- Email: [email protected]
Change of an anode's microstructure morphology during the fuel starvation of an anode-supported solid oxide fuel cell
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Change of an anode's microstructure morphology during the fuel starvation of an anode-supported solid oxide fuel cell Grzegorz Brus a,b,*, Kota Miyoshi a, Hiroshi Iwai a, Motohiro Saito a, Hideo Yoshida a a b
Kyoto University, Department of Aeronautics and Astronautics, Kyoto, Japan AGH University of Science and Technology, Faculty of Energy and Fuels, Krakow, Poland
article info
abstract
Article history:
A RedOx cycle is one of the most dangerous factors of anode degradation due to micro-
Received 30 January 2015
structure changes. In this paper, the evolution of anode microstructure during a RedOx
Received in revised form
cycle caused by a fuel starvation process was studied using electrochemical measure-
24 March 2015
ments. After a power generation experiment, the anode microstructure was reconstructed
Accepted 28 March 2015
using a combination of focused ion beam and scanning electron microscopy (FIBeSEM).
Available online 22 April 2015
The microstructure changes were quantified for parameters such as the tortuosity factor, tripe phase boundary density, volume fraction, connectivity and average grain size. The
Keywords:
three dimensional microstructure reconstruction indicated that the fuel starvation con-
Solid oxide fuel cells
dition might cause significant changes in microstructure morphology.
Redox cycle
Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Fuel starvation
reserved.
Focused ion beam Microstructure morphology
Introduction Solid Oxide Fuel Cells (SOFCs) are electrochemical devices that can convert the chemical potential of fuels directly into electricity. Because of their high working temperature (600 Ce1000 C), SOFCs have the highest energy conversion rate among all types of fuel cells. A solid oxide fuel cell consists of two porous ceramic electrodes (cathode and anode) separated by a dense solid ceramic electrolyte made of yttria stabilized zirconia phase (YSZ). The electrode microstructure morphology is an important factor determining its electrochemical performance. A typical anode consists of porous Ni/
YSZ and a typical cathode is made from porous lanthanum strontium cobalt ferrite (LSCF). Each component plays a unique and important role in the transport process by providing a pathway for different species; a YSZ phase for oxygen ions, Ni and LSCF phases for electrons and a pore phase for gases. While the SOFC slowly reaches its commercialization, it becomes a key issue to guarantee long term and safe operation. It is generally accepted that to satisfy the market, 40 000 h of safe operation is required [1,2]. One of the most dangerous causes of anode degradation is the oxidation of Ni particles. Because NiO occupies more volume than Ni, oxidation results in internal stresses inside an anode. A further reduction of NiO to Ni results in shrinking of the
* Corresponding author. Kyoto University, Department of Aeronautics and Astronautics, Thermal Engineering Laboratory, Kyoto daigaku-Katsura, Nishikyo-ku, Kyoto 615-8530, Japan. Tel./fax: þ81 075 383 3652. E-mail addresses: [email protected], [email protected] (G. Brus). http://dx.doi.org/10.1016/j.ijhydene.2015.03.143 0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
6928
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
particle volume. This expansion and contraction of Ni particles is called in the literature a RedOx cycle and may cause damage to a cell microstructure. There are several possible causes of a RedOx cycle such as i) the thermal cycle of a SOFC system during shut down and start up, ii) the stop of gas supply in the case of an emergency situation, iii) oxidation from steam in the case of SOFC with a reforming system, iv) leakage through the sealing, v) local fuel starvation caused by a load following operation, vi) too high a hydrogen conversion rate and vii) high fuel dilution. From a reaction point of view, they can be classified into two groups. One is the redox by chemical reaction (Ni þ O2 / NiO) and/or by electrochemical reaction (Ni þ O2 / NiO). The first group, redox by chemical reaction, is often studied in laboratory scale experiments by stopping the fuel supply and blowing air to the anode side. After the oxidation of nickel particles, air is stopped and hydrogen is fed again to the anode to finish the RedOx cycle. This approach was studied for both electrolyte supported cells [3e20] and anode supported cells [21e26]. Conclusions were different from one paper to another. Some authors report massive degradation, other authors report moderate or find no significant degradation. In the second group, the anode is electrochemically oxidized. Electrochemical oxidation will happen rather in the vicinity of the anode-electrolyte interface. For laboratory scale experiments, current is applied to the cell to force oxygen ions to go through the electrolyte and oxidize Ni at the anode-electrolyte interface. In the case of the electrolyte supported cell, Takagi et al. [27] found moderate degradation that was not fatal for the cell. In the case of the anode supported cell, the conclusions are in contradiction with each other. Sarantaridis et al. [28] has reported a catastrophic effect of the redox cycle on the cell whereas Hatae et al. [29] reported no electrolyte crack or OCV drops. An outstanding research to conclude all the above mentioned cases was conducted by Laurencin et al. [20,30,31]. In their research they investigated both anode supported [30] and electrolyte supported [20] cells as well as different types of oxidation: chemical and electrochemical [30]. Impedance spectroscopy was used to estimate the electrochemical degradation. Their results indicated probability of delamination at the anode-electrolyte interface and mechanical damage of the cell as a result of stress [31]. To estimate the probability of the cell failure, the thermo-mechanical model has been proposed [31]. Recent research suggests that fuel starvation might occur during transient operations, if fuel is consumed in the cell faster than it can be supplied [32,33]. When fuel starvation occurs during a real SOFC operation, the cell might be strongly polarized and oxygen ions keep coming out from the cathode side which causes electrochemical oxidation of Ni in the anode. In this study we focus on this electrochemical oxidation of Ni by fuel starvation. Recent results by Fang et al. [34] show the significance of this problem. In their research, a stack of cells was investigated under high current density and high fuel utilization factor. Their research indicates that the upstream part of the anode was oxidized as a result of fuel starvation. At the laboratory scale, this situation was studied recently by different research groups [35,36]. In their research using diluted hydrogen as fuel, it was observed that cell performance drops rapidly at very low hydrogen concentration
(below 2%). Further investigation unraveled the oxidation of nickel near the anode-electrolyte interface [35,36]. Chen et al. [36] have reported the fatal failure of an anode-supported cell after exposing the cell to fuel starvation conditions [36]. Throughout the research an explanation of the cell failure by fuel starvation was provided from a macroscopic point of view. However the impact of fuel starvation on the electrode microstructure remains unclear. The most precise information about cell microstructure can be derived from real structural analysis. Recently, the combination of a focused ion beam and scanning electron microscope (FIBeSEM) as well as the X-ray tomography technique brought a breakthrough in the direct 3D observation of porous structure [37e43]. 3D structure reconstruction was introduced to the field of SOFCs by Wilson et al., in 2006 [44]. The method enables the observation of many sequential 2D images of a porous microstructure and reconstructs it into a 3D structure using advance image processing. From the reconstructed microstructure, it is possible to evaluate the microstructure parameters. These parameters directly obtained from the real electrode structure are the key to considering the relationships between porous microstructures and the cell power generation performance. There were attempts to use the 3D microstructure observation to investigate microstructure evaluation during the RedOx cycle [9,27,45], but to the authors' best knowledge, it has not been applied to the fuel starvation problem. The literature review clearly unravels the gap in published data that the authors aim to fill by providing detailed information about microstructure morphology change, during the fuel starvation of an anode supported solid oxide fuel cell. The novelty lays in using state-of-the-art techniques such as FIBeSEM, combined with advanced methods of deriving microstructure parameters to report the changes caused in microstructure during the fuel starvation process.
Experiment Experimental set-up A conventional double-tube type test apparatus for SOFC button cell evaluation was used (BEL Japan, Inc., BEL-SOFC). The SOFC sample was located between two ceramic tubes in the electric furnace. The ceramic tubes have a double co-axial structure. The fuel/air is supplied through the inner-tube and exhausted through the gap between the inner-tube and the outer-tube. The furnace was heated-up to 800 C. Fuel was fed to the anode via flow controllers omitting a humidifier. Air was used as an oxidant on the cathode side. The anode and the cathode were connected to the measuring devices with 0.5 mm platinum wires. Platinum meshes were welded with platinum wires and connected with each electrode. An electrochemical Interface (Solartron analytical, model 1287A) was used for the currentevoltage characteristic measurements.
