- Email: [email protected]
oxygen-enriched yttria-stabilized zirconia
Journal of Power Sources 293 (2015) 635e641
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
The sulfur poisoning of the nickel/oxygen-enriched yttria-stabilized zirconia Yanxing Zhang a, Zhengyang Wan a, Zongxian Yang a, b, * a b
College of Physics and Electronic Engineering, Henan Normal University, Xinxiang, Henan 453007, People's Republic of China Collaborative Innovation Center of Nano Functional Materials and Applications, Henan Province, People's Republic of China
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
The adsorption and diffusion properties of sulfur on the Ni/YSZ þ O are studied. The adsorbed sulfur doesn't favor to be located at the Ni/YSZ þ O interface. The extra O in YSZ weakens the S adsorption at the vacancy site of Ni/ YSZ-Ov. The extra O in YSZ improves the diffusion of S out of the vacancy of Ni/YSZ-Ov. The Ni/YSZ þ O can help to alleviate the sulfur poisoning as compared with the Ni/YSZ.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 February 2015 Received in revised form 28 April 2015 Accepted 13 May 2015 Available online xxx
The sulfur poisoning properties of the nickel/oxygen-enriched yttria-stabilized zirconia (denoted as Ni/ YSZ þ O) with or without interface O vacancy are studied using the first-principles method based on density functional theory. The effects of the extra O atom at the subsurface vacancy of Ni/YSZ are focused. It is found that S at the Ni/YSZ þ O can diffuse easily away from the interface oxygen to the top Ni layer sites. With the formation of O vacancy at the Ni/YSZ þ O interface (denoted as Ni/(YSZ þ O)-Ov), the adsorbed S prefers to diffuse back to the Ni/YSZ interface O vacancy. Compared with Ni/YSZ-Ov, the Ni/ (YSZ þ O)-Ov can effectively not only weaken the S adsorption at the interface O vacancy site, but also improve the diffusion of S out of the interface O vacancy. Therefore, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface O vacancy site as compared with the Ni/YSZ. © 2015 Elsevier B.V. All rights reserved.
Keywords: Solid oxide fuel cell Sulfur poisoning Nickel/yttrium-stabilized zirconia anode Oxygen vacancy
1. Introduction Solid oxide fuel cells (SOFCs) are expected to be a crucial technology in the future power generation [1,2]. SOFCs offer many
* Corresponding author. College of Physics and Electronic Engineering, Henan Normal University, Xinxiang, Henan 453007, People's Republic of China. E-mail address: [email protected] (Z. Yang). http://dx.doi.org/10.1016/j.jpowsour.2015.05.044 0378-7753/© 2015 Elsevier B.V. All rights reserved.
desirable advantages compared to other types of fuel cells and conversion devices due to their use of solid electrolytes, lack of moving parts, ability to circumvent precious metal use, high efficiency, low pollution, and fuel flexibility. The conventional anode for a SOFC, denoted as Ni/YSZ, is consisted of nickel and yttria-stabilized zirconia (YSZ). However, a major issue in the long-term stability and activity of the anode catalyst is its poor resistance toward poisonous compounds
636
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
presented in the feed stream of fuels. Trace amounts of H2S presented in biomass generated syngas streams are enough to deactivate the catalyst [3,4], which is called sulfur poisoning. Many previous experimental studies indicated that the S poisoning behavior is characterized by two stages, fast initial degradation and long term degradation [4e10]. Liu et al. [7,11] showed that S poisoning observed in the low concentration of H2S at elevated temperatures is originated from the dissociation of sulfur-containing species and the adsorption of atomic sulfur on the anode surface. The adsorbed H2S on Ni surfaces has been shown to dissociate above 300 K [12], with only S remaining on the surface. These studies clearly suggest that the elemental sulfur strongly adsorbs on the Ni surfaces with a small activation barrier (Ea) of H2S dissociation and a large exothermic enthalpy (DH). The strongly adsorbed S species block the active sites on the anode surface and thus increase the resistance to electrochemical oxidation of the fuel. The calculations [3] also suggest that the adsorbed sulfur species exist primarily in the form of atomic sulfur instead of molecular species, e.g., H2S. In fact, the faster or sluggish kinetics is related to the barriers of reactions. The very small Ea means that the barrier for H2S dissociation is small, but the large exothermic DH indicates that the barrier of the reverse reaction of the H2S dissociation is large. The very small Ea and large exothermic DH further imply fast kinetics for sulfur adsorption (as a result of H2S dissociation) and sluggish kinetics for sulfur removal, which is consistent with the experimental observation of the instant drop in performance upon exposure to H2S and a very slow recovery in performance after clean hydrogen is switched back. However, experimental results indicated that sulfur tolerance was in fact improved by using Ni/Sc2O3 [8] or Ni/Gd2O3-doped ZrO2 [13] anodes, which suggest at least, that sulfur tolerance depends strongly on anode and electrolyte materials besides nickel itself. Zeng et al. [14] have studied the mechanisms governing the sulfur poisoning of the triple-phase boundary (TPB) of Ni/XSZ (X2O3 stabilized zirconia) anodes using density functional theory. The calculated sulfur adsorption energies reveal a clear correlation between the size of the cation dopant X3þ and the sulfur tolerance of the Ni/XSZ anode. Malyi et al. [15] found that S addition to zirconia, either by doping or through gas diffusion, increases both the formation energy and migration barrier of the oxygen vacancies. Since the fuel oxidation is believed to take place at the TPB made of Ni, YSZ and fuel gas, oxygen vacancies may be created at the Ni/YSZ interface. As known, the anode of SOFC is the Ni/YSZ composite, instead of the Ni itself. The ideal Ni (111) only represents the anode region beyond the TPB. Therefore, people should give special attention to the Ni/YSZ system and the effects of the interface O vacancy. In our recent work [16], we studied the sulfur poisoning at the TPB region of the Ni/YSZ. We found that the adsorbed sulfur does not favor to be located at the stoichiometric Ni/YSZ interface. With O vacancy at the Ni/YSZ interface, the adsorbed S diffuses to the Ni/YSZ interface and is oxidized to S2 and trapped at the oxygen vacancy. The trapped S is very difficult to be removed by the fuel (e.g., H2) and therefore blocks the pathway for the O ion transfer. As a result, the resistance for O ion transfer would increase and the SOFC performance would drop. Trace amounts of S would block the O vacancy sites at the interface and induce the instant and significant drop in performance of SOFC. To alleviate the sulfur adsorption at the O vacancy site and/ or enhance the diffusion of S out of the interface oxygen vacancy site would help to enhance the resistance to S poisoning at the oxygen vacancy site. To improve the sulfur tolerance of SOFCs, alternative anode
