An in operando study of chemical expansion and oxygen surface exchamge rate in epitaxial GdBaCo2O5.5 electrodes in a solid state electrochemical cell by time-resolved X-ray diffraction

This report explores the fundamental characteristics of epitaxial thin films of mixed ionic electronic conducting GdBaCo2O5.5±δ (GBCO) material with layered perovskite structure, relevant for use as an active electrode for the oxygen reduction and evolution reactions in electrochemical devices. Time-resolved X-raydiffraction in combination with voltage step chrono-amperometric measurements in a solid state electrochemical cell provide a deeper insight into the chemical expansion mechanism in GBCO electrode. The chemical expansion coefficient along c-axis, αc, shows a negative value upon the compound oxidation contrary to standard perovskite materials with disordered oxygen vacancies. Chemical expansion also shows a remarkable asymmetry from αc= -0.037 to -0.014 at δ<0 and δ>0, respectively . This change in chemical expansion is an indication of a different mechanism of the structure changes associated with the variable Co cation oxidation state from CoCoCo. Since the redox reactions are dominated by the oxygen surface exchange between the GBCO electrode and gas atmosphere, monitoring the time response of the structure changes allows for direct determination of oxygen reduction and evolution reaction kinetics. The reaction kinetics are progressively slowed down upon reduction in the δ<0 oxygen stoichiometry region, while they maintain a high catalytic activity in the δ>0 region, in agreement with the structural changes and the electronic carrier delocalization when crossing δ=0. This work validates the time-resolved XRD technique for fast and reversible measurements of electrode activity in a wide range of oxygen non-stoichiometry in a solid-state electrochemical cell operating under realistic working conditions.


Introduction
In electrochemical devices like solid oxide fuel cells (SOFCs) and oxygen permeation membranes the kinetics of the Oxygen Reduction Reaction (ORR), related to the oxygen surface exchange between the gas atmosphere and the electrode materials, often limits the overall performance of the device. 1,2 Large activation energies of these reactions at the SOFC cathodes cause large polarization losses when reducing operating temperatures below 800 °C. 3 Therefore, gaining a deeper knowledge of ORR mechanisms and rates of cathode materials has been a major challenge in the development of competitive intermediate-temperature SOFC technology. 4,5 To this end, a variety of cathode materials have been proposed from composite materials [6][7][8] to mixed ionic-electronic materials. [9][10][11][12][13] Studies of oxygen surface exchange in cathode materials are generally performed by electrochemical impedance spectroscopy, 14 conductivity relaxation 15,16 or isotopic exchange depth profiling combined with SIMS or LEIS analysis. [17][18][19][20][21] Recently, time-resolved X-ray diffraction, [22][23][24][25][26][27] wafer curvature relaxation monitoring 28 as well as optical absorption experiments, 29 have emerged as powerful methods to determine surface exchange kinetics in thin film electrodes through monitoring subtle crystal structure and optical transient changes in the material induced by oxide stoichiometry changes when exposed to sudden changes in the pO 2 gas atmosphere. Other external stimuli, such as the application of an electric field between the electrode material and a solid electrolyte, typically Yttria-stabilized zirconia (YSZ) material, can also cause a perturbation from the equilibrium changing the oxygen stoichiometry The study of the transient structure changes by in-situ XRD after electrical stimuli have also proven to be an effective method to explore redox reaction mechanisms in SrCoO x films 30 and to determine kinetics of oxygen surface exchange reactions in classical perovskite electrode materials such as La 1-x Sr x Co 1-y Fe y O 3-δ . [22][23][24] The present study focuses on the determination of the mechanisms of  32 In oxygen permeation membranes LnBaCo 2 O 5+δ ceramics have exhibited reduced oxygen permeation. 35 However, the fact that these compounds could show a reduced chemical expansion compared to classical perovskites, still makes them ideal attractive for oxygen permeation membranes, where large chemical potential gradients often cause mechanical failure by crack formation or delamination during operation. 36,37 GdBaCo 2 O 5.5±δ compound has a layered perovskite structure with GdO and BaO layers alternating in the A-sites of the perovskite structure along c-axis direction. This produces a doubling of the cell parameter. 38 46 . This also guarantees the achievement of sufficient oxygen exchange at the interface between the top GBCO and bottom Ag electrodes and the gas atmosphere, which is also necessary for the electrochemical stabilisation of a current flow.
Positive and negative voltage bias steps from 0, ±100, ±200 and ±300 mV were applied to the electrodes and the electrical current transients were monitored continuously. Special care was taken to avoid sample heating by Joule effect at large voltages, which could add a thermal expansion contribution to the chemical expansion. This is the reason why the voltage was limited only to ±300mV. Voltage was first changed from 0V to +100mV and maintained for 20-30 min until a stationary regime was established. The applied voltage was then reset back to 0V for 20-30 min. A subsequent cycle at -100mV was set with the same time span. This procedure was followed for increasing voltages of ±200 mV and ±300mV.
Simultaneously, the structure cell parameters of the GBCO material (c-axis parameter) were continuously monitored by XRD by following the angular shift of one particular diffraction peak (004) in fast 2θ static scans every 10 sec making use of a multichannel X-ray detector (PIXcel from Panalytical).
For the best accuracy of the cell parameter determination we used a 2 x Ge(220) monochromator to use only CuKα 1 radiation. A complete description of the time-resolved X-ray diffraction procedure has been previously described in. 25-27  Reflections 004 and 008 corresponding to the CGO buffer layer were also observed close to the intense 00L peaks from YSZ single crystal substrate.