Experimental method In the experiment, three commercial samples provided by SOFC Power were used. The Ni/YSZeYSZeGDC/LSCF samples
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
were cut by a laser technique from a single piece of the large size cell. Therefore, the microstructure of each anode is assumed to be the same. The technical informations related to the SOFC preparation such as the initial weight percentage of NiO to YSZ, sintering temperature, pore formers etc. are protected by intellectual property rights of the company and therefore confidential. Sample 1 was reduced using a mixture of 60% hydrogen and 40% nitrogen at 800 C for 1 h. The hydrogen concentration was increased gradually to reach 60% at the beginning of reduction process. Next, the cell was kept under a polarized condition of 0.8 V and fed with 60 ml/min of H2 and 40 ml/ min N2 for 10 min and then cooled down at OCV conditions, under reducing atmosphere. After the sample had cooled down, it cooling down the sample was investigated using FIBeSEM and the data was used to obtain a 3D reconstruction of the sample. Sample 2 was reduced at the same condition described in the previous paragraph. Then the cell was kept under a polarized condition of 0.8 V and fed 60 ml/min of H2 and 40 ml/ min N2. After stabilization, the hydrogen supply was stopped. When the current dropped to zero, hydrogen was fed again. After the electrochemical measurements, the sample was cooled down to room temperature using only a supply of pure nitrogen. After the experiment, the sample was investigated using FIBeSEM and the data was used to obtain a 3D reconstruction of the sample's microstructure. Sample 3. To determine if the nickel was oxidized during fuel starvation, an additional experiments were conducted. In this experiment the sample was kept under the same conditions as described in the previous paragraph. After reaching stabilization, the hydrogen valve was closed. When the current dropped to zero, the oxygen valve was also closed on the cathode side. The cell potential was kept constant at 0.8 [V] until the current dropped to zero then polarization was stopped. Next the furnace was cooled down to room temperature, with pure nitrogen flowing on both sides; the anode and the cathode. After cooling down to room temperature, the anode-electrolyte interface was investigated using EnergyDispersive X-ray spectroscopy (EDX). Additionally, several points from the EDX mapping analysis were selected to conduct a more precise point analysis (see Table 1).
sand paper and diamond paste to prepare the sample for the FIBeSEM observation.
FIBeSEM observation A three dimensional structure of the anode was observed using the FIBeSEM system installed at Kyoto University, Japan. The configuration of the FIBeSEM setup is presented in Fig. 1. The procedure can be summarized by the following steps: i) a trench is fabricated to give access to the sample's intersection. ii) a SEM picture of an observed intersection is taken using an in-lens secondary electron detector (Mag ¼ 3 k, EHT ¼ 1.5 kV, WD ¼ 5 mm). iii) FIB uses a beam of Ga þ ion to mill the sample to expose another intersection and iv) a “cut and see” procedure is repeated to obtain the sequence of 2D images. The idea of the “cut and see” procedure and the example of SEM images are presented in Fig. 1.
Phase separation and 3D reconstruction The SEM image segmentation is a crucial step in the analysis of the data. A phase separation was conducted based on image brightness using AVIZO software (Mercury Computers Systems, Inc.). Fig. 2 presents an example of the phase separation process. The pixel was 48.51 nm. After the phase separation, the 3D microstructure was reconstructed using a set of 2D images. Fig. 3(a) and (b) show the reconstructed microstructure of the porous anode.
Quantification of porous structure The microstructural parameters of the anode were evaluated using three dimensional structure reconstruction.
Volume fraction (ε) The volume fraction of each phase was measured by counting the number of voxels corresponding to Ni, YSZ and pores, respectively [46].
Average grain size (d) The average grain size was evaluated by the three dimensional version of the line-intercept method [46]. In this
FIBeSEM imaging Sample preparation for the FIBeSEM observation After the power generation experiment, Sample 1 and Sample 2 were infiltrated using epoxy resin (Marumoto Struers KK) under vacuum conditions. Filling the pores with resin is important for recognizing the pore region during SEM observation. The infiltrated samples were cut and polished using Table 1 e Summarized properties of the anode before and after fuel starvation experiment.
Point 1 Point 2 Point 3 Point 3
Ni
O
Zr
Species
92.09 74.31 27.67 0.17
3.08 23.86 68.68 68.04
4.82 1.83 3.65 31.79
Nickel Nickel oxide Nickel oxide YSZ
6929
Fig. 1 e Schematic view of FIBeSEM settings and the measuring procedure.
6930
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
Fig. 3 e Three dimensional reconstruction of a) Sample 1 (reference sample) b) Sample 2 (after the RedOx cycle). The green color represents Ni, the yellow is YSZ and the red are pores. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 2 e An example of image analysis of the epoxy impregnated NieYSZ (sample 2).
media is lower than that obtained in a free space. The degree of reduction is described quantitatively using the tortuosity factor. The details of the procedure are described by Kishimoto et al. in reference [47].
Phase boundary length density (ltpb) method lines are drawn in three directions along orthogonal coordinates. The length of the intercept lines is measured and the average value is calculated to obtain a mean particle size.
Tortuosity factor (t) The tortuosity factor was quantified by a diffusion based simulation based on a random walk process [47]. In this method, a large number of random walkers are stochastically distributed in each phase voxels. In the next time step each walker randomly migrates into one of its neighbor voxels. However, if the selected neighboring voxel is not the same phase then the walker stays in the current position and waits for the next time step. The process is repeated and the mean square displacement of random walkers is calculated. Due to the obstacles, the mean square displacement in the porous
The Tripe Phase Boundary length density was determined using the volume expansion method [38] and AVIZO software (Mercury Computers Systems, Inc.). In this method each phase is slightly expanded in a computational domain. The overlapped regions form string-like volumes that contained TPB lines inside. Centerlines of those strings are taken as lines that represent the TPBs, and their lengths are measured and divided by the volume of the investigated sample giving the value of the TPB density [38].
Phase connectivity To discuss the transport phenomena via each phase, only the connected phase should be taken into consideration. The isolated part of a phase, which does not contribute to the transport, should be excluded. In the presented study, 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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
isolated phase was excluded by indicating the phase that penetrates the reconstructed structure in a certain direction [47].
Results and discussion The results of the electrochemical measurements for Sample 2 are presented in Fig. 4(a). As can be see in Fig. 4(a) measurements can be divided into the following zones: i) normal operation ii) after the hydrogen stops, the current drops insignificantly and hydrogen leftovers are used iii) the oxidation of nickel particles iv) a short recovery after flowing back hydrogen v) the decay of properties vi) the acceleration of decay vii) end of cell's life. The experiment was repeated twice and confirmed the reproducibility of the results. An analogical trend has been observed in the past [36] and was associated with the oxidation of nickel particles. The results of the electrochemical measurements for Sample 3 are presented in Fig. 4(b). As can be seen in Fig. 4(b), Sample 3 was exposed only to fuel starvation without feeding hydrogen to finish the redox cycle. After the electrochemical measurements Sample 3 was
Fig. 4 e Electrochemical measurements.
