Proton-conducting electrolyte materials

Perovskite oxides

The most studied HTPCs are oxides with the perovskite-type structure (ABO₃), where the A-cation is an alkaline earth element, such as Ba, Sr, and Ca, while the B-cation is a tetravalent element, usually Ce or Zr. To promote protonic conductivity, the B-site is partially substituted with trivalent elements such as Y, In, or Gd, leading to the formation of oxygen vacancies. Proton conduction occurs mostly through the dissociation of water from the gas phase into a hydroxide ion and a proton; the former fills an oxygen vacancy, while the latter forms a hydrogen bond with lattice oxygen.^{Reference Fabbri, Pergolesi and Traversa3} The reaction generates two protonic defects. These protons can be highly mobile and migrate by a hopping mechanism from one oxygen atom to the next (Figure 1a).

Figure 1.

(a) Sketch of a possible proton transfer mechanism from the oxygen ion I to oxygen ion II in a high-temperature proton conductor (HTPC) perovskite oxide. The red sphere represents the B site cation of the perovskite structure, the larger blue spheres represent oxygen ions, while the small blue spheres represent protons. Reproduced with permission from Reference 3. © 2010 Royal Society of Chemistry. (b) Conductivities versus reciprocal temperature for selected HTPC electrolytes: BaZr_0.8Y_0.2O_3–δ (BZY) large-grained pellet^{Reference Yamazaki, Hernandez-Sanchez and Haile8} and film,^{Reference Pergolesi, Fabbri, D’Epifanio, Di Bartolomeo, Tebano, Sanna, Licoccia, Balestrino and Traversa9} 1% Ca-doped LaNbO₄ (Ca-LNO) pellet,^{Reference Haugsrud and Norby11} and La_28–xW_4+xO_54+δ (x ∼ 0.85) (LWO) pellet.^{Reference Magraso, Frontera, Marrero-Lopez and Nunez17}

According to the most accepted model,^{Reference Kreuer4–Reference Islam, Davies and Gales6} thermal vibrations of the perovskite BO₆ octahedra lead to chaotic fluctuations in the vicinity of the O atoms involved in the hydrogen bonding. Then, the distance between two adjacent oxygen atoms can briefly become short enough to allow the hydrogen to overcome the potential barrier and hop to the neighboring oxygen (proton transfer). At the same time, thermal vibrations also favor local elongations of the O–H link, which becomes weaker and eventually breaks, leaving the proton free to bond with another oxygen atom (re-orientation and rotational diffusion). These two mechanisms complement each other, resulting in continuous low-energy proton-diffusion pathways.^{Reference Merinov and Goddard5} The energetics of these two mechanisms is consistent with an activation energy of about 0.15 eV or lower. However, in cubic perovskite oxides, the experimental activation energy is ∼0.4–0.6 eV.^{Reference Kreuer4} Such a discrepancy has been experimentally rationalized by the effect of proton trapping.^{Reference Yamazaki, Blanc, Okuyama, Buannic, Lucio-Vega, Grey and Haile7} In the case of Y-doped barium zirconate, the incorporated protons are trapped at the dopant sites, adding an additional potential barrier for long-range diffusion.

The most investigated HTPCs are 10–20 at.% Y-doped barium zirconate (BZY) and barium cerate (BCY).^{Reference Fabbri, Pergolesi and Traversa3} BZY exhibits good chemical stability in fuel cell operating conditions and high grain interior (bulk) conductivity, but the grain boundary regions block the proton migration. Moreover, BZY is a highly refractory material, and generally very high temperatures (above 1700°C) are required to sinter dense ceramic pellets. Such high temperatures lead to BaO evaporation, and the Ba loss favors the Y substitution into the A sites of the perovskite structure. As a result, oxygen vacancies are consumed, thereby reducing the number of charge carriers. Lower sintering temperatures, suitable for preserving the required composition, lead to high porosity and/or low average grain size with extended grain boundary regions. As a consequence, the total electrical conductivity of BZY sintered pellets is too low for practical applications. On the other hand, BCY sintered pellets have overall good conducting properties, but the main drawback is the low chemical stability in the operating environment for SOFCs, since BCY reacts with water vapor and CO₂.^{Reference Fabbri, Pergolesi and Traversa3}

