Comparison of electrochemical hydrogen oxidation on different metal/ceramic model anodes and mechanistic implications

Micro-patterned metal/ceramic model thin film electrodes were used to investigate the hydrogen oxidation mechanism on various metal/ion conductor combinations such as Ni/yttria-stabilized zirconia (YSZ), Pt/YSZ, Ni/scandia-stabilized zirconia, Ni/titania/YSZ. For all electrode types besides those with a continuous titania interlayer a triple phase boundary (TPB) and an area pathway could be separated. Moreover, activation energies as well as hydrogen and water reaction orders were determined for both pathways and compared with previously published data on Ni/YSZ. It was found that when varying the electronically conducting phases on YSZ the activation energy of the TPB path remains virtually the same. However, changing the ion conductor significantly influenced the activation energy, indicating a rate-limiting step on the electrolyte. However, the stark difference in water reaction order of Ni/YSZ and Pt/YSZ electrodes shows that the metal phase is also actively participating in the reaction chain. It is suggested that the electrolyte surface is responsible for the rate-limiting electrochemical redox reaction while the metal interacts with the gas phase molecules and thus supplies the reactants. For samples with a thin mixed conducting titania layer on YSZ, the electrochemical behavior drastically changed, which can be explained by an extension of the reaction zone away from the TPB.


Introduction
Solid oxide fuel cells (SOFCs) are a promising technology for the conversion of chemical energy to electrical energy due to their high efficiency compared to Carnot-limited engines, low pollutant emission, and much higher fuel-flexibility than proton exchange membrane fuel cells. In the last 25 years, much research effort has been put into increasing the power density and lifetime of SOFCs by optimization of cathodes and introduction of novel cathode materials [1,2]. Owing to the related achievements on cathode performance, research dealing with the resistive losses at the anode has again come more into focus, especially for applications employing carbon and sulfur containing fuels. Despite many studies investigating the electrochemical properties of SOFC anodes in general and Ni/yttria stabilized zirconia (YSZ) in particular , the basic reaction mechanisms at metal/ceramic anodes are still not completely understood.
Fundamental studies dealing with the electrochemistry of Ni/YSZ anodes often utilized micro-structured thin film electrodes, which offer the advantage of a much better defined geometry compared to porous cermet electrodes. With such model electrodes, convoluting factors like gas-or oxide ion diffusion can be avoided [3, 10-12, 14, 15, 17, 18, 27-29]. Moreover, electrodes with well-defined geometry can be used to compare different materials without geometrical uncertainties due to processing parameters. In a previous study, Ni/YSZ electrodes with varying triple phase boundary (TPB) length or surface area were electrochemically characterized in H 2 /H 2 O atmosphere, leading to the discovery that Ni/YSZ electrodes are not purely TPB active, but also Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. exhibit a reaction path through the electrode bulk including charge transfer at the Ni/YSZ interface [30]. Moreover, activation energies and hydrogen/water reaction orders could be determined for both reaction pathways. Other metal/YSZ systems in H 2 /H 2 O are much less investigated and only little comparative data exists [31].
The present study aims at a deeper and more general understanding of the reaction mechanisms on metal/ ceramic composite electrodes in hydrogen/water atmosphere by extending the parameter space from geometrical variations to the variation of materials. In particular, the material combinations of Pt/YSZ, Ni/ scandia stabilized zirconia (SSZ), and Ni/ titania/YSZ are investigated and compared to the previously studied Ni/YSZ. Impedance spectroscopy measurements were performed on micro-patterned thin-film model electrodes of these material combinations and the measurement temperature as well as the partial pressures of H 2 and H 2 O in the atmosphere were systematically varied. This revealed information on the activation energy of the polarization resistance and the reaction orders of the reaction for all material combinations. Based on this comparison, mechanistic conclusions can be drawn.

