MATERIALS TRANSACTIONS
Online ISSN : 1347-5320
Print ISSN : 1345-9678
ISSN-L : 1345-9678
Influence of NiO Reduction on Residual Strain in NiO/Ni-YSZ
Fumitada IguchiSarana AkraseveeYutaro Miyoshi
Author information
JOURNALS FREE ACCESS FULL-TEXT HTML

2018 Volume 59 Issue 1 Pages 27-32

Details
Abstract

The residual strains in the composites of nickel oxide (NiO) and yttria-stabilized zirconia (YSZ) and in the cermets of reduced nickel (Ni) and YSZ, which were used as anodes for solid oxide fuel cells (SOFCs), were measured using X-ray diffraction. The influence of Ni reduction on the residual strain was evaluated. Tensile and compressive residual strains caused by thermal strain were observed in NiO and YSZ phases, respectively. They clearly depended on the volume fraction of NiO and YSZ, and changed proportionally. The YSZ phase in the Ni-YSZ cermets also showed a similar dependence on the volume fraction of NiO. The compressive strain increased as the NiO increased; however, a local maximum was observed for NiO 50 vol%, beyond which it decreased with increasing amount of NiO. Compressive strain in the YSZ phase in the Ni-YSZ with NiO of 60 vol%, which is a common volume fraction of SOFC anodes, was only 30% of that in the NiO-YSZ. Plastic deformation of the Ni phase near the interfaces, and relaxation of the compressive strain in the YSZ phase were responsible for this phenomenon. This revealed the difference in the residual strain in the YSZ phase after reduction.

1. Introduction

Typical applications of high-temperature solid state ionics are solid oxide fuel cells (SOFCs) and solid oxide electrolysis cells (SOECs). These cells are already in the early commercial stages. However, some issues regarding further social implementation exist, which require improvement of the mechanical durability and reliability of these cells as power conversion devices, because they are required to operate for extended periods of time1). To address these issues, the inherent strain and stress conditions during an operational lifetime (from fabrication to construction, and operation) are fundamental information that has been studied extensively28).

It has long been known that the mechanical properties of materials, which are defined as the response of materials to applied forces, show dependence on temperature and chemical composition9,10). Studies conducted over two decades additionally revealed that the defect structures of ionic conductors and mixed ionic and electric conductors, which are influenced by the surrounding chemical potentials, related to the mechanical properties of these conductors such as the Young's modulus, creep rates, and fracture strengths. In other words, the mechanical properties are dependent on the surrounding oxygen partial pressure1118). Furthermore, the change of the defect structure also causes volume changes, which make strain and stress analyses in SOFCs and SOECs more difficult than in automobiles and airplanes at high operating temperatures and for small cell sizes. Despite the rapid progress of computational mechanics nowadays, experimental observation of strain and stress is still required in the front line of solid state ionics1924).

For these reasons, we focused on the anode of an SOFC. Conventionally, the anode consists of nickel oxide (NiO) and yttria-stabilized zirconia (YSZ), with the NiO being reduced to Ni metal during the cell operation. The reduction of NiO to Ni accompanied by volume shrinkage changes the mechanical properties of the anode drastically12,2530). However, the geometrical complexity of the continuous NiO, Ni, and YSZ networks constructed in the anodes prevents us from obtaining a proper understanding of the strain and stress behavior. Recently, with computational analyses based on 3D-models constructed using a scanning electron microscope with a focused ion beam milling (FIB-SEM), the strain and stress distributions in the anode have been determined31). However, the influence of NiO reduction on the strain and stress conditions is still unclear.

Therefore, in this study, the residual strain in the composites of NiO and YSZ (NiO-YSZ) and the cermets of Ni and YSZ (Ni-YSZ), which are obtained by reducing the NiO-YSZ composites, is measured by X-ray diffraction (XRD). Furthermore, the influence of the NiO reduction on the residual strain is evaluated.

2. Experimental Procedure

The NiO-YSZ composites with various volume fractions were prepared by the solid-state sintering method. The Ni-YSZ cermets were obtained by reducing the composites in SOFC operation condition. The YSZ network in Ni-YSZ was extracted by etching Ni from Ni-YSZ (YSZ-Frame).

