Investigation into the Effect of Sulfate and Borate Incorporation on the Structure and Properties of SrFeO_3-δ

Abstract

In this paper, we demonstrate the successful incorporation of sulfate and borate into SrFeO_3-δ, and characterise the effect on the structure and conductivity, with a view to possible utilisation as a cathode material in Solid Oxide Fuel Cells. The incorporation of low levels of sulfate/borate is sufficient to cause a change from a tetragonal to a cubic cell. Moreover, whereas heat treatment of undoped SrFeO_3-δ under N₂ leads to a transformation to brownmillerite Sr₂Fe₂O₅ with oxygen vacancy ordering, the sulfate/borate-doped samples remain cubic under the same conditions. Thus, sulfate/borate doping appears to be successful in introducing oxide ion vacancy disorder in this system.

1. Introduction

Research into solid oxide fuel cells (SOFCs) as alternate energy materials has grown due to their high efficiency and consequent reduction in greenhouse gas emissions. Specifically, research into perovskite materials has attracted significant interest for potential SOFC materials including cathode, electrolyte and anode materials (see, for example, the review articles [1,2]). Traditionally, doping strategies for perovskite materials has involved the introduction of cations of a similar size e.g., Sr²⁺ for La³⁺, Mg²⁺ for Ga³⁺ to give the oxide ion conducting electrolyte La_1−xSr_xGa_1−yMg_yO_{3−x/2−y/2} [3]. More recently, we have applied oxyanion (silicate, phosphate, sulfate, borate) doping strategies to improve the properties of solid oxide fuel cell materials, initially demonstrating the successful incorporation into Ba₂In₂O₅ [4]. Research on oxyanion incorporation into perovskites was initially reported for superconducting cuprate materials. This work showed that a wide range of oxyanions (carbonate, borate, nitrate, sulfate, and phosphate) could be incorporated into the perovskite structure [5,6,7,8,9,10,11,12]. In this doping strategy, the central ion of the oxyanion group is located on the perovskite B cation site, with the oxygens of this group occupying either three (borate, carbonate, nitrate) or four (sulfate, phosphate) of the available six anion positions around this site, albeit suitably displaced to achieve the correct coordination for the oxyanion group. As an extension to this work on cuprate superconductors, we have demonstrated successful oxyanion doping of perovskite materials for SOFC applications. This has included oxyanion (sulfate, silicate, phosphate) doping of Ba₂(In/Sc)₂O₅ electrolyte materials [13,14,15,16]. Doping with these oxyanions introduces disorder on the oxygen sublattice, leading to improvements in the ionic conductivity. This work has also been extended to potential solid oxide fuel cell electrode materials, with successful oxyanion doping in the perovskite systems SrCoO₃ [17,18,19,20], SrCo_0.85Fe_0.15O₃ [21], SrMnO₃ [18], CaMnO₃ [22] and SrFeO_3-δ [23], with the results suggesting improved performance/stability.

Following the prior work demonstrating the successful incorporation of silicate into SrFeO_3-δ [23], we report in this paper an investigation into the effect of sulfate and borate doping. The undoped system SrFeO_3-δ has attracted interest due to its high ionic and electronic conductivity, however under low p(O₂) the oxide ion (and electronic conductivity) are significantly reduced. This is due to oxygen loss leading to a transformation to the oxygen vacancy ordered brownmillerite structure, Sr₂Fe₂O₅ (Figure 1). We have therefore investigated the effect of borate/sulfate doping on this transformation.

2. Results and Discussion

2.1. SrFe_1−xS_xO_3-δ

2.1.1. X-ray Diffraction Data

A range of samples of SrFe_1−xS_xO_3-δ with increasing sulfate content (0 ≤ x ≤ 0.1) were prepared. X-ray diffraction analysis showed that without sulfate doping SrFeO_3-δ forms a tetragonal perovskite in line with prior reports. This is illustrated in Figure 2 (expanded region 2$\mathsf{\theta }$ = 45° to 60°) where peak splitting can clearly be observed. Upon doping with sulfate, there is a transformation to a cubic cell (0.025 ≤ x ≤ 0.075), where no peak splitting is now observed (Figure 2). Above x = 0.075, small SrSO₄ impurities appear, suggesting that the solubility limit of sulfate in SrFe_1−xS_xO_3-δ is x ≈ 0.075.

In order to provide further support for the incorporation of sulfate, equivalent Fe-deficient samples were prepared without addition of sulfate. The X-ray diffraction data for SrFe_0.95O_3-δ (Fe-deficient, no sulfate) and SrFe_0.95S_0.05O_3-δ are compared in Figure 3. These data show the presence of impurities for SrFe_0.95O_3-δ along with peak splitting, consistent with a tetragonal cell, as for the undoped SrFeO_3-δ parent phase. In contrast, impurities are not observed for the sulfate-containing phase SrFe_0.95S_0.05O_3-δ, and the cell is now cubic. Therefore, this comparison provides further evidence for the successful incorporation of sulfate.