Experimental
All electrodes were prepared on 10×10×0.5 mm 3 (111) oriented YSZ single crystals (9.5 mol% Y 2 O 3 , Crystec, GER). A porous Ni/YSZ counter electrode was applied on the unpolished bottom side of these substrates by consecutively screen printing NiO/YSZ and Ni paste (Heraeus, GER) and subsequently sintering the sample at 1250°C for 2 h in air. On the polished YSZ top side, the following different types of working electrodes were grown by pulsed laser deposition (PLD) and/or magnetron sputtering: • Pt/YSZ: 400 nm of Pt (99.95%, Ögussa, AT) was grown on the polished top side of the YSZ crystal by sputter deposition (MED020 Coating system, BAL-TEC, LIE) at 2×10 −2 mbar Ar with a current of 100 mA. The metal film was then recrystallized at 800°C in a reducing atmosphere containing 2.5 kPa H 2 /0.15 kPa H 2 O/balance Ar (denoted H2.5W10 in the following). Micro-structuring of the film was done by photolithography (photoresist: ma-N 1420, developer: ma-D 533S, both: Micro Resist Technology GmbH, GER) and Ar ion etching (3 kV, about 2 mA ion current).
• Ni/SSZ: first, 60 nm of scandia stabilized zirconia (6 mol% Sc 2 O 3 , Treibacher, AT) were deposited on the top side of the YSZ crystal via PLD at 650°C substrate temperature in 4×10 −2 mbar O 2 with a target-to-substrate distance of 6 cm. For the target ablation a KrF excimer laser (Compex Pro 201F KrF Excimer laser, Coherent, USA) with 400 mJ/pulse and 5 Hz pulse frequency was used; the deposition time was 20 min. Then, 1200 nm of Ni (99.99%, Advent RM, UK) were deposited by magnetron sputtering at 8×10 −3 mbar Ar with a current of 150 mA and the film was recrystallized at 800°C in H2.5W10 atmosphere. Finally, the Ni film was microstructured by photolithography (as described above) and chemical etching in aqua regia.
• Ni/TiO 2 /YSZ: 30 nm TiO 2 were deposited on the top side of the YSZ crystal by sputtering a Ti target (BAL-TEC) at 7×10 −3 mbar Ar and equilibrating the obtained layer for 3 h at 800°C in air. In [32] it is shown that anatase is formed by this procedure. Afterwards, 1200 nm Ni were deposited, recrystallized, and structured as described for Ni/SSZ.
Micro-structuring of the metal films led to samples covered with many working electrodes with nominally constant TPB length of 4.03 mm but varying area as described in [30,33]. A sketch of the electrode shapes is given in figure 1(a). After film growth and micro-patterning all samples were annealed at 800°C in H2.5W10 atmosphere for 10 6 s (i.e. ca. 278 h) to stabilize the electrochemical properties. (see [34] for more details). Subsequently, the pattern electrodes were geometrically characterized by optical microscopy. Electrochemical impedance spectroscopy (Alpha-A high performance frequency analyzer with POT/GAL interface, Novocontrol, GER) was performed in a symmetrically heated micro-contact setup (Huber Scientific, AT), which offers the opportunity of contacting four electrodes at once and addressing them subsequently. For details regarding this setup the reader is referred to [35]. Several exemplary impedance spectra are shown in figure 1(b) for Ni/SSZ microelectrodes of different shape and area (but identical TPB length) and for identical shapes but different materials (Pt/YSZ, Ni/SSZ and, for comparison, also Ni/YSZ from [30]). All spectra could be fitted to a simple equivalent circuit (see figure 1) consisting of an electrolyte spreading resistance R spread and a resistor in parallel to a constant phase element. The latter resistor R Pol revealed the polarization resistance of the electrochemical reaction, see also [30]. Spectra for Ni/TiO 2 /YSZ including their analysis are described in the corresponding chapter.

Pt/YSZ
3.1.1. Geometry dependence Pt electrodes with constant TPB length but varying area were electrochemically characterized by impedance spectroscopy at 800°C in H2.5W10 atmosphere (2.5 kPa H 2 , 0.15 kPa H 2 O, balance Ar). The resulting inverse polarization resistances (electrode conductances) are plotted as a function of the electrode area in figure 2(a) (black squares). Similar to the reference material Ni/YSZ (gray triangles, data from [30]), a y-axis intercept as well as a linear correlation between conductance and electrode area is found. Therefore, the electrode cannot be purely TPB active, since an increase in electrode area at constant TPB length also increases the activity/ conductance of the electrode. In accordance with [30] this geometry dependence of the conductance is interpreted in terms of two distinct parallel reaction pathways for hydrogen oxidation being present for Pt/YSZ electrodes, a 'TPB pathway', represented by the y-axis intercept in figure 2(a), and an 'area pathway', which is reflected by the slope in the conductance versus area plot. The respective geometry-normalized conductances for both pathways are summarized in table 1 and further discussed below.

