Anode fabrication for solid oxide fuel cells: Electroless and electrodeposition of nickel and silver into doped ceria scaffolds

A novel fabrication method using electroless and electrodeposited Ni/Ag/GDC for SOFC anodes is presented. First a porous Ce0.9Gd0.1O2 x (GDC) scaffold was deposited on a YSZ electrolyte by screen printing and sintering. The scaffold was then metallized with silver using Tollens' reaction, followed by electrodeposition of nickel from a Watt's bath. The electrodes (Ni/Ag/GDC) were tested in both symmetrical and fuel cell configurations. The microstructures of the Ni/Ag/GDC anodes were analyzed using scanning electron microscopy (SEM) and energy dispersive x-ray spectroscopy (EDX). Nano-particles of Ni formed in the porous GDC scaffold provided triple phase boundaries (TPB). The electronic conductivity of the Ni/Ag/GDC (3.5/24.7/71.8 vol%) electrode was good even at relatively low Ni volume fractions. The electrochemical performance was examined in different concentrations of humidified hydrogen (3% H2O) and over a range of temperatures (600e750 C). The total area specific resistance (ASR) of the anode at 750 C in humidified 97 vol% H2 was 1.12 U cm, with low-frequency polarization (R_l) as the largest contributor. The electrodes were successfully integrated into a fuel cell and operated in both H2 and syngas. Copyright © 2016, The Authors. Published by Elsevier Ltd on behalf of Hydrogen Energy Publications, LLC. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/).


Introduction
Solid oxide fuel cells (SOFCs) could play a significant role in sustainable energy futures because of their ability to convert a diverse range of fuels to electricity at high efficiency. A typical SOFC consists of dense yttria-stabilized zirconia (YSZ) as the oxygen anion conducting electrolyte and Ni-cermets as the anode. A variety of materials are used as the cathode, including doped manganites, cobaltites and ferrites. Ni is usually chosen as the metallic component of the anode because of its high electronic conductivity and stability, excellent reforming characteristics, and good electrocatalytic activity for electrochemical oxidation [1]. However, Ni can catalyze the deposition of carbon on its surface when hydrocarbon fuels are used, leading to cell deactivation under some conditions [2,3].
The performance of SOFC anodes is linked to the length of the triple phase boundary (TPB), where the electrochemical reactions occur. This region is dependent upon the optimization of anode porosity, microstructure and composition [4].
The reaction sites can also be extended by using Ni/ Ce 0.9 Gd 0.1 O 2Àx (Ni/GDC) as compared to the widely used Ni/ YSZ anode materials. GDC exhibits mixed ionic and electronic conductivity in the reducing conditions found in an SOFC anode, providing a GDC/gas interface or so-called doublephase boundary (2 PB) that shows electrocatalytic activity [5e7]. Moreover, GDC has the ability to enhance the electrocatalysis of hydrocarbon oxidation, and enables improved sulfur tolerance, via the Ce 4þ /Ce 3þ redox transition. It also exhibits higher ionic conductivity than YSZ in the temperature range from 300 to 700 C [8e13].
Ni/GDC anodes are commonly prepared by mechanically mixing powders of NiO and GDC, followed by sintering at 1300 C and reducing in a hydrogen rich atmosphere [14,15] However, this method requires a large volume of Ni (>30 vol %) to achieve adequate electronic conductivity in the electrodes. The impregnation or infiltration method enables the formation of nano-sized oxide or metal particles uniformly in a porous scaffold matrix [16e18]. It allows the fabrication of anode and cathode composites with a wide range of compositions, particularly with metals that cannot be processed at high temperature (e.g. Cu) [19]. However, the impregnation method is time consuming as it requires multiple steps to achieve sufficient conductivity and hence may not be suitable for mass production. Based on previous studies, ten successive metal impregnation steps were required to achieve sufficient electronic conductivity throughout the electrode, demanding more than 10 h for cell fabrication [20e22]. A faster impregnationirradiation technique (60 s/cycle) to accelerate conventional impregnation methods has shown well distributed Ni in the scaffold, however, repeated cycles can damage the thin scaffolds [23]. Therefore in this paper, we metallize the scaffold by electroless and electrodeposition as an alternative to impregnation as a fabrication method of scaffold electrodes.