Results and Discussion
There is no indication of any chemical reaction at the interface forming the oxygen exchange rates at the film electrode/air surface and at the film electrode/electrolyte interface before reaching a steady state, as described in ref [22]. When resetting the voltage to 0V the current dropped to a negative value -3.1 µA and exponentially went to zero recovering the initial state.
When a negative voltage step from 0V to -100 mV was applied the current initially decreased to -17.0 µA and exponentially stabilized at -12.7 µA. The increase in the voltage produced a proportional increase in the initial current peak and subsequent current intensity decay. At V= ±300 mV the current showed fluctuations of a large magnitude that precluded extraction of any reliable current value. However, the corresponding transient current values when switching off the voltage to 0V showed perfect exponential decays in the full voltage range from 100 to 300 mV. Their respective time responses were analysed by fitting simple exponential decay curves.

B. Cell parameter changes
The corresponding cell parameter changes at 350 °C were monitored during the same voltage steps as shown in Fig. 5a. When applying a positive bias from 0V to +100mV (anodic potential) GBCO c-axis parameter increased from When the potential was switched off from +100 mV to 0 V, the c-parameter decreased exponentially to the initial equilibrium value. This means that oxygen is released from the GBCO film lattice (reduction process) in a reversible way. When applying a negative bias voltage step from 0V to -100mV (cathodic potential) the reverse situation occurred and the GBCO caxis parameter shrank to 7.5975 Å, corresponding to a ∆c/c= -0.028%. Fig.   5b depicts the measured ∆c/c relative change for the different bias voltage.
As can be observed the c-parameter proportionally increases with anodic potential and it tends to saturate at +300 V where c-parameter changes reach ∆c/c = +0.044% expansion. However, for the negative bias there is a much steeper reduction of the cell parameter reaching ∆c/c = -0.102% at V= -300 mV, which is more than twice the cell parameter change at a positive bias of the same magnitude. This is an indication of a clear asymmetry in the GBCO cell expansion.
The asymmetry in the chemical strain could be related to two possible causes: i) either the change in oxygen stoichiometry ∆δ could be different for a given magnitude of the bias voltage depending on its sign, or ii) in the case of a similar oxygen stoichiometry change, the corresponding chemical expansion (or more generally the chemical strain, i.e. the relative change in cell volume per ∆δ) may not be exactly symmetric, corresponding to a different mechanism for oxidation and reduction.

C. Oxygen non-stoichiometry determination
In order to differentiate between these two possibilities the chronoamperometry was used to analyse the oxygen stoichiometry changes, (1) Therefore, the transported charge Q for the whole film volume will be directly related to the stoichiometry change ∆δ per unit cell volume: where n=2 is the number of electrons per oxygen O 2ion, and e is the electron charge.
Note that it is still not possible to estimate the absolute value of the oxygen stoichiometry -only the relative change from the equilibrium state at 350 °C in air (PO 2 = 0.21 atm).
As an example of the calculation, Fig. 6a shows a detail of the exponential current decay after applying a positive voltage bias step from 0V to +100mV.

D. Chemical expansion coefficient
Taking advantage of the calculated oxygen stoichiometry changes Fig. 6c depicts the relative c-axis variations, the strain ε c observed along the c-axis, as a function of the change in ∆δ N . The slope represents the chemical expansion coefficient as defined in [47]. The most remarkable result is that in the whole stoichiometry range analysed from GBCO 5.34 to GBCO 5 The overall c-axis variation is a balance between the change in both the slabs and it seems to be dominated by the [GdO δ ] expansion making the GBCO behave differently from typical oxygen vacancy disordered perovskites and more similarly to some layered oxides with K 2 NiF 4 structure.
Returning to the experimental results depicted in Fig. 6c, it is clear that the chemical expansion coefficient shows a nonlinear dependence in the oxygen stoichiometry range explored by the electrochemical reduction and oxidation. where HS and LS are high and low spin states, respectively. Although the coordination and spin state of the Co ions in the GBCO structure are far more complex that the compounds from which the ionic radii were extracted, this approach can still serve as a qualitative guide to estimate the changes in the ionic radii. In the GBCO films it can be assumed that the equilibrium oxygen stoichiometry is close to GdBaCo 2 O 5.5 (δ∼0), which corresponds to a pure Co 3+ oxidation state, in either LS or HS state. Then, a positive bias will favour the Co 3+ Co 4+ oxidation, while a negative bias will cause Co 3+ Co 2+ . In the case that the change in Co ionic radius was the dominant effect in the overall cell expansion variations, and assuming the [GdO δ ] expansion to behave uniformly in the whole ∆δ, any substantial change in the Co 2+ Co 3+ or Co 3+ Co 4+ steps will produce a certain asymmetry between positive and negative bias. If we consider only the high spin values, the difference in the ionic radii between Co 2+ (HS)Co 3+ (HS) is -18%, while for Co 3+ (HS)Co 4+ (HS) it is -11%. In this case, the oxidation causes a larger shrinkage of the [BaCoO] block, a factor of 1.6 times larger in the negative bias than for positive bias.