Activation energies
The slopes k and intercepts d from the conductance-area diagrams at different temperatures (figure 2(b)) can be plotted in an Arrhenius diagram as shown for Pt/YSZ and Ni/YSZ in figure 2(c). The slopes k (in S m −2 ) correspond to the area specific conductance of the area pathway and the intercept d (in S) is the inverse resistance of the TPB pathway of the corresponding electrode type with 4.03 mm TPB length. The activation energies are summarized in table 1. The Arrhenius plot shows that for the TPB pathway (intercepts, d) the activation energy of Pt/YSZ (2.19±0.08 eV) is very similar to the activation energy in the Ni/YSZ system (1.96±0.07 eV [30]). The area pathway, on the other hand, shows a significantly different activation energy with 1.49±0.01 eV in the Pt/YSZ system compared to 1.04±0.04 eV for Ni/YSZ [30].

H 2 /H 2 O/O 2 reaction orders
To obtain partial reaction orders of the electrochemical reaction were measured in varying partial pressures of hydrogen and water. When assuming a simple power law, the reaction order may be determined from a double-logarithmic diagram of the inverse polarization resistance of each pathway versus partial pressure according to the equation z denoting the partial reaction order and p x the corresponding partial pressure. When plotting slopes k (area specific conductance of the area pathway) and intercepts d (conductance of the TPB pathway) obtained from the conductance-area diagrams (see figure 2(d)) in such double-logarithmic diagrams, reaction orders of the respective pathways can thus be obtained. The resulting collection of reaction order diagrams for Pt/YSZ is  Table 1. Summary of measured geometry related electrodes conductances (LSC= line specific conductance (d/4.03 mm), ASC= area specific conductance (k), both at 800°C in H2.5W10), activation energies, and hydrogen, water, and oxygen reaction orders (RO) of both pathways obtained in this study and from an expanded data set from [30] shown in figure 3. The figure shows that reaction order values for hydrogen and water are in the range of +1/2 and −1/4, respectively, irrespective of the reaction pathway. Moreover, a nominal oxygen partial pressure can be determined from p (the equilibrium constant K was calculated from thermodynamic data [36]) and for both pathways a simple reaction order is found also for oxygen (∼−1/5). From this result, one may speculate that on Pt/YSZ H 2 and H 2 O do not affect the polarization resistance independently. Rather, the influence of the H 2 and H 2 O partial pressure on the polarization resistance possibly occurs via the equivalent equilibrium oxygen partial pressure. This is in constrast to Ni/YSZ where even the sign of the water reaction order (+0.35) is different [30]. The calculated reaction orders are summarized in table 1.

Geometry dependence
Analogously to Pt/YSZ, also Ni/SSZ micro-electrodes were measured at 800°C in H2.5W10 atmosphere. Exemplary spectra are depicted in figures 1(b) and (c) and show the same shape as for Pt/YSZ and the reference Ni/YSZ. The area dependency of the conductance is given in figure 4(a), together with a linear fit of the data. In accordance with Pt/YSZ and Ni/YSZ [30], again two parallel reaction pathways are found-one scaling with the TPB length, the other one with the electrode area. The absolute values of the area specific conductance of the area pathway and line specific conductance of the TPB pathway are summarized in table 1.

Activation energies
Impedance measurements of the Ni/SSZ electrodes were also conducted at varying temperatures. For each measured temperature, slopes k (area pathway) and intercepts d (TPB pathway) were extracted from conductance-area diagrams and plotted in an Arrhenius diagram, which is depicted in figure 4(c). From that plot activation energies of 1.27±0.11 eV and 1.42±0.07 eV can be extracted for the area and TPB pathway of Ni/SSZ, respectively. It is remarkable that the activation energy of the TPB pathway is more than 0.5 eV lower than for both YSZ-related systems mentioned above (Pt/YSZ, Ni/YSZ). This significantly lower activation energy of Ni/SSZ was also confirmed in a second identical measurement series [34], thus excluding a measurement error or a fit artifact as possible reason for this difference. Rather, this observation points towards an important role of the electrolyte in the TPB pathway-further details are discussed below.

H 2 /H 2 O reaction orders
Again, reaction orders of both reaction pathways on Ni/SSZ were extracted by determining the slopes k (area pathway) and intercepts d (TPB pathway) in conductance-area diagrams for different gas compositions (see figure 4(d)) and plotting the resulting data set in double logarithmic diagrams against partial pressures, see figure 5. As can be seen in the conductance-area plot (figure 4(d)) some data for the water partial pressure variation scatter significantly (the relative standard deviation of the area pathway activity of H2.5W100 is 340%,  [30] are shown in (g) (area pathway) and (h) (TPB pathway) for comparison. Equilibrium oxygen partial pressures are calculated using NIST data [36]. H20W80=20 kPa H 2 , 1.2 kPa H 2 O, balance Ar. while for H2.5W10 it is only 15%), which results in substantial uncertainty of the water reaction orders. The reliably obtained H 2 reaction orders were extracted according to equation (1) and are summarized in table 1.

