Surface reactivity and cation non-stoichiometry in BaZr_1−xY_xO_3−δ (x = 0–0.2) exposed to CO₂ at elevated temperature

Abstract

The reactivity of BaZr_1−xY_xO_3−δ (x = 0–0.2) ceramics under 1 atm CO₂ at 650 °C for up to 1000 h was investigated in order to elucidate possible degradation processes occurring when the material is applied as a proton-conducting electrolyte in electrochemical devices. The annealed ceramics were characterized by a range of techniques (SEM, TEM, GIXRD, XPS and SIMS) with respect to changes in the phase composition and microstructure. Formation of BaCO₃ was observed on the surfaces of the annealed samples and the amount increased with time and was higher for the Y-doped compositions. The subsurface regions were found to be deficient in Ba and, in the case of the Y-doped compositions, enriched in Y in two distinct chemical states as identified by XPS. First-principles calculations showed that they were Y residing on the Zr and Ba-sites, respectively, and that local enrichment of Y both in bulk and on the surface attained a structure similar to Y₂O₃. Overall, it was substantiated that the reaction with CO₂ mainly proceeded according to a defect chemical reaction involving transfer of Y to the Ba-site and consumption of BaZrO₃ formula units. It was suggested that a similar degradation mechanism may occur in the case of Ba(OH)₂ formation under high steam pressure conditions.

---

1. Introduction

A^IIB^IVO₃ perovskites have been widely studied as materials for solid state electrochemical devices.^1–3 Alkali earths such as strontium and barium on the A-site implies the possibility of reaction with CO₂ to form stable carbonates at low to intermediate temperatures in electrochemical systems for the conversion of hydrocarbons.^1,3–10 The CO₂-sensitivity of the materials can originate from thermodynamic phase instability or surface reactivity, and minor compositional modifications can affect the chemical stability. ¹¹ Consequently, traditional solid-state reaction synthesis can be insufficient to avoid compositional heterogeneities that exhibit poorer resistance to CO₂ ¹¹ and powder processing can therefore also have a significant impact on stability as well as sinterability.^12–14

The high proton conductivity of Y-doped BaZrO₃ (BZY) has made it the state-of-the-art solid-state electrolyte in proton ceramic fuel cells (PCFCs)^3,15 and electrochemical membrane reactors.^16,17 BZY has a cubic crystal structure,¹⁸ and BZY containing 10 mol% yttria (BZY10) has become one of the most studied compositions.¹⁹ Ce-containing compositions, BaZr_1−x−yCe_xY_yO₃ (BZCY), possess higher proton conductivity and improved sintering properties.²⁰ However, the chemical stability towards CO₂-, H₂O-, or H₂S-containing atmospheres is reduced with increasing Ce-content.²¹

The presence of carbonates can have a significant effect on the transport properties of proton conductors,^22,23 as well as on the catalytic activity in the presence of hydrocarbons.⁹ Nevertheless, a recent study by Duan et al. showed excellent coking resistance and sulfur tolerance of BaZrO₃-based PCFCs.²⁴ Structural and chemical degradation has been reported even in yttria stabilized zirconia (YSZ) in carbon-rich fuel gases such as CH₄, CO or syngas, inducing coking/graphitization and (oxy)carbide formation^25–28 that led to irreversible changes in the transport properties.^27,28 Thus, a thorough understanding of potential degradation processes at the surface of BZY materials is highly desirable, and could lead to valuable insights into the surface chemistry and structure of this important proton-conducting electrolyte.

While bulk BZY has been shown to be chemically stable in 1 atm CO₂ above about 600 °C in accordance with thermodynamic considerations,^29,30 the surface of the material may still exhibit some reactivity. Computational studies have recently shown that CO₂ exhibits strong chemisorption on BaZrO₃ surfaces by formation of carbonate species and that a carbonate overlayer of BaCO₃ can be thermodynamically preferred even in 400 ppm CO₂ above 100 °C.³¹ The reactivity between BZY and pure CO₂ has been demonstrated to result in the formation of BaCO₃ and deterioration of the mechanical properties at 550–750 °C.³² It was determined that the reaction with CO₂ resulted in Ba-deficiency in the exposed surface rather than formation of ZrO₂ crystallites,³² and the reaction could thereby reasonably be described according to

where and denote Ba- and O-vacancies, respectively, in Kröger–Vink notation.³³ Similar degradation has also been reported for Sr-based perovskites.^5,34 Alternative reaction pathways can be hypothesized based on further consideration of the phase stability and cation defect chemistry of BZY. In this respect, the coexistence of a Y-rich composition has been reported within single grains of Ba-deficient BZY.^35–37 Han et al. showed by synchrotron X-ray diffraction that excess Y can occupy the A-site, i.e., .¹⁸ Another possible reaction mechanism between BZY and CO₂ towards BaCO₃ and Ba-deficient BZY can therefore be expressed by relocation of Y from the Zr to the Ba-site and consumption of one BaZrO₃ formula unit according to

In this study, we present a thorough investigation of the reactivity of dense BaZr_1−xY_xO_3−δ (x = 0–0.2) ceramics with CO₂ as the gas relevant for anodic fuel conversion electrochemical devices. The BZY samples were exposed to 1 atm CO₂ at 650 °C for 10–1000 h, and the extent of BaCO₃ formation and changes in the BZY surface microstructure and chemical composition were investigated by using several techniques. Reactions (1) and (2) both result in Ba-deficient BZ(Y) according to the change in A/B-site ratios, Ba/Zr and Ba/(Zr + Y), respectively. However, the reactions may be differentiated by the presence of the species as well as the level of deterioration of the BZY grains due to the consumption of a BaZrO₃ formula unit in reaction (2). Furthermore, density functional theory (DFT) calculations were performed to evaluate relevant point defect energetics and to compare the thermodynamic feasibility of reactions (1) and (2).

