We present a unique platform for characterizing electrode surfaces in solid oxide fuel cells (SOFCs) that allows simultaneous performance of multiple characterization techniques (e.g. in situ Raman spectroscopy and scanning probe microscopy alongside electrochemical measurements). Complementary information from these analyses may help to advance toward a more profound understanding of electrode reaction and degradation mechanisms, providing insights into rational design of better materials for SOFCs.
Solid oxide fuel cells (SOFCs) are potentially the most efficient and cost-effective solution to utilization of a wide variety of fuels beyond hydrogen 1-7. The performance of SOFCs and the rates of many chemical and energy transformation processes in energy storage and conversion devices in general are limited primarily by charge and mass transfer along electrode surfaces and across interfaces. Unfortunately, the mechanistic understanding of these processes is still lacking, due largely to the difficulty of characterizing these processes under in situ conditions. This knowledge gap is a chief obstacle to SOFC commercialization. The development of tools for probing and mapping surface chemistries relevant to electrode reactions is vital to unraveling the mechanisms of surface processes and to achieving rational design of new electrode materials for more efficient energy storage and conversion2. Among the relatively few in situ surface analysis methods, Raman spectroscopy can be performed even with high temperatures and harsh atmospheres, making it ideal for characterizing chemical processes relevant to SOFC anode performance and degradation8-12. It can also be used alongside electrochemical measurements, potentially allowing direct correlation of electrochemistry to surface chemistry in an operating cell. Proper in situ Raman mapping measurements would be useful for pin-pointing important anode reaction mechanisms because of its sensitivity to the relevant species, including anode performance degradation through carbon deposition8, 10, 13, 14 (“coking”) and sulfur poisoning11, 15 and the manner in which surface modifications stave off this degradation16. The current work demonstrates significant progress towards this capability. In addition, the family of scanning probe microscopy (SPM) techniques provides a special approach to interrogate the electrode surface with nanoscale resolution. Besides the surface topography that is routinely collected by AFM and STM, other properties such as local electronic states, ion diffusion coefficient and surface potential can also be investigated17-22. In this work, electrochemical measurements, Raman spectroscopy, and SPM were used in conjunction with a novel test electrode platform that consists of a Ni mesh electrode embedded in an yttria-stabilized zirconia (YSZ) electrolyte. Cell performance testing and impedance spectroscopy under fuel containing H2S was characterized, and Raman mapping was used to further elucidate the nature of sulfur poisoning. In situ Raman monitoring was used to investigate coking behavior. Finally, atomic force microscopy (AFM) and electrostatic force microscopy (EFM) were used to further visualize carbon deposition on the nanoscale. From this research, we desire to produce a more complete picture of the SOFC anode.
1. Fabrication of a YSZ-embedded Mesh Anode Cell
2. Exposure, Polishing, and Modification of Ni Mesh Electrode
3. Preparation and Electrochemical Testing of Full Cells
*Figure 1 presents a schematic of the YSZ-embedded Ni mesh cell, along with a typical photograph and optical micrograph of the embedded mesh.
*For our investigations, cells were electrochemically characterized with an EG&G PAR potentiostat (model 273A) coupled with a Solartron 1255 HF frequency response analyzer using CorrWare and ZPlot softwares (Scribner and Associates). Linear sweep voltammetry and constant-voltage amperometry were used to characterize cell performance, and impedance spectra were acquired in the frequency range of 100 kHz to 0.1 Hz with an amplitude of 10 mV. For the sulfur poisoning study, a certified gas mixture of 100 ppm H2S in H2 was mixed into the fuel gas stream with pure H2 to obtain a 20 ppm H2S/H2 mixture.