compositions have been proposed. Copper/ceria/zirconia [17] anodes were reported to be stable in fuel gases containing up to 450 ppm H2S. Other anode compositions showing good sulfur tolerance include a lanthanum doped strontium titanate [18], Pdimpregnated titanate/cerate composition [19], lanthanum molybdate [20], gold/molybdenum disulfide [21], Ni/YSZ modified with niobia [22], and lanthanum vanadium oxide [23]. For considerations including cost, processability, and stability, minor modifications to the widely used Ni/YSZ anode may be preferred to the more exotic compositions and forms. For maintaining charge neutrality of the YSZ system, the oxygen vacancies are created during the substitution of Zr4þ by Y3þ in the YSZ lattice. Under practical operating conditions of SOFCs, these O vacancies are mobile to give rise to oxygen ionic conductivity via vacancy diffusion mechanism, and can be filled by O2 migrating through the crystal lattice of YSZ, to form an O-enriched YSZ surface, YSZ þ O (111). The O2 ions supplied to the intrinsic vacancy of YSZ originate from the electrochemical reaction of O2 at an electrode, which depends on the actual SOFC fuel gas operation and polarization conditions in the anode polarization measurement [24e26]. It is found that the favored hydrogen oxidation reaction process on Ni/YSZ is water formation on an O atom at the interface and this is also the most facile hydrogen oxidation reaction process on Ni/(YSZ þ O) [27]. In this work, we study the sulfur poisoning at the TPB region of the Ni/(YSZ þ O) -Ov, focusing on the effects of the extra O atom at the subsurface vacancy of Ni/YSZ-Ov on the adsorption and the diffusion of S at the interface O vacancy site. 2. Model and computation method All calculations presented in this work are performed employing the periodic density functional theory (DFT) method implemented in the Vienna Ab-Initio Simulation Package (VASP) [28]. The exchangeecorrelation interactions are treated with the PerdeweBurkeeErnzerhof (PBE) functional [29]. Spin-polarized calculations are applied throughout. The electroneion interactions are treated using the projector augmented wave (PAW) method [30,31]. The wave functions are expanded in plane waves with a cut off energy of 408 eV. The model of the Ni/YSZ cermet with the horizontal dimensions of 12.56 7.25 Å as that used in the Shishkin and Ziegler's work [32] is adopted as the substrate. The Ni/YSZ model with two intrinsic vacancies used in this work includes two building blocks of the 9% mol YSZ, which has 10 ZrO2 and 1 Y2O3 and can be considered as an elementary building unit of YSZ with 9% mol concentration of yttria shown in Fig. 1(a). A vacuum layer of 15 Å is used to separate the periodic images in the direction perpendicular to the surface. The model has been successfully used in our previous researches [16,33e37]. In the proposed structure, both Ni and YSZ face each other by the (111) crystallographic planes, with a small lattice mismatch of 3% in the direction with sustained translational symmetry. Experimentally, Abe et al. [38] and other researchers [39,40] fabricated and characterized the Ni/YSZ anode cermet, which has the structure with the (111) planes of the Ni part parallel to the (111) planes of YSZ. Using the transmission electron microscopy technique (TEM), the authors have shown a clear absence of amorphous phases at the interface with a (111)//(111) orientation relationship between Ni and YSZ. The Monkforst-Pack [41] kpoint mesh of 2 3 1 is used for the Brillouin zone (BZ) sampling. The atoms in the bottom multilayer are kept fixed for all calculations. Structural optimization of all systems is performed until the atomic forces drop below 0.02 eV Å1. The climbing image nudged elastic band (CI-NEB) [42] method is
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
employed to calculate the transition states and migration barriers. The adsorption energy of a S atom is defined by
637
[45], and hybride functionals (HSE06, PBE0, B3LYP) [46,47], etc. However, these methods demand greater computational effort
Eads ¼ ES þ ENi=YSZOv or Ni=ðYSZþOÞOv ESNi=YSZOv or SNi=ðYSZþOÞOv
(1)
Fig. 1. The Ni/YSZ (a) and Ni/YSZ þ O (b) models. The black circle represents the intrinsic O vacancy.
where ES is the energy of a single S atom simulated in the 8 8 8 Å box; ES-Ni/YSZ-Ov or S-Ni/(YSZþO)-Ov and ENi/YSZ-Ov or Ni/ (YSZþO)-Ov are the total energies of Ni/YSZ-Ov or Ni/(YSZ þ O)-Ov with and without the S adsorbate, respectively. The 3s23p4 of S, 2s22p4 of O, 2s22p3 of N, 2s22p5 of F, 3d84s2 of Ni, 4d25s2 of Zr and 4s24p65s24 d1 of Y are treated as valence electrons in the DFT calculations. The Bader charge [43] analysis scheme is applied to determine the atomic charges and charge transfer. Pure GGA functionals underestimate the band gap and, as a consequence, may affect the other properties. This drawback can be remedied in higher levels of theory, e.g. the weighted density approximation (WDA), screened exchange (sX) [44], GW approximation
Fig. 2. The S adsorption sites at Ni/YSZ þ O and Ni/(YSZ þ O)-Ov.
and are not always feasible for large models and extensive sampling. For this reason, in our work we apply the GGA approximation, which is known to give good energetics, and the qualitative description of the electronic structure. We try to use the international system of units (SI) in this paper. However, we keep some of the popular units used in the micro world and give their equivalent in SI for clarity, e.g. 1 eV ¼ 1.60217733 1019 J; 1 Å ¼ 1010 m; 1 e ¼ 1.60217733 1019 C.
3. Results and discussion 3.1. The adsorption and diffusion of S atom on the Ni/YSZ þ O Fig. 1(a) and (b) show the models of Ni/YSZ and Ni/YSZ þ O (the oxygen-enriched Ni/YSZ), respectively. Oxygen ion is believed to be transferred through the intrinsic vacancies. So the extra oxygen atom is added to the intrinsic vacancies of the stoichiometric YSZ of the Ni/YSZ cermet to model the Ni/YSZ þ O system. There are two intrinsic oxygen vacancies, one is under the Ni part denoted as V1, and the other is not under the Ni part denoted as V2. We have compared the two Ni/YSZ þ O systems, one with an extra oxygen atom in the V1 site and the other with an extra oxygen atom in the V2 site. It is found that the Ni/YSZ þ O with an extra oxygen atom in the V2 site is more stable (lower in energy by 0.52 eV) than that with an extra oxygen in the V1 site. So we select the Ni/YSZ þ O system with an extra oxygen in the V2 site as the model of Ni/ YSZ þ O shown in Fig. 1(b), which is in agreement with the Ni/ YSZ þ O model used in Shishkin and Ziegler's works [27,32]. The charge on the Ni part in Ni/YSZ þ O decreases by 1.04 e as compared with that in the Ni/YSZ based on the Bader charge [43] analysis.