6931
analyzed using EDX to investigate possible presence of nickel oxide. The results of the EDX mapping is presented in Fig. 5. As can be seen in Fig. 5 nickel was oxidized near the electrolyte interface even though oxygen was not introduced to the anode side. The oxygen ions must have come from the cathode side during polarization. Since hydrogen was not present on the anode side, oxygen reacted with the nickel creating nickel oxide until the electrochemical reactions stopped. An EDX point measurement was performed at 4 different positions marked in Fig. 5 and results are summarized in Table 1 it confirms that nickel oxide is present in the vicinity of the anode-electrolyte interface. The characteristic shape of nickel oxide can be recognized using scanning electron microscopy (SEM) [48]. Fig. 6 presents SEM images of the anode microstructure near the anode surface Fig. 6(a) and near the vicinity of the anode-electrolyte interface 6(b). As can be seen in Fig. 6 the morphology of the Ni particles is completely different. Fig. 6(b) indicates typical nickel oxide structures, not present in Fig. 6(a). A closer view at the nickel oxide particles can be seen in Fig. 7 which supports the presented conclusions. It should be noted that the OCV of all investigated samples reached initially 1.2 V at 800 C in 60% H2 and 40% N2. This indicates that the YSZ thin electrolyte prevented crossover of the gases and the cell was well sealed. The results of the microstructure evaluation of Sample 1 (reference sample) and Sample 2 (after fuel starvation) are presented in Table 2. As can be seen in Table 2, the RedOx cycle causes significant changes in the sample microstructure. The tortuosity factor of the nickel phase increases from 7.24 to 14.91. At the same time, the connectivity of the nickel phase drops from 97 % to 89 %. The isolated nickel particles before and after the fuel starvation experiment are presented in Fig. 8. Those results can be explained by the expansion and contraction of nickel particles which results in particles
Fig. 5 e EDX mapping of the region near the anodeelectrolyte interface taken from sample 3. The green color represents Zr, the red is oxygen and the blue Ni. Details of the observation: Mag. 16 k, EHT ¼ 10 kV, WD ¼ 5.0. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
6932
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
Fig. 6 e SEM images of sample 3.
Fig. 7 e SEM image of a nickel oxide particle near the anode-electrolyte interface taken from sample 3.
Fig. 8 e Isolated nickel particles a) reference sample b) after fuel starvation condition.
fragmentation. This phenomenon can be confirmed by the decrease in nickel particle size and connectivity. When connectivity is lost, the isolated nickel which no-longer contributes to the transport phenomena is excluded from the tortuosity factor calculation. As a consequence, the tortuosity factor increases. During the redox cycle, the generated steam causes multi-cracks and the tortuosity of pore size becomes smaller. As a consequence of microstructure fragmentation, the TPB length become higher. The tortuosity factor values agree with those published in the literature when a similar phase volume fraction is considered [49]. The tortuosity factor is affected by the volume fraction and its effect is particularly strong at low volume fractions below 30%. A typical value of the tortuosity factor is between 1.5 and 4. However, in cases of low porosity of the anode, it can easily reach a value as high as 10 [49]. Compared to the Ni and pore phases, the
Table 2 e Summarized microstructure properties of the anode before and after fuel starvation experiment.
Ni YSZ Pore ltpb
ti [e]
d [mm]
εi [%]
Connectivity [%]
Before / after
Before / after
Before / after
Before / after
7.24 / 14.91 2.05 / 2.26 7.18 / 4.54 3.1 / 4.0
0.88 / 0.84 0.95 / 0.86 0.81 / 0.76 [mm2]
29 / 27 45 / 43 25 / 30 ltpb,active
97 / 89 100 / 100 99 / 99 2.1 / 3.0
[mm2]
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
microstructure change of YSZ phase is small as can be seen in Table 2. It supports a common understanding that the YSZ phase is relatively stable. The expansion and contraction of Ni phase, therefore, mainly affects the microstructure of Ni phase itself and the pore phase as a result. In Table 2, it is also shown that the particle diameters of three phases have decreasing tendency after the redox cycle. It is consistent with the increase of the TPB density shown in the same table because a higher TPB density is generally expected for an electrode with finer particles. The total TPB density increases from 3.1 [mm2] to 4.0 [mm2]. The comparison with already published data indicates some similarities and differences with the case where air was blown to the anode to cause the redox cycle [45]. An average Ni particles gets smaller after the redox cycle. Same tendency can be observed in the case of average pore size. Those changes lead in both cases to increase of TPB length density after redox cycle. Also in both cases, fuel starvation and air blow lead to decrease of Ni volume fraction after single redox cycle. However the noticeable difference is in connectivity of Ni particles. In case of air blow the connectivity slightly increased after redox cycle. In contradiction, a significant drop of Ni connectivity was observed in case of fuel starvation. An observation using a scanning electron microscope indicates that nickel near the anode-electrolyte interface is oxidized under fuel starvation conditions. The number of isolated nickel particles increased after one redox cycle. But generally, the microstructure of the anode recovers, as can be seen in Table 2. From a microstructure point of view, the anode after one redox cycle should have almost the same performance as the anode before the redox operation. The short recovery of the performance observed in Sample 2 indirectly supports it. However, Sample 2 died after the fuel starvation experiment. This is inconsistence with microstructure measurements. Therefore the cell failure observed in Fig. 4(a) cannot be explained only from microstructure change. The obtained data can support thermo-mechanical model that should be further investigated.
Conclusions The presented paper reports the microstructure changes of an anode electrode of an anode-supported solid oxide fuel cell after a RedOx cycle. The redox cycle was conducted by stopping fuel supply at polarized condition to simulate a fuel starvation scenario of a SOFC operation. The microstructure of the sample after a RedOx cycle and the reference sample were reconstructed using the FIBeSEM technique and microstructure parameters were evaluated. The presented results depict the possible degradation effects after fuel starvation that can happen at the downstream of the anode. The connectivity of the Ni particles was significantly lost in the conducted experiment. The lost of connectivity was coupled with an increase of the tortuosity factor of nickel and a decrease in the tortuosity factor of pore phase. The average grain size of each phase dropped suggesting strong fragmentation of microstructure as a consequence of the volume change of nickel particles over a RedOx cycle.
6933
Acknowledgments The first author was supported by a JSPS Postdoctoral Fellowship for Foreign Researchers. Additionally, the present work was partially supported by the PAN-JSPS Joint Research Project “Thermal Interaction between Stack and Reformer in Small Scale SOFC” and partially by the New Energy and Industrial Technology Development Organization (NEDO).
references