Many efforts have emerged to either improve the chemical stability of BCY or the sinterability of BZY. The latter strategy turned out to be particularly effective, since an experimental protocol for the synthesis of BZY ceramic pellets without the use of sintering aids and leading to an average grain size in the micrometer range has been recently reported.^{Reference Yamazaki, Hernandez-Sanchez and Haile8} The literature also reported the fabrication of highly ordered (grain boundary free) BZY thin films grown by pulsed laser deposition, which allows direct access to the very high bulk conductivity of this material.^{Reference Pergolesi, Fabbri, D’Epifanio, Di Bartolomeo, Tebano, Sanna, Licoccia, Balestrino and Traversa9,Reference Kim, Gur, Jung, Kang, Sinclair and Prinz10} The proton conductivities in the LT range of BZY (as sintered pellets with micrometer grain size and as an epitaxial thin film) are reported in Figure 1b.

Electrode materials

In H⁺-SOFCs, water is produced at the cathode side; the cathodic oxygen reduction reaction has much slower kinetics than the hydrogen reduction at the anode side (when H₂ is used as the fuel), and therefore the cathode is generally the source of the largest overpotential associated with the electrode reactions. The oxygen reduction at the cathode comprises several reactions steps, summarized in Table I, which can vary depending on the cathode material.^{Reference Fabbri, Pergolesi and Traversa20} As shown in Figure 2, molecular oxygen adsorption occurs at the cathode surface followed by oxygen dissociation. In the case of a mixed O^2–/e^– conducting cathode (Figure 2a) or a composite cathode made of a mixed O^2–/e^– conductor and a proton conductor (Figure 2c), the adsorbed oxygen can diffuse on the surface of the mixed O^2–/e^– conductor toward the reaction area, where it reacts with two electrons and two protons to form water. For a mixed H⁺/e⁻ conducting cathode (Figure 2b), the oxygen does not need to migrate toward the reaction site because protons and electrons can migrate within the whole cathode material.

Table I.

Possible elementary cathode reaction steps in a H⁺-solid-oxide fuel cell.

The elementary reaction steps described are also illustrated in Figure 2 for different cathode materials.

Figure 2.

Possible reactions at the cathode using (a) a mixed O₂-ion/e⁻ conductor, (b) a mixed H⁺/e⁻ conductor, and (c) a composite cathode made of a proton conductor phase and a mixed O^2–/e⁻ conductor phase. The dark gray spheres represent a mixed O^2–/e⁻ conductor phase, the light blue sphere is a mixed H⁺/e^– conductor, while the light gray sphere is a H⁺ conductor. The red semicircles represent the reaction sites where the oxygen reduction takes place. Figure adapted from Reference 20.

Only recently have research efforts been oriented toward the development of specifically designed cathode materials for H⁺-SOFCs.^{Reference Fabbri, Pergolesi and Traversa20} While the use of a mixed oxygen-ion/electron (O^2–/e^–) conductor results in confinement of the reaction sites to the electrode/electrolyte interface (Figure 2a), a mixed proton/electron (H⁺/e^–) conductor would allow extending the reaction area to the whole cathode surface, thus facilitating the evaporation of the water by-product (Figure 2b). An alternative way to extend the reaction area is the development of composite cathodes made of a mixed O^2–/e^– conductor and a H⁺-conducting oxide (Figure 2c).^{Reference Fabbri, Pergolesi and Traversa20}