638
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
Table 1 The adsorption properties of sulfur on the systems of (1) Ni/YSZ þ O or (2) Ni/ (YSZ þ O)-Ov. Eads: the adsorption energy of sulfur, the values in the parentheses are the adsorption energy of sulfur on the Ni/YSZ in (1) or Ni/YSZ-Ov system in (2); dS-Ni or Zr: the bond length of the SeNi(Zr); S-Chg(e): the net charge of the adsorbed sulfur. Eads (eV) (1) Ni/YSZ þ O 1 4.44 (4.60) 2 5.08 (5.08) 3 5.01 (4.79) 4 4.97 (4.88) 5 5.22 (5.22) 6 5.32 (5.30) (2) Ni/(YSZ þ O)-Ov V 5.67 (6.21) 1 5.51 (5.46) 2 5.21 (5.26) 3 4.93 (4.75) 4 4.94 (4.89) 5 5.15 (5.26) 6 5.39 (5.35)
dS-Ni(Å)
dS-Zr(Å)
2.10, 2.11, 2.11, 2.12, 2.12, 2.12,
2.18, 2.13, 2.11, 2.11, 2.18, 2.18,
2.20 2.18 2.12 2.16 2.30 2.30
2.14, 2.11, 2.11, 2.13, 2.11, 2.12, 2.14,
2.15 2.11, 2.15, 2.13, 2.11, 2.22, 2.14,
2.24 2.18 2.13 2.13 2.36 2.18
S-Chg(e) 0.51 0.57 0.48 0.53 0.64 0.60
2.89, 2.90
1.00 0.56 0.57 0.49 0.53 0.64 0.60
The optimized adsorption sites at the Ni/YSZ þ O for S atom are shown in Fig. 2(a). The adsorption properties are summarized in Table 1. From Table 1, it is found that the adsorption of S on the Ni/ YSZ þ O has the same trend as that on the Ni/YSZ system, i.e., the adsorbed S does not favor to be adsorbed at the site near the interface Oxygen due to the repulsion between the adsorbed S and the interface O2. The values of the adsorption energies on the Ni/ YSZ þ O and Ni/YSZ systems are very close to each other. The calculated diffusion paths of S atom at the Ni/YSZ þ O and Ni/YSZ are shown in Fig. 3(a) and (b), respectively. From Fig. 3(a), it is found that the favorable path for S diffusion at the Ni/YSZ þ O from site 1 to 6 is 1 / 2/5 / 6 and the favorable reverse S diffusion path is 6 / 5/4 / 3/2 / 1. It is also found that, at the Ni/YSZ þ O, S can diffuse easily away from the interface oxygen to the top Ni layer sites with the barriers no larger than 0.24 eV, while the reverse
barrier is as large as 0.65 eV. Similar diffusion properties can be found for the diffusion of S atom at the Ni/YSZ as shown in Fig. 3(b).
3.2. The adsorption and diffusion of S atom on the Ni/(YSZ þ O)-Ov The hydrogen oxidation reaction occurred on the Ni/YSZ þ O would induce the formation of O vacancy at the interface. The Ni/ YSZ þ O with an interface O vacancy is denoted as Ni/(YSZ þ O)-Ov. The optimized adsorption sites for the S atom at the Ni/(YSZ þ O)Ov are shown in Fig. 2(b). The adsorption properties are summarized in Table 1. From Table 1, it is found that the adsorption of the S atom at the Ni/YSZ þ O has the same trend as that at the Ni/YSZ-Ov system, i.e., the adsorbed S favors to be adsorbed at the interface Oxygen vacancy site due to the disappearance of the repulsion between the adsorbed S and the interface O2. However, it is found that the adsorption of S at the interface O vacancy site of the Ni/(YSZ þ O)-Ov is lower in energy by 0.54 eV than that of Ni/YSZOv. Based on the Bader charge [43] analysis shown in Fig. 4(a) and (b), it is found that the adsorbed S has developed static attraction with the neighboring Zr4þ cation and the charge of adsorbed S atom at the vacancy site of Ni/(YSZ þ O)-Ov decreased by 0.27 e as compared with that at the vacancy site of Ni/YSZ-Ov. The weakening for the adsorption of S might be attributed to the decrease of charge on the adsorbed S, which may cause the weakening of the static attraction between S and the neighboring Zr4þ. To this end, we also have substituted the extra O with an N or a F atom. The calculated results are shown in Fig. 4 (c) and (d). As shown in Fig. 4(e), a linear correlation clearly exists between the adsorption energy and the charge of the S atom, i.e., the lower the charge of the adsorbed S, the lower the static attraction of S with the neighboring Zr4þ cations, and the lower the adsorption energy of S at the interface O vacancy site. The calculated the diffusion paths of S atom at the Ni/(YSZ þ O)Ov and Ni/YSZ-Ov are shown in Fig. 5(a) and (b), respectively. From Fig. 5(a), it is found that the favorable path for S diffusion from site 6
Fig. 3. The MEPs for the different diffusion paths of S on the Ni/YSZ þ O and Ni/YSZ.
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
639
Fig. 4. The S adsorption at the interface O vacancy site of (a) Ni/YSZ-Ov, (b) Ni/(YSZ þ O)-Ov, (c) Ni/(YSZ þ F)-Ov and (d) Ni/(YSZ þ N)-Ov, and (e) the plot showing the correlation between the charge of adsorbed S atom and the adsorption energy of S at the interface O vacancy site.
Fig. 5. The MEPs for the different diffusion paths of S on the Ni/(YSZ þ O)-Ov and Ni/YSZ-Ov.
to V at the Ni/(YSZ þ O)-Ov is 6 / 5/4 / 3 / 2 / V and the favorable reverse S diffusion path is V/1 / 2/3 / 4/5 / 6. Therefore, the S at the Ni/(YSZ þ O)-Ov prefers to diffuse back to the interface oxygen vacancy site from the top Ni layer sites with the barriers no larger than 0.47 eV, with the reverse barriers no larger than 0.47 eV. From Fig. 5(b), it is found that the favorable path for S diffusion from site 6 to V at the Ni/YSZ-Ov is 6 / 5/4 / 3 / 2 / V and the favorable reverse S diffusion path
is V / 1 / 2/5 / 6. It is also found that the S at the Ni/YSZ-Ov can diffuse back to the interface oxygen vacancy site from the top Ni layer sites with the barriers no larger than 0.37 eV, while the reverse barrier is as large as 0.77 eV. In order to elucidate the poisoning effect of H2S to the interface O vacancy of Ni/(YSZ þ O)-Ov, we have studied the adsorption of H2S on Ni/(YSZ þ O)-Ov and found that the most stable adsorption site of H2S is located at the Ni part as shown in Fig. 6. We have also
640
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
Fig. 6. The MEPs for the dissociation path of H2S on the Ni/(YSZ þ O)-Ov.
