these carbonates vanish in an oxygen atmosphere at operating temperature 15. Fig. S6 Comparison of XPS spectra for (a) Co2p, (b) Sr3d, (c) O1s and (d) valence band edge region of asdeposited and modified LSC thin film surfaces. NATURE MATERIALS 7

8 LSC surface imaged by FESEM before and after decoration SEM images obtained after 88 pulses (Fig. S7) and especially after 888 pulses (Fig. S8) suggest that Sr and Sr (sc) decorations form a rather dense layer on top of the nm wide columnar grains of the as deposited LSC thin film. On the other hand, for Co and La surface modifications small grains/islands seem to be distributed over the entire surface of the original LSC grain microstructure. It was shown in an earlier study that LSC films prepared by our procedure at >550 C are already terminated by a monolayer of SrO 3. It is very likely that the primarily Sr terminated surface of LSC is very well wetted by material deposited from the Sr or Sr (sc) target, e.g. SrO on top of a SrO/SrOH/SrCO 3 termination. La island may thus form because of stronger adatom adatom than adatom LSC interactions and could leave SrO terminated as prepared LSC surface exposed to the gas. This remaining free surface becomes then improved by Co as seen in Fig. 6. Please note that Co oxide island formation was only found in SEM after depositing an amount that corresponds to ~7 monolayers. Fig. S7 SEM images showing the surface of as prepared LSC WE before and after deposition of different target materials using 88 laser pulses. The microstructure of the as prepared thin film (columnar LSC grains with a nm ø) is still visible after the deposition of La, Co and even for Sr (sc). For the Sr target the LSC surface is clearly covered by particles. NATURE MATERIALS 8

9 Fig. S8 SEM images showing the surface of as prepared LSC WE before and after deposition of different target materials using 888 laser pulses. Shadows of the microstructure of the as prepared thin film can still be found below the fine distributed islands of Co and La and particles. The surfaces of Sr (sc) and Sr decorated samples look almost molten and completely dense. XRD diffraction patterns of Sr, Co and La target Fig. S9 X ray diffraction patterns of polycrystalline La, Co and Sr target. La and Co target consists exclusively of La 2 O 3 and Co 3 O 4, respectively. For the Sr target different phases were found: 87 at% SrO, ~10 at% Sr(OH) 2 and 3 at% Sr 4 PtO 6. The Sr 4 PtO 6 was formed during calcination of the SrO powder or sintering of the target pellet in the Pt crucible. NATURE MATERIALS 9

10 XRD diffraction pattern of as prepared working electrode Fig. S10 X ray diffraction patterns of an as prepared WE (GDC/LSC, black) and a GDC layer (red) deposited on a YSZ single crystal. All peaks could be assigned to either LSC, GDC or YSZ. C axis oriented growth of GDC and LSC in (001) direction on top of the (001) oriented YSZ single crystal is observed. Laue oscillations are found close to the high intensity substrate peaks and indicate the presence of smooth interfaces. Cross section of as prepared LSC thin film Fig. S11 SEM image of sample X section showing the 150 nm thin LSC film on top of the 55 nm thin GDC film grown on a YSZ single crystal NATURE MATERIALS 10

11 References 1 Opitz, A. K. & Fleig, J. Investigation of O 2 reduction on Pt/YSZ by means of thin film microelectrodes: The geometry dependence of the electrode impedance. Solid State Ionics 181, (2010). 2 Rupp, G. M., Schmid, A., Nenning, A. & Fleig, J. The Superior Properties of La 0.6 Ba 0.4 CoO 3 δ Thin Film Electrodes for Oxygen Exchange in Comparison to La 0.6 Sr 0.4 CoO 3 δ. Journal of The Electrochemical Society 163, F564 F573 (2016). 3 Rupp, G. M. et al. Surface chemistry of La 0.6Sr 0.4CoO 3 δ thin films and its impact on the oxygen surface exchange resistance. Journal of Materials Chemistry A 3, (2015). 4 Januschewsky, J., Ahrens, M., Opitz, A. K., Kubel, F. & Fleig, J. Optimized La 0.6Sr 0.4CoO 3 δ Thin Film Electrodes with Extremely Fast Oxygen Reduction Kinetics. Advanced Functional Materials 19, (2009). 5 Rupp, G. M. et al. Correlating surface cation composition and thin film microstructure with the electrochemical performance of lanthanum strontium cobaltite (LSC) electrodes. Journal of Materials Chemistry A 2, (2014). 6 Scofield, J. H. Hartree Slater subshell photoionization cross sections at 1254 and 1487 ev. Journal of Electron Spectroscopy and Related Phenomena 8, (1976). 7 Tanuma, S., Powell, C. J. & Penn, D. R. Calculation of electron inelastic mean free paths (IMFPs) VII. Reliability of the TPP 2M IMFP predictive equation. Surface and Interface Analysis 35, (2003). 8 Cai, Z., Kubicek, M., Fleig, J. & Yildiz, B. Chemical Heterogeneities on La 0.6Sr 0.4CoO 3 δ Thin Films Correlations to Cathode Surface Activity and Stability. Chemistry of Materials 24, (2012). 9 Lu, Q., Chen, Y., Bluhm, H. & Yildiz, B. Electronic Structure Evolution of SrCoO x during Electrochemically Driven Phase Transition Probed by in Situ X ray Spectroscopy. The Journal of Physical Chemistry C 120, (2016). 10 Langell, M., Anderson, M., Carson, G., Peng, L. & Smith, S. Valence band electronic structure of Co 3O 4 epitaxy on CoO (100). Physical Review B 59, 4791 (1999). 11 Petitto, S. C., Marsh, E. M., Carson, G. A. & Langell, M. A. Cobalt oxide surface chemistry: the interaction of CoO (100), Co 3 O 4 (110) and Co 3 O 4 (111) with oxygen and water. Journal of Molecular Catalysis A: Chemical 281, (2008). 12 A.K. Opitz et al. In Preparation. 13 Petitto, S. C., Marsh, E. M., Carson, G. A. & Langell, M. A. Cobalt oxide surface chemistry: the interaction of CoO (100), Co 3O 4 (110) and Co 3O 4 (111) with oxygen and water. Journal of Molecular Catalysis A: Chemical 281, (2008). 14 Sunding, M. et al. XPS characterisation of in situ treated lanthanum oxide and hydroxide using tailored charge referencing and peak fitting procedures. Journal of Electron Spectroscopy and Related Phenomena 184, (2011). 15 Nenning, A. et al. Ambient Pressure XPS Study of Mixed Conducting Perovskite Type SOFC Cathode and Anode Materials under Well Defined Electrochemical Polarization. The Journal of Physical Chemistry C 120, (2016). NATURE MATERIALS 11