An all perovskite direct methanol solid oxide fuel cell with high resistance to carbon formation at the anode

Abstract

A chemically stable perovskite material Sr₂Fe_1.5Mo_0.5O₆ (SFMO) is employed as the anode of a solid oxide fuel cell (SOFC). An electrolyte-supported single cell with anode, electrolyte and cathode all made of perovskite structured materials and with a configuration of SFMO|La_0.8Sr_0.2Ga_0.83Mg_0.17O₃|Ba_0.5Sr_0.5Co_0.8Fe_0.2O₃ (SFM|LSGM|BSCF) is fabricated by a screen printing method. The single cell gives a maximum power density of 391 mW cm⁻² for CH₃OH, and 520 mW cm⁻² for H₂ as the fuel, respectively, at 1073 K with oxygen as the oxidant gas. The mass spectra of the flue gas out of the test reactor confirm that methanol thermally decomposes inside the anode chamber and generates mainly CO and H₂ at 1023 K. Analysis of the after-test cell tells that the anode surface has no carbon formation under reaction with methanol as the feed for 3 h. The carbon resistance is attributed to the fact that the anode is in oxide state which cannot facilitate the formation of bulk carbon with graphite structure. The fast activation and gasification of the carbon species by the oxidative atmosphere around the anode surface are also beneficial factors. The test results indicate also that the activation of CH₃OH is much more difficult than that of H₂.

---

1. Introduction

The direct electrochemical-oxidation of carbon-containing fuels in a solid oxide fuel cell (SOFC) has been a great challenge because of the high tendency of carbon deposition on the anode surface.^1–3 Carbon deposition on the anode leads to the coverage of the active sites, blockage of the pores and destruction of the microstructure, which ends the lifetime of the SOFC.⁴ It has been well known that the carbon-containing fuels have a large thermodynamic potential for coke formation in many catalytic processes.^5,6 However, the formation of bulk phase carbon is only kinetically favored when suitable catalytic sites exist. Metal catalysts often facilitate fibrous carbon formation.^7,8 In the direct carbon-containing fuel electrochemical-oxidization SOFCs, metals or reducible metallic oxides have often been employed as the catalytic component of the anode composite material. As a consequence, coking at the anode has been a major obstacle for the commercial application of SOFCs.^9,10

Methanol as a fuel is carbon dioxide neutral, renewable and sustainable if produced from biomass or from a solar-energy powered catalytic process.^11,12 In a liquid state at ambient conditions, methanol is of high energy density so that greatly eases the effort for storage, transportation and distribution.^10–13 Furthermore, methanol as a readily available and low cost oil processing product, has been investigated extensively in reforming reactions and it is commonly considered that it is more active than the most of the other carbon-containing fuels benefiting a desired intermediate-temperature operation.^14–17

Under the direct feed of a dry carbon-containing fuel, an ideal anode catalytic material should be highly active for the catalytic dissociation of the fuel molecule and highly suppressive for C–C formation as well as highly inductive for producing CO and CO₂.¹⁸ A metallic Ni-based anode fails due to its high activity for the formation of C–C bonds and graphite structured carbon fibers and tubes. Nevertheless, nano-sized Ni particles have a better tolerance to coking and highly dispersed Ni on perovskite has shown a good performance.¹⁹ Perovskite materials also share a prominent anti-coking capability in the methane and ethanol steam reforming reactions.^20–23 Perovskite has a cubic unit cell with A ions at the body center, B ions at the corners and oxygen ions at the center of the faces. In the perovskite structure, the A cation is coordinated with twelve oxygen ions and the B cation with six. The perovskite has a moderate activity towards the activation of carbon-containing fuels and is carbon-resistive because graphitic carbon formation needs an ensemble of several metal atoms at the catalytic site.⁴ Huang et al.² reported that perovskite Sr₂Mg_1−xMn_xMoO_6−δ as an SOFC anode has a long-term stability with tolerance to sulfur. They got a single-cell performance with a maximum power density of 438 mW cm⁻² with dry methane as the feed at 1073 K with an electrolyte of 300-μm-thick La_0.8Sr_0.2Ga_0.83Mg_0.17O₃ (LSGM).² Wang et al.²⁴ employed Sr₂FeMoO₆ as the anode and got a maximum power density of 605 mW cm⁻² with dry methane as the fuel at 1123 K. However, this material has to be synthesized in a H₂ atmosphere and reduced for 5 h at 1023 K before testing. Sr₂Fe_1.5Mo_0.5O₆ (SFMO) is stable in an oxidative atmosphere and has been employed as the electrodes in a symmetrical SOFC and as the cathode in a normal SOFC.^25,26 SFMO has high ionic and electronic conductivity as well as high oxygen surface exchange and diffusion constants.^25,27 Based on the existing data, it is assumed that SFMO can be a good material for the anode of the direct methanol fuel cell due to its ability to dissociate hydrocarbons and fast transfer of oxygen ions from the bulk to the surface.