calculated the dissociation of H2S at the most stable adsorption site on Ni/(YSZ þ O)-Ov shown below. It is found that the adsorbed H2S dissociates easily to HS and H with a barrier of 0.35 eV. The HS diffuses easily to interface O vacancy site (with a barrier of 0.01 eV) and dissociates easily to H and S (with a barrier of 0.29 eV), while the reverse barrier for the formation of HS from H and S, HS diffusion out of O vacancy site, as well as the formation of H2S are very large (1.03, 1.18, 0.68 eV, respectively), indicating the sulfur poisoning at the interface O vacancy of Ni/(YSZ þ O)-Ov. Compared with S diffusion properties at the Ni/YSZ-Ov, the diffusion of S out of the interface O vacancy site at the Ni/(YSZ þ O)Ov is much easier, with lower barriers from site 1 to site 2 (0.23 vs 0.77 eV) or from site 1 to site 3 (0.63 vs 1.20 eV). So we propose that compared with the Ni/YSZ, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface oxygen vacancy site. 4. Conclusion The sulfur poisoning properties of the Ni/YSZ þ O with or without interface O vacancy are studied using the first-principles method based on density functional theory, focusing on the effects of the extra O atom at the subsurface vacancy of Ni/YSZ. It is found that S at the Ni/YSZ þ O can diffuse easily away from the interface oxygen to the top Ni layer sites. With the formation of O vacancy at the Ni/YSZ þ O interface, the adsorbed S diffuses back to the Ni/YSZ interface O vacancy. Compared with Ni/YSZ-Ov, the Ni/ (YSZ þ O)-Ov can effectively not only weaken the S adsorption at the interface O vacancy site, but also improve the diffusion of S out
of the interface O vacancy. Therefore, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface O vacancy site as compared with the Ni/YSZ. Oxygen ion is believed to be transferred through the intrinsic vacancies. The DFT results on the sulfur poisoning properties of the Ni/YSZ þ O would be helpful for understanding the properties of Ni/YSZ cermet at real working condition, in which extra oxygen atoms exist in some of the intrinsic vacancies. Acknowledgments This work was supported by the National Natural Science Foundation of China (Grant Nos. 11174070 and 11474086). References [1] R.M. Ormerod, Chem. Soc. Rev. 32 (2003) 17e28. [2] M.C. Williams, J.P. Strakey, W.A. Surdoval, L.C. Wilson, Solid State Ionics 177 (2006) 2039e2044. [3] Z. Cheng, J.H. Wang, Y.M. Choi, L. Yang, M. Lin, M. Liu, Energy Environ. Sci. 4 (2011) 4380e4409. [4] Y. Matsuzaki, I. Yasuda, Solid State Ionics 132 (2000) 261e269. [5] S. Fang, L. Bi, X. Wu, H. Gao, C. Chen, W. Liu, J. Power Sources 183 (2008) 126e132. [6] S. Zha, Z. Cheng, M. Liu, J. Electrochem Soc. 154 (2007) B201eB206. [7] L. Yang, Z. Cheng, M. Liu, L. Wilson, Energy Environ. Sci. 3 (2010) 1804e1809. [8] K. Sasaki, K. Susuki, A. Iyoshi, M. Uchimura, N. Imamura, H. Kusaba, Y. Teraoka, H. Fuchino, K. Tsujimoto, Y. Uchida, J. Electrochem Soc. 153 (2006) A2023eA2029. [9] Z. Cheng, S. Zha, M. Liu, J. Power Sources 172 (2007) 688e693. [10] Z. Cheng, M. Liu, Solid State Ionics 178 (2007) 925e935.
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641 [11] J.-H. Wang, M. Liu, Electrochem Commun. 9 (2007) 2212e2217. [12] E. Hardegree, P. Ho, J. White, Surf. Sci. 165 (1986) 488e506. [13] L.L. Zheng, X. Wang, L. Zhang, J.-Y. Wang, S.P. Jiang, Int. J. Hydrogen Energy 37 (2012) 10299e10310. [14] Z. Zeng, M.E. Bjorketun, S. Ebbesen, M.B. Mogensen, J. Rossmeisl, Phys. Chem. Chem. Phys. 15 (2013) 6769e6772. [15] O.I. Malyi, P. Wu, V.V. Kulish, K. Bai, Z. Chen, Solid State Ionics 212 (2012) 117e122. [16] Y. Zhang, Z. Lu, Z. Yang, T. Woo, J. Power Sources 237 (2013) 128e131. [17] H. He, R.J. Gorte, J.M. Vohs, Electrochem Solid State Lett. 8 (2005) A279eA280. [18] R. Mukundan, E.L. Brosha, F.H. Garzon, Electrochem Solid State Lett. 7 (2004) A5eA7. [19] X.C. Lu, J.H. Zhu, Z. Yang, G. Xia, J.W. Stevenson, J. Power Sources 192 (2009) 381e384. [20] X. Lu, J. Zhu, J. Electrochem Soc. 155 (2008) B1053eB1057. [21] Z.-R. Xu, J.-L. Luo, K.T. Chuang, J. Power Sources 188 (2009) 458e462. [22] S. Choi, J. Wang, Z. Cheng, M. Liu, J. Electrochem Soc. 155 (2008) B449eB454. [23] L. Aguilar, S. Zha, Z. Cheng, J. Winnick, M. Liu, J. Power Sources 135 (2004) 17e24. [24] T. Pramananda Perumal, V. Sridhar, K.P.N. Murthy, K.S. Easwarakumar, S. Ramasamy, Phys. A Stat. Mech. its Appl. 309 (2002) 35e44. [25] V. Yurkiv, A. Gorski, W.G. Bessler, H.-R. Volpp, Chem. Phys. Lett. 543 (2012) 213e217. e, H.-R. Volpp, W.G. Bessler, ECS Trans. [26] V. Yurkiv, A. Utz, A. Weber, E. Ivers-Tiffe 35 (2011) 1743e1751. [27] M. Shishkin, T. Ziegler, J. Phys. Chem. C 114 (2010) 11209e11214.
[28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47]
641
G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865e3868. €chl, Phys. Rev. B 50 (1994) 17953. P.E. Blo G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. M. Shishkin, T. Ziegler, J. Phys. Chem. C 113 (2009) 21667e21678. Y. Zhang, Z. Fu, S. Dong, Z. Yang, Phys. Chem. Chem. Phys. 16 (2014) 1033e1040. Y. Zhang, Z. Fu, M. Wang, Z. Yang, J. Power Sources 261 (2014) 136e140. Y. Zhang, Z. Yang, J. Power Sources 271 (2014) 516e521. Y. Zhang, Z. Yang, M. Wang, J. Power Sources 279 (2015) 759e765. Y. Zhang, Z. Yang, J. Power Sources 279 (2015) 224e230. H. Abe, K. Murata, T. Fukui, W.J. Moon, K. Kaneko, M. Naito, Thin Solid Films 496 (2006) 49e52. ~ oz, S. Gallego, J.I. Beltr M.C. Mun an, J. Cerd a, Surf. Sci. Rep. 61 (2006) 303e344. T. Sasaki, K. Matsunaga, H. Ohta, H. Hosono, T. Yamamoto, Y. Ikuhara, Mater. Trans. 45 (2004) 2137e2143. H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188e5192. nsson, J. Chem. Phys. 113 (2000) 9901. G. Henkelman, B.P. Uberuaga, H. Jo nsson, Comput Mate Sci. 36 (2006) G. Henkelman, A. Arnaldsson, H. Jo 354e360. K. Xiong, J. Robertson, M. Gibson, S. Clark, Appl. Phys. Lett. 87 (2005), 183505183505-183503. lik, E.K. Chang, S.G. Louie, Phys. Rev. B 57 (1998) 7027. B. Kra A. Alkauskas, P. Broqvist, F. Devynck, A. Pasquarello, Phys. Rev. Lett. 101 (2008) 106802. P. Broqvist, A. Pasquarello, Appl. Phys. Lett. 89 (2006) 262904.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
The sulfur poisoning of the nickel/oxygen-enriched yttria-stabilized zirconia Yanxing Zhang a, Zhengyang Wan a, Zongxian Yang a, b, * a b
College of Physics and Electronic Engineering, Henan Normal University, Xinxiang, Henan 453007, People's Republic of China Collaborative Innovation Center of Nano Functional Materials and Applications, Henan Province, People's Republic of China
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
The adsorption and diffusion properties of sulfur on the Ni/YSZ þ O are studied. The adsorbed sulfur doesn't favor to be located at the Ni/YSZ þ O interface. The extra O in YSZ weakens the S adsorption at the vacancy site of Ni/ YSZ-Ov. The extra O in YSZ improves the diffusion of S out of the vacancy of Ni/YSZ-Ov. The Ni/YSZ þ O can help to alleviate the sulfur poisoning as compared with the Ni/YSZ.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 February 2015 Received in revised form 28 April 2015 Accepted 13 May 2015 Available online xxx
The sulfur poisoning properties of the nickel/oxygen-enriched yttria-stabilized zirconia (denoted as Ni/ YSZ þ O) with or without interface O vacancy are studied using the first-principles method based on density functional theory. The effects of the extra O atom at the subsurface vacancy of Ni/YSZ are focused. It is found that S at the Ni/YSZ þ O can diffuse easily away from the interface oxygen to the top Ni layer sites. With the formation of O vacancy at the Ni/YSZ þ O interface (denoted as Ni/(YSZ þ O)-Ov), the adsorbed S prefers to diffuse back to the Ni/YSZ interface O vacancy. Compared with Ni/YSZ-Ov, the Ni/ (YSZ þ O)-Ov can effectively not only weaken the S adsorption at the interface O vacancy site, but also improve the diffusion of S out of the interface O vacancy. Therefore, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface O vacancy site as compared with the Ni/YSZ. © 2015 Elsevier B.V. All rights reserved.