The raw powders of NiO (Wako Pure Chemical Industries, Ltd., >99%), and 8 mol% yttria-doped zirconia (Tosho Co., Ltd, TZ-8Y) were added to obtain various volume ratios ranging from NiO:YSZ = 10:90 vol% to 90:10 vol% in 10 vol% intervals (N10Y90 to N90Y10). It was then grinded by a planetary ball mill (Fritsch, P-6) at 270 rpm in ethanol for 24 h to obtain a homogeneous mixture of the powders. The dried and sieved powders were formed into a pellet by hydrostatic pressing at 100 MPa, following uniaxial pressing at 25 MPa. The pellets were sintered at 1600℃ for 10 h in laboratory air. The surface of the sintered pellets was mechanically polished and rounded specimens of the NiO-YSZ composites were obtained. Some of the specimens were reduced in humidified hydrogen at 800℃ for 2 h to obtain the Ni-YSZ cermets. This reduction process was repeated until the XRD peak of NiO disappeared. For the XRD measurements, an X-ray diffractometer (PANalytical, X'pert Pro) using Cu Kα X-rays, and the Bragg-Brentano configuration was used. Finally, some of the Ni-YSZ cermets were soaked in diluted nitric acid (30 vol%) for 72 h to dissolve the nickel and obtain the YSZ-Frame.

The residual strains in the NiO-YSZ and the Ni-YSZ were measured using XRD. Generally, the residual strain in the SOFCs was measured by the side inclination method using the difference in strain between the perpendicular and parallel directions to the irradiated surface because the residual strain was considered as plane stress in each component. However, for the composites and the cermets, the residual strain was considered as isostatic strain. In addition, the specimens consisted of small grains (<1 μm), and the penetration depth of Cu Kα into the specimens (10 s μm) was larger than the grain sizes. It was suggested that the difference of the residual strain between the two directions was small and difficult to be observed. Therefore, in this study, residual strain was calculated by comparing the lattice spacing of the specimens with the residual strain for NiO-YSZ and Ni-YSZ and the strain-free specimen YSZ-Frame. In addition, this method was also expected to cancel the influence of the volume change of YSZ due to slight chemical composition fluctuation.

The lattice spacing d of YSZ (620), NiO (333, 511), and Ni (331) were measured in three types of specimens. From the lattice spacing (d), the residual strain ε of YSZ, NiO and Ni in NiO-YSZ and Ni-YSZ was calculated by eqs. (1), (2), and (3), respectively.   

\[ \varepsilon_{YSZ1,2} = \frac{d_{YSZ1,2} - d_{YSZ3}}{d_{YSZ3}} \](1)
  
\[ \varepsilon_{NiO} = \frac{d_{NiO} - d_{ICDD - NiO}}{d_{ICDD - NiO}} \](2)
  
\[ \varepsilon_{Ni} = \frac{d_{Ni} - d_{ICDD - Ni}}{d_{ICDD - Ni}} \](3)
where suffixes YSZ1, YSZ2, and YSZ3 indicate NiO-YSZ, Ni-YSZ, and YSZ-Frame, respectively. As strain-free lattice spacing of NiO and Ni in NiO-YSZ and Ni-YSZ, the lattice spacing from the International Centre for Diffraction Date (ICDD) database of 00-0247-1049 for NiO and 00-004-0850 for Ni was used because strain-free NiO and Ni was difficult to obtain, and the lattice spacing deviation from the database, caused by chemical composition fluctuation, was not expected to occur in them. Finally, the residual strain ε was converted to residual volume strain εV as described in eq. (4).   
\[ \varepsilon_{\rm V} = 3 \cdot \varepsilon \](4)

Additionally, Raman scattering spectroscopy was performed to confirm the crystal structure of the YSZ-Frame specimens because the yttria stabilized zirconia has three crystal symmetries: namely monoclinic, tetragonal, and cubic, in accordance with its yttrium concentration. The details of a measurement setup and the apparatus were explained elsewhere20).

3. Results and Discussions

The prepared NiO-YSZ specimens were sufficiently dense, with relative densities of over 95%. The color of NiO-YSZ, Ni-YSZ and YSZ-Frame are summarized in Table 1. For the specimen N10Y90, the observed color was green, indicating that it had not changed by reduction. In the case of N20Y80, the inside of the YSZ-Frame was green and covered with a thick white layer although the color of Ni-YSZ was gray. However, after the reduction, XRD peaks of NiO was disappeared in the Ni-YSZ specimens. These results indicated that the gas permeabilities in the specimens with low NiO volume fractions were poor and NiO (green) was reduced to Ni (gray) only at the surface area of the specimens, retaining the NiO-YSZ core. Considering that the percolation limit in composites consisting of similar grains is ~30 vol%, the NiO networks in N10Y90 and N20Y80 are not connected. In addition, the gas flow channels fabricated by the reduction of NiO are not connected and do not show sufficient gas permeability to the inside of the specimens.