2. Experimental

2.1 Sample preparation

Dense BaZr_1−xY_xO_3−δ (x = 0–0.2) ceramics were prepared by the procedure described in two recent contributions.^32,38 Powders made by spray pyrolysis (CerPoTech, Norway) were calcined at 950 °C for 12 h, ball milled and pressed into pellets by uniaxial pressing followed by cold isostatic pressing at 200 MPa. The pellets were sintered at 1600 °C for 10 h in a powder bed consisting of 90 wt% BZY and 10 wt% BaCO₃. The pellets were polished with SiC papers and diamond suspensions down to 1 μm in order to obtain surfaces with low roughness. The density of the materials was measured by the Archimedes method using isopropanol at room temperature. The polished disc samples were broken into several pieces and annealed under 1 atm of dry flowing CO₂ at 650 °C for total exposure times of 10, 20, 100 and 1000 h. The 10 and 20 h annealing experiments were performed in an alumina tube furnace, and the longer annealings of 100–1000 h were performed in a ProboStat measurement cell (NORECS, Norway) with a sample holder of alumina and platinum wires. The samples are identified by their yttrium content (BZ, BZY10 and BZY20 for 0, 10 and 20% Y doping, respectively) and their exposure time to CO₂ at 650 °C.

X-ray powder diffraction (XRD) of the calcined powders and sintered materials was performed using a Bruker D8 Advance DaVinci diffractometer using CuK_α1 radiation. Grazing incidence XRD (GIXRD) was performed using CuK_α1 and parallel beam optics on pristine samples and samples exposed to CO₂ to characterize the formation of BaCO₃ on the surface. Rietveld refinement of the XRD patterns for powders was carried out using TOPAS V4.1 software using a cubic structure model (Pm3̄m) for all the BZ(Y) materials.

The microstructure of the materials was investigated with Hitachi S3400N and Hitachi FEG Zeiss Ultra scanning electron microscopes (SEM) on gold coated samples. The chemical composition of phases was investigated by energy dispersive spectroscopy (EDS) with Oxford Instruments AZtec Energy analysis software.

2.2 Transmission electron microscopy

Cross-sectional transmission electron microscopy (TEM) samples were made using a focused ion beam on a dual-beam JEOL SEM. TEM was carried out using a FEI Titan G2 60-300 operated at 300 kV with a high brightness XFEG, probe corrector, Gatan imaging filter for electron energy loss spectroscopy (EELS) analysis and super-X EDS detector. EELS results were acquired with STEM spectral imaging, using dual EELS, acquiring both high loss and low loss, for energy referencing.

2.3 Secondary ion mass spectrometry

The distribution of the isotopes was determined by secondary ion mass spectrometry (SIMS) using a Cameca IMS 4f instrument. A primary ion beam of 10 keV Cs⁺ was applied while qualitative optimization of sputtering and mass intensity was performed with a 12.5 keV O₂⁺ ion beam. The primary ion current varied between 30 and 60 nA, and the secondary ion intensity ranged between 5 and 6 nA. Depending on the secondary ion intensities, an area with a diameter of ∼30 μm in the center of a sputtered 100 × 100 μm² area was gated for signal detection and analysis in order to avoid negative interference (edge and wall effects). An electron shower with an acceleration of 4.5 kV for ∼1.2 mA sample current was used to charge compensate the insulating samples. Dynamic transfer optics setting (DTOS) of 60% was used. The secondary molecular ion signals were measured as a function of cycles per second (cps) and sputter time.

2.4 X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) was performed on both pristine and CO₂-exposed BZY samples using a Kratos Axis Ultra^DLD XPS system with an incident monochromated Al Kα X-ray beam. The working pressure in the XPS chamber at room temperature was below 5 × 10⁻⁹ mbar. High resolution scans were recorded for the Ba 3d and 4d, Zr 3d, Y 3d, O 1s and C 1s energy regions, with a step size of 0.1 eV. All of the obtained binding energies (BE) were calibrated based on the C 1s peak for aliphatic bonds from adventitious carbon, set to 284.8 eV BE. The resulting XPS data were analysed using CasaXPS software. A Gaussian–Lorentzian line shape GL(30) was employed to fit the XPS peaks after background subtraction based on the Shirley algorithm. The peak separation and area ratio of the Y 3d_5/2 and the Y 3d_3/2 peaks were constrained to be 2.05 eV and 3 : 2, respectively.