Cell parameters for all these phases were determined using the Rietveld method (an example fit is shown in Figure 4). The variation of the cell parameters for SrFe_1−xS_xO_3-δ with increasing sulfate content is given in

Table 1. Lattice parameters and Fe/S site occupancies obtained from Rietveld refinement using XRD data for SrFe_1−xS_xO_3-δ. The structure of SrFeO_3-δ was refined using a tetragonal space group (P4/mmm). The structures of the sulfate-doped samples were refined using a cubic space group (Pm$\overline{3}$m).

SrFe1−xSxO3-δ
S (x)	0	0.025	0.05	0.075
a (Å)	3.8648(1)	3.8723(1)	3.8776(1)	3.8766(1)
c (Å)	3.8487(1)	-	-	-
V (Å3)	57.486(4)	58.066(2)	58.303(4)	58.260(4)
Rwp (%)	1.84	1.67	2.01	1.97
Rexp (%)	0.92	0.92	0.90	0.90
Fe occupancy	1	0.98(1)	0.96(1)	0.94(1)
S occupancy	-	0.02(1)	0.04(1)	0.06(1)
Fe/S Uiso	0.003(1)	0.009(1)	0.011(1)	0.008(1)

. The data show a small general increase with increasing sulfate content, and this will be discussed in more detail in Section 2.1.4.

In addition to the determination of the unit cell parameters, site occupancies were refined for the Fe/S site. These occupancies are given in Table 1, and show that the refined values are in good agreement with the expected values.

2.1.2. Stability under N₂

The effect of heating the SrFe_1−xS_xO_3-δ samples under N₂ was then examined. Upon heating under N₂ to 950 °C, the X-ray diffraction data showed that SrFeO_3-δ transforms to the oxygen vacancy ordered brownmillerite type Sr₂Fe₂O₅ (Figure 1 and Figure 5). This would be expected to be unfavorable for fuel cell applications due to the ordering of oxygen vacancies, which is expected to lower the oxide ion conductivity. In contrast, for the sulfate-doped samples, the disordered cubic perovskite is retained under a similar heat treatment. For the x = 0.025 sample, there are some very weak peaks (see expanded XRD figure) associated with the brownmillerite structure, but for the higher sulfate contents (x ≥ 0.05), a single phase cubic cell is observed. In line with the reduction in the Fe oxidation state towards 3+, there is an increase in the cell parameters associated with the larger size of Fe³⁺ versus Fe⁴⁺ (

Table 2. Lattice parameters for SrFe_1−xS_xO_3-δ after heating in air and N₂.

SrFe1−xSxO3-δ
S (x)	0.025		0.05		0.075
	Air	Dry N2	Air	Dry N2	Air	Dry N2
a (Å)	3.8723(1)	3.9231(1)	3.8776(1)	3.9256(1)	3.8766(1)	3.9280(1)
c (Å)	-	-	-	-	-	-
V (Å3)	58.066(2)	60.379(1)	58.303(4)	60.496(1)	58.260(4)	60.606(1)
Rwp (%)	1.67	3.10	2.01	3.09	1.97	3.20
Rexp (%)	0.92	2.59	0.90	2.51	0.90	2.50

).

2.1.3. Thermogravimetric Analysis

Thermogravimetric analysis (TGA) (heat treatment in N₂ to reduce the Fe oxidation state to 3+) was then utilized to determine the oxygen contents of the samples as prepared in air. This analysis resulted in an interesting observation associated with Mass spectrometry analysis of the evolved gas. For the undoped sample SrFeO_3-δ, the loss of mass is only associated with oxygen as indicated by the mass spectrometry data. However, for the sulfate-doped samples, a mass loss associated with CO₂ was observed in addition to the expected mass loss due to O₂.

Thus, the results indicated the presence of some carbonate in the sulfate-doped samples. In order to remove this carbonate, heat treatment in O₂ (up to 900 °C) was carried out for these SrFe_1−xS_xO_3-δ samples. After this oxygen treatment, TGA analysis indicated no presence of carbonate. This is illustrated in Figure 6, where a mass loss associated with CO₂ is not observed after heat treatment in O₂ for SrFe_0.95S_0.05O_3-δ.

The TGA results indicating the presence of carbonate may at first glance suggest the existence of a small amount of SrCO₃ impurity. However, the temperature at which the CO₂ is lost is significantly lower than would be expected for SrCO₃. In order to illustrate this, TGA data for SrCO₃ were also collected and compared to the data for the SrFe_0.95S_0.05O_3-δ. This experiment shows a significant difference in the temperature at which the loss of CO₂ occurs for SrFe_0.95S_0.05O_3-δ and SrCO₃. In particular, the starting temperature for CO₂ loss for SrFe_0.95S_0.05O_3-δ occurs at a significantly lower temperature (≈490 °C vs. ≈760 °C for SrCO₃), which may suggest that this carbonate is present in the perovskite structure, i.e., we have a mixed sulfate/carbonate-doped sample—SrFe_1−x−yS_xC_yO_3-δ. Further work is required to investigate this possibility, although as noted in the introduction, carbonate has been shown previously to be accommodated in perovskite materials.