Table 1 Color of the NiO-YSZ, Ni-YSZ and YSZ-Frame specimens
  Color
Abbreviation NiO-YSZ Ni-YSZ YSZ-Frame
N10Y90 Green Green Green
N20Y80 Green Gray W+Green
N30Y70 Green Gray White
N40Y60 Green Gray White
N50Y50 Green Gray White
N60Y40 Green Gray White
N70Y30 D.Green Gray White
N80Y20 D.Green Gray White
N90Y10 D.Green Gray White

D.Green: Dark green, W+Green: White (outer surface), Green (inside)

Figures 1 (a), (b), and (c) show transmission electron microscopy (TEM) images of the NiO-YSZ specimens N10Y90, N50Y50, and N90Y10, respectively. In addition, a scanning electron microscopy (SEM) image of NiO-YSZ (N50Y50) is shown in Fig. 2. As shown in the Figs. 1 (a), and (c), every interface between YSZ and NiO were connected smoothly, and a secondary phase was not observed. The NiO and YSZ phases in N10Y90 and N90Y10 were isolated, as expected from the percolation limit. By contrast, in N50Y50, both phases made continuous networks, as shown in Fig. 1(b). Chemical composition analysis by energy-dispersive X-ray (EDX) spectroscopy was performed during TEM observation. The results indicated that a non-negligible amount of NiO was dissolved into YSZ and the amount of NiO in the YSZ increased as the NiO volume ratio increased. The observed values were ~1%, 8%, and 10% for N10Y90, N50Y50, and N90Y10, respectively. Although the accuracy of EDX has to be confirmed, the Ni solution and its dependence on the NiO volume ratio were experimentally determined, and the influence of the Ni solution on the YSZ was evaluated.

Fig. 1

TEM images of the NiO-YSZ specimens at (a) N10Y90, (b) N50Y50, (c) N90Y10.

Fig. 2

SEM image of the NiO-YSZ specimens at N50Y50.

Fig. 3

Raman spectra of the YSZ-Frame specimens.

Raman scattering spectroscopy was performed to confirm the influence of the Ni solution on the crystal structure. Figure 3 shows the Raman spectra of the YSZ-Frame specimens. Since YSZ-Frame specimens could not be obtained for N10Y90 and N20Y80, and that of N90Y10 was powdery, their Raman spectra were not obtained. Comparing the Raman spectra of the YSZ-Frame specimens with that of pure YSZ showed that the almost same spectra consisted of the main peak due to a T2g vibration mode, and defect-induced peaks32). These results indicated that the crystal symmetry of the specimens remained cubic although Ni was dissolved into the YSZ lattice. Therefore, strain analysis was performed with the assumption that the crystal symmetry of every specimen was cubic.

Figure 4 shows the three XRD patterns of the NiO-YSZ, Ni-YSZ, and YSZ-Frame specimens at N50Y50. In NiO-YSZ, the peak of YSZ (620) was broader than that of Ni-YSZ and YSZ-Frame, and its position was located at a greater angle as compared to the others. Since reduction (800℃) and etching (room temperature) were performed at relatively lower temperatures than sintering, grain growth, grain boundary sliding, and chemical inter-diffusion, which change the peak width and position, were not expected. Therefore, these peak changes are concluded to be the change in their residual strain. The direction of the peak position shift from the NiO-YSZ to YSZ-Frame indicates that the strain in the NiO-YSZ and Ni-YSZ is compressive, and its polarity corresponds to thermal strain because the thermal expansion coefficient of NiO and Ni is higher than that of YSZ. The broader peaks also indicated that the residual strain was not uniform and distributed in each phase. Because thermal strain is introduced via interfaces between the NiO, Ni, and YSZ phases and relaxed from the interfaces to the inside of the phases, these distributions are reasonable phenomena.

Fig. 4

XRD patterns of N50Y50 in (a) NiO-YSZ, (b) Ni-YSZ, and (c) YSZ-Frame.

The calculated peak positions of YSZ in three types, and those of NiO and Ni in NiO-YSZ, and Ni-YSZ are shown in Figs. 5 and 6. Some data points of N10Y90, and N20Y80 were not obtained because NiO could not be reduced in specimens with a low NiO volume fraction. In Fig. 5(a), the YSZ peak positions in YSZ-Frame showed a local minimum around a NiO volume fraction of 50%, and shifted to a higher angle by 0.2° at N90Y10. This shift is equivalent to a lattice constant change of −0.06%. This shrinkage is speculated to be the influence of the Ni solution. In the NiO-YSZ, and Ni-YSZ specimens, the peak positions show a linear dependence to the NiO volume fraction, i.e. the residual strain in them can be explained by thermal stress, except for that of YSZ in Ni-YSZ.

Fig. 5

Peak position shift of YSZ in NiO-YSZ, Ni-YSZ, and Etched YSZ as a function of NiO ratio.