Herein, we report for the first time a SOFC, with SFMO as the anode, for directly utilizing methanol as the fuel and compare the performance with that of hydrogen. Efforts are made to illuminate how the methanol reacts on the perovskite anode in the SOFC.

2. Experimental

SFMO powder was synthesized by a solid-state reaction. Stoichiometric SrCO₃, Fe₂O₃ and MoO₃ were thoroughly mixed with an agate mortar and then calcined at 1173 K for 10 h. The calcined powder was pelletized at 200 MPa and annealed at 1523 K in air for 5 h to obtain a pure perovskite structure. For XRD measurement, the pellets were ground and ball milled into very fine powders. The XRD patterns of the fresh sample and the reduced one, in H₂ at 1273 K for 5 h, were recorded at room temperature using a D/max 2500 v/pc instrument (Rigaku Corp. Japan) with Cu-Kα radiation, 40 kV and 200 mA, at a scanning rate of 5° min⁻¹. The surface properties of the fresh and the after-test SFMO were tested by a PHI 1600 X-ray photoelectron spectroscopy (XPS) system with Mg-Kα radiation. The preparations of Ba_0.5Sr_0.5Co_0.8Fe_0.2O₃ (BSCF) cathode and LSGM electrolyte materials have been described elsewhere.^28,29

The electrolyte-supported single cell was fabricated by a screen-printing technique. LSGM powder was pressed to a pellet under 200 MPa and sintered at 1723 K for 20 h to the dense state with a thickness of 300 μm. SFMO and BSCF were both mixed with a binder (V006, Heraeus Ltd.) to make into slurries. The SFMO slurry was firstly coated on one side of the electrolyte and sintered at 1423 K in air for 1 h. The BSCF slurry was then screen-printed on the other side of the electrolyte and sintered at 1273 K in air for 1 h. Ag paste was coated on both sides to function as the current collector.

A home-made setup was employed for the performance test of the single cell at 1023 K and 1073 K under atmosphere. Liquid methanol is gasified at 423 K before being fed into the anode. The hydrogen or pure methanol stream, with a flow rate of equivalent 80 ml min⁻¹, passed through an alumina tube to reach the anode chamber. Pure oxygen fed the cathode chamber with a flow rate of 100 ml min⁻¹. The cell was operated so that H₂ firstly was fed as the fuel and then the feed switched to methanol after all tests of H₂ were done. To test the decomposition of methanol at the inlet channel and anode chamber, mass spectra of the after-condensed outlet flue gases of the test reactor are recorded for the cell with and without the anode by a Hiden Analytical Instrument MS. The I–V characteristics of the single cell were measured by an IT-30 electronic load made by Ningbo BaTe Technology Co., Ltd., China. The impedance spectra of the single cell at open circuit (OC) conditions were recorded with an Ametek Electrochemical Work Station. The SEM images and EDX of the anode surface before and after test were obtained with a Genesis XM2 instrument to reflect any change induced by the methanol feed.