3. Computational methods

Y-doped BaZrO₃ was modeled as a 3 × 3 × 3 supercell (135 atoms) containing 4 yttrium acceptors and 2 oxygen vacancies for charge neutrality, corresponding to a dopant concentration of approx. 15% (denoted as BZY15). Interactions between and were thereby taken into account for a specific configuration of the highly doped material. The binding energies between isolated defects were calculated based on the single fully ionized defects in 4 × 4 × 4 supercells. The BZY15 cell was constructed with a pair of on next neighbor sites and two on second next neighbor sites. Based on the calculated association energy between isolated and of −0.29 eV, the two charge-compensating were placed in the most stable neighboring sites to . The same procedure was followed for introducing additional barium Schottky defect pairs, or the introduction of two and removal of the least stable , according to eqn (1) and (2), respectively. The tendency of the defects to segregate to the surface was investigated using 11-layer (0 0 1) BaO-terminated slabs with 3 × 3 expansion perpendicular to the surface. Defect segregation energies were calculated as the total energy difference between charge neutral cells with the defects residing in the bulk and surface regions, respectively. The entropies of reactions (1) and (2) were estimated from the tabulated entropies of CO₂ and the defect-free solid phases.^39,40

The DFT calculations were performed with VASP⁴¹ based on the projector-augmented wave (PAW) method⁴² and the generalized gradient approximation (GGA) functional proposed by Perdew, Burke and Ernzerhof.⁴³ Geometric optimization of the Y-doped supercell was performed with a 500 eV plane-wave energy cut-off and the subsequent calculations of the Ba-deficient cells were performed with fixed lattice parameters to represent bulk continuation to the surface. A 2 × 2 × 2 Monkhorst–Pack scheme was used for k-point sampling of the supercells.⁴⁴ The ionic positions were relaxed until the residual forces were within 0.05 eV Å⁻¹, while the self-consistent electronic optimization was performed with an energy convergence criterion of 10⁻⁶ eV. Core level shifts for Y 3d in the different configurations and clusters were obtained according to the initial state approximation as implemented in VASP.

4. Results

4.1 Formation of BaCO₃

The extent of BaCO₃ formation on the surface of BZ and BZY samples after annealing in CO₂ at 650 °C for 10 to 1000 h was investigated by GIXRD and SEM. Fig. 1a and b show the GIXRD patterns of the pristine and annealed specimens. For both compositions, reflections corresponding to BaCO₃ (Pmcn) appear after annealing, and the amount of BaCO₃ increases with annealing time. On the other hand, there is no clear evidence of formation of ZrO₂ other than possibly small amounts for the BZY10 sample after 1000 h. Based on the relative intensities of the main reflections, the formation of BaCO₃ was more prominent in the Y-doped composition. As shown in the insets, there is a continuous shift of the BZ and BZY Bragg reflections towards smaller lattice parameters with increasing annealing time and BaCO₃ formation. Fig. 1c and d show the surface of BZY10 before and after exposure to CO₂ at 650 °C for 10 h. Crystals of BaCO₃ were formed on the surface by exposure to CO₂, resulting in a rough surface. The average size of the crystals was about 4 μm.

Fig. 1 GIXRD patterns of BZ (a) and BZY10 (b) after CO₂ exposure at 650 °C for 10–1000 h, and SEM micrographs of the BZY10 surface before (c) and after exposure to CO₂ at 650 °C for 10 h (d).

4.2 Cation depth profiles

Fig. 2a and b show cross-sectional EDS line scans of BZY10 as a function of depth from the surface for different annealing times normalized to the Zr-signal. The barium line scans show a clear deficiency towards the surface, which increases with increasing annealing time. The yttrium concentration profile in Fig. 2b appears relatively constant throughout the analyzed region with no significant variations as a function of annealing time. The same result was observed when Y was normalized to the nominal B-site ratio, i.e., Y/(Zr + Y) ratio (not shown).

Fig. 2 Cross-sectional EDS line scans of BZY10 as a function of depth from the surface for different annealing times: A/B-site ratio (a) and Y/B-site ratio (b). The SIMS depth profiles of Ba (c) and Y (d) normalized to the Zr intensity in BZ, BZY10 and BZY20 annealed in CO₂ for 1000 h.

SIMS analyses were performed on the BZ and BZY samples annealed in CO₂ at 650 °C for 1000 h. The depth profiles of the elements normalized to Zr intensity are given in Fig. 2c and d. According to these profiles, the BZ sample exhibits Ba-deficiency at the surface, while BZY10 and BZY20 seem to be quite stoichiometric. However, considering Y-segregation to the surface region shown in Fig. 2b, the normalization to Zr does not reveal a consistent A/B-ratio due to the change in all of the elements from the surface towards the bulk.

The carbon concentration profiles in BZY10 are shown in Fig. 3. The amount and depth of carbon increase as a function of annealing time in CO₂. Due to the low atomic number, the carbon intensities were quite low compared to the intensities of Ba, Y and Zr.

Fig. 3 Concentration profile of C measured by EDS in BZY10.

4.3 TEM investigation of the surface microstructure and phases

TEM was used to perform microstructure and phase characterization and in particular to investigate possible formation of secondary phases at the surface. Fig. 4 shows a surface cross-sectional TEM image of BZY20 annealed for 1000 h in CO₂ at 650 °C, and electron diffraction patterns from the surface region and 150 nm into the bulk. The grains corresponding to BaCO₃ are clearly visible on the surface. Other secondary phases like ZrO₂ or Y₂O₃ were not observed.