Metallizing the scaffold by electroless and electrodeposition methods could be a scalable process due to its ability to add a large amount of metal (including the low melting point metals) to the porous ceramic scaffold in a fast process. An electroless deposition involves chemically reducing metallic ions or complexes onto a substrate in a controlled fashion [24]. Electrodeposition is a versatile technique since a metallic coating is obtained on the surface of another metal or conductive surface by the electroreduction of metal ions from aqueous solutions [25]. This approach allows control of the morphology of metal deposits and may be particularly useful for the production of multilayer microstructures.
Electroless Ni deposition onto YSZ particles for Ni/YSZ cermet anode fabrication has been carried out in several studies [26e28]. Among these studies, the lowest Ni content that has been reported to result in sufficient electrical conductivities for SOFC anode application is 20% vol Ni (~160 S cm À1 at 750 C) [26]. However, their development as SOFC anodes is still at an early stage.
Electrodeposition techniques for SOFC anode fabrication are usually applied to provide bimetallic electrodes especially to avoid carbon formation. It has been shown that this technique is a promising way to add Cu into porous Ni-based anodes at low temperature (<100 C) to suppress carbon deposition when direct hydrocarbon fuels are used in the SOFC [19,29]. Electrodeposition requires a conducting surface. This can be prepared by several techniques, for example exposing n-butane onto scaffolds at elevated temperature (850 C) to form an electrically conducting carbon layer [30], or pre-incorporating ceramic-metallic composites (Ni/YSZ cermets) [29]. In this study, we focused on the fabrication of Ni/GDC anodes using a combination of electroless and electrodeposition techniques. We used Ag to provide a thin but electronically conducting layer for Ni electrodeposition, deposited by an electroless process to form Ni/Ag/GDC anodes. This is part of our efforts to fabricate anodes by metallizing, and a continuation from our previous study in this direction [22]. The fabrication of the scaffold provides additional means of microstructural control, since modification of the screen printing ink with pore formers to control porosity is feasible. In this paper, we present a study with a GDC scaffold to fabricate electroless and electrodeposited Ni/Ag/GDC anodes, assessing electrochemical performance using symmetrical cells and a fuel cell.

GDC ink and scaffold preparation
The GDC ink was prepared based on formulations developed by Somalu [31] and used in our previous studies [20,32]. Ce 0.9 Gd 0.1 O 1.95 powder (Fuel Cell Materials) and dispersant (Hypermer KD15) were thoroughly mixed with ethanol using ball milling for 24 h. After milling the mixture was dried in an oven at 90 C for 24 h. Then, terpineol solvent (Sigma Aldrich) and ethylcelulose N7 grade binder (IMCD UK/Hercules) were added into the milled mixture and stirred continuously to form an ink. The ink was homogenized using a triple roll mill (EXAKT, 80E, Germany) for 20 min. The ink preparation is summarized in Fig. 1.
The GDC scaffold was screen printed onto an 8 mol %-yttria-stabilized zirconia disc with thickness of 290 mm (8YSZ, Fuel Cell Materials). The scaffolds were sintered at

Electroless and electrodeposition
Silver was deposited into the porous GDC scaffold using Tollens' method to provide an electronically conductive layer for electrodeposition. Ni was plated onto the silver layers using an Autolab PGSTAT302N from Ni 2þ aqueous solution i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 1 (Watt's bath) at 70 C. The full details of the process have been described previously [22]. The overall fabrication method is summarized in Fig. 2. Fig. 3 shows the symmetrical and fuel cells configurations used in this study. Symmetrical cells of Ag/GDC (S003) and Ni/ Ag/GDC (S004) were also prepared to further understand the electrode response and to measure the electrical conductivity of the anode respectively. The metal content was determined by the weight change after metal incorporation. The final compositions of the cells are listed in Table 1. A silver paste and platinum mesh were used as the current collector for the fuel cell and symmetrical cell respectively. Both symmetrical and fuel cells were heated in air to oxidize Ni before they were annealed in N 2 and then reduced in 10 vol% H 2 e 90 vol% N 2 at 600 C to minimize the microstructural changes in the nickel-metal layer.