Given the dual (O₂ and CO₂) mass loss for the air-synthesized SrFe_1−xS_xO_3-δ samples, it is not possible to determine reliable oxygen contents for these samples. In addition, as detailed by Starkov et al., the determination of oxygen contents in partially substituted ferrites is non-trivial without a reliable fixed reference point [24]. Since it is possible that there may still be a small amount of Fe⁴⁺ in these samples after the N₂ treatment, there is not a conclusive fixed reference oxygen point, and so any calculated oxygen contents would only be rough approximations.

2.1.4. Heat Treatment under O₂

The samples (described above) heated under O₂ were also examined by X-ray diffraction (Figure 7). All the sulfate-doped SrFe_1−xS_xO_3-δ samples were shown to retain their original cubic structure while undoped SrFeO_3-δ remained tetragonal.

Cell parameters and site occupancies were determined by Rietveld refinement using the X-ray diffraction data. From these structure refinements (

Table 3. Lattice parameters and Fe/S site occupancies for SrFe_1−xS_xO_3-δ, obtained from Rietveld refinement using XRD data for SrFe_1−xS_xO_3-δ after heating in O₂.

SrFe1−xSxO3-δ Heated in O2
S (x)	0	0.025	0.05	0.075
a (Å)	3.8651(1)	3.8641(1)	3.8692(1)	3.8691(1)
c (Å)	3.8477(1)	-	-	-
V (Å3)	57.349(3)	57.694(2)	57.924(2)	57.922(3)
Rwp (%)	4.16	3.29	3.79	3.93
Rexp (%)	3.71	2.81	2.73	2.79
Fe occ	1(-)	0.97(2)	0.94(2)	0.93(2)
S occ	-	0.03(2)	0.06(2)	0.07(2)

), there is a decrease in the unit cell parameters upon heating in O₂. This can be more clearly seen in Figure 8 where the variation in cell volume versus sulfate content is compared for samples heated in O₂ and those just heated in air. These data show a reduction in cell volume on heating in oxygen, which can be correlated with a greater concentration of the smaller Fe⁴⁺ as a result of an increase in oxygen content. The cell parameter data also show an interesting variation with sulfate content, with an approximately linear increase in cell volume up to x = 0.05. The fact that there is no further increase for x = 0.075 may therefore suggest that the sulfate solubility limit is closer to x = 0.05.

The increase in cell volume with increasing sulfate content is at first glance unexpected given that S⁶⁺ is significantly smaller than Fe^3+/4+. One possible explanation could relate to changes in the Fe³⁺/Fe⁴⁺ ratio on sulfate incorporation. However, this cell volume increase is also seen for the N₂ treated samples (Table 2), where we only have Fe³⁺, and thus another factor must be significant. In this respect, the cell volume increase may be associated with the extra oxygen associated with this dopant. Thus, if we take the case of the N₂ treated samples, we are effectively replacing Fe(III)O_1.5 with SO₃, which means the introduction of an extra 1.5 oxide ions per sulfate, which might be expected to contribute to an expanded cell size.

2.1.5. Conductivity Data

Following the successful incorporation of sulfate, the conductivities of the SrFe_1−xS_xO_3-δ samples were examined in air. In general, the SrFe_1−xS_xO_3-δ samples were found to have similar conductivities, with the exception of a notable decrease at lower temperatures for the higher sulfate content (x = 0.075) sample (Figure 9). This observation of a decrease in conductivity for higher dopant levels is comparable to the silicon-doped SrFeO_3-δ system where it was proposed that at higher doping levels the silicate disrupts the Fe-O network resulting in a decrease in conductivity. [23]. Another factor, however, could relate to low levels of insulating impurities in this high sulfate content sample, given that the cell parameter data suggest that the sulfate solubility limit may be closer to x = 0.05.

2.2. SrFe_1−xB_xO_3-δ

2.2.1. X-ray Diffraction Data

The possible incorporation of borate into SrFeO_3-δ was then examined, with samples of SrFe_1−xB_xO_3-δ prepared for 0 ≤ x ≤ 0.15. The results showed that a higher borate content was achievable compared to the sulfate-doped samples, with single phase borate-doped samples for x ≤ 0.1 (Figure 10). In terms of the effect of borate incorporation on the structure, similar results were observed as for the SrFe_1−xS_xO_3-δ samples. In particular, borate doping resulted in a similar cubic cell as for the sulfate-doped samples.

Cell parameters and site occupancies were determined using the Rietveld method (an example