Fig. 6

Peak position shift of Ni in NiO-YSZ, and Ni in Ni-YSZ as a function of NiO ratio.

Figure 7 shows the calculated volume strain as a function of the NiO volume ratio by eqs. (1) to (3), and (4). The influence of Ni solution into YSZ as shown in Fig. 5 is canceled using dYSZ3 calculated from the peak position of the YSZ-Frame at each volume fraction. These values In the NiO-YSZ specimens, the volume strains of NiO and YSZ show similar linear dependence with opposite polarities. This relationship is the typical volume fraction dependency of residual strain based on thermal strain and is observed in many composites33). Furthermore, considering bulk modulus, derived from the thermal expansion coefficient (Table 2)34,35), the thermal volume strains in NiO and YSZ for N50Y50 are simply calculated to be 0.25 and −0.27, respectively. These values show good agreement with the observed results. Generally, conventional SOFCs use anodes with a NiO volume ratio of ~60%. Hence, in conventional anodes before reduction, tensile strain greater than 0.1% and compressive strain greater than 0.2% are thought to exist in the NiO and YSZ phases at room temperature, respectively.

Fig. 7

Volume strain of (a) NiO, YSZ in NiO-YSZ, and (b) Ni, YSZ in Ni-YSZ as a function of NiO ratio.

Table 2 Mechanical, and thermal properties of YSZ, NiO, and Ni.
Material YSZ34) NiO35) Ni35)
Young’s modulus, E/GPa 190 220 205
Poisson’s ratio, ν/- 0.308 0.3 0.29
Bulk modulus, K/GPa 165 183 163
Thermal expansion coefficient, TEC/× 10−6 K−1 10.8 14.5 12.5

The trend of residual strain in YSZ changes drastically upon reduction of NiO to metallic Ni. The volume strain increases with increasing NiO volume fraction, and show a local maximum at 50 vol%. Then, it decreases significantly to one-third of the volume strain in NiO-YSZ at 60 vol%. Since the temperature difference between reduction (800℃) and room temperature is smaller than that between sintering (1400℃) and room temperature, and porosity is produced in the cermets by the NiO reduction, the thermal strain is expected to be smaller than that of NiO-YSZ. However, the significant volume strain change between 50 vol% and 60 vol% cannot be explained only in terms of temperature and porosity differences.

Plastic deformation of Ni and the strain relaxation of YSZ are responsible for the observed drastic volume strain change. Nickel is a ductile metal, and its yield stress decreases as temperature increases. At the same time, the tensile strain in the Ni phase increases as temperature decreases from the reduction temperature to room temperature. Since Ni around the interfaces shows a maximum tensile strain during cooling after the reduction process, the tensile stress in the Ni phase around the interfaces is greater than the yield stress, leading to plastic deformation. Plastic deformation of Ni, which occurs due to the movement of dislocations, relaxes elastic strain in YSZ; however, elastic strain inside of the Ni remained, as schematically illustrated in Fig. 8.

Fig. 8

Schematic illustration of Ni plastic deformation near the interfaces between Ni and YSZ phases.

In this study, the systematic study of the residual strain in NiO-YSZ and Ni-YSZ reveals that the residual strain in YSZ is drastically changed by the reduction of NiO to Ni, and can be attributed to the ductility of the Ni phase. We usually focus on the macroscopic residual strain and stress in the electrolyte, and other components. In addition, these results suggest that the microscopic residual strain and stress of the YSZ networks in Ni-YSZ anodes, which influence their mechanical strength, are drastically changed.

4. Conclusion

The residual strain in NiO-YSZ, and Ni-YSZ was systematically studied. By using Ni-YSZ, which was directly obtained by NiO-YSZ reduction, and YSZ-Frame, which was obtained by the etching of Ni-YSZ, the volume fraction dependence of the residual strain was measured by XRD. The obtained results revealed the influence of NiO reduction on the residual strain in the NiO, Ni, and YSZ phases. Thermal strain was the primary source of the residual strain. However, the residual strain of YSZ in Ni-YSZ was relaxed due to Ni plastic deformation caused by the ductility of Ni. As a result, the Ni-YSZ cermet with (NiO:YSZ = 60:40 vol%), which is the common volume fraction in the conventional SOFC anode, showed one-third of the YSZ residual strain of that in NiO-YSZ. This strain change influences the mechanical reliability of the YSZ networks in Ni-YSZ cermets.

Acknowledgments

This work was carried out as part of the New Energy and Industrial Technology Development Organization (NEDO) project, “Development of Systems and Elemental Technology on SOFC.” The authors would like to thank Editage (www.editage.jp) for the English language review.

REFERENCES
 
© 2017 The Japan Institute of Metals and Materials
feedback
Top