DC conductivity
The DC electrical conductivity of an electrodeposited Ni/Ag/ GDC (S004) anode was measured under reducing atmospheres of humidified-10 vol% H 2 (balance N 2 ) at 500e750 C after preoxidation and annealing in N 2 (as described in Section Electroless and Electrodeposition). The conductivity data was collected using the Van der Pauw method by connecting the 4point probe to an Autolab PGSTAT302N.

Characterization procedure
The electrochemical performance of symmetrical cells with and without Ni-electrodeposit (S001 and S003), including a fuel cell (S002) were measured. In symmetrical cell mode, impedance spectroscopy was used to study anode performance from 600 to 750 C in humidified-H 2 (10e97 vol% H 2 , 3 vol% H 2 O). The impedance data were obtained using a potentiostat (Autolab PGSTAT302N), in the frequency range 10 À1 e 10 5 Hz using an AC signal amplitude of 20 mV and 10 mV for symmetrical and fuel cell modes respectively. Experiments were performed using a two-electrode configuration in the temperature range 550e750 C in fuel cell mode at various concentrations of humidified H 2 and humidified syngas (15 vol% H 2 : 25 vol% CO).
The microstructure of the electrodes was examined using a Scanning Electron Microscopy (LEO Gemini 1525 FEGSEM) and EDX was performed using a JEOL-6400. The value of open porosity for the electrodeposited Ni/GDC was about 50e53 vol% with pore diameters around 100e400 nm. The porosity of the scaffold was calculated from its density, obtained from the mass and volume, and then compared to the theoretical GDC density (7.21 g cm À3 ). Fig. 2b depicts a uniform coverage of Ag metal onto the GDC scaffold as well as a layer of Ni formed after electrodeposition (Fig. 2c). Around 1 h was required to metallize the GDC scaffold with both Ag and Ni; around 7 times faster than current impregnation methods (5 cycles of 0.5 h of decomposition at 500 C and~1 h for heating and cooling) leading tõ 7 vol% of Ni incorporation in the scaffold [20]. The metal content of the cells prepared in this study were obtained by weight difference (Table 1) with a measurement uncertainty in the nickel content up to 25% linked to the instrument accuracy of 0.1 mg and the relatively low metal content. The amount of electrodeposited Ni in the porous GDC scaffolds varied from one cell to another, which may reflect variability in the electrodeposition technique and/or error in the measurement. Despite this uncertainty, the electrodes showed an adequate electrical conductivity (see below) with Ni contents lower than those reported in the literature, which are in the  i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 1 range of 20e60 vol% [26,33,34]. The amount of Ni in cell S001 (7.5 vol%) almost reached the same amount of Ni as that obtained using the impregnation method (8.2e13 vol%) [16,35], however with a significantly faster fabrication method and comparable formation of Ni particles (will be shown later in Figs. 10a and 11a).