Fig. 4 TEM cross-sectional image of the surface of BZY20 (a) and diffraction patterns from the surface and bulk (b).

The bulk (red) diffraction pattern shows the 〈100〉 zone axis of the BZY perovskite structure (Pm3̄m) with a lattice parameter of 4.1 Å. The diffraction pattern from the surface (blue) shows the same crystal structure, albeit with a slightly increased lattice parameter of 4.4 Å, in addition to diffraction patterns from BaCO₃ and/or Ba(OH)₂. The increase in the BZY lattice parameter might suggest a higher Y concentration in the top 150 nm. However, no difference in Y concentration was detected by EDS. One might speculate whether the apparent increase in C content towards the surface observed by EDS profiling may reflect interstitial carbon species and associated defects that expand the lattice. In this respect, DFT calculations have shown that carbon interstitials can be stabilized as carbonate species in either Ba- or Zr-vacancies in BaZrO₃.²² While their equilibrium concentrations were determined to be negligible in the acceptor doped material, the reaction may be facilitated by pre-existing cation vacancies as has been shown for dissolution of interstitial Ni into BaZrO₃ from NiO.⁴⁵

In Fig. 5a, a magnified high angled annular dark field (HAADF) STEM image shows that smaller pores of 5–20 nm were present inside the grains near the surface of BZY20 annealed for 1000 h in CO₂ at 650 °C. This is also shown in Fig. 5b, in addition to structural dislocations. These nanoscopic voids exhibit high surface energies and cannot be expected to be stable after sintering of the material, and it is reasonable to assume that they may have been formed during the annealing in CO₂. This can point to reaction (2), which involves consumption of BaZrO₃ formula units, rather than reaction (1) which involves formation of Ba-deficient BZY. Additional HAADF images and electron energy loss spectroscopy (EELS) characterization are provided in the ESI.†

Fig. 5 HAADF images of (a) pores, and (b) pores and dislocations in the material.

4.4 XPS and computational analyses of chemical states

XPS was performed to characterize changes in the chemical state of Y in the BZY samples after reaction with CO₂ to emulate possible formation of according to reaction (2). The XPS spectra of Ba 4d and Y 3d are compared for BZY10 and BZY20 annealed in CO₂ for 100 h and 1000 h in Fig. 6. The bulk spectra were taken from the annealed samples (1000 h) fractured under vacuum, and they may have been affected by long-range diffusion processes during the annealing although the main changes in stoichiometry were observed within 10 μm from the surface based on the EDS and SIMS profiles. The Ba 4d peak shows that at least two chemical states are present (Fig. 6a). The lower binding energy (BE) peak can be related to BZY while the higher BE peak is typical of BaCO₃ and Ba(OH)₂. The intensity of the latter peak clearly increases with exposure time to CO₂, well in agreement with the BaCO₃ formation on the surface of BZY10 samples as detected by GIXRD and EELS (Fig. S2†). The Zr element is quite stable in CO₂ atmosphere, showing the unchanged position of Zr 3d peaks with increasing exposure time (Fig. S3†). On the other hand, the Y 3d peak shows a complex structure that changes with increasing time of CO₂ exposure (Fig. 6b and c). Two spin–orbit pairs describing at least two different chemical states were required for a satisfactory fitting of the Y 3d energy region. The spin–orbit pair at the highest BE exhibited a core level shift of about 2.5 eV relative to the pair at the lowest BE for BZY10. In the case of BZY20, the core level shift was quite similar for the bulk, while it decreased to about 1.6 eV after exposure to CO₂ (Fig. 6c). The peak intensities for the chemical state at lower BE decreased with CO₂ exposure time in favor of the chemical state at higher BE (particularly evident for BZY20).

Fig. 6 High resolution XPS (a) BZY10 Ba 4d, (b) BZY10 Y 3d and (c) BZY20 Y 3d spectra for the bulk (black), and the surface of the 100 h (green) and 1000 h (red) annealed samples.

The relative Y 3d core level shift of various Y defects and clusters was investigated computationally for comparison with results from XPS. Fig. 7 shows several configurations of (Fig. 7b–d) and (Fig. 7e–g) with the calculated Y 3d core level shifts (Δ) relative to the fully coordinated octahedra (Fig. 7a). The Y 3d core level shift for associated with a protonic defect, a nearest neighbor and/or an oxygen vacancy was 0.18–0.88 eV to higher binding energy relative to the isolated . In comparison, the core level shift for was more significant, 2.21–2.40 eV towards higher binding energy. Similar Y 3d core level shifts were obtained for and defects and clusters on the (0 0 1) BaO-terminated surface, i.e., there was no significant difference between the core level shifts for the bulk and surface.

Fig. 7 Local structural environment and Y 3d core level shift (Δ) for various configurations of (a–d) and (e–g) from 3 × 3 × 3 supercells.