3. Results and discussion

3.1. Characterization

Fig. 1 depicts the XRD patterns of the fresh SFMO and the one reduced under H₂ at 1223 K for 5 h. Both the fresh and the reduced SFMO samples display clearly and identically a pure-perovskite structure with only small shifts of the peak positions for the two samples. SFMO has a cubic crystal structure, with a lattice constant a equalling 0.7851 nm and 0.7859 nm for the fresh sample and the reduced one, respectively. The lattice constants here are calculated with an XRD analysis software, i.e., Jade. It can be remarked that the SFMO structure is stable both in oxidative and reductive atmospheres. The results agree with those reported by Liu et al.²⁸

Fig. 1 XRD patterns of (upper line) fresh SFMO and (lower line) reduced SFMO in H₂ at 1273 K for 5 h.

Fig. 2 shows the chemical states of Fe and Mo elements in the fresh and after-test SFMO by the Fe-2p and Mo-3d core-level spectra, the atom ratios for each element are summarized in Table 1. Metals such as Fe or Mo did not appear. The results showed a slight change in the atomic ratios of different species, i.e., an increase in Fe³⁺ and Mo⁵⁺, and a decrease in Fe²⁺ and Mo⁶⁺. This slight difference is assumed to be caused by the electro-oxidation reactions, but further verification on this difference need to be done. XPS results further confirm that this material is chemically stable under the operation conditions. This property promises the redox-stability of the anode and easy treatment during the fabrication of the cell. It has been reported that, since the Mo⁵⁺/Mo⁶⁺ redox band overlapped with the Fe³⁺/Fe²⁺ couple, the Fe³⁺ survived consequently from being fully reduced in the perovskite structure.²⁹ Unlike Sr₂FeMoO₆ which is famous for its magnetoresistance property with an almost regular arrangement of Fe and Mo atoms at B sites, the prepared SFMO in this work may have a much more random arrangement of B site metal ions due to the inequality of the metal ratios.^30,31 Analysis of the XPS indicates that the theoretical stoichiometries of oxygen is 5.977 and 5.98 for the fresh and the after-test SFMO respectively, thus it can be concluded that there exist substantial oxygen vacancies in the SFMO lattice, promising a good oxygen ionic conductivity. The electronic conductivity of SFMO is reported to be high both in air and hydrogen atmospheres and it is expected that the large iron content in this material permits the formation of Fe–O–Fe bonds as the pathways for the movement of electrons or electron holes.²⁹ In summary, the co-existence of Fe³⁺–Mo⁵⁺, Fe²⁺–Mo⁶⁺ pairs in the SFMO structure results in a rich oxygen vacancy as well as an excellent electronic conductivity.

Fig. 2 (a) Fe-2p and (b) Mo-3d core-level spectra of SFMO at room temperature. (⁃•⁃): before-test; (⁃■⁃) after-test.

Table 1 Summary of the atom ratios of Fe and Mo by XPS analysis

	Fe^2+	Fe^3+	Mo^5+	Mo^6+
Fresh SFMO	0.539	0.461	0.507	0.493
After-test SFM	0.497	0.503	0.543	0.457