Results and discussion
Conductivity of electrodeposited Ni/Ag/GDC Fig. 4 presents the conductivity of S004. Also shown in this figure is the electronic conductivity of Ni and Ag metals [36] as well as GDC [37] for reference. The conductivity decreased to 19 S cm À1 when the temperature was increased to 750 C; indicating the metallic nature of the electrode. This value is lower than values reported in the literature (150e620 S cm À1 ) for electrodes with two to three times higher Ni loading. Although Ni islands in the scaffold limit the electronic conductivity of the electrode, the MIEC properties of GDC in reducing atmospheres could help the conductivity of Ni islands during the measurement [5,38,39]. Furthermore, the contribution of the Ni network on the surface of the electrode (Fig. 10b) to the electronic conductivity needs to be considered in interpreting the Van der Pauw data. Despite the relatively small amounts of metal, the conductivity of the metallized electrode obtained in this study was comparable to the value achieved by combustion synthesized Ni/YSZ (20 S cm À1 ) with 30 vol% of Ni [40], though not as good as that achieved by Ni/YSZ and Ni/GDC cermets using mechanical mixing (~400 S cm À1 , 30e40 vol% Ni) [33,41] and conventional impregnation methods with higher Ni fractions (600e900 S cm À1 , 12e16 vol% Ni) [20,42].

Symmetrical cell mode measurements
Two semicircles were observed in the impedance spectra at most temperatures and atmospheres. The values of polarization resistance were estimated by fitting the data to the equivalent electrical circuit shown in Fig. 5. The intercept of the impedance spectra with the real axis at high frequencies corresponds to the ohmic resistance (R ohm ). The total area specific resistance (ASR) of the electrode was calculated by the difference of the low frequency intercept of the impedance curve with the real axis and the R ohm . Then the value was extracted from the fitting and normalized by dividing by two and multiplying by the electrode area. Fig. 6 indicates a relatively constant R ohm at the same temperature with varying atmospheres. As expected, at higher temperatures R ohm decreases from 8.0 to 5.4 U (600e750 C), since the ionic conductivity of the electrolyte is thermally activated. The values of experimental R ohm were higher than the expected values calculated from the YSZ thickness and known conductivities. This difference might be due to contact resistance, and/or an ohmic contribution from the GDC scaffold, and/or a contribution from a reaction layer at the GDC-YSZ interface [43]. This interdiffusion layer (GDC-YSZ) may create lower ion conductivity, almost two orders of magnitude lower than that of YSZ and GDC at 800 C [44].
The total ASR, which characterizes the electrochemical performance of the electrodes, are tabulated in the inset tables of Fig. 7 and Fig. 8. The low frequency impedance response, R_l was observed to dominate the total ASR at all concentrations of H 2 and at different temperatures. The lowest total ASR reached 1.12 U cm 2 at 750 C in 97 vol% H 2 . The total ASR obtained in this study is higher than state of the art conventional and impregnated Ni/GDC and Ni/YSZ cermets in the literature, which are usually in the range of 0.15e0.35 U cm 2 [20,34,45e47]. Although a faster metallization    1 mm). The line obtained in this study was extrapolated at temperatures <500 C. Also shown in the graph are the conductivity of Ni and Ag metals, and impregnated 12 vol%-Ni/GDC [20].
process of the scaffolds has been demonstrated in this study, it must be noted that the amount and distribution of Ni electrodeposited in the scaffold (7.5 vol%) still requires further optimization before state of the art electrode performance can be achieved. However, this work demonstrates the promise of using electroless and electrodeposition technique for the fabrication of scaffold electrodes. Fig. 7 displays the effect of H 2 concentration on the complex impedance spectra on S001 and S003 cells (impedance data was normalized to zero) at fixed T ¼ 750 C. The polarization associated with the intermediate-frequencies arc (10 2 e10 3 Hz), R_m was observed to be constant at 750 C in all H 2 concentrations with a value of 0.06 ± 0.01 U cm 2 for S001. A blank symmetrical cell (without Ni, S003), was prepared to measure the electrochemical response of the Ag/GDC scaffold. The R_m value of S003 (0.04 U cm 2 ) was very close to the values attained in S001, and these values are in accordance to those reported in the literature, typically at <0.1 U cm 2 [48,49]. This is clearly shown by the inset of Fig. 7. In Fig. 8, the R_m values significantly decreased from 0.25 to 0.05 U cm 2 when the temperature was increased to 750 C. These results suggest that the R_m was associated with the charge transfer kinetics of hydrogen oxidation at the GDC scaffold.