Despite the rather large difference in size between Y and 12-coordinated Ba, attained a 6-coordinated configuration by relaxing towards the edge (Fig. 7f) or face (Fig. 7g) of the unit cell. The resulting Y–O bond lengths, 2.3–2.4 Å, were similar to those for the various configurations of , 2.1–2.3 Å. Notably, the local structure surrounding resembled that of Y₂O₃ (Ia3̄) comprising distorted YO₆ octahedra and three-coordinated oxide ions similar to Fig. 7e–g. Overall, Y exhibited distinctly different 3d core level shifts when substituted on the Ba and Zr-sites in BaZrO₃: 0–0.88 eV to higher binding energy for and about 2.3 eV to higher binding energy for . The difference in chemical shift of 1.4–2.3 eV between and species corresponds well with the separation of about 1.6–2.5 eV observed by XPS. In the case of dehydrated materials due to dry atmosphere, and a relatively low concentration of oxygen vacancies due to reaction (2), the calculated shift would be closer to 2.4 eV (Fig. 7a and e).

4.5 DFT calculations of reaction thermodynamics

DFT calculations were used to calculate and compare the enthalpy of reactions (1) and (2), while the reaction entropies were estimated from the tabulated entropies of CO₂ and the solid phases. The relaxed structures of the computational cells with an optimized configuration of additional defects are shown in Fig. 8. The BZY15 reference cell, shown in Fig. 8a, contains one oxygen vacancy in between two nearest-neighbor , i.e., (Y_Zr–v_O–Y_Zr)^×, and one oxygen vacancy associated with a single .

Fig. 8 Relaxed structures of the initial defective cell containing 4 Y-acceptors and 2 oxygen vacancies (a), the Ba-deficient cell containing an additional barium Schottky pair (b), and the Y-rich cell containing 2 Y on the Ba-site (c).

The cell in Fig. 8b contains a Schottky pair where the most favored site for was adjacent to the (Y_Zr–v_O–Y_Zr)^× complex, with the oxygen vacancies associated with the remaining fully coordinated . The cell in Fig. 8c contains two , both of which were found to be most stable adjacent to (Y_Zr–v_O–Y_Zr)^×.

The total energy differences between the cells in Fig. 8b and c with respect to the reference cell in Fig. 8a were used to evaluate the enthalpies of reactions (1) and (2). Table 1 shows the calculated Gibbs energies for the reactions at 650 °C based on cells containing different amounts of defects as indicated by the degree of Ba-deficiency, x. While the obtained Gibbs energy of reaction (1) was positive for all the considered cells and configurations, it was significantly lower and close to zero for reaction (2). This indicates that reaction (2) involving formation of BaCO₃ by accommodating Ba-deficiency and Y-enrichment in the bulk perovskite structure – and hence consumption of BaZrO₃ – can occur to a significant extent at 650 °C in 1 atm CO₂.

Table 1 Thermodynamic parameters for reactions (1) and (2) at 650 °C showing the defects formed and consumed in the reactions, and the extent of Ba-deficiency (x) in the defective cells. The reaction enthalpies were obtained from DFT and the reaction entropies were estimated from the tabulated entropies of CO₂ and the solid phases

Reaction	Defects	x	ΔH/eV	TΔS/eV	ΔG/eV
1		0.040	0.29	−1.26	1.55
		0.070	0.40		1.66
		0.110	0.55		1.81
2		0.016	−1.89	−1.86	−0.03
		0.074	−1.92		−0.06

---

The tendency of yttrium to segregate to the surface was evaluated for BaO-terminated (0 0 1). The calculated segregation energy of was −0.33 eV, indicating a considerable stabilization of the defect at the surface, shown in Fig. S4 (ESI†). The segregation energy for an additional to the at the surface was −0.50 eV, resulting in an overall segregation energy of −0.82 eV for the associated defect pair from the bulk to the surface (see the ESI† for additional details).

5. Discussion

Observations by several techniques (SEM, TEM, and GIXRD) show that BaCO₃ crystallites form on the surfaces of BZ and BZY after annealing in 1 atm CO₂ at 650 °C. The amount of BaCO₃ increases with annealing time from 10 to 1000 h and it becomes larger for Y-doped compositions. The surface becomes Ba-deficient based on SIMS and SEM-EDS analysis of BZ and BZY, respectively (it is difficult to assess Ba-stoichiometry in BZY by SIMS since there is no constant reference due to simultaneous changes in the Y and Zr contents). The Ba-deficiency reaches several micrometers into the material, and it is likely that Ba is transported along grain boundaries.⁴⁶ Moreover, the corresponding carbon profile from SEM-EDS indicates BaCO₃ formation along grain boundaries and/or dissolution of carbon species into the perovskite.

The BZY s-urfaces become enriched in Y as indicated by SIMS and XPS, and the Y-enrichment appears to extend several micrometers into the material, i.e., similar to the corresponding deficiency in Ba. XPS shows two distinctly different chemical states of Y, which based on DFT calculations were identified as Y substituted on the A-site, , and B-site, , respectively. Furthermore, the DFT calculations indicate that these Y-species tend to cluster, both in the bulk and at the surface. The resulting local enrichment in Y becomes structurally similar to Y₂O₃ and may explain previous observations of Y-rich regions within BZY grains.^35–37 The change in stoichiometry near the surface is accompanied by a substantial increase in the lattice parameter of the remaining perovskite, as observed by TEM.

All in all, formation of BaCO₃ on BZY in CO₂ atmosphere at 650 °C appears to mainly proceed according to reaction (2), in which ends up as a charge compensating donor, , while a BaZrO₃ formula unit is consumed. The latter fits with the observation of small intragrain pores by TEM and the general absence of Y₂O₃ or ZrO₂ secondary phases. In comparison to BZ, where BaCO₃ formation is lower, the Y-dopant thereby provides an extra compositional degree of freedom that facilitates BaCO₃ formation according to reaction (2).