---

3.2. Performance

3.2.1. Characters of the output parameters. Fig. 3 presents the performance of the SOFC with a configuration of SFMO|LSGM|BSCF at 1023 K and 1073 K. The maximum power densities with H₂ as the fuel are 376 mW cm⁻² at 1023 K and 520 mW cm⁻² at 1073 K, respectively. For methanol, the values are 308 mW cm⁻² at 1023 K and 391 mW cm⁻² at 1073 K, respectively. The open circuit potential (OCP) of the cell is around 1.10 V for H₂ which is close to the value calculated from the Nernst equation 1.16 V (1073 K), and about 0.92 V for CH₃OH which is lower than the Nernst value 1.18 V (1073 K). Faro et al. reported that with a Ni-modified La_0.6Sr_0.4Fe_0.8Co_0.2O₂/Ce_0.9Ga_0.1O₂ anode the maximum power output is 350 mW cm⁻² and the OCP is 0.78 V at 1073 K.³² Clearly our system achieved a better performance. It is worth to note that for H₂ as the fuel, the voltage for the maximum power density is about 0.71 V, while for CH₃OH it is around 0.51 V. Under the same polarized conditions, i.e., the difference between the OCP and the discharging potential is the same, the cell with H₂ as the fuel has a much higher current density than that with methanol. For example, when the polarization value equals 0.2 V at 1073 K, the current density reaches 378 mW cm⁻² for H₂, while the value is 230 mW cm⁻² for methanol. With the same configuration of the cell, ohmic loss and cathode polarization of the cell with H₂ are theoretically much larger than that with methanol since polarizations increase with the current density. Since the total polarization of a single cell is composed by ohmic loss, anode polarization and cathode polarization, the anode polarization of the cell using H₂ as the fuel is much smaller than that using methanol as the fuel. The difference in the anode polarization may be explained by fact that the electro-oxidation of H₂ is not restricted by the activation process, but the electro-oxidation of CH₃OH is much more difficult than that of H₂. The output result of the single cell illuminated that the perovskite SFMO has excellent activity towards the electro-oxidation of both H₂ and methanol. Interpretation of this phenomenon is up to now unclear and here an assumption of the reaction process on perovskite is proposed. Different from the three-phase-boundary of the Ni-based composite electrode, the electrochemical reaction region of the perovskite anode is two-phase-boundary, i.e., the boundary between gas phase and solid phase, due to the mixed conduction property of perovskite. Since the Ni-based composite electrode and perovskite electrode have similar specific surface area of ∼2 m² g and the microstructure, thus it is very likely that the perovskite electrode has a much larger active area than that of the composite electrodes and further benefits the reaction rate.^9,10 As to the ability for the activation of the fuel, it is usually attributed to the B site metals, i.e., Fe or Mo, for perovskite.²⁷ If considering the activated species attached to the B site metals, electrochemical oxidation of these species is very likely to be carried out by the oxygen ion just around the B site metals, which is contributed from the substantial oxygen vacancies and the high ionic conductivity of SFMO. These activation and electro-oxidation processes on SFMO ensure a final reasonable performance of the single cell.

Fig. 3 Performance of the built single cell. The measurement starts form OCP and each discharging potential is kept for 20 s to obtain the steady output parameters.

3.2.2. Influence of the thermal and catalytic decomposition of methanol. Fig. 4 is the mass spectra recorded for the after-condensed outlet flue gases of the test reactor with and without anode catalyst at 1023 K. The result of Fig. 4a shows that methanol pyrolysis takes place at 1023 K, generating mainly CO and H₂ as products. Fig. 4b shows the spectrum of the product with SFMO anode. For the case with the anode in the cell, a hydrogen pressure of 1.3 × 10⁻⁷ torr was measured, which is higher than that, of 0.9 × 10⁻⁷ torr, measured without the anode. This result explains that the catalytic decomposition occurs at the surface of the anode. The ratios of the condensed liquid volume to the inlet total volume of methanol in the (a) and (b) conditions are 80% and 65%, respectively. Neglecting the water content in the condensed liquid, roughly 20% and 15% of methanol are thermally and catalytically decomposed, respectively, at 1023 K. Due to the decomposition reactions, the gas reacting on the anode surface turns out to be a gas mixture of methanol, CO, H₂, etc. Thus, there will be competitive adsorption among methanol, H₂, CO, and even H₂O and CO₂etc. Consequently a very complex reaction system is formed. Unfortunately, the accurate quantitative analysis of the gas mixture is difficult.

Fig. 4 Mass spectra of the after-condensed anode side flue gas coming out of the reactor. (a) the cell without the anode in the reactor; (b) complete single cell in the reactor. The main fragments appearing in the Figure are: H (m/z = 1), H₂ (m/z = 2), CH₂ (m/z = 14), CH₃ (m/z = 15), O (m/z = 16), OH (m/z = 17), H₂O (m/z = 18), CO (m/z = 28), CH₃OH (m/z = 32), CO₂ (m/z = 44).