Keywords: Solid oxide fuel cell Sulfur poisoning Nickel/yttrium-stabilized zirconia anode Oxygen vacancy
1. Introduction Solid oxide fuel cells (SOFCs) are expected to be a crucial technology in the future power generation [1,2]. SOFCs offer many
* Corresponding author. College of Physics and Electronic Engineering, Henan Normal University, Xinxiang, Henan 453007, People's Republic of China. E-mail address: [email protected] (Z. Yang). http://dx.doi.org/10.1016/j.jpowsour.2015.05.044 0378-7753/© 2015 Elsevier B.V. All rights reserved.
desirable advantages compared to other types of fuel cells and conversion devices due to their use of solid electrolytes, lack of moving parts, ability to circumvent precious metal use, high efficiency, low pollution, and fuel flexibility. The conventional anode for a SOFC, denoted as Ni/YSZ, is consisted of nickel and yttria-stabilized zirconia (YSZ). However, a major issue in the long-term stability and activity of the anode catalyst is its poor resistance toward poisonous compounds
636
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
presented in the feed stream of fuels. Trace amounts of H2S presented in biomass generated syngas streams are enough to deactivate the catalyst [3,4], which is called sulfur poisoning. Many previous experimental studies indicated that the S poisoning behavior is characterized by two stages, fast initial degradation and long term degradation [4e10]. Liu et al. [7,11] showed that S poisoning observed in the low concentration of H2S at elevated temperatures is originated from the dissociation of sulfur-containing species and the adsorption of atomic sulfur on the anode surface. The adsorbed H2S on Ni surfaces has been shown to dissociate above 300 K [12], with only S remaining on the surface. These studies clearly suggest that the elemental sulfur strongly adsorbs on the Ni surfaces with a small activation barrier (Ea) of H2S dissociation and a large exothermic enthalpy (DH). The strongly adsorbed S species block the active sites on the anode surface and thus increase the resistance to electrochemical oxidation of the fuel. The calculations [3] also suggest that the adsorbed sulfur species exist primarily in the form of atomic sulfur instead of molecular species, e.g., H2S. In fact, the faster or sluggish kinetics is related to the barriers of reactions. The very small Ea means that the barrier for H2S dissociation is small, but the large exothermic DH indicates that the barrier of the reverse reaction of the H2S dissociation is large. The very small Ea and large exothermic DH further imply fast kinetics for sulfur adsorption (as a result of H2S dissociation) and sluggish kinetics for sulfur removal, which is consistent with the experimental observation of the instant drop in performance upon exposure to H2S and a very slow recovery in performance after clean hydrogen is switched back. However, experimental results indicated that sulfur tolerance was in fact improved by using Ni/Sc2O3 [8] or Ni/Gd2O3-doped ZrO2 [13] anodes, which suggest at least, that sulfur tolerance depends strongly on anode and electrolyte materials besides nickel itself. Zeng et al. [14] have studied the mechanisms governing the sulfur poisoning of the triple-phase boundary (TPB) of Ni/XSZ (X2O3 stabilized zirconia) anodes using density functional theory. The calculated sulfur adsorption energies reveal a clear correlation between the size of the cation dopant X3þ and the sulfur tolerance of the Ni/XSZ anode. Malyi et al. [15] found that S addition to zirconia, either by doping or through gas diffusion, increases both the formation energy and migration barrier of the oxygen vacancies. Since the fuel oxidation is believed to take place at the TPB made of Ni, YSZ and fuel gas, oxygen vacancies may be created at the Ni/YSZ interface. As known, the anode of SOFC is the Ni/YSZ composite, instead of the Ni itself. The ideal Ni (111) only represents the anode region beyond the TPB. Therefore, people should give special attention to the Ni/YSZ system and the effects of the interface O vacancy. In our recent work [16], we studied the sulfur poisoning at the TPB region of the Ni/YSZ. We found that the adsorbed sulfur does not favor to be located at the stoichiometric Ni/YSZ interface. With O vacancy at the Ni/YSZ interface, the adsorbed S diffuses to the Ni/YSZ interface and is oxidized to S2 and trapped at the oxygen vacancy. The trapped S is very difficult to be removed by the fuel (e.g., H2) and therefore blocks the pathway for the O ion transfer. As a result, the resistance for O ion transfer would increase and the SOFC performance would drop. Trace amounts of S would block the O vacancy sites at the interface and induce the instant and significant drop in performance of SOFC. To alleviate the sulfur adsorption at the O vacancy site and/ or enhance the diffusion of S out of the interface oxygen vacancy site would help to enhance the resistance to S poisoning at the oxygen vacancy site. To improve the sulfur tolerance of SOFCs, alternative anode