Now, the question is to what extent the formation of BaCO₃ by reaction (2) will proceed. Under the conditions used it is possible that it will proceed until complete decomposition to BaCO₃ and for instance Y-saturated ZrO₂ and Zr-saturated Y₂O₃. This process will take thousands of hours to complete, while larger BZY grains will probably retard the transport of Ba to the surface and hence also the decomposition and degradation. Another alternative is that it stops at equilibrium when a certain content of is reached. The observation that there is a certain Ba-deficiency and Y excess quite far into the material – instead of complete decomposition at the very surface – may support this alternative. These and other degradation paths could be studied by the reaction between for instance BaCO₃ and YSZ in 1 atm CO₂.

In any case, while we have shown that BZY suffers from severe BaCO₃ formation in 1 atm CO₂ and probably degradation of its functional and thermomechanical properties, many applications of BZY electrolytes may not be significantly affected. Lower CO₂ partial pressures or impurity levels of CO₂ should mitigate the degradation processes in hydrogen fuel cells and steam electrolyzers. However, such devices operating at steam pressures within the stability regime of Ba(OH)₂ may exhibit similar degradation phenomena and mechanisms as we have observed here.⁴⁷

A recent study on CO₂ and H₂O coadsorption has shown preferential adsorption of H₂O on BZY and indicated that the presence of H₂O prevents the formation of a carbonate overlayer on BZY surfaces.⁴⁸ For applications with high CO₂ partial pressures due reforming of hydrocarbons, the presence of steam may therefore limit the BaCO₃ formation and the associated degradation processes. Further degradation studies should therefore take into account both CO₂ and H₂O. Temperatures lower than 650 °C are desirable for many applications of BZY, which worsens the thermodynamics of stability but slows down diffusion, so the effect of temperature may go both ways.

6. Conclusions

Over a time-scale of 1000 h, BaZr_1−xY_xO_3−δ (x = 0–0.2) ceramics exhibit severe degradation under 1 atm CO₂ at 650 °C due to formation of BaCO₃ on the surfaces and possibly along the grain boundaries, accompanied by Ba-deficiency in the subsurface regions. For the Y-doped compositions, the surface and subsurface become enriched in Y, which may be ascribed to a degradation reaction in which Y ends up on the Ba-site while BaZrO₃ formula units are consumed. Accordingly, the degradation proceeds faster with Y as an additional compositional degree of freedom and a more favourable degradation reaction. The local enrichment of Y both in the bulk and on the surface attains a structure similar to Y₂O₃. Electrochemical devices based on Y-doped BaZrO₃ electrolytes may be stable towards BaCO₃ formation under lower CO₂ partial pressures and/or in the presence of H₂O, while further work is necessary to be conclusive.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Financial support from The Research Council of Norway under the program NANO2021 to the project “Functional oxides for clean energy technologies: fuel cells, gas separation membranes and electrolysers” (FOXCET, 228355) conducted by SINTEF, University of Oslo and The Norwegian University of Science and Technology (NTNU), is gratefully acknowledged in addition to FME NCCS (257579). Computational resources were provided by the Norwegian Metacenter for Computational Science (NOTUR) under the project nn9259k.