4. Post-test Raman Spectromicroscopic Mapping
Raman spectromicroscopy was performed using a Renishaw RM1000 system equipped with a Modu-Laser StellarPro 514 nm Ar-ion laser (5 mW) and a Thorlabs HRP170 633 nm He-Ne laser (17 mW). The system is equipped with an X-Y-Z motorized stage (Prior Scientific H101RNSW) and a 50X objective lens, which together allow for ~2 μm mapping resolution. Renishaw WiRE 2.0 software was used in conjunction with the hardware. Data was processed using MATLAB (MathWorks).
5. In situ Raman Monitoring of Coking8
*The in situ Raman analysis was performed with a custom-modified Harrick Scientific high-temperature reaction chamber.The chamber is equipped with a quartz window cap, gas connections, and a cooling line. A schematic and photograph is provided in Figure 2.
CAUTION: Cooling water should be used to protect the optical microscope on the Raman system from heating!
6. Nanoscale Visualization of Coking by AFM and EFM
Sulfur Poisoning Analysis
Shown in Figure 4 are typical I-V and I-P curves of a cell with a Ni mesh electrode under H2 and 20 ppm H2S condition. Clearly, the introduction of even just a few ppm of H2S can poison the Ni-YSZ anode and cause considerable performance degradation.
In order to more intensively understand the poisoning behavior of the Ni-YSZ anode, AC impedance spectroscopy of the cell was performed under open circuit voltage (OCV) conditions. Shown in Figure 5 are the Nyquist plots for the cell before and after the anode were exposed to fuels containing 0 and 20 ppm H2S at 767 °C. Two-electrode impedance was used in this study, since the cell impedance is dominated by that of the working electrode; a reference electrode was therefore not needed. The impedance spectra show that the bulk resistance remained the same while the interfacial resistance increased dramatically after the anode was exposed to H2S-containing fuel.
Shown in Figure 6 is the typical poisoning and recovery behavior of the cell at 767 °C under 20 ppm H2S. Both the poisoning and recovery processes seem to finish and reach steady state within a few minutes, which is very different behavior from thick Ni-YSZ anodes that previously studied, where a much longer timeframe was needed especially for the recovery process 25, 26.
Raman intensity mapping was used to gain more chemical information about how the electrode surface interacts with sulfur in fuel gas. Figure 7 displays an optical micrograph of the Ni/YSZ interface along with a Raman map of the same area plotting the intensity of a mode associated with SOx species (integrated intensity between 960-1000 cm-1). The species were exclusively observed on the Ni surface and were generally concentrated more away from the TPBs.
In situ Raman Analysis of Coking on YSZ-embedded Ni Mesh
Figure 8a displays an optical micrograph of an unmodified Ni mesh electrode at 625 °C in reducing atmosphere (to prevent undesired Ni oxidation) directly before introduction of C3H8. After C3H8 began flowing into the chamber, spectra were collected periodically from a spot on the surface of a sample of Ni mesh. This spot is marked by the green circle in Figure 8b, which shows the mesh after 15 hr of wet C3H8 exposure at 625 °C. Carbon is only detectable on the Ni mesh surface; the spectrum collected from the substrate (red square) shows only YSZ features (Figure 8c). A plot of the change in relative carbon signal intensity over the first few hours of exposure time is shown in Figure 8d. The intensity of the carbon signal increased as carbon built up on the surface of the Ni from coking. The amount of detectable carbon signal eventually leveled off after a few hours. A Ni mesh sample modified by BaO as described in Protocol 2.5 was subjected to the same experimental conditions. A micrograph of and Raman spectra collected from the surface of the modified mesh sample at the same 15-hour exposure mark as Figures 8b/8c are shown in Figure 8e and Figure 8f, respectively. Carbon signal was not detectable even on the Ni surface in this case.
SPM Analysis of Coking on Ni
Inhomogeneous dark patches formed on the surface of the Ni coupon after it was exposed to C3H8-containing gas, which are depicted in the SEM images in Figure 9. Shown in Figure 10 are AFM/EFM images taken from both these light and dark regions.
Figure 1. Schematic diagram of the Ni-pattern electrode cell fabrication and characterization27.