3.2.3. Resistance of the single cell. The impedance spectroscopy measured under OC conditions with H₂ and dry CH₃OH as the anode feed is shown in Fig. 5. The intercept with the x-axis at high frequency is normally attributed to the electrolyte resistance, while the intercept with the x-axis at lower frequency refers to the total resistance of the cell. For the same cell, the electrolyte resistance should be same. At 1023 K, the electrolyte resistance seems a little different for H₂ and CH₃OH. The fact is that the data points for CH₃OH are not as semicircular as that for H₂ because of the discharging process in the electro-oxidation of methanol is less stable than that in the electro-oxidation of H₂, especially at a lower temperature. There may be some errors in the intercept and by line-fitting it may look like the intercept is changed. The total resistance of the cell when fed by CH₃OH is larger (1.32 Ohm at 1023 K and 0.90 Ohm at 1073 K) than that when fed by H₂ (1.03 Ohm at 1023 K and 0.79 Ohm at 1073 K). The different values of the total resistance for the same cell with different fuels can be roughly considered as that the variation caused by the activation process in the anode. The resistance induced by the electrolyte is more than half of the total resistance. Therefore, the cell with a thinner electrolyte is expected to have a much smaller resistance and better performance.

Fig. 5 Impedance spectra of the single cell using H₂ and CH₃OH as the fuels under OC state at 1023 K and 1073 K. The intersections of the impedance curves and the x-axis at low and high frequencies are attributed to the electrolyte and total resistances, respectively. The electrolyte resistances at 1023 K and 1073 K are 0.57 Ω and 0.74 Ω, respectively.

3.2.4. Carbon tolerance of the anode. Coking resistance is an important parameter for an anode material. Fig. 6 and Fig. 7 illustrate the EDX and SEM images of the anode surface after reacting with methanol for 3 h at OC configuration under which coke formation is favored thermodynamically. The results obviously show that no carbon formed on the surface of the SFMO anode under the most favorable operation conditions for coking, i.e. OC state. Therefore, the anode under charging conditions is expected to be also carbon-free. The formation of bulk phase carbon usually demands an ensemble of several metal atoms, while the chemical stability of SFMO ensures no such metal clusters exist on the surface and thus prevents the appearance of carbon. Furthermore, considering the mixed conduction property of the SFMO, the active sites such as B site metal ions for activating fuel molecules are just around the oxygen vacancy/site where electro-oxidation occurs. Therefore, the tolerance towards carbon is also very likely to be caused by the instant formation–consumption of the activated carbon species. Furthermore, as competitive adsorption of CO, H₂O, etc. on the surface exist; the initial carbon species may be consumed by the gasification.

Fig. 6 EDX scanning of the after-test anode surface; the right Figure indicates there is no carbon element in the anode material.

Fig. 7 SEM images of the after-test anode surface; the morphology shows no carbon fiber or carbon whiskers attached to the anode material.

4. Conclusion

The SFMO material prepared by the solid-state reaction technique shows an excellent chemical stability under both oxidative and reductive atmospheres. Employing SFMO as the anode, a SOFC with the direct electrochemical oxidation of methanol, gave promising power densities, i.e. 308 mW cm⁻² and 391 mW cm⁻² at 1023 K and 1073 K, respectively. Impedance spectra with I–V curves of the single cell indict that the activation of methanol is much more difficult than that of hydrogen. A number of different measurements clearly confirmed that the anode survived from carbon formation. This result of carbon resistance as well as activity towards electro-oxidation of the fuel promises a reasonable application of SFMO as catalysts for the direct electrochemical oxidation of carbon-containing fuels.

Acknowledgements

The financial support of NSF of China under contract numbers 21076150 and 21120102039 is gratefully acknowledged. The work has been also supported in part by the Program of Introducing Talents to the University Disciplines under file number B06006, and the Program for Changjiang Scholars and Innovative Research Teams in Universities under file number IRT 0641.