In the case of perovskite-type electrolytes, growing interest has recently been directed toward the development of mixed H⁺/e^– conducting oxides. Mixed H⁺/e^– conductivity in a wet oxidizing atmosphere can be obtained by doping the B-sites of BaZrO₃ with Y and Pr, due to the mixed 3+ and 4+ valence state of Pr. As an example, the composition BaZr_0.6Pr_0.3Y_0.2O_3−δ showed good chemical stability and a mixed H⁺/e^– conductivity in humidified air.^{Reference Fabbri, Markus, Bi, Pergolesi and Traversa21} Consistent results were also obtained doping BaZrO₃ with Pr and Gd, with an optimized composition of BaZr_0.6Pr_0.3Gd_0.1O_3–δ.^{Reference Magraso, Frontera, Gunnaes, Tarancon, Marrero-Lopez, Norby and Haugsrud22,Reference Magraso, Kjolseth, Haugsrud and Norby23} Following a similar strategy, transition metals such as Co and Fe were introduced into the lattice of BaCeO₃ and BaZrO₃ perovskites.^{Reference Shang, Tong and O’Hayre24–Reference Zhang and Zhao26} However, the value of the partial electronic conductivity, where reported, of such mixed H⁺/e^– conductors is too low for application as a single-phase cathode material. Recently, it has been suggested that typical cathode materials used in SOFCs based on oxygen-ion conducting electrolytes, besides being mixed O^2–/e^– conductors, might also present some level of proton conductivity.^{Reference Grimaud, Mauvy, Bassat, Fourcade, Marrony and Grenier27,Reference Grimaud, Mauvy, Bassat, Fourcade, Rocheron, Marrony and Grenier28} In particular, it has been proposed that perovskites with significant oxygen deficiency, such as Ba_0.5Sr_0.5Co_0.8Fe_0.2O_3–δ and PrBaCo₂O_5–δ, might allow the largest degree of proton insertion.^{Reference Grimaud, Mauvy, Bassat, Fourcade, Rocheron, Marrony and Grenier28} However, the suggestion of proton conduction in mixed O^2–/e^– conductors is still controversial, and further investigations are required to unambiguously prove the presence of a protonic contribution to the total conductivity.

Encouraging results were also reported on the development of composite cathode materials. For example, a relatively low area specific resistance (ASR) of 0.16 Ω cm² was obtained at 600°C for a composite cathode made of La_0.5Sr_0.5Co_0.6Fe_0.4O_3–δ and BaZr_0.5Pr_0.3Y_0.2O_3–δ,^{Reference Fabbri, Bi, Pergolesi and Traversa29} where the former is a mixed O^2–/e^– conductor, and the latter is a mixed H⁺/e^– conductor.

The microstructure also plays an important role in achieving low overpotentials. Advanced cathode microstructures were obtained at relatively low processing temperatures, avoiding cation interdiffusion, by infiltration of LaCoO₃ into a porous backbone of Ba-Ce-Zr-based oxide.^{Reference Ricote, Bonanos, Lenrick and Wallenberg30}

Regarding the anodic electrode, most of the H⁺-SOFCs utilize composite anodes fabricated by mixing NiO with the same material used for the electrolyte. Conversely, anodes for the LWO electrolyte cannot be prepared in this “traditional” manner, as it reacts with NiO, and alternative fabrication routes have been reported.^{Reference Solis, Vert, Balaguer, Escolastico, Roitsch and Serra31} Two general drawbacks are associated with the use of Ni for the fabrication of the anode. First, Ni tends to agglomerate, resulting in deterioration of anode performance with time. Additionally, when hydrocarbons are used as the fuel, coke formation strongly hinders Ni-based anode performance. Moreover, the chemical stability of the H⁺-conducting phase in the anode is obviously a further important issue when carbon-containing compounds are used as a fuel.

Fuel cell performance

As outlined previously, the main issues related with reducing SOFC operating temperature are the increased electrolyte resistivity and electrode polarization. In addition to improving the electrolyte conduction properties and electrode kinetics, modifying the fuel cell design might also assist in overcoming these drawbacks. Making the anode the supporting structure for SOFC fabrication allows the electrolyte to be implemented as a thin layer, thereby minimizing ohmic loss. Several ceramic processing methods have been proposed to fabricate anode-supported H⁺-SOFCs. Most of the tested H⁺-SOFCs on the laboratory scale are fabricated by a co-pressing and co-firing technique,^{Reference Fabbri, Bi, Pergolesi and Traversa32} which generally allows a minimum electrolyte thickness (capable of preventing reactant cross-over) of ∼15 μm to be supported on a porous anode structure, as shown in Figure 3.^{Reference Fabbri, Bi, Tanaka, Pergolesi and Traversa33}

Figure 3.

Scanning electron microscope image of the fractured cross-section of an anode-supported H⁺-SOFC (solid-oxide fuel cell). Four layers are clearly visible: the supporting anode, the anode functional layer, the electrolyte (BaZr_0.7Pr_0.1Y_0.2O_3–δ), and the composite cathode. Reproduced with permission from Reference 33. © 2011 Wiley.