The low-frequency (<10 2 Hz) arcs were always larger than the intermediate-frequency arcs, and became more significant at low temperatures and low H 2 concentrations as shown in Figs. 7 and 8 respectively. The lowest low-frequency polarization (R_l) obtained in this study was 1.07 U cm 2 for 97 vol % H 2 at 750 C. However, this value is higher than those reported for GDC-impregnated Ni [50], Ni/GDC cermets [34] and Ni-impregnated GDC [20] ranging from 0.1 to 0.4 U cm 2 at temperatures of 690e800 C. There are several reasons that could explain these observations: (i) The high R_l values could be linked to the limited electronic conductivity of the electrodes, closely related to the amount of metal deposited and the mixed conductivity of GDC. The value of R_l for S003 is three times higher than S001 in the same operating conditions, but with no Ni. This indicates that the electronic conductivity of the GDC scaffold contributes to the electrochemical activity. (ii) The "chemical capacitance" effect that originates from the variation in the oxygen non-stoichiometry (oxygen vacancies) of GDC in reducing atmospheres, complicates the interpretation of the R_l data [7,51]. (iii) The contribution of diffusional processes to the R_l cannot be ignored as the R_l changed with gas composition (Fig. 7). This is probably attributed to: (1) the diffusion of reactant gas to and reaction products from the reaction sites (TPB and 2 PB) in the porous scaffold [52]; (2) Knudsen effects due to the small pore size of the scaffold (~0.25 mm) [49,53]; and/or (3) gas conversion impedance (sensitive to gas composition and temperature) [34,48,54].
Further investigation to resolve these effects is on-going. The slope of the Arrhenius plots of the electrode polarization resistances (R_m and R_l) were used to calculate the activation energy for S001 and S003 as depicted in Fig. 9. From the inset table (Fig. 9), the activation energy for R_m changed from 1.49 to 0.85 eV (97 vol% e humidified H 2 ) when small amounts of Ni were added to the GDC scaffold. The activation energy for R_m of S001 did not show any significant variation (0.85 ± 0.03 eV) at different concentrations of humidified H 2 eN 2 (10e97 vol%). This shows that the kinetics of hydrogen oxidation on the surface of the electrode is improved by adding Ni.
For R_l, the activation energies were lower than for R_m, and significantly decreased from 0.7 to 0.19 ± 0.04 eV when Ni was added. This may be attributed to the limited electronic conductivity of the GDC scaffold and/or diffusional processes among other effects [55,56]. The small change in the activation energies of R_l with temperature (600e750 C) might be  attributed to surface diffusion and/or adsorption/dissociation of hydrogen species. These are restricted by the conductivity of the GDC scaffold [34,57]. However, further investigation is required to understand this behavior. Overall, the activation energy obtained in this study is comparable to the values reported in the literature with different fabrication methods and anodes (see Table 2).
The microstructure of S001 was examined after electrochemical testing (see Fig. 10). Since there was not enough Ni (7.5 vol%) deposited inside the scaffold to form a continuous film, individual nano-particles of Ni could be observed (Fig. 10a). The Ni distribution and particle sizes (0.05e0.2 mm) are similar to the electrode fabricated by five cycles of the impregnation process (Fig. 11a). Furthermore these particle sizes are considerably smaller than those found in the conventional powder mixing method (1.4e5 mm) [62,63]. This shows that decoupling scaffold fabrication and metal incorporation into the scaffold provides independent control of metal particle size, porosity and TPB density compared to conventional electrodes fabrication (powder mixing). Percolating and homogenous Ni distribution inside the electrodes could be achieved with 50 vol% and 10e20 vol% of Ni by conventional and impregnated electrodes respectively, much higher than the amount of Ni in this study. The challenge to the method proposed here is to increase the amount of Ni added into the GDC pores at lower temperatures ( 70 C) and with greater speed than the impregnation method. Fig. 10b displays a Ni network on the GDC surface. If this had formed throughout the porous GDC scaffold, it would have provided better electronic pathways from the reaction sites to the current collector. Moreover, the thickness of the scaffold obtained in this study is about 7 mm (Fig. 10c), within the electrochemically active reaction limit ( 10 mm) [64,65]; increasing its thickness may further improve the overall performance.