References

1. S. Tao and J. T. S. Irvine, Nat. Mater., 2003, 2, 320–323.
2. J. B. Goodenough, Y. H. Huang, R. I. Dass and Z. L. Xing, Science, 2006, 312, 254–257.
3. E. P. Murray, T. Tsai and S. A. Barnett, Nature, 1999, 400, 649–651.
4. C. H. Bartholomew, Catal. Rev. Sci. Eng., 1982, 24, 67–112.
5. D. L. Trimm, Catal. Today, 1999, 49, 3–10.
6. J. Barbier, Appl. Catal., 1986, 23, 225–243.
7. L. Yang, S. Z. Wang, K. Blinn, M. F. Liu, Z. liu, Z. Cheng and M. L. Liu, Science, 2009, 326, 126–129.
8. Y. D. Li, D. X. Li and G. W. Wang, Catal. Today, 2011, 162, 1–48.
9. R. M. Ormerod, Chem. Soc. Rev., 2003, 32, 17–28.
10. S. McIntosh and R. J. Gorte, Chem. Rev., 2004, 104, 4845–4865.
11. M. Gattrell, N. Gupta and A. Co, Energy Convers. Manage., 2007, 48, 1255–1265.
12. M. E. Himmel, S. Y. Ding, D. K. Johnson, W. S. Adney, M. R. Nimlos, J. W. Brady and T. D. Foust, Science, 2007, 315, 804–807.
13. J. Xuan, M. K. H. Leung, D. Y. C. Leung and M. Ni, Renewable Sustainable Energy Rev., 2009, 13, 1301–1313.
14. J. M. Ogden, M. M. Steinbugler and T. G. Kreutz, J. Power Sources, 1999, 79, 143–168.
15. S. Assabumrungrat, N. Laosiripojana, V. Pavarajarn, W. Sangtongkitcharoen, A. Tangjitmatee and P. Praserthdam, J. Power Sources, 2005, 139, 55–60.
16. D. Cocco and V. Tola, Proceedings of the ASME Turbo Expo, 2006, 4(2006), 257–266.
17. D. Cocco and V. Tola, Proceedings of the 8th Biennial Conference on Engineering Systems Design and Analysis, 2006, 1, 137–147.
18. E. Nikolla, A. Holewinski, J. Schwank and S. Linic, J. Am. Chem. Soc., 2006, 128, 11354–11355.
19. V. Sadykov, N. Mezentseva, G. Alikina, R. Bunina, V. Pelipenko, A. Lukashevich, S. Tikhov, V. Usoltsev, Z. Vostrikov, O. Bobrenok, A. Smirnova, J. Ross, O. Smorygo and B. Rietveld, Catal. Today, 2009, 146, 132–140.
20. A. L. Sauvet, J. Fouletier, F. Gaillard and M. Primet, J. Catal., 2002, 209, 25–34.
21. P. Vernoux, E. Djurado and M. Guillodo, J. Am. Ceram. Soc., 2001, 84, 2289–2295.
22. J. Akikusa, K. Adachi, K. Hoshino, T. Ishihara and Y. Takita, J. Electrochem. Soc., 2001, 148, A1275–A1278.
23. S. Q. Chen, Y. D. Li, Y. Liu and X. Bai, Int. J. Hydrogen Energy, 2011, 36, 5849–5856.
24. Z. M. Wang, Y. Tian and Y. D. Li, J. Power Sources, 2011, 196, 6104–6109.
25. L. Zhang, Y. Liu, Y. Zhang, G. Xiao, F. Chen and C. Xia, Electrochem. Commun., 2011, 13, 711–713.
26. G. Xiao, Q. Liu, X. Dong, K. Huang and F. Chen, J. Power Sources, 2010, 195, 8071–8074.
27. G. Xiao, Q. Liu, F. Zhao, L. Zhang, C. Xia and F. Chen, J. Electrochem. Soc., 2011, 158, B455.
28. P. Zeng, Z. Chen, W. Zhou, H. Gu, Z. Shao and S. Liu, J. Membr. Sci., 2007, 291, 148–156.
29. Q. Liu, X. Dong, G. Xiao, F. Zhao and F. Chen, Adv. Mater., 2010, 22, 5478–5482.
30. S. Vasala, M. Lehtimäki, Y. H. Huang, H. Yamauchi, J. B. Goodenough and M. Karppinen, J. Solid State Chem., 2010, 183, 1007–1012.
31. J. Navarro, Mater. Res. Bull., 2003, 38, 1477–1486.
32. M. Lo Faro, A. Stassi, V. Antonucci, V. Modafferi, P. Frontera, P. Antonucci and A. S. Aricò, Int. J. Hydrogen Energy, 2011, 36, 9977–9986.

---