compositions have been proposed. Copper/ceria/zirconia [17] anodes were reported to be stable in fuel gases containing up to 450 ppm H2S. Other anode compositions showing good sulfur tolerance include a lanthanum doped strontium titanate [18], Pdimpregnated titanate/cerate composition [19], lanthanum molybdate [20], gold/molybdenum disulfide [21], Ni/YSZ modified with niobia [22], and lanthanum vanadium oxide [23]. For considerations including cost, processability, and stability, minor modifications to the widely used Ni/YSZ anode may be preferred to the more exotic compositions and forms. For maintaining charge neutrality of the YSZ system, the oxygen vacancies are created during the substitution of Zr4þ by Y3þ in the YSZ lattice. Under practical operating conditions of SOFCs, these O vacancies are mobile to give rise to oxygen ionic conductivity via vacancy diffusion mechanism, and can be filled by O2 migrating through the crystal lattice of YSZ, to form an O-enriched YSZ surface, YSZ þ O (111). The O2 ions supplied to the intrinsic vacancy of YSZ originate from the electrochemical reaction of O2 at an electrode, which depends on the actual SOFC fuel gas operation and polarization conditions in the anode polarization measurement [24e26]. It is found that the favored hydrogen oxidation reaction process on Ni/YSZ is water formation on an O atom at the interface and this is also the most facile hydrogen oxidation reaction process on Ni/(YSZ þ O) [27]. In this work, we study the sulfur poisoning at the TPB region of the Ni/(YSZ þ O) -Ov, focusing on the effects of the extra O atom at the subsurface vacancy of Ni/YSZ-Ov on the adsorption and the diffusion of S at the interface O vacancy site. 2. Model and computation method All calculations presented in this work are performed employing the periodic density functional theory (DFT) method implemented in the Vienna Ab-Initio Simulation Package (VASP) [28]. The exchangeecorrelation interactions are treated with the PerdeweBurkeeErnzerhof (PBE) functional [29]. Spin-polarized calculations are applied throughout. The electroneion interactions are treated using the projector augmented wave (PAW) method [30,31]. The wave functions are expanded in plane waves with a cut off energy of 408 eV. The model of the Ni/YSZ cermet with the horizontal dimensions of 12.56 7.25 Å as that used in the Shishkin and Ziegler's work [32] is adopted as the substrate. The Ni/YSZ model with two intrinsic vacancies used in this work includes two building blocks of the 9% mol YSZ, which has 10 ZrO2 and 1 Y2O3 and can be considered as an elementary building unit of YSZ with 9% mol concentration of yttria shown in Fig. 1(a). A vacuum layer of 15 Å is used to separate the periodic images in the direction perpendicular to the surface. The model has been successfully used in our previous researches [16,33e37]. In the proposed structure, both Ni and YSZ face each other by the (111) crystallographic planes, with a small lattice mismatch of 3% in the direction with sustained translational symmetry. Experimentally, Abe et al. [38] and other researchers [39,40] fabricated and characterized the Ni/YSZ anode cermet, which has the structure with the (111) planes of the Ni part parallel to the (111) planes of YSZ. Using the transmission electron microscopy technique (TEM), the authors have shown a clear absence of amorphous phases at the interface with a (111)//(111) orientation relationship between Ni and YSZ. The Monkforst-Pack [41] kpoint mesh of 2 3 1 is used for the Brillouin zone (BZ) sampling. The atoms in the bottom multilayer are kept fixed for all calculations. Structural optimization of all systems is performed until the atomic forces drop below 0.02 eV Å1. The climbing image nudged elastic band (CI-NEB) [42] method is
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
employed to calculate the transition states and migration barriers. The adsorption energy of a S atom is defined by
637
[45], and hybride functionals (HSE06, PBE0, B3LYP) [46,47], etc. However, these methods demand greater computational effort
Eads ¼ ES þ ENi=YSZOv or Ni=ðYSZþOÞOv ESNi=YSZOv or SNi=ðYSZþOÞOv
(1)
Fig. 1. The Ni/YSZ (a) and Ni/YSZ þ O (b) models. The black circle represents the intrinsic O vacancy.
where ES is the energy of a single S atom simulated in the 8 8 8 Å box; ES-Ni/YSZ-Ov or S-Ni/(YSZþO)-Ov and ENi/YSZ-Ov or Ni/ (YSZþO)-Ov are the total energies of Ni/YSZ-Ov or Ni/(YSZ þ O)-Ov with and without the S adsorbate, respectively. The 3s23p4 of S, 2s22p4 of O, 2s22p3 of N, 2s22p5 of F, 3d84s2 of Ni, 4d25s2 of Zr and 4s24p65s24 d1 of Y are treated as valence electrons in the DFT calculations. The Bader charge [43] analysis scheme is applied to determine the atomic charges and charge transfer. Pure GGA functionals underestimate the band gap and, as a consequence, may affect the other properties. This drawback can be remedied in higher levels of theory, e.g. the weighted density approximation (WDA), screened exchange (sX) [44], GW approximation
Fig. 2. The S adsorption sites at Ni/YSZ þ O and Ni/(YSZ þ O)-Ov.
and are not always feasible for large models and extensive sampling. For this reason, in our work we apply the GGA approximation, which is known to give good energetics, and the qualitative description of the electronic structure. We try to use the international system of units (SI) in this paper. However, we keep some of the popular units used in the micro world and give their equivalent in SI for clarity, e.g. 1 eV ¼ 1.60217733 1019 J; 1 Å ¼ 1010 m; 1 e ¼ 1.60217733 1019 C.
3. Results and discussion 3.1. The adsorption and diffusion of S atom on the Ni/YSZ þ O Fig. 1(a) and (b) show the models of Ni/YSZ and Ni/YSZ þ O (the oxygen-enriched Ni/YSZ), respectively. Oxygen ion is believed to be transferred through the intrinsic vacancies. So the extra oxygen atom is added to the intrinsic vacancies of the stoichiometric YSZ of the Ni/YSZ cermet to model the Ni/YSZ þ O system. There are two intrinsic oxygen vacancies, one is under the Ni part denoted as V1, and the other is not under the Ni part denoted as V2. We have compared the two Ni/YSZ þ O systems, one with an extra oxygen atom in the V1 site and the other with an extra oxygen atom in the V2 site. It is found that the Ni/YSZ þ O with an extra oxygen atom in the V2 site is more stable (lower in energy by 0.52 eV) than that with an extra oxygen in the V1 site. So we select the Ni/YSZ þ O system with an extra oxygen in the V2 site as the model of Ni/ YSZ þ O shown in Fig. 1(b), which is in agreement with the Ni/ YSZ þ O model used in Shishkin and Ziegler's works [27,32]. The charge on the Ni part in Ni/YSZ þ O decreases by 1.04 e as compared with that in the Ni/YSZ based on the Bader charge [43] analysis.