Impedance spectrum
The impedance spectrum of a Ni/TiO 2 /YSZ sample with continuous TiO 2 interlayer is shown in figure 6(a). It can be seen that the Nyquist plot clearly shows two arcs. These spectra were fitted to the equivalent circuit shown in the inset of the figure in the frequency range of 100 Hz-5 mHz. Additional electrode features at higher frequencies are neglected in this fit. The total polarization resistance was calculated by summing all resistance values in the equivalent circuit (R Pol =R1+R2+R3); please note that for the sake of simplicity the small contribution of the electrolyte was not separated in this case. The obtained polarization resistance is about two orders of magnitude lower than for all other electrode types. Reasons for this behavior are discussed below.
Removing the uncovered TiO 2 layer by Ar ion bombardment completely changed the electrochemical behavior, yielding again an impedance response very similar to Ni/YSZ, Ni/SSZ or Pt/YSZ-i.e. an electrode feature consisting of one slightly asymmetric semicircle; compare figure 5(b) with figures 1(b) and (c). Also, the same fitting approach could be successfully applied as for the metal/zirconia based electrodes.

Geometry dependence
Again, electrodes with varying area but constant TPB length were measured at 800°C in H2.5W10 atmosphere. The inverse polarization resistances obtained for continuous TiO 2 layer samples are compared with Ni/YSZ in figure 7(a). From this plot it is obvious that the electrode conductances were much higher than those of Ni/YSZ electrodes, which may already indicate a completely different reaction mechanism on that type of samples.
Removal of the titania interlayer not being covered by Ni changes the polarization resistance drastically. A plot of the electrode conductance 1/R Pol of such an etched samples versus the electrode area is shown in figure 7(b). It can be seen that the conductances of the etched samples are still somewhat higher than for Ni/YSZ electrodes, but at least in the same order of magnitude. Analysis of the electrode conductance values does not confirm statistical significance for the existence of an area pathway. In both cases the slope does not deviate from zero with p<0.05, therefore the average conductance of all measured electrodes was summarized as TPB pathway activity in table 1.

Activation energy
Measurements on etched and non-etched Ni/TiO 2 /YSZ samples were also conducted at different temperatures. Since no area pathway could be confirmed, only activation energies of the polarization resistance of individual electrodes were derived. The Arrhenius diagram comparing results on both etched and non-etched samples is depicted in figure 7(c). This plot shows that after removing the uncovered TiO 2 layer the electrodes exhibit an activation energy of 2.10±0.14 eV, which is close to that of the reference Ni/YSZ TPB activation energy  (1.96±0.07 eV). The non-etched electrode, on the other hand, is activated by only 0.48±0.03 eV. Such a severe difference of activation energies strongly suggests a completely different reaction mechanism-see discussion below.

H 2 /H 2 O reaction orders
Analogously to the previous sections, water and hydrogen reaction orders were determined for the entire polarization resistance of etched and non-etched Ni/TiO 2 /YSZ samples. The corresponding reaction order diagrams are shown in figure 8 and the extracted reaction orders are summarized in table 1. Differences are particularly pronounced with respect to H 2 O.