References

1. J. W. Phair and S. P. S. Badwal, Materials for separation membranes in hydrogen and oxygen production and future power generation, Sci. Technol. Adv. Mater., 2006, 7, 792–805.
2. D. A. Medvedev, J. G. Lyagaeva, E. V. Gorbova, A. K. Demin and P. Tsiakaras, Advanced materials for SOFC application: Strategies for the development of highly conductive and stable solid oxide proton electrolytes, Prog. Mater. Sci., 2016, 75, 38–79.
3. N. Kochetova, I. Animitsa, D. Medvedev, A. Demin and P. Tsiakaras, Recent activity in the development of proton-conducting oxides for high-temperature applications, RSC Adv., 2016, 6, 73222–73268.
4. J. A. Kilner and M. Burriel, Materials for Intermediate-Temperature Solid-Oxide Fuel Cells, Annu. Rev. Mater. Res., 2014, 44, 365–393.
5. J. D. Baniecki, M. Ishii, K. Kurihara, K. Yamanaka, T. Yano, K. Shinozaki, T. Imada, K. Nozaki and N. Kin, Photoemission and quantum chemical study of SrTiO₃(001)surfaces and their interaction with CO₂, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 195415.
6. J. LÜ, L. Wang, L. Fan, Y. Li, L. Dai and H. Guo, Chemical stability of doped BaCeO₃-BaZrO₃ solid solutions in different atmospheres, J. Rare Earths, 2008, 26, 505–510.
7. X. Ma, J. Dai, H. Zhang and D. E. Reisner, Protonic conductivity nanostructured ceramic film with improved resistance to carbon dioxide at elevated temperatures, Surf. Coat. Technol., 2005, 200, 1252–1258.
8. C. S. Tu, R. R. Chien, V. H. Schmidt, S. C. Lee, C. C. Huang and C. L. Tsai, Thermal stability of Ba(Zr_0.8−xCe_xY_0.2)O_2.9 ceramics in carbon dioxide, J. Appl. Phys., 2009, 105, 103504.
9. N. Zakowsky, S. Williamson and J. Irvine, Elaboration of CO₂ tolerance limits of BaCe_0.9Y_0.1O_3–δ electrolytes for fuel cells and other applications, Solid State Ionics, 2005, 176, 3019–3026.
10. D. Medvedev, J. Lyagaeva, S. Plaksin, A. Demin and P. Tsiakaras, Sulfur and carbon tolerance of BaCeO₃–BaZrO₃ proton-conducting materials, J. Power Sources, 2015, 273, 716–723.
11. A. Brandão, J. F. Monteiro, A. V. Kovalevsky, D. P. Fagg, V. V. Kharton and J. R. Frade, Guidelines for improving resistance to CO₂ of materials for solid state electrochemical systems, Solid State Ionics, 2011, 192, 16–20.
12. P. Duran, J. Tartaj and C. Moure, Sintering behaviour and microstructural evolution of agglomerated spherical particles of high-purity barium titanate, Ceram. Int., 2003, 29, 419–425.
13. B.-K. Yoon, E.-Y. Chin and S.-J. L. Kang, Dedensification During Sintering of BaTiO₃ Caused by the Decomposition of Residual BaCO₃, J. Am. Ceram. Soc., 2008, 91, 4121–4124.
14. I. Antunes, A. Brandão, F. M. Figueiredo, J. R. Frade, J. Gracio and D. P. Fagg, Mechanosynthesis of nanopowders of the proton-conducting electrolyte material Ba(Zr,Y)O_3−δ, J. Solid State Chem., 2009, 182, 2149–2156.
15. K. D. Kreuer, Proton-Conducting Oxides, Annu. Rev. Mater. Res., 2003, 33, 333–359.
16. H. Malerød-Fjeld, D. Clark, I. Yuste-Tirados, R. Zanón, D. Catalán-Martinez, D. Beeaff, S. H. Morejudo, P. K. Vestre, T. Norby, R. Haugsrud, J. M. Serra and C. Kjølseth, Thermo-electrochemical production of compressed hydrogen from methane with near-zero energy loss, Nat. Energy, 2017, 2, 923–931.
17. S. H. Morejudo, R. Zanon, S. Escolastico, I. Yuste-Tirados, H. Malerod-Fjeld, P. K. Vestre, W. G. Coors, A. Martinez, T. Norby, J. M. Serra and C. Kjolseth, Direct conversion of methane to aromatics in a catalytic co-ionic membrane reactor, Science, 2016, 353, 563–566.
18. D. Han, K. Kishida, K. Shinoda, H. Inui and T. Uda, A comprehensive understanding of structure and site occupancy of Y in Y-doped BaZrO₃, J. Mater. Chem. A, 2013, 1, 3027.
19. M. A. Laguna-Bercero, Recent advances in high temperature electrolysis using solid oxide fuel cells: a review, J. Power Sources, 2012, 203, 4–16.
20. D. Medvedev, A. Murashkina, E. Pikalova, A. Demin, A. Podias and P. Tsiakaras, BaCeO₃: Materials development, properties and application, Prog. Mater. Sci., 2014, 60, 72–129.
21. J.-W. Jhuang, K.-R. Lee, J.-k. Chang, C.-T. Shen, Y.-H. Lee, S.-W. Lee and C.-J. Tseng, Chemical stability and electrical and mechanical properties of BaZr_xCe_0.8−xY_0.2O₃ with CeO₂ protection method, Int. J. Hydrogen Energy, 2017, 42, 22259–22265.
22. F. Trobec and V. Thangadurai, Transformation of proton-conducting Perovskite-type into fluorite-type fast oxide ion electrolytes using a CO₂ capture technique and their electrical properties, Inorg. Chem., 2008, 47, 8972–8984.
23. A. L. Cabrera, F. Vargas and R. A. Zarate, Adsorption of carbon dioxide by barium titanate: Evidence of adsorption process mediated by a dipole-dipole interaction, J. Phys. Chem. Solids, 1994, 55, 1303–1307.