638
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
Table 1 The adsorption properties of sulfur on the systems of (1) Ni/YSZ þ O or (2) Ni/ (YSZ þ O)-Ov. Eads: the adsorption energy of sulfur, the values in the parentheses are the adsorption energy of sulfur on the Ni/YSZ in (1) or Ni/YSZ-Ov system in (2); dS-Ni or Zr: the bond length of the SeNi(Zr); S-Chg(e): the net charge of the adsorbed sulfur. Eads (eV) (1) Ni/YSZ þ O 1 4.44 (4.60) 2 5.08 (5.08) 3 5.01 (4.79) 4 4.97 (4.88) 5 5.22 (5.22) 6 5.32 (5.30) (2) Ni/(YSZ þ O)-Ov V 5.67 (6.21) 1 5.51 (5.46) 2 5.21 (5.26) 3 4.93 (4.75) 4 4.94 (4.89) 5 5.15 (5.26) 6 5.39 (5.35)
dS-Ni(Å)
dS-Zr(Å)
2.10, 2.11, 2.11, 2.12, 2.12, 2.12,
2.18, 2.13, 2.11, 2.11, 2.18, 2.18,
2.20 2.18 2.12 2.16 2.30 2.30
2.14, 2.11, 2.11, 2.13, 2.11, 2.12, 2.14,
2.15 2.11, 2.15, 2.13, 2.11, 2.22, 2.14,
2.24 2.18 2.13 2.13 2.36 2.18
S-Chg(e) 0.51 0.57 0.48 0.53 0.64 0.60
2.89, 2.90
1.00 0.56 0.57 0.49 0.53 0.64 0.60
The optimized adsorption sites at the Ni/YSZ þ O for S atom are shown in Fig. 2(a). The adsorption properties are summarized in Table 1. From Table 1, it is found that the adsorption of S on the Ni/ YSZ þ O has the same trend as that on the Ni/YSZ system, i.e., the adsorbed S does not favor to be adsorbed at the site near the interface Oxygen due to the repulsion between the adsorbed S and the interface O2. The values of the adsorption energies on the Ni/ YSZ þ O and Ni/YSZ systems are very close to each other. The calculated diffusion paths of S atom at the Ni/YSZ þ O and Ni/YSZ are shown in Fig. 3(a) and (b), respectively. From Fig. 3(a), it is found that the favorable path for S diffusion at the Ni/YSZ þ O from site 1 to 6 is 1 / 2/5 / 6 and the favorable reverse S diffusion path is 6 / 5/4 / 3/2 / 1. It is also found that, at the Ni/YSZ þ O, S can diffuse easily away from the interface oxygen to the top Ni layer sites with the barriers no larger than 0.24 eV, while the reverse
barrier is as large as 0.65 eV. Similar diffusion properties can be found for the diffusion of S atom at the Ni/YSZ as shown in Fig. 3(b).
3.2. The adsorption and diffusion of S atom on the Ni/(YSZ þ O)-Ov The hydrogen oxidation reaction occurred on the Ni/YSZ þ O would induce the formation of O vacancy at the interface. The Ni/ YSZ þ O with an interface O vacancy is denoted as Ni/(YSZ þ O)-Ov. The optimized adsorption sites for the S atom at the Ni/(YSZ þ O)Ov are shown in Fig. 2(b). The adsorption properties are summarized in Table 1. From Table 1, it is found that the adsorption of the S atom at the Ni/YSZ þ O has the same trend as that at the Ni/YSZ-Ov system, i.e., the adsorbed S favors to be adsorbed at the interface Oxygen vacancy site due to the disappearance of the repulsion between the adsorbed S and the interface O2. However, it is found that the adsorption of S at the interface O vacancy site of the Ni/(YSZ þ O)-Ov is lower in energy by 0.54 eV than that of Ni/YSZOv. Based on the Bader charge [43] analysis shown in Fig. 4(a) and (b), it is found that the adsorbed S has developed static attraction with the neighboring Zr4þ cation and the charge of adsorbed S atom at the vacancy site of Ni/(YSZ þ O)-Ov decreased by 0.27 e as compared with that at the vacancy site of Ni/YSZ-Ov. The weakening for the adsorption of S might be attributed to the decrease of charge on the adsorbed S, which may cause the weakening of the static attraction between S and the neighboring Zr4þ. To this end, we also have substituted the extra O with an N or a F atom. The calculated results are shown in Fig. 4 (c) and (d). As shown in Fig. 4(e), a linear correlation clearly exists between the adsorption energy and the charge of the S atom, i.e., the lower the charge of the adsorbed S, the lower the static attraction of S with the neighboring Zr4þ cations, and the lower the adsorption energy of S at the interface O vacancy site. The calculated the diffusion paths of S atom at the Ni/(YSZ þ O)Ov and Ni/YSZ-Ov are shown in Fig. 5(a) and (b), respectively. From Fig. 5(a), it is found that the favorable path for S diffusion from site 6
Fig. 3. The MEPs for the different diffusion paths of S on the Ni/YSZ þ O and Ni/YSZ.
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
639
Fig. 4. The S adsorption at the interface O vacancy site of (a) Ni/YSZ-Ov, (b) Ni/(YSZ þ O)-Ov, (c) Ni/(YSZ þ F)-Ov and (d) Ni/(YSZ þ N)-Ov, and (e) the plot showing the correlation between the charge of adsorbed S atom and the adsorption energy of S at the interface O vacancy site.
Fig. 5. The MEPs for the different diffusion paths of S on the Ni/(YSZ þ O)-Ov and Ni/YSZ-Ov.
to V at the Ni/(YSZ þ O)-Ov is 6 / 5/4 / 3 / 2 / V and the favorable reverse S diffusion path is V/1 / 2/3 / 4/5 / 6. Therefore, the S at the Ni/(YSZ þ O)-Ov prefers to diffuse back to the interface oxygen vacancy site from the top Ni layer sites with the barriers no larger than 0.47 eV, with the reverse barriers no larger than 0.47 eV. From Fig. 5(b), it is found that the favorable path for S diffusion from site 6 to V at the Ni/YSZ-Ov is 6 / 5/4 / 3 / 2 / V and the favorable reverse S diffusion path
is V / 1 / 2/5 / 6. It is also found that the S at the Ni/YSZ-Ov can diffuse back to the interface oxygen vacancy site from the top Ni layer sites with the barriers no larger than 0.37 eV, while the reverse barrier is as large as 0.77 eV. In order to elucidate the poisoning effect of H2S to the interface O vacancy of Ni/(YSZ þ O)-Ov, we have studied the adsorption of H2S on Ni/(YSZ þ O)-Ov and found that the most stable adsorption site of H2S is located at the Ni part as shown in Fig. 6. We have also
640
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641
Fig. 6. The MEPs for the dissociation path of H2S on the Ni/(YSZ þ O)-Ov.