Discussion
The results of non-etched Ni/TiO 2 /YSZ samples differ strongly from all other experiments. They show a surprisingly high electrode conductance and a very low activation energy (0.48 eV). However, the situation changes when removing the parts of the titania layer not covered by Ni. The conductance sharply decreases and approaches the values of pure Ni/YSZ. This very different behavior of non-etched films can be explained by an expansion of the reaction zone along the free titania surface and is discussed in more detail below. All electrodes with YSZ being part of the TPB, i.e. Ni/YSZ, Pt/YSZ, and etched Ni/TiO 2 /YSZ, exhibit very similar activations energies in the range of 2 eV. The TPB pathway of Ni on scandia stabilized zirconia, on the other hand, shows a significantly different activation energy compared to the YSZ-based TPBs. This indicates that the activation energy depends primarily on the electrolyte phase and suggests that the rate limiting step is located on/in the electrolyte rather than on the metal. Possibly, the rate limiting step on/in the electrolyte includes an electronrelated process such as charge transfer to a surface species and/or electron transport in the electrolyte.
This importance of the electrolyte, however, does not exclude any mechanistic role of the metal phase on YSZ. Information on the role of the metal is gained from the reaction orders of the TPB pathway in different systems. The hydrogen reaction order of the TPB pathway varies only slightly for Ni/YSZ, Ni/SSZ, Ni/TiO 2 /YSZ and Pt/YSZ electrodes (between 0.35 and 0.49), but the water reaction order differs significantly -partly even with different algebraic signs. From these differences we conclude that the metal phase does not only act as a current collector but actively takes part in the reaction chain of the TPB path, most probably via interaction with the gas phase, i.e. adsorption/desorption and/or dissociation. This hypothesis is further supported by the results found for the area pathway.
The available activation energies of the area pathway depend on the metal phase, which is hardly surprising, as all physically meaningful reaction mechanisms for this pathway include a role of the metal [30]). The hydrogen reaction orders of the area pathway are the same for Pt/YSZ and Ni/YSZ. However, strongly different H 2 O reaction orders are found, partly with different signs. On the other hand, a close similarity exists for the measured reaction orders of TPB and area pathway of each individual system, i.e. for Ni/YSZ and for Pt/YSZ. Particularly, the water reaction order of TPB and area pathways is virtually identical for a given metal (Ni or Pt). This indicates that the water related part of the TPB and area pathway is probably the same. We therefore suggest that for both pathways the interaction with the gas phase goes via the metal. Hence, the TPB activity of metal/ ceramic electrodes seems to rely on a kind of job sharing: the main electrochemical reaction takes place on the electrolyte and the metal is responsible for the supply (or removal) of adsorbed gas species. This adsorbate supply/removal seems to work even if a thin TiO 2 layer is still between Ni and YSZ, see sketch in figure 7.
Finally, the very different behavior of non-etched Ni/TiO 2 /YSZ electrodes is discussed in more detail. We suppose that TiO 2 acts as a mixed ionic and electronic conducting electrode under the present experimental conditions. Accordingly, hydrogen is oxidized on larger parts of the TiO 2 surface and oxide ions move across the titania layer, while electrons move in-plane and are collected by the Ni. The main conductance increase of the non-etched Ni/TiO 2 /YSZ samples is thus caused by the much larger active area for the electrochemical reaction due to in-plane electron conduction in TiO 2 . A very similar situation can be found for ceria based mixed conducting anodes [37,38]. The rate-limiting process is probably a combination of lateral transport of electrons and electrochemical reaction at the surface of mixed conducting TiO 2 . The interpretation of lateral electron transport in titania interlayers being relevant for the polarization resistance of non-etched Ni/TiO 2 /YSZ samples is further supported by the reaction orders. While no simple oxygen reaction order can be found for the etched Ni/TiO 2 /YSZ, non-etched electrodes show a continuous increase of the conductance with decreasing p O 2 , see figure 8. From the defect chemical model of undoped TiO 2 [39] an oxygen reaction order of −1/6 would be expected for the electronic conductivity of the titania interlayer, which is rather close to the value observed here.

Conclusion
In this study the electro-catalytic properties of micro-structured thin film model electrodes were investigated for the metal/electrolyte combinations Ni/YSZ, Pt/YSZ, Ni/SSZ, Ni/TiO 2 /YSZ. By variation of the electrode area while keeping the TPB length constant, an area and a TPB related reaction pathway could be separated for electrodes without TiO 2 interlayer. By variation of temperature as well as hydrogen and water partial pressure, activation energies and reaction orders were determined for both pathways. The resulting data set was compared with previously published data on Ni/YSZ. The comparison of activation energies showed that for all samples with uncovered YSZ electrolyte (i.e. Ni/YSZ, Pt/YSZ, and etched Ni/TiO 2 /YSZ) the activation energies were very similar around 2 eV. For Ni/SSZ only 1.4 eV was found. This behavior suggests that the rate-limiting step of the TPB pathway is located on the electrolyte. However, the electronically conducting phase is not an inert current collector, since the metal significantly influences the water reaction order (with even different algebraic sign on Ni/YSZ and Pt/YSZ). Therefore, the metal is suggested to be relevant for the adsorption/desorption of species and their transport to/from the reaction zone on the electrolyte. Deposition of a thin TiO 2 layer onto the YSZ electrolyte before applying the Ni electrodes caused an increase of the electrode activity by about two orders of magnitude. This can be explained by the mixed conducting properties of the titania film at the given experimental conditions, which leads to an extension of the electrochemically active region along the free titania surface.