The development of anode-supported H⁺-SOFCs based on doped-BaZrO₃ electrolytes is complicated by the poor sinterability of this material. This problem has been addressed following different approaches; by using doping elements able to favor the sinterability (i.e., BaZr_0.7Pr_0.1Y_0.3O_3–δ,^{Reference Fabbri, Bi, Rupp, Pergolesi and Traversa34} BaZr_0.7Sn_0.1Y_0.3O_3–δ,^{Reference Sun, Liu and Liu35} BaZr_0.8/0.7In_0.2/0.3O_3–δ^{Reference Sun, Zhu, Shi and Liu36,Reference Bi, Fabbri, Sun and Traversa37} ), by the addition of sintering aids,^{Reference Sun, Fabbri, Bi and Traversa38–Reference Tong, Clark, Hoban and O’Hayre40} or by using anodes able to promote electrolyte sinterability.^{Reference Bi, Fabbri, Sun and Traversa41,Reference Bi, Fabbri and Traversa42} Several H⁺-SOFCs have also been developed using a solid solution of barium zirconate and barium cerate (e.g., BaCe_0.7Zr_0.1Y_0.2O_3–δ) as an electrolyte.^{Reference Fabbri, Bi, Pergolesi and Traversa32} However, it has been recently shown that even by lowering the Ce content down to 50 at.%, the solid solution between barium zirconate and cerate presents severe stability limitations in a wet atmosphere.^{Reference Slodczyk, Sharp, Upasen, Colomban and Kilner43}

The highest power density outputs for H⁺-SOFCs using chemically stable BaZrO₃-based electrolytes are around 250 mW cm^–2 at 600°C^{Reference Fabbri, Bi, Pergolesi and Traversa32,Reference Sun, Liu and Liu35} —still not sufficiently competitive with the performance of SOFCs based on oxygen-ion conducting electrolytes.^{Reference Wachsman and Lee1}

The performance of fuel cells based on Ca-doped LaNbO₄ as the electrolyte is even lower, mainly ascribed to the poor kinetics of several tested cathode materials^{Reference Magraso, Fontaine, Bredesen, Haugsrud and Norby13} and the lower protonic conductivity of this material compared to the best perovskites. In addition, the fabrication of a complete cell using LaNbO₄ avoiding detrimental phase segregation is very challenging. Unless these remaining issues are addressed, LaNbO₄ will not be the ultimate electrolyte choice for fuel cell applications. Concerning the LWO electrolyte, tailor-made anodes and cathodes with ASRs ∼0.3–0.5 Ω cm² have been proposed,^{Reference Solis, Vert, Balaguer, Escolastico, Roitsch and Serra31,Reference Balaguer, Solis, Bozza, Bonanos and Serra44,Reference Quarez, Oumellal and Joubert45} but electrode-supported fuel cells have not yet been tested.

Using different fuel cell designs, promising results for H⁺-SOFCs have been obtained by fabricating polycrystalline BZY freestanding membranes on Si chips by thin-film deposition and microfabrication technologies. A peak power density output of 186 mW cm^–2 at 450°C has been achieved.^{Reference Kim, Gur, Kang, Jung, Sinclair and Prinz46} These results, coupled with the high conductivity of BZY epitaxial thin films with reduced grain boundary regions, open new possibilities for the development of miniaturized H⁺-SOFCs.

Summary and perspectives

HTPC-sintered polycrystalline pellets, as well as highly ordered thin films, have shown conductivity values in the same range or even larger than those of the best performing (and chemically stable) oxygen-ion conductors at low temperatures. Moreover, composite cathode materials, mixed protonic and electronic conductors, specifically designed for HTPC electrolytes, have been developed showing comparable performance as the typical cathodes used for oxygen-ion conductors.

In spite of the good performance of the individual cell components, the power density output reported so far for H⁺-SOFCs is significantly lower compared to SOFCs based on oxygen-ion conductors. The reason for this relatively limited performance is not clear, but it might be due to poorer electrode kinetics, especially in the cathode. It must be highlighted here that research and development on H⁺-SOFCs is still relatively limited. Improved cell design and microstructural optimization may render the key solution. The good performance of the individual cell components holds the promise of obtaining larger power density output, and clearly indicates the need for more scientific and technological efforts for full exploitation of the potential of HTPCs.