[1] Comminges C, Fu QX, Zahid M, Steiner NY, Bucheli O. Monitoring the degradation of a solid oxide fuel cell stack during 10,000 h via electrochemical impedance spectroscopy. Electrochim Acta 2012;59:367e75. [2] Schuler JA, Wuillemin Z, Hessler-Wyser A, Comminges C, Steiner NY. Cr-poisoning in (La,Sr)(Co,Fe)O3 cathodes after 10,000 h SOFC stack testing. J Power Sources 2012;211:177e83. [3] Cassidy M, Lindsay G, Kendall K. The reduction of nickel zirconia cermet anodes and the effects on supported thin electrolytes. J Power Sources 1996;61:189e92. [4] Waldbillig D, Wood A, Ivey DG. Electrochemical and microstructural characterization of the redox tolerance of solid oxide fuel cell anodes. J Power Sources 2005;145:206e15. [5] Modena S, Ceschini S, Tomasi A, Montinaro D, Sglavo VM. Reduction and reoxidation processes of NiOYSZ composite for solid oxide fuel cell anodes. J Fuel Cell Sci Technol 2006;3:487e91. [6] Marrero-Lopez D, Pena-Martinez J, Ruiz-Morales JC, Gabas M, Nunez P, Aranda M, et al. Redox behaviour, chemical compatibility and electrochemical performance of Sr2 MgMoO6d as SOFC anode. Solid State Ionics 2010;180:1672e82. [7] Iwanschitz B, Sfeir J, Mai A, Schutze M. Degradation of SOFC anodes upon redox cycling: a comparison between Ni/YSZ and Ni/CGO. J Electrochem Soc 2010;157:B269e78. [8] Sumi H, Lee Y-H, Muroyama H, Matsui T, Eguchi K. Comparison between internal steam and CO2 reforming of methane for Ni-YSZ and Ni-SCSZ SOFC anodes. J Electrochem Soc 2010;157:B1118e25. [9] Muroyama H, Sumi H, Kishida R, Kim J-Y, Matsui T, Eguchi K. Correlation between microstructure and electrochemical characteristics of Ni-YSZ anode subjected to redox cycles. ECS Trans 2011;35:1379e87. [10] Sarantaridis D, Chater RJ, Atkinson A. Changes in physical and mechanical properties of SOFC Ni-YSZ composites caused by redox cycling. J Electrochem Soc 2008;155:B467e72. [11] Glauche A, Betz T, Mosch S, Trofimenko N, Kusnezoff M. Long-term, redox and Thermal cycling stability of electrolyte supported cells. In: 216th ECS meeting. ECS; 2009. p. 411e9. [12] Klemenso T, Mogensen M. Ni-YSZ solid oxide fuel cell anode behavior upon redox cycling based on electrical characterization. J Am Ceram Soc 2007;90:3582e8. [13] Klemenso T, Appel CC, Mogensen M. In Situ observations of microstructural changes in SOFC anodes during redox cycling. Electrochem Solid State Lett 2006;9:A403e7. [14] Zhang Y, Liu B, Tu B, Dong Y, Cheng M. Redox cycling of NiYSZ anode investigated by TPR technique. Solid State Ionics 2005;176:2193e9. [15] Nakamura K, Somekawa T, Baba Y, Horiuchi K, Matsuzaki Y, Yoshimoto M. Expansion and shrinkage in the redox process of (Mg, Ni) O solid solutions for preparation of steam
6934
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
reforming catalysts in SOFCs. J Ceram Soc Jpn 2009;117:166e70. Pihlatie M, Kaiser A, Mogensen M. Redox stability of SOFC: thermal analysis of NiYSZ composites. Solid State Ionics 2009;180:1100e12. Shearing PR, Bradley RS, Gelb J, Tariq F, Withers PJ, Brandon NP. Exploring microstructural changes associated with oxidation in NiYSZ SOFC electrodes using high resolution X-ray computed tomography. Solid State Ionics 2012;216:69e72. Stathis G, Simwonis D, Tietz F, Moropoulou A, Naoumides A. Oxidation and resulting mechanical properties of Ni/þY2O3stabilized zirconia anode substrate for solid-oxide fuel cells. J Mater Res 2002;17:951e8. Blennow P, Hansen KK, Wallenberg L, Mogensen M. Electrochemical characterization and redox behavior of Nbdoped SrTiO3. Solid State Ionics 2009;180:63e70. Laurencin J, Delette G, Sicardy O, Rosini S, Lefebvre-Joud F. Impact of redox cycles on performances of solid oxide fuel cells: case of the electrolyte supported cells. J Power Sources 2010;195:2747e53. Malzbender J, Wessel E, Steinbrech RW. Reduction and reoxidation of anodes for solid oxide fuel cells. Solid State Ionics 2005;176:2201e3. Hatae T, Matsuzaki Y, Yamazaki Y. Study on electrochemical re-oxidation of anode-supported solid oxide fuel cells using oxide ion current. Solid State Ionics 2008;179:274e81. Ettler M, Timmermann H, Malzbender J, Weber A, Menzler NH. Durability of Ni anodes during reoxidation cycles. J Power Sources 2010;195:5452e67. Faes A, Nakajo A, Hessler-Wyser A, Dubois D, Brisse A, Modena S. Redox study of anode-supported solid oxide fuel cell. J Power Sources 2009;193:55e64. Ihringer R. 2R-CellTM: reliability of anode supported thin electrolyte upon multi thermo and redox cycles. In: 216th ECS Meeting. ECS; 2009. p. 473e83. Monzon H, Laguna-Bercero MA. Redox-cycling studies of anode-supported microtubular solid oxide fuel cells. Int J Hydrogen Energy 2012;37:7262e70. Takagi N, Nakamura Y, Shikazono N, Kasagi N. Microstructure change of SOFC anode caused by electrochemical redox cycles. In: Proceedings of the 9th European solid oxide fuel cell forum; 2010. p. 7e135. Sarantaridis D, Rudkin RA, Atkinson A. Oxidation failure modes of anode-supported solid oxide fuel cells. J Power Sources 2008;180:704e10. Hatae T, Matsuzaki Y, Yamashita S, Yamazaki Y. Current density dependence of changes in the microstructure of SOFC anodes during electrochemical oxidation. Solid State Ionics 2009;180:1305e10. Laurencin J, Delette G, Morel B, Lefebvre-Joud F, Dupeux M. Solid oxide fuel cells damage mechanisms due to Ni-YSZ reoxidation: case of the anode supported cell. J Power Sources 2009;192:344e52. Laurencin J, Roche V, Jaboutian C, Kieffer I. Ni-8YSZ cermet re-oxidation of anode supported solid oxide fuel cell: from kinetics measurements to mechanical damage prediction. Int J Hydrogen Energy 2012;37:12557e73. Komatsu Y, Brus G, Kimijima S, Szmyd JS. The effect of overpotentials on the transient response of the 300W SOFC cell stack voltage. Appl Energy 2014;115:352e9.
[33] Fardadi M, McLarty DF, Brouwer J, Jabbari F. Enhanced performance of counter flow SOFC with partial internal reformation. Int J Hydrogen Energy 2014;39:1975366. [34] Fang Q, Blum L, Peters R, Peksen M, Batfalsky P, Stolten D. SOFC stack performance under high fuel utilization. Int J Hydrogen Energy 2015;40:112836. [35] Brus G, Miyawaki K, Iwai H, Saito M, Yoshida H. Tortuosity of an SOFC anode estimated from saturation currents and a mass transport model in comparison with a real microstructure. Solid State Ionics 2014;265:13e21. [36] Chen G, Guan G, Abliz S, Kasai Y, Abudula A. Rapid degradation mechanism of Ni-CGO anode in low concentrations of H2 at a high current density. Int J Hydrogen Energy 2011;36:8461e7. [37] Smith JR, Chen A, Gostovic D, Hickey D, Kundinger D, Duncan KL, et al. Evaluation of the relationship between cathode microstructure and electrochemical behavior for SOFCs. Solid State Ionics 2009;180:90e8. [38] Iwai H, Shikazono N, Matsui T, Teshima H, Kishimoto M, Kishida R, et al. Quantification of SOFC anode microstructure based on dual beam FIB-SEM technique. J Power Sources 2010;195:955e61. [39] Soldati AL, Baque L, Troiani H, Cotaro C, Schreiber A, Caneiro A, et al. High resolution FIB-TEM and FIB-SEM characterization of electrode/electrolyte interfaces in solid oxide fuel cells materials. Int J Hydrogen Energy 2011;36:9180e8. [40] Cronin JS, Chen-Wiegart Y-CK, Wang J, Barnett SA. Threedimensional reconstruction and analysis of an entire solid oxide fuel cell by full-field transmission X-ray microscopy. J Power Sources 2013;233:174e9. [41] Delette G, Laurencin J, Usseglio-Viretta F, Villanova J, Bleuet P, Lay-Grindler E, et al. Thermo-elastic properties of SOFC/SOEC electrode materials determined from threedimensional microstructural reconstructions. Int J Hydrogen Energy 2013;38:1237991. [42] Celik S, Ibrahimoglu B, Toros S, Mat MD. Three dimensional stress analysis of solid oxide fuel cell anode micro structure. Int J Hydrogen Energy 2014;39:19119e31. [43] Zheng K, Zhang Y, Li L, Ni M. On the tortuosity factor of solid phase in solid oxide fuel cell electrodes. Int J Hydrogen Energy 2015;40:665e9. [44] Wilson JR, Kobsiriphat W, Mendoza R, Chen H-Y, Hiller JM, Miller DJ, et al. Three-dimensional reconstruction of a solidoxide fuel-cell anode. Nat Mater 2006;5:541e4. [45] Shimura T, Jiao Z, Hara S, Shikazono N. Quantitative analysis of solid oxide fuel cell anode microstructure change during redox cycles. J Power Sources 2014;267:58e68. [46] Kishimoto M. Three-dimensional microstructure of solid oxide fuel cell anode: observation, quantification, and application to numerical analysis. Kyoto University; 2013. [47] Kishimoto M, Iwai H, Saito M, Yoshida H. Quantitative evaluation of transport properties of SOFC porous anode by random walk process. ECS Trans 2009;25:1887e96. [48] Hanasaki M, Uryu C, Taniguchi S, Shiratori Y, Sasaki K. Durability of SOFC against Thermal and redox cycling. ECS Trans 2013;57:691e7. [49] Kishimoto M, Miyawaki K, Iwai H, Saito M, Yoshida H. Effect of composition ratio of Ni-YSZ anode on distribution of effective three-phase boundary and power generation performance. Fuel Cells 2013;13:476e86.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Change of an anode's microstructure morphology during the fuel starvation of an anode-supported solid oxide fuel cell Grzegorz Brus a,b,*, Kota Miyoshi a, Hiroshi Iwai a, Motohiro Saito a, Hideo Yoshida a a b
Kyoto University, Department of Aeronautics and Astronautics, Kyoto, Japan AGH University of Science and Technology, Faculty of Energy and Fuels, Krakow, Poland
article info
abstract
Article history:
A RedOx cycle is one of the most dangerous factors of anode degradation due to micro-
Received 30 January 2015
structure changes. In this paper, the evolution of anode microstructure during a RedOx
Received in revised form
cycle caused by a fuel starvation process was studied using electrochemical measure-
24 March 2015
ments. After a power generation experiment, the anode microstructure was reconstructed
Accepted 28 March 2015
using a combination of focused ion beam and scanning electron microscopy (FIBeSEM).