24. C. Duan, R. J. Kee, H. Zhu, C. Karakaya, Y. Chen, S. Ricote, A. Jarry, E. J. Crumlin, D. Hook, R. Braun, N. P. Sullivan and R. O'Hayre, Highly durable, coking and sulfur tolerant, fuel-flexible protonic ceramic fuel cells, Nature, 2018, 557, 217–222.
25. M. Kogler, E. M. Kock, T. Bielz, K. Pfaller, B. Klotzer, D. Schmidmair, L. Perfler and S. Penner, Hydrogen Surface Reactions and Adsorption Studied on Y₂O₃, YSZ, and ZrO₂, J. Phys. Chem. C, 2014, 118, 8435–8444.
26. E. M. Kock, M. Kogler, T. Bielz, B. Klotzer and S. Penner, In Situ FT-IR Spectroscopic Study of CO₂ and CO Adsorption on Y₂O₃, ZrO₂, and Yttria-Stabilized ZrO₂, J. Phys. Chem. C, 2013, 117, 17666–17673.
27. E. M. Kock, M. Kogler, T. Gotsch, B. Klotzer and S. Penner, Structural and chemical degradation mechanisms of pure YSZ and its components ZrO₂ and Y₂O₃ in carbon-rich fuel gases, Phys. Chem. Chem. Phys., 2016, 18, 14333–14349.
28. M. Kogler, E.-M. Köck, B. Klötzer, L. Perfler and S. Penner, Surface Reactivity of YSZ, Y₂O₃, and ZrO₂ toward CO, CO₂, and CH₄: A Comparative Discussion, J. Phys. Chem. C, 2016, 120, 3882–3898.
29. K. H. Ryu and S. M. Haile, Chemical stability and proton conductivity of doped BaCeO₃–BaZrO₃ solid solutions, Solid State Ionics, 1999, 125, 355–367.
30. E. Fabbri, A. D'Epifanio, E. Di Bartolomeo, S. Licoccia and E. Traversa, Tailoring the chemical stability of Ba(Ce_0.8−xZr_x)Y_0.2O_3−δ protonic conductors for Intermediate Temperature Solid Oxide Fuel Cells (IT-SOFCs), Solid State Ionics, 2008, 179, 558–564.
31. J. M. Polfus, B. Yildiz, H. L. Tuller and R. Bredesen, Adsorption of CO₂ and Facile Carbonate Formation on BaZrO₃ Surfaces, J. Phys. Chem. C, 2018, 122, 307–314.
32. R. Sažinas, C. Bernuy-López, M. A. Einarsrud and T. Grande, Effect of CO₂ Exposure on the Chemical Stability and Mechanical Properties of BaZrO₃-Ceramics, J. Am. Ceram. Soc., 2016, 99, 3685–3695.
33. F. A. Kröger and H. J. Vink, Relations between the Concentrations of Imperfections in Crystalline Solids, Solid State Phys., 1956, 3, 307–435.
34. I. Kaus, K. Wiik, B. Krogh, M. Dahle, K. H. Hofstad and S. Aasland, Stability of SrFeO₃-Based Materials in H₂O/CO₂-Containing Atmospheres at High Temperatures and Pressures, J. Am. Ceram. Soc., 2007, 90, 2226–2230.
35. Y. Yamazaki, R. Hernandez-Sanchez and S. M. Haile, Cation non-stoichiometry in yttrium-doped barium zirconate: phase behavior, microstructure, and proton conductivity, J. Mater. Chem., 2010, 20, 8158.
36. A. K. Azad, C. Savaniu, S. Tao, S. Duval, P. Holtappels, R. M. Ibberson and J. T. S. Irvine, Structural origins of the differing grain conductivity values in BaZr_0.9Y_0.1O_2.95 and indication of novel approach to counter defect association, J. Mater. Chem., 2008, 18, 3414.
37. F. Giannici, M. Shirpour, A. Longo, A. Martorana, R. Merkle and J. Maier, Long-Range and Short-Range Structure of Proton-Conducting Y: BaZrO₃, Chem. Mater., 2011, 23, 2994–3002.
38. R. Sažinas, M. A. Einarsrud and T. Grande, Toughening of Y-doped BaZrO₃ proton conducting electrolytes by hydration, J. Mater. Chem. A, 2017, 5, 5846–5857.
39. M. Chase, NIST-JANAF Thermochemical Tables, J. Phys. Chem. Ref. Data. Monograph, 4th edn, 1998, p. 1952.
40. K. Kurosaki, R. J. M. Konings, F. Wastin and S. Yamanaka, The low-temperature heat capacity and entropy of SrZrO₃ and BaZrO₃, J. Alloys Compd., 2006, 424, 1–3.
41. G. Kresse and D. Joubert, From ultrasoft pseudopotentials to the projector augmented-wave method, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
42. P. E. Blöchl, Projector augmented-wave method, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
43. J. P. Perdew, K. Burke and M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 77, 3865–3868.
44. H. J. Monkhorst and J. D. Pack, Special points for Brillouin-zone integrations, Phys. Rev. B: Condens. Matter Mater. Phys., 1976, 13, 5188–5192.
45. J. M. Polfus, M.-L. Fontaine, A. Thøgersen, M. Riktor, T. Norby and R. Bredesen, Solubility of transition metal interstitials in proton conducting BaZrO₃ and similar perovskite oxides, J. Mater. Chem. A, 2016, 4, 8105–8112.
46. R. Sažinas, I. Sakaguchi, I. Hasle, J. M. Polfus, R. Haugsrud, M. A. Einarsrud and T. Grande, Tracer diffusion of ⁹⁶Zr and ¹³⁴Ba in polycrystalline BaZrO₃, Phys. Chem. Chem. Phys., 2017, 19, 21878–21886.
47. A. Slodczyk, M. D. Sharp, S. Upasen, P. Colomban and J. A. Kilner, Combined bulk and surface analysis of the BaCe_0.5Zr_0.3Y_0.16Zn_0.04O_3−δ (BCZYZ) ceramic proton-conducting electrolyte, Solid State Ionics, 2014, 262, 870–874.
48. J. M. Polfus, J. Yang and B. Yildiz, Interplay between H₂O and CO₂ coadsorption and space-charge on Y-doped BaZrO₃ surfaces, J. Mater. Chem. A, 2018, 6, 24823–24830.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ta11021b

---