Influence of Y-substitution on phase composition and proton uptake of self-generated Ba(Ce,Fe)O3-δ-Ba(Fe,Ce)O3-δ composites

Self-generated composites from the series BaCe1-(x+z)FexYzO3-ẟ with z=0.2 for 0.1 x 0.6 ≤ ≤ and z=0 for Ce:Fe = 1 were obtained by one-pot synthesis. The composites consist of proton and electron conducting phases and are interesting as electrode materials for protonic ceramic fuel and electrolyser cells. X-ray diffraction with quantitative phase analysis and scanning electron microscopy with energy-dispersive X-ray spectroscopy showed that the materials consist of Fe-rich phases and a Ce-rich perovskite phase, which are present in the corresponding proportion depending on the precursor composition (Ce-Fe ratio). Substitution with Y leads to a narrowing of the miscibility gap compared to BaCe1-xFexO3-ẟ composites, thus favouring transformation of the composites into single cubic phases at temperatures above 1000°C. Further, Y influences the mutual solubility of Fe3+/4+ and Ce4+ in the Ce-rich and Fe-rich phase, respectively, as shown elemental mapping via scanning transmission electron microscopy. As only a small proportion of the Y dissolves in the electrolyte-type phase, the increased proton uptake resulting from the incorporation of Y in the Ce-rich phase is limited. Strategies to overcome this limitation by substitution with ions with similar ionic radii, but different basicity, are discussed.


Introduction
Solid oxide fuel cells (SOFCs) convert chemical energy contained in various fuels such as hydrogen, methane and other hydrocarbons, into electrical energy. The conversion takes place with high efficiency and without emission of NO x . 1, 2 State-of-the-art SOFCs use oxygen ionic conductors (as electrolytes) and mixed oxygen ionic-electronic conductors (as cathodes). 1-3 In order to obtain sufficiently fast ionic transport, cells based on these materials require relatively high operating temperatures of 700-1000°C, which causes materials compatibility and degradation issues. With regard to operating temperature and efficiency, protonic ceramic fuel cells (PCFCs) are an interesting alternative to SOFCs 4-6 . PCFCs use proton-conducting oxides, offering the advantages of sufficiently fast ionic transport at relatively low temperatures (300-600°C) and significantly smaller activation energies than oxygen ion conductors. 7 However, as the operating temperature of the cell decreases, the electrochemical performance does as well, due to the reduced catalytic activity of the air electrode. Mixed oxygen ionic-electronic conductors, especially Co-containing perovskites such as BaCo 0.7 Fe 0.22 Y 0.08 O 3-δ 8 , Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ 9 and La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3-δ 10 , have been investigated as possible air electrodes in order to improve the catalytic activity at lower temperatures. However, these materials suffer from a comparably low proton uptake. An important goal therefore is to improve the proton uptake and conductivity of potential PCFC cathode materials such that the active zone for oxygen reduction is expanded beyond the gas/electrolyte/cathode triple phase boundary. 11,12 Different strategies for further optimisation of PCFC cathodes were investigated e.g. by using composites of proton-conducting and electron/hole-conducting materials. 13,14 With regard to fabrication of composite cathodes, some disadvantages are associated with conventional mixing of the two phases, such as retention of inhomogeneities and small active areas between the phases (large grain sizes). 15,16 Alternatively, using one-pot self-assembly methods allows one to obtain two homogeneously distributed nanocrystalline phases with complementary functionalities during calcination. [17][18][19][20][21][22][23] Cheng et al. used Fe substituted BaCeO 3 as a model substance for their proof-of-concept study because of its good proton uptake capacity. 24 BaCe 0.5 Fe 0.5 O 3-δ , which was used as a precursor, decomposes into a Ce-rich (BaCe 0.85 Fe 0.15 O 3δ ) and a Fe-rich (BaCe 0.15 Fe 0.85 O 3-δ ) thermodynamically stable phase. A membrane fabricated from the dual-phase composite thus obtained showed hydrogen permeation flux superior to a membrane obtained by conventional mixing and procession of the two single-phase compounds. 24 In the present study substitution of BaCe 0.5 Fe 0.5 O 3-δ with Y is investigated with the aim of increasing the oxygen vacancy concentration and the basicity of the materials, thereby enhancing the proton uptake capacity. Self-generated nanocomposites obtained from BaCe 0.5 Fe 0.5 O 3-ẟ and BaCe 1-(x+0.2) Fe x Y 0.2 O 3-ẟ (0.1 x 0.6) precursors are characterised with ≤ ≤ respect to fundamental material properties in order to gain further insights into phase composition and -distribution, as well as proton uptake capacity. Therefore, the obtained trends in proton uptake are correlated with ionic radii and basicity of various B-site substituents (Y, Yb, Sm, Gd) to suggest promising compositions for further studies. (SG#221), see magnified (011) peak in Fig. 1(b)), the other perovskite phase belongs to the orthorhombic GdFeO 3 structure (Pmcn (SG#62), magnified (002) peak in Fig. 1(b)) for the Yfree material. For Y-containing samples it acquires a trigonal structure (R-3c (SG#167), magnified (110) peak in Fig. 1b).