Figure 2. Schematic of environmental chamber setup for in situ Raman analysis (left) along with a photograph of the chamber (right). The yellow tubes in the photograph are cooling water lines which are not pictured in the schematic on the left.
Figure 3. Schematic of the working principle of electrostatic force microscopy (EFM). (A) The schematic showing bias applied onto the sample. (B) The phase angle shift of the AFM tip plot vs. the vibration frequency; strong electrostatic interaction incurs larger degrease in the phase angle. (C) The schematic electron energy band diagram of the sample (nickel and carbon) and tip when the sample is biased positively w.r.t. the tip and (D) when the sample is biased negatively w.r.t. the tip28.
Figure 4. Typical performance of the cell operated under pure H2 and H2 with 20 ppm H2S at 767 °C 27.
Figure 5. Impedance spectra of the cell with pure H2 and H2 with 20 ppm H2S as fuels measured at 767 °C under open circuit voltage 27.
Figure 6. Typical (A) poisoning and (B) recovery behavior of the cell in H2 with 20 ppm H2S at 767 °C operated at 0.75V 27.
Figure 7. Optical micrograph of corner of patterned Ni mesh electrode in YSZ that was operated in H2S-containing fuel and aged in air for 18 days (left) along with a Raman map of the SOx band in the same area (right)27.
Figure 8. (a) Optical micrograph of Ni mesh embedded in YSZ. (b) Same embedded Ni mesh after exposure to C3H8-containing gas at 625 °C for 15 hr. (c) Raman spectra collected in situ from the spots marked in (b) at the 15-hour mark of C3H8-containing gas exposure. (d) Plot of change in carbon Raman signal intensity collected over time from green circle spot on Ni mesh marked in (b). (e) Optical micrograph near interface of BaO-modified Ni mesh and YSZ during the same C3H8 treatment. (f) Raman spectra collected in situ at the 15-hour mark from spots marked in (e). (Reproduced from 8). Click here to view larger figure.
Figure 9. SEM images of coked Ni showing the morphological differences caused by inhomogeneous coking.
Figure 10. (A) AFM image of the light region shown in Figure 9 and (B) EFM image of the same area. (C) AFM image of the dark region shown in Figure 10 and (D) EFM image of the same area. The sample was biased to -1V with respect to the AFM tip, and thus giving a brighter color (weaker tip-sample interaction) to the region covered with carbon.
Sulfur Poisoning Analysis
The impedance spectra shown in Figure 5 suggest that sulfur poisoning is a surface or interfacial phenomenon rather than one that affects the bulk of the material. Specifically, the quick poisoning of the Ni mesh electrode (Figure 6) might result from the direct exposure of Ni electrode to the fuel gas and subsequent sulfur adsorption; gas diffusion would not limit the rate of this process as much as in the case of a thick porous Ni/YSZ anode. Strongly adsorbed sulfur at or near the triple phase boundary (TPB) between Ni, YSZ, and fuel would likely block the active sites for electrochemical oxidation of H2 at the TPB, resulting in quick performance loss. On the other hand, during the recovery process, the current passing through the cell can help the electrochemical oxidation of the sulfur adsorbed on anode surface to SO2, especially at or near the TPB. Once sulfur is oxidized to SO2, it will quickly desorb from anode surface, leading to the re-exposure of the Ni/YSZ interface to the gas phase (regeneration of TPBs) and virtually full recovery of the performance 26. However, the current might not be efficient enough to fully remove the adsorbed sulfur far away from the TPB regions, which may remain on the surface even after full recovery of the performance. This possibility was explored by aging the electrode in air to form SOx groups that can be detected by Raman spectroscopy (Figure 7). The Raman mapping findings support this hypothesis. In addition, no other Ni-S phases could be detected by Raman spectroscopy, which supports the notion that the poisoning process at the studied temperature does not involve new bulk sulfide phase formation.