Available online 22 April 2015
The microstructure changes were quantified for parameters such as the tortuosity factor, tripe phase boundary density, volume fraction, connectivity and average grain size. The
Keywords:
three dimensional microstructure reconstruction indicated that the fuel starvation con-
Solid oxide fuel cells
dition might cause significant changes in microstructure morphology.
Redox cycle
Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Fuel starvation
reserved.
Focused ion beam Microstructure morphology
Introduction Solid Oxide Fuel Cells (SOFCs) are electrochemical devices that can convert the chemical potential of fuels directly into electricity. Because of their high working temperature (600 Ce1000 C), SOFCs have the highest energy conversion rate among all types of fuel cells. A solid oxide fuel cell consists of two porous ceramic electrodes (cathode and anode) separated by a dense solid ceramic electrolyte made of yttria stabilized zirconia phase (YSZ). The electrode microstructure morphology is an important factor determining its electrochemical performance. A typical anode consists of porous Ni/
YSZ and a typical cathode is made from porous lanthanum strontium cobalt ferrite (LSCF). Each component plays a unique and important role in the transport process by providing a pathway for different species; a YSZ phase for oxygen ions, Ni and LSCF phases for electrons and a pore phase for gases. While the SOFC slowly reaches its commercialization, it becomes a key issue to guarantee long term and safe operation. It is generally accepted that to satisfy the market, 40 000 h of safe operation is required [1,2]. One of the most dangerous causes of anode degradation is the oxidation of Ni particles. Because NiO occupies more volume than Ni, oxidation results in internal stresses inside an anode. A further reduction of NiO to Ni results in shrinking of the
* Corresponding author. Kyoto University, Department of Aeronautics and Astronautics, Thermal Engineering Laboratory, Kyoto daigaku-Katsura, Nishikyo-ku, Kyoto 615-8530, Japan. Tel./fax: þ81 075 383 3652. E-mail addresses: [email protected], [email protected] (G. Brus). http://dx.doi.org/10.1016/j.ijhydene.2015.03.143 0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
6928
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
particle volume. This expansion and contraction of Ni particles is called in the literature a RedOx cycle and may cause damage to a cell microstructure. There are several possible causes of a RedOx cycle such as i) the thermal cycle of a SOFC system during shut down and start up, ii) the stop of gas supply in the case of an emergency situation, iii) oxidation from steam in the case of SOFC with a reforming system, iv) leakage through the sealing, v) local fuel starvation caused by a load following operation, vi) too high a hydrogen conversion rate and vii) high fuel dilution. From a reaction point of view, they can be classified into two groups. One is the redox by chemical reaction (Ni þ O2 / NiO) and/or by electrochemical reaction (Ni þ O2 / NiO). The first group, redox by chemical reaction, is often studied in laboratory scale experiments by stopping the fuel supply and blowing air to the anode side. After the oxidation of nickel particles, air is stopped and hydrogen is fed again to the anode to finish the RedOx cycle. This approach was studied for both electrolyte supported cells [3e20] and anode supported cells [21e26]. Conclusions were different from one paper to another. Some authors report massive degradation, other authors report moderate or find no significant degradation. In the second group, the anode is electrochemically oxidized. Electrochemical oxidation will happen rather in the vicinity of the anode-electrolyte interface. For laboratory scale experiments, current is applied to the cell to force oxygen ions to go through the electrolyte and oxidize Ni at the anode-electrolyte interface. In the case of the electrolyte supported cell, Takagi et al. [27] found moderate degradation that was not fatal for the cell. In the case of the anode supported cell, the conclusions are in contradiction with each other. Sarantaridis et al. [28] has reported a catastrophic effect of the redox cycle on the cell whereas Hatae et al. [29] reported no electrolyte crack or OCV drops. An outstanding research to conclude all the above mentioned cases was conducted by Laurencin et al. [20,30,31]. In their research they investigated both anode supported [30] and electrolyte supported [20] cells as well as different types of oxidation: chemical and electrochemical [30]. Impedance spectroscopy was used to estimate the electrochemical degradation. Their results indicated probability of delamination at the anode-electrolyte interface and mechanical damage of the cell as a result of stress [31]. To estimate the probability of the cell failure, the thermo-mechanical model has been proposed [31]. Recent research suggests that fuel starvation might occur during transient operations, if fuel is consumed in the cell faster than it can be supplied [32,33]. When fuel starvation occurs during a real SOFC operation, the cell might be strongly polarized and oxygen ions keep coming out from the cathode side which causes electrochemical oxidation of Ni in the anode. In this study we focus on this electrochemical oxidation of Ni by fuel starvation. Recent results by Fang et al. [34] show the significance of this problem. In their research, a stack of cells was investigated under high current density and high fuel utilization factor. Their research indicates that the upstream part of the anode was oxidized as a result of fuel starvation. At the laboratory scale, this situation was studied recently by different research groups [35,36]. In their research using diluted hydrogen as fuel, it was observed that cell performance drops rapidly at very low hydrogen concentration
(below 2%). Further investigation unraveled the oxidation of nickel near the anode-electrolyte interface [35,36]. Chen et al. [36] have reported the fatal failure of an anode-supported cell after exposing the cell to fuel starvation conditions [36]. Throughout the research an explanation of the cell failure by fuel starvation was provided from a macroscopic point of view. However the impact of fuel starvation on the electrode microstructure remains unclear. The most precise information about cell microstructure can be derived from real structural analysis. Recently, the combination of a focused ion beam and scanning electron microscope (FIBeSEM) as well as the X-ray tomography technique brought a breakthrough in the direct 3D observation of porous structure [37e43]. 3D structure reconstruction was introduced to the field of SOFCs by Wilson et al., in 2006 [44]. The method enables the observation of many sequential 2D images of a porous microstructure and reconstructs it into a 3D structure using advance image processing. From the reconstructed microstructure, it is possible to evaluate the microstructure parameters. These parameters directly obtained from the real electrode structure are the key to considering the relationships between porous microstructures and the cell power generation performance. There were attempts to use the 3D microstructure observation to investigate microstructure evaluation during the RedOx cycle [9,27,45], but to the authors' best knowledge, it has not been applied to the fuel starvation problem. The literature review clearly unravels the gap in published data that the authors aim to fill by providing detailed information about microstructure morphology change, during the fuel starvation of an anode supported solid oxide fuel cell. The novelty lays in using state-of-the-art techniques such as FIBeSEM, combined with advanced methods of deriving microstructure parameters to report the changes caused in microstructure during the fuel starvation process.
Experiment Experimental set-up A conventional double-tube type test apparatus for SOFC button cell evaluation was used (BEL Japan, Inc., BEL-SOFC). The SOFC sample was located between two ceramic tubes in the electric furnace. The ceramic tubes have a double co-axial structure. The fuel/air is supplied through the inner-tube and exhausted through the gap between the inner-tube and the outer-tube. The furnace was heated-up to 800 C. Fuel was fed to the anode via flow controllers omitting a humidifier. Air was used as an oxidant on the cathode side. The anode and the cathode were connected to the measuring devices with 0.5 mm platinum wires. Platinum meshes were welded with platinum wires and connected with each electrode. An electrochemical Interface (Solartron analytical, model 1287A) was used for the currentevoltage characteristic measurements.