Under the experimental conditions, silver does not alloy with nickel [66,67] and does not have a high wettability on oxide substrates [68]. Ag films on YSZ become thermally unstable and eventually the film segregates and agglomerates to form isolated islands at high temperatures [69,70]. These islands are clearly shown in Fig. 11b when a Ni/Ag film was heated in air at 800 C for 1 h. The anode is usually preoxidized at high temperature and annealed in nitrogen before its measurement in reducing atmospheres (Step 4 in Fig. 2). The identification of elements in Figs. 10 and 11 was possible in combination with the results from EDX (not shown). In the areas analyzed (Fig. 10), despite the initial presence of silver (15.1 vol%), this metal was not detected (EDX detection limit >0.05 vol%). This suggests that the Ag evaporated during operation at high temperatures, as observed by Wang et al. [68] in air. Evaporation of Ag might be further exacerbated in reducing atmospheres [71].  i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 1

Fuel cell mode measurements
As a proof of fuel cell concept using electrodeposited Ni/Ag/ GDC, S002 was operated under 50 vol% H 2 -50 vol% N 2 and syngas at 750 C. The voltage and power densities as a function of current density are shown in Fig. 12a. Most losses were ohmic in nature, dominated by the thick YSZ electrolyte (290 mm). Reflecting this, the maximum power densities were modest and reached about 56 mW cm À2 at~0.7 V under 50 vol% H 2 -50 vol% N 2 with an ASR including electrolyte and electrodes of 4.79 U cm 2 . This value is consistent with the total cell polarization obtained from the total impedance in Fig. 12b of   4.81 U cm 2 . The maximum power density achieved by S002 operating under syngas was 33 mW cm À2 at~0.5 V. Future work will focus on depositing these electrodes onto much thinner electrolytes.

Conclusions
The successful deposition of silver and nickel by electroless and electrolytic methods into porous GDC scaffold was achieved. This technique is much faster than conventional impregnation methods, however the amount and  i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 1 distribution of Ni deposited in the scaffold needs to be further optimized to achieve state of the art electrode performance. Metallizing the scaffold took place at temperatures lower than 100 C, well below the temperatures used to process impregnated anodes. After operation at higher temperatures (600e750 C), there is evidence for the formation of islands of nickel inside the scaffold that increase the TPB of the electrode, when used on a mixed conducting GDC scaffold. On the electrode surface, the nickel forms an interconnected network that can be used as a current collector. The main role of Ag is to provide a conductive surface for Ni electrodeposition. Its contribution to the electronic conductivity of the scaffold and electrochemical reactions might be minimal. In regions analyzed by EDX, silver was not detected, probably indicating agglomeration in other regions, migration to the current collector or evaporation due to high temperature and reducing environments.
In the impedance measurements, there were two types of electrode response. The first appeared at intermediate frequencies (100e1000 Hz), and the second at low frequencies (<100 Hz), showing a polarization resistance of 0.05 U cm 2 and 1.07 U cm 2 respectively, with a total ASR of 1.12 U cm 2 at 750 C (symmetrical cell mode, S001). The activation energy of the electrode at R_m is always higher than at R_l, being 0.85 ± 0.03 eV and 0.19 ± 0.03 eV respectively. The cells are able to be operated under both H 2 and syngas (fuel cell mode). Whilst the distribution and amount of Ni needs to be further improved, this fabrication technique offers an alternative for processing SOFC anodes, especially for large scale production.