In situ Raman Analysis of Coking on YSZ-embedded Ni Mesh
The Raman results provide clear evidence that coking occurs preferentially on the Ni surfaces when a Ni-YSZ anode is operated under a hydrocarbon fuel. This preference towards Ni for carbon deposition is likely due to Ni’s higher catalytic activity for such reaction over YSZ. On the other hand, carbon deposition was apparently staved off considerably by BaO modification. The modification treatment was similar to the one used in the work done by Yang et al. 16, which found that BaO nanoparticles on the Ni surface can stave off coking in the presence of water. The present work supports those findings.
SPM Analysis of Coking on Ni
SEM analysis revealed detailed shape and morphology of the different degrees of carbon deposition (Figure 9). The darker patches were likely regions of heavier carbon deposition. Electrostatic force microscopy (EFM) was used in combination with AFM to help confirm this and map the carbon deposition on the nanoscale (< 25 nm resolution). While AFM is capable of distinguishing the morphological variation after hydrocarbon treatment, EFM simultaneously gauges the surface potential through which the surface phases are identified (Figure 3). When the sample is biased positively or negatively vs. the tip, the region with lower or higher surface potential (Evac-Ef), respectively, will incur stronger interaction with the tip. Since nickel and carbon, which are the two species present on the sample, each have a different surface potential, EFM is capable of separating them. In the case of our study, the sample was biased to -1 V with respect to the tip, so carbon had weaker interaction with the tip. As shown in Figure 10, the light region had less roughness in terms of topography, and only a few spots associated with carbon show up in the EFM image, confirming very light carbon deposition. Meanwhile, the dark region inspected has a great amount of surface roughness, and the majority of the surface is covered by species having weak interaction with the tip, indicating much heavier coking.
Nanoscale species mapping based on EFM is a promising approach to study the impact of nano-sized catalysts applied onto the materials. For example, by applying BaO onto the nickel surface, the coverage and morphology of coking patches changed drastically. By comparing the EFM images that show the local carbon deposition behavior for different catalysts, their resistance capabilities can be better evaluated.
The authors have nothing to disclose.
This work was supported by the HeteroFoaM Center, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences (BES) under Award Number DE-SC0001061.
Name of Reagent/Material | Company | Catalog Number | Comments |
Nickel mesh | Alfa Aesar | CAS: 7440-02-0 | |
Ni Foil | Alfa Aesar | CAS: 7440-02-0 | |
YSZ powder | TOSOH | Lot No:S800888B | |
Ag paste | Heraeus | C8710 | |
Barium oxide | Sigma-Aldrich | 1304-28-5 | |
Silver wire | Alfa Aesar | 7440-22-4 | |
Acetone | VWR | 67-64-1 | |
Ethanol | Alfa Aesar | 64-17-5 | |
UHP H2 | Airgas | 99.999% purity | |
100 ppm H2S/H2 | Airgas | Certified custom mix | |
n-type Si AFM tip | MikroMasch | NSC16 | 10 nm tip radius |
Au coated AFM tip | MikroMasch | CSC11/Au/Cr | 20-30 nm tip radius |
Raman Spectrometer | Renishaw | RM1000 | |
Ar Ion laser | ModuLaser | StellarPro 150 | |
He-Ne laser | Thorlabs | HPL170 | |
Atomic Force Microscope | Veeco | Nanoscope IIIA | |
Moving Raman Stage | Prior Scientific | H101RNSW | |
Optical Microscope | Leica | DMLM | |
Scanning Electron Microscope | LEO | 1550 | |
Tube Furnace | Applied Test Systems | 2110 | |
Polisher | Allied High Tech Products | MetPrep | |
6 μm Grinding media | Allied High Tech Products | 50-50040M | |
3 μm Polishing media | Allied High Tech Products | 90-30020 | |
1 μm Polishing media | Allied High Tech Products | 90-30015 | |
0.1 μm Polishing media | Allied High Tech Products | 90-32000 | |
Raman chamber | Harrick Scientific | HTRC |