Experimental method In the experiment, three commercial samples provided by SOFC Power were used. The Ni/YSZeYSZeGDC/LSCF samples
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
were cut by a laser technique from a single piece of the large size cell. Therefore, the microstructure of each anode is assumed to be the same. The technical informations related to the SOFC preparation such as the initial weight percentage of NiO to YSZ, sintering temperature, pore formers etc. are protected by intellectual property rights of the company and therefore confidential. Sample 1 was reduced using a mixture of 60% hydrogen and 40% nitrogen at 800 C for 1 h. The hydrogen concentration was increased gradually to reach 60% at the beginning of reduction process. Next, the cell was kept under a polarized condition of 0.8 V and fed with 60 ml/min of H2 and 40 ml/ min N2 for 10 min and then cooled down at OCV conditions, under reducing atmosphere. After the sample had cooled down, it cooling down the sample was investigated using FIBeSEM and the data was used to obtain a 3D reconstruction of the sample. Sample 2 was reduced at the same condition described in the previous paragraph. Then the cell was kept under a polarized condition of 0.8 V and fed 60 ml/min of H2 and 40 ml/ min N2. After stabilization, the hydrogen supply was stopped. When the current dropped to zero, hydrogen was fed again. After the electrochemical measurements, the sample was cooled down to room temperature using only a supply of pure nitrogen. After the experiment, the sample was investigated using FIBeSEM and the data was used to obtain a 3D reconstruction of the sample's microstructure. Sample 3. To determine if the nickel was oxidized during fuel starvation, an additional experiments were conducted. In this experiment the sample was kept under the same conditions as described in the previous paragraph. After reaching stabilization, the hydrogen valve was closed. When the current dropped to zero, the oxygen valve was also closed on the cathode side. The cell potential was kept constant at 0.8 [V] until the current dropped to zero then polarization was stopped. Next the furnace was cooled down to room temperature, with pure nitrogen flowing on both sides; the anode and the cathode. After cooling down to room temperature, the anode-electrolyte interface was investigated using EnergyDispersive X-ray spectroscopy (EDX). Additionally, several points from the EDX mapping analysis were selected to conduct a more precise point analysis (see Table 1).
sand paper and diamond paste to prepare the sample for the FIBeSEM observation.
FIBeSEM observation A three dimensional structure of the anode was observed using the FIBeSEM system installed at Kyoto University, Japan. The configuration of the FIBeSEM setup is presented in Fig. 1. The procedure can be summarized by the following steps: i) a trench is fabricated to give access to the sample's intersection. ii) a SEM picture of an observed intersection is taken using an in-lens secondary electron detector (Mag ¼ 3 k, EHT ¼ 1.5 kV, WD ¼ 5 mm). iii) FIB uses a beam of Ga þ ion to mill the sample to expose another intersection and iv) a “cut and see” procedure is repeated to obtain the sequence of 2D images. The idea of the “cut and see” procedure and the example of SEM images are presented in Fig. 1.
Phase separation and 3D reconstruction The SEM image segmentation is a crucial step in the analysis of the data. A phase separation was conducted based on image brightness using AVIZO software (Mercury Computers Systems, Inc.). Fig. 2 presents an example of the phase separation process. The pixel was 48.51 nm. After the phase separation, the 3D microstructure was reconstructed using a set of 2D images. Fig. 3(a) and (b) show the reconstructed microstructure of the porous anode.
Quantification of porous structure The microstructural parameters of the anode were evaluated using three dimensional structure reconstruction.
Volume fraction (ε) The volume fraction of each phase was measured by counting the number of voxels corresponding to Ni, YSZ and pores, respectively [46].
Average grain size (d) The average grain size was evaluated by the three dimensional version of the line-intercept method [46]. In this
FIBeSEM imaging Sample preparation for the FIBeSEM observation After the power generation experiment, Sample 1 and Sample 2 were infiltrated using epoxy resin (Marumoto Struers KK) under vacuum conditions. Filling the pores with resin is important for recognizing the pore region during SEM observation. The infiltrated samples were cut and polished using Table 1 e Summarized properties of the anode before and after fuel starvation experiment.
Point 1 Point 2 Point 3 Point 3
Ni
O
Zr
Species
92.09 74.31 27.67 0.17
3.08 23.86 68.68 68.04
4.82 1.83 3.65 31.79
Nickel Nickel oxide Nickel oxide YSZ
6929
Fig. 1 e Schematic view of FIBeSEM settings and the measuring procedure.
6930
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
Fig. 3 e Three dimensional reconstruction of a) Sample 1 (reference sample) b) Sample 2 (after the RedOx cycle). The green color represents Ni, the yellow is YSZ and the red are pores. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 2 e An example of image analysis of the epoxy impregnated NieYSZ (sample 2).
media is lower than that obtained in a free space. The degree of reduction is described quantitatively using the tortuosity factor. The details of the procedure are described by Kishimoto et al. in reference [47].
Phase boundary length density (ltpb) method lines are drawn in three directions along orthogonal coordinates. The length of the intercept lines is measured and the average value is calculated to obtain a mean particle size.
Tortuosity factor (t) The tortuosity factor was quantified by a diffusion based simulation based on a random walk process [47]. In this method, a large number of random walkers are stochastically distributed in each phase voxels. In the next time step each walker randomly migrates into one of its neighbor voxels. However, if the selected neighboring voxel is not the same phase then the walker stays in the current position and waits for the next time step. The process is repeated and the mean square displacement of random walkers is calculated. Due to the obstacles, the mean square displacement in the porous
The Tripe Phase Boundary length density was determined using the volume expansion method [38] and AVIZO software (Mercury Computers Systems, Inc.). In this method each phase is slightly expanded in a computational domain. The overlapped regions form string-like volumes that contained TPB lines inside. Centerlines of those strings are taken as lines that represent the TPBs, and their lengths are measured and divided by the volume of the investigated sample giving the value of the TPB density [38].
Phase connectivity To discuss the transport phenomena via each phase, only the connected phase should be taken into consideration. The isolated part of a phase, which does not contribute to the transport, should be excluded. In the presented study, 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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
isolated phase was excluded by indicating the phase that penetrates the reconstructed structure in a certain direction [47].
Results and discussion The results of the electrochemical measurements for Sample 2 are presented in Fig. 4(a). As can be see in Fig. 4(a) measurements can be divided into the following zones: i) normal operation ii) after the hydrogen stops, the current drops insignificantly and hydrogen leftovers are used iii) the oxidation of nickel particles iv) a short recovery after flowing back hydrogen v) the decay of properties vi) the acceleration of decay vii) end of cell's life. The experiment was repeated twice and confirmed the reproducibility of the results. An analogical trend has been observed in the past [36] and was associated with the oxidation of nickel particles. The results of the electrochemical measurements for Sample 3 are presented in Fig. 4(b). As can be seen in Fig. 4(b), Sample 3 was exposed only to fuel starvation without feeding hydrogen to finish the redox cycle. After the electrochemical measurements Sample 3 was
Fig. 4 e Electrochemical measurements.
6931
analyzed using EDX to investigate possible presence of nickel oxide. The results of the EDX mapping is presented in Fig. 5. As can be seen in Fig. 5 nickel was oxidized near the electrolyte interface even though oxygen was not introduced to the anode side. The oxygen ions must have come from the cathode side during polarization. Since hydrogen was not present on the anode side, oxygen reacted with the nickel creating nickel oxide until the electrochemical reactions stopped. An EDX point measurement was performed at 4 different positions marked in Fig. 5 and results are summarized in Table 1 it confirms that nickel oxide is present in the vicinity of the anode-electrolyte interface. The characteristic shape of nickel oxide can be recognized using scanning electron microscopy (SEM) [48]. Fig. 6 presents SEM images of the anode microstructure near the anode surface Fig. 6(a) and near the vicinity of the anode-electrolyte interface 6(b). As can be seen in Fig. 6 the morphology of the Ni particles is completely different. Fig. 6(b) indicates typical nickel oxide structures, not present in Fig. 6(a). A closer view at the nickel oxide particles can be seen in Fig. 7 which supports the presented conclusions. It should be noted that the OCV of all investigated samples reached initially 1.2 V at 800 C in 60% H2 and 40% N2. This indicates that the YSZ thin electrolyte prevented crossover of the gases and the cell was well sealed. The results of the microstructure evaluation of Sample 1 (reference sample) and Sample 2 (after fuel starvation) are presented in Table 2. As can be seen in Table 2, the RedOx cycle causes significant changes in the sample microstructure. The tortuosity factor of the nickel phase increases from 7.24 to 14.91. At the same time, the connectivity of the nickel phase drops from 97 % to 89 %. The isolated nickel particles before and after the fuel starvation experiment are presented in Fig. 8. Those results can be explained by the expansion and contraction of nickel particles which results in particles
Fig. 5 e EDX mapping of the region near the anodeelectrolyte interface taken from sample 3. The green color represents Zr, the red is oxygen and the blue Ni. Details of the observation: Mag. 16 k, EHT ¼ 10 kV, WD ¼ 5.0. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
6932
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
Fig. 6 e SEM images of sample 3.