Results and discussion
In addition to the change in the space group, the positions of the reflections attributed to the cubic phase, shift towards smaller diffraction angles with decreasing Fe content (Fig. 1b). The positions of the reflections ascribed to the orthorhombic/trigonal phases are almost independent of the Ce-Fe ratio. Similar trends are observed as a function of the lattice parameters on the Fe content (Fig. 2a). According to energy-dispersive X-ray spectroscopy (EDXS) (see chapter 2.2), the cubic phase is rich in Fe and the orthorhombic/trigonal phase is rich in Ce. Following Vegard's law 25 , the lattice parameters of the cubic Fe-rich phases (Tab. S-2) decrease linearly with increasing Fe content of the precursor (Fig. 2a With the help of spark plasma sintering (SPS), the precursor powders can be compacted to dense ceramic pellets at comparably low temperature. This allows us to follow the lattice parameters and phase distribution as a function of sintering temperature ( Fig. S-1 The determination of grain size distribution of self-generated composites was carried out by using the image analysis software ImageJ. 30 The

Trends in water uptake
The water uptake of perovskites is primarily represented by the acid-base reaction in R.
(1) (hydration reaction). 27 R. (1) Two protonic defects are formed by dissociative incorporation of water into an oxygen OH  (2)) yields proton uptake by a redox reaction ("hydrogenation"): 31 27 The proton concentration was obtained from the change in sample mass after pH 2 O steps at constant T, under the assumption that R. (1) predominates (which is ensured by the experimental conditions used). However, the distribution of the water uptake between the two phases in the composite is not easily accessible (thus also no H hydrat 0 , S hydrat 0 can be extracted).  3 35 , and has also been shown for Ba(Zr 0.88-x Fe x Y 0.12 )O 3 12 . This effect is attributed to the higher covalence of Fe-O bonds (electron density is drawn from oxygen to iron and thus the basicity of the oxygen is reduced). 36 In comparison with single-phase proton-conducting ferrates/cobaltates ( 37 ; all at T=500°C and pH 2 O=17 mbar), the proton uptake of the self-assembled BaCe 1-(x+0.2) Fe x Y 0.2 O 3-ẟ composites is considerably higher (0.6-1.4 mol% at T=500°C and pH 2 O=17 mbar). A better understanding of the cation distribution within the two phases of the selfgenerated composites could lead to a further optimization of the proton uptake by tailored compositional variations. Substitution with Y 3+ increases the oxygen vacancy concentration and the basicity in both Ce-and Fe-rich phases. If Y is dissolved preferentially in the Fe-rich phase, only a small fraction of the Y will occur in the Ce-rich phase (which is the phase with the higher degree of hydration) and thus the proton uptake of the composite is only slightly improved (see  Figure 7). The key for the proton uptake is the distribution of the acceptor between the Ce-rich and the Fe-rich phase. The mismatch in the ionic radii of Y 3+ (0.9 Å) and Ce 4+ (0.87 Å) can lead to strain in the Ce-rich phase, although the size effect is more severe for Y in the Fe-rich phase. Intermixing of the cations, including dissolution of part of the Y 3+ in the Fe-rich phase, may be a means to reduce this strain 38 , but leads to a decrease in the proton uptake. Ytterbium was explored as an alternative substituent ion for the Ba(Ce,Fe)O 3-ẟ series to counteract straininduced cation intermixing, since Yb 3+ (0.868 Å) has almost the same ionic radius as Ce 4+ .
However, this approach did not result in any significant increase in proton uptake, see Fig. 7a.
Subsequently, two substituents with relatively large ionic radii, Gd 3+ (0.94 Å) and Sm 3+ (0.96 Å) were tried, but again no significant change in proton uptake was observed. Thus, it appears that the ionic radius of the substituent ion plays a subordinate role for the distribution of the acceptor, and therefore for the proton uptake of the composite. Another driving force for the formation of the self-assembled composites is the balance in basicity of the two phases due to the solubility of Ce in the Fe-rich phase and vice versa and the Y distribution. 39 The acceptor distribution is governed by an acid-base reaction, whereby the larger (more basic) ion experience a stronger driving force for the incorporation into the Fe-rich phase. As evident from  which were not aligned along major zone axes with respect to the electron beam in order to avoid channelling effects during the measurement. 43

Thermogravimetry
Thermogravimetry (STA449C Jupiter, Netzsch, Germany) was carried out with approximately 0.5-1 g of sample (sintered pellet crushed to particles) in an alumina crucible with an N 2 flow of 60 ml/min. In order to implement humid conditions the water partial pressure (pH 2 O) was adjusted by bubbling the gas through a thermostated water evaporator (the flow of 10 ml/min "protective gas" through the balance compartment of the STA449 was always kept dry). For isothermal pH 2 O changes, the measured buoyancy effects were negligible. In order to "freeze" the oxygen exchange reaction, the particles were quenched from 700°C in dry N 2 so that all Fe is present as Fe 3+ and then cooled rapidly (20 K/min) to the desired temperature at which the proton uptake is to be determined. An appropriate particle size is essential for the suppression of the surface controlled reaction of oxygen incorporation. The optimal particle size (here: 100-250 µm) is a compromise between reduced surface reaction (low pO 2 and low T) and a reasonable time (10-100 min) for the kinetically more facile water uptake reaction to reach equilibrium (see Fig. S-10). This method avoids complications from hole-proton defect interactions and the long equilibration times at low temperatures as described in Zohourian et al. 27 .