calculated the dissociation of H2S at the most stable adsorption site on Ni/(YSZ þ O)-Ov shown below. It is found that the adsorbed H2S dissociates easily to HS and H with a barrier of 0.35 eV. The HS diffuses easily to interface O vacancy site (with a barrier of 0.01 eV) and dissociates easily to H and S (with a barrier of 0.29 eV), while the reverse barrier for the formation of HS from H and S, HS diffusion out of O vacancy site, as well as the formation of H2S are very large (1.03, 1.18, 0.68 eV, respectively), indicating the sulfur poisoning at the interface O vacancy of Ni/(YSZ þ O)-Ov. Compared with S diffusion properties at the Ni/YSZ-Ov, the diffusion of S out of the interface O vacancy site at the Ni/(YSZ þ O)Ov is much easier, with lower barriers from site 1 to site 2 (0.23 vs 0.77 eV) or from site 1 to site 3 (0.63 vs 1.20 eV). So we propose that compared with the Ni/YSZ, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface oxygen vacancy site. 4. Conclusion The sulfur poisoning properties of the Ni/YSZ þ O with or without interface O vacancy are studied using the first-principles method based on density functional theory, focusing on the effects of the extra O atom at the subsurface vacancy of Ni/YSZ. It is found that S at the Ni/YSZ þ O can diffuse easily away from the interface oxygen to the top Ni layer sites. With the formation of O vacancy at the Ni/YSZ þ O interface, the adsorbed S diffuses back to the Ni/YSZ interface O vacancy. Compared with Ni/YSZ-Ov, the Ni/ (YSZ þ O)-Ov can effectively not only weaken the S adsorption at the interface O vacancy site, but also improve the diffusion of S out
of the interface O vacancy. Therefore, the Ni/YSZ þ O can help to alleviate the sulfur poisoning at the interface O vacancy site as compared with the Ni/YSZ. Oxygen ion is believed to be transferred through the intrinsic vacancies. The DFT results on the sulfur poisoning properties of the Ni/YSZ þ O would be helpful for understanding the properties of Ni/YSZ cermet at real working condition, in which extra oxygen atoms exist in some of the intrinsic vacancies. Acknowledgments This work was supported by the National Natural Science Foundation of China (Grant Nos. 11174070 and 11474086). References [1] R.M. Ormerod, Chem. Soc. Rev. 32 (2003) 17e28. [2] M.C. Williams, J.P. Strakey, W.A. Surdoval, L.C. Wilson, Solid State Ionics 177 (2006) 2039e2044. [3] Z. Cheng, J.H. Wang, Y.M. Choi, L. Yang, M. Lin, M. Liu, Energy Environ. Sci. 4 (2011) 4380e4409. [4] Y. Matsuzaki, I. Yasuda, Solid State Ionics 132 (2000) 261e269. [5] S. Fang, L. Bi, X. Wu, H. Gao, C. Chen, W. Liu, J. Power Sources 183 (2008) 126e132. [6] S. Zha, Z. Cheng, M. Liu, J. Electrochem Soc. 154 (2007) B201eB206. [7] L. Yang, Z. Cheng, M. Liu, L. Wilson, Energy Environ. Sci. 3 (2010) 1804e1809. [8] K. Sasaki, K. Susuki, A. Iyoshi, M. Uchimura, N. Imamura, H. Kusaba, Y. Teraoka, H. Fuchino, K. Tsujimoto, Y. Uchida, J. Electrochem Soc. 153 (2006) A2023eA2029. [9] Z. Cheng, S. Zha, M. Liu, J. Power Sources 172 (2007) 688e693. [10] Z. Cheng, M. Liu, Solid State Ionics 178 (2007) 925e935.
Y. Zhang et al. / Journal of Power Sources 293 (2015) 635e641 [11] J.-H. Wang, M. Liu, Electrochem Commun. 9 (2007) 2212e2217. [12] E. Hardegree, P. Ho, J. White, Surf. Sci. 165 (1986) 488e506. [13] L.L. Zheng, X. Wang, L. Zhang, J.-Y. Wang, S.P. Jiang, Int. J. Hydrogen Energy 37 (2012) 10299e10310. [14] Z. Zeng, M.E. Bjorketun, S. Ebbesen, M.B. Mogensen, J. Rossmeisl, Phys. Chem. Chem. Phys. 15 (2013) 6769e6772. [15] O.I. Malyi, P. Wu, V.V. Kulish, K. Bai, Z. Chen, Solid State Ionics 212 (2012) 117e122. [16] Y. Zhang, Z. Lu, Z. Yang, T. Woo, J. Power Sources 237 (2013) 128e131. [17] H. He, R.J. Gorte, J.M. Vohs, Electrochem Solid State Lett. 8 (2005) A279eA280. [18] R. Mukundan, E.L. Brosha, F.H. Garzon, Electrochem Solid State Lett. 7 (2004) A5eA7. [19] X.C. Lu, J.H. Zhu, Z. Yang, G. Xia, J.W. Stevenson, J. Power Sources 192 (2009) 381e384. [20] X. Lu, J. Zhu, J. Electrochem Soc. 155 (2008) B1053eB1057. [21] Z.-R. Xu, J.-L. Luo, K.T. Chuang, J. Power Sources 188 (2009) 458e462. [22] S. Choi, J. Wang, Z. Cheng, M. Liu, J. Electrochem Soc. 155 (2008) B449eB454. [23] L. Aguilar, S. Zha, Z. Cheng, J. Winnick, M. Liu, J. Power Sources 135 (2004) 17e24. [24] T. Pramananda Perumal, V. Sridhar, K.P.N. Murthy, K.S. Easwarakumar, S. Ramasamy, Phys. A Stat. Mech. its Appl. 309 (2002) 35e44. [25] V. Yurkiv, A. Gorski, W.G. Bessler, H.-R. Volpp, Chem. Phys. Lett. 543 (2012) 213e217. e, H.-R. Volpp, W.G. Bessler, ECS Trans. [26] V. Yurkiv, A. Utz, A. Weber, E. Ivers-Tiffe 35 (2011) 1743e1751. [27] M. Shishkin, T. Ziegler, J. Phys. Chem. C 114 (2010) 11209e11214.
[28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47]
641
G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865e3868. €chl, Phys. Rev. B 50 (1994) 17953. P.E. Blo G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. M. Shishkin, T. Ziegler, J. Phys. Chem. C 113 (2009) 21667e21678. Y. Zhang, Z. Fu, S. Dong, Z. Yang, Phys. Chem. Chem. Phys. 16 (2014) 1033e1040. Y. Zhang, Z. Fu, M. Wang, Z. Yang, J. Power Sources 261 (2014) 136e140. Y. Zhang, Z. Yang, J. Power Sources 271 (2014) 516e521. Y. Zhang, Z. Yang, M. Wang, J. Power Sources 279 (2015) 759e765. Y. Zhang, Z. Yang, J. Power Sources 279 (2015) 224e230. H. Abe, K. Murata, T. Fukui, W.J. Moon, K. Kaneko, M. Naito, Thin Solid Films 496 (2006) 49e52. ~ oz, S. Gallego, J.I. Beltr M.C. Mun an, J. Cerd a, Surf. Sci. Rep. 61 (2006) 303e344. T. Sasaki, K. Matsunaga, H. Ohta, H. Hosono, T. Yamamoto, Y. Ikuhara, Mater. Trans. 45 (2004) 2137e2143. H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188e5192. nsson, J. Chem. Phys. 113 (2000) 9901. G. Henkelman, B.P. Uberuaga, H. Jo nsson, Comput Mate Sci. 36 (2006) G. Henkelman, A. Arnaldsson, H. Jo 354e360. K. Xiong, J. Robertson, M. Gibson, S. Clark, Appl. Phys. Lett. 87 (2005), 183505183505-183503. lik, E.K. Chang, S.G. Louie, Phys. Rev. B 57 (1998) 7027. B. Kra A. Alkauskas, P. Broqvist, F. Devynck, A. Pasquarello, Phys. Rev. Lett. 101 (2008) 106802. P. Broqvist, A. Pasquarello, Appl. Phys. Lett. 89 (2006) 262904.