Fig. 7 e SEM image of a nickel oxide particle near the anode-electrolyte interface taken from sample 3.
Fig. 8 e Isolated nickel particles a) reference sample b) after fuel starvation condition.
fragmentation. This phenomenon can be confirmed by the decrease in nickel particle size and connectivity. When connectivity is lost, the isolated nickel which no-longer contributes to the transport phenomena is excluded from the tortuosity factor calculation. As a consequence, the tortuosity factor increases. During the redox cycle, the generated steam causes multi-cracks and the tortuosity of pore size becomes smaller. As a consequence of microstructure fragmentation, the TPB length become higher. The tortuosity factor values agree with those published in the literature when a similar phase volume fraction is considered [49]. The tortuosity factor is affected by the volume fraction and its effect is particularly strong at low volume fractions below 30%. A typical value of the tortuosity factor is between 1.5 and 4. However, in cases of low porosity of the anode, it can easily reach a value as high as 10 [49]. Compared to the Ni and pore phases, the
Table 2 e Summarized microstructure properties of the anode before and after fuel starvation experiment.
Ni YSZ Pore ltpb
ti [e]
d [mm]
εi [%]
Connectivity [%]
Before / after
Before / after
Before / after
Before / after
7.24 / 14.91 2.05 / 2.26 7.18 / 4.54 3.1 / 4.0
0.88 / 0.84 0.95 / 0.86 0.81 / 0.76 [mm2]
29 / 27 45 / 43 25 / 30 ltpb,active
97 / 89 100 / 100 99 / 99 2.1 / 3.0
[mm2]
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
microstructure change of YSZ phase is small as can be seen in Table 2. It supports a common understanding that the YSZ phase is relatively stable. The expansion and contraction of Ni phase, therefore, mainly affects the microstructure of Ni phase itself and the pore phase as a result. In Table 2, it is also shown that the particle diameters of three phases have decreasing tendency after the redox cycle. It is consistent with the increase of the TPB density shown in the same table because a higher TPB density is generally expected for an electrode with finer particles. The total TPB density increases from 3.1 [mm2] to 4.0 [mm2]. The comparison with already published data indicates some similarities and differences with the case where air was blown to the anode to cause the redox cycle [45]. An average Ni particles gets smaller after the redox cycle. Same tendency can be observed in the case of average pore size. Those changes lead in both cases to increase of TPB length density after redox cycle. Also in both cases, fuel starvation and air blow lead to decrease of Ni volume fraction after single redox cycle. However the noticeable difference is in connectivity of Ni particles. In case of air blow the connectivity slightly increased after redox cycle. In contradiction, a significant drop of Ni connectivity was observed in case of fuel starvation. An observation using a scanning electron microscope indicates that nickel near the anode-electrolyte interface is oxidized under fuel starvation conditions. The number of isolated nickel particles increased after one redox cycle. But generally, the microstructure of the anode recovers, as can be seen in Table 2. From a microstructure point of view, the anode after one redox cycle should have almost the same performance as the anode before the redox operation. The short recovery of the performance observed in Sample 2 indirectly supports it. However, Sample 2 died after the fuel starvation experiment. This is inconsistence with microstructure measurements. Therefore the cell failure observed in Fig. 4(a) cannot be explained only from microstructure change. The obtained data can support thermo-mechanical model that should be further investigated.
Conclusions The presented paper reports the microstructure changes of an anode electrode of an anode-supported solid oxide fuel cell after a RedOx cycle. The redox cycle was conducted by stopping fuel supply at polarized condition to simulate a fuel starvation scenario of a SOFC operation. The microstructure of the sample after a RedOx cycle and the reference sample were reconstructed using the FIBeSEM technique and microstructure parameters were evaluated. The presented results depict the possible degradation effects after fuel starvation that can happen at the downstream of the anode. The connectivity of the Ni particles was significantly lost in the conducted experiment. The lost of connectivity was coupled with an increase of the tortuosity factor of nickel and a decrease in the tortuosity factor of pore phase. The average grain size of each phase dropped suggesting strong fragmentation of microstructure as a consequence of the volume change of nickel particles over a RedOx cycle.
6933
Acknowledgments The first author was supported by a JSPS Postdoctoral Fellowship for Foreign Researchers. Additionally, the present work was partially supported by the PAN-JSPS Joint Research Project “Thermal Interaction between Stack and Reformer in Small Scale SOFC” and partially by the New Energy and Industrial Technology Development Organization (NEDO).
references
[1] Comminges C, Fu QX, Zahid M, Steiner NY, Bucheli O. Monitoring the degradation of a solid oxide fuel cell stack during 10,000 h via electrochemical impedance spectroscopy. Electrochim Acta 2012;59:367e75. [2] Schuler JA, Wuillemin Z, Hessler-Wyser A, Comminges C, Steiner NY. Cr-poisoning in (La,Sr)(Co,Fe)O3 cathodes after 10,000 h SOFC stack testing. J Power Sources 2012;211:177e83. [3] Cassidy M, Lindsay G, Kendall K. The reduction of nickel zirconia cermet anodes and the effects on supported thin electrolytes. J Power Sources 1996;61:189e92. [4] Waldbillig D, Wood A, Ivey DG. Electrochemical and microstructural characterization of the redox tolerance of solid oxide fuel cell anodes. J Power Sources 2005;145:206e15. [5] Modena S, Ceschini S, Tomasi A, Montinaro D, Sglavo VM. Reduction and reoxidation processes of NiOYSZ composite for solid oxide fuel cell anodes. J Fuel Cell Sci Technol 2006;3:487e91. [6] Marrero-Lopez D, Pena-Martinez J, Ruiz-Morales JC, Gabas M, Nunez P, Aranda M, et al. Redox behaviour, chemical compatibility and electrochemical performance of Sr2 MgMoO6d as SOFC anode. Solid State Ionics 2010;180:1672e82. [7] Iwanschitz B, Sfeir J, Mai A, Schutze M. Degradation of SOFC anodes upon redox cycling: a comparison between Ni/YSZ and Ni/CGO. J Electrochem Soc 2010;157:B269e78. [8] Sumi H, Lee Y-H, Muroyama H, Matsui T, Eguchi K. Comparison between internal steam and CO2 reforming of methane for Ni-YSZ and Ni-SCSZ SOFC anodes. J Electrochem Soc 2010;157:B1118e25. [9] Muroyama H, Sumi H, Kishida R, Kim J-Y, Matsui T, Eguchi K. Correlation between microstructure and electrochemical characteristics of Ni-YSZ anode subjected to redox cycles. ECS Trans 2011;35:1379e87. [10] Sarantaridis D, Chater RJ, Atkinson A. Changes in physical and mechanical properties of SOFC Ni-YSZ composites caused by redox cycling. J Electrochem Soc 2008;155:B467e72. [11] Glauche A, Betz T, Mosch S, Trofimenko N, Kusnezoff M. Long-term, redox and Thermal cycling stability of electrolyte supported cells. In: 216th ECS meeting. ECS; 2009. p. 411e9. [12] Klemenso T, Mogensen M. Ni-YSZ solid oxide fuel cell anode behavior upon redox cycling based on electrical characterization. J Am Ceram Soc 2007;90:3582e8. [13] Klemenso T, Appel CC, Mogensen M. In Situ observations of microstructural changes in SOFC anodes during redox cycling. Electrochem Solid State Lett 2006;9:A403e7. [14] Zhang Y, Liu B, Tu B, Dong Y, Cheng M. Redox cycling of NiYSZ anode investigated by TPR technique. Solid State Ionics 2005;176:2193e9. [15] Nakamura K, Somekawa T, Baba Y, Horiuchi K, Matsuzaki Y, Yoshimoto M. Expansion and shrinkage in the redox process of (Mg, Ni) O solid solutions for preparation of steam
6934
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
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 4 0 ( 2 0 1 5 ) 6 9 2 7 e6 9 3 4
reforming catalysts in SOFCs. J Ceram Soc Jpn 2009;117:166e70. Pihlatie M, Kaiser A, Mogensen M. Redox stability of SOFC: thermal analysis of NiYSZ composites. Solid State Ionics 2009;180:1100e12. Shearing PR, Bradley RS, Gelb J, Tariq F, Withers PJ, Brandon NP. Exploring microstructural changes associated with oxidation in NiYSZ SOFC electrodes using high resolution X-ray computed tomography. Solid State Ionics 2012;216:69e72. Stathis G, Simwonis D, Tietz F, Moropoulou A, Naoumides A. Oxidation and resulting mechanical properties of Ni/þY2O3stabilized zirconia anode substrate for solid-oxide fuel cells. J Mater Res 2002;17:951e8. Blennow P, Hansen KK, Wallenberg L, Mogensen M. Electrochemical characterization and redox behavior of Nbdoped SrTiO3. Solid State Ionics 2009;180:63e70. Laurencin J, Delette G, Sicardy O, Rosini S, Lefebvre-Joud F. Impact of redox cycles on performances of solid oxide fuel cells: case of the electrolyte supported cells. J Power Sources 2010;195:2747e53. Malzbender J, Wessel E, Steinbrech RW. Reduction and reoxidation of anodes for solid oxide fuel cells. Solid State Ionics 2005;176:2201e3. Hatae T, Matsuzaki Y, Yamazaki Y. Study on electrochemical re-oxidation of anode-supported solid oxide fuel cells using oxide ion current. Solid State Ionics 2008;179:274e81. Ettler M, Timmermann H, Malzbender J, Weber A, Menzler NH. Durability of Ni anodes during reoxidation cycles. J Power Sources 2010;195:5452e67. Faes A, Nakajo A, Hessler-Wyser A, Dubois D, Brisse A, Modena S. Redox study of anode-supported solid oxide fuel cell. J Power Sources 2009;193:55e64. Ihringer R. 2R-CellTM: reliability of anode supported thin electrolyte upon multi thermo and redox cycles. In: 216th ECS Meeting. ECS; 2009. p. 473e83. Monzon H, Laguna-Bercero MA. Redox-cycling studies of anode-supported microtubular solid oxide fuel cells. Int J Hydrogen Energy 2012;37:7262e70. Takagi N, Nakamura Y, Shikazono N, Kasagi N. Microstructure change of SOFC anode caused by electrochemical redox cycles. In: Proceedings of the 9th European solid oxide fuel cell forum; 2010. p. 7e135. Sarantaridis D, Rudkin RA, Atkinson A. Oxidation failure modes of anode-supported solid oxide fuel cells. J Power Sources 2008;180:704e10. Hatae T, Matsuzaki Y, Yamashita S, Yamazaki Y. Current density dependence of changes in the microstructure of SOFC anodes during electrochemical oxidation. Solid State Ionics 2009;180:1305e10. Laurencin J, Delette G, Morel B, Lefebvre-Joud F, Dupeux M. Solid oxide fuel cells damage mechanisms due to Ni-YSZ reoxidation: case of the anode supported cell. J Power Sources 2009;192:344e52. Laurencin J, Roche V, Jaboutian C, Kieffer I. Ni-8YSZ cermet re-oxidation of anode supported solid oxide fuel cell: from kinetics measurements to mechanical damage prediction. Int J Hydrogen Energy 2012;37:12557e73. Komatsu Y, Brus G, Kimijima S, Szmyd JS. The effect of overpotentials on the transient response of the 300W SOFC cell stack voltage. Appl Energy 2014;115:352e9.
[33] Fardadi M, McLarty DF, Brouwer J, Jabbari F. Enhanced performance of counter flow SOFC with partial internal reformation. Int J Hydrogen Energy 2014;39:1975366. [34] Fang Q, Blum L, Peters R, Peksen M, Batfalsky P, Stolten D. SOFC stack performance under high fuel utilization. Int J Hydrogen Energy 2015;40:112836. [35] Brus G, Miyawaki K, Iwai H, Saito M, Yoshida H. Tortuosity of an SOFC anode estimated from saturation currents and a mass transport model in comparison with a real microstructure. Solid State Ionics 2014;265:13e21. [36] Chen G, Guan G, Abliz S, Kasai Y, Abudula A. Rapid degradation mechanism of Ni-CGO anode in low concentrations of H2 at a high current density. Int J Hydrogen Energy 2011;36:8461e7. [37] Smith JR, Chen A, Gostovic D, Hickey D, Kundinger D, Duncan KL, et al. Evaluation of the relationship between cathode microstructure and electrochemical behavior for SOFCs. Solid State Ionics 2009;180:90e8. [38] Iwai H, Shikazono N, Matsui T, Teshima H, Kishimoto M, Kishida R, et al. Quantification of SOFC anode microstructure based on dual beam FIB-SEM technique. J Power Sources 2010;195:955e61. [39] Soldati AL, Baque L, Troiani H, Cotaro C, Schreiber A, Caneiro A, et al. High resolution FIB-TEM and FIB-SEM characterization of electrode/electrolyte interfaces in solid oxide fuel cells materials. Int J Hydrogen Energy 2011;36:9180e8. [40] Cronin JS, Chen-Wiegart Y-CK, Wang J, Barnett SA. Threedimensional reconstruction and analysis of an entire solid oxide fuel cell by full-field transmission X-ray microscopy. J Power Sources 2013;233:174e9. [41] Delette G, Laurencin J, Usseglio-Viretta F, Villanova J, Bleuet P, Lay-Grindler E, et al. Thermo-elastic properties of SOFC/SOEC electrode materials determined from threedimensional microstructural reconstructions. Int J Hydrogen Energy 2013;38:1237991. [42] Celik S, Ibrahimoglu B, Toros S, Mat MD. Three dimensional stress analysis of solid oxide fuel cell anode micro structure. Int J Hydrogen Energy 2014;39:19119e31. [43] Zheng K, Zhang Y, Li L, Ni M. On the tortuosity factor of solid phase in solid oxide fuel cell electrodes. Int J Hydrogen Energy 2015;40:665e9. [44] Wilson JR, Kobsiriphat W, Mendoza R, Chen H-Y, Hiller JM, Miller DJ, et al. Three-dimensional reconstruction of a solidoxide fuel-cell anode. Nat Mater 2006;5:541e4. [45] Shimura T, Jiao Z, Hara S, Shikazono N. Quantitative analysis of solid oxide fuel cell anode microstructure change during redox cycles. J Power Sources 2014;267:58e68. [46] Kishimoto M. Three-dimensional microstructure of solid oxide fuel cell anode: observation, quantification, and application to numerical analysis. Kyoto University; 2013. [47] Kishimoto M, Iwai H, Saito M, Yoshida H. Quantitative evaluation of transport properties of SOFC porous anode by random walk process. ECS Trans 2009;25:1887e96. [48] Hanasaki M, Uryu C, Taniguchi S, Shiratori Y, Sasaki K. Durability of SOFC against Thermal and redox cycling. ECS Trans 2013;57:691e7. [49] Kishimoto M, Miyawaki K, Iwai H, Saito M, Yoshida H. Effect of composition ratio of Ni-YSZ anode on distribution of effective three-phase boundary and power generation performance. Fuel Cells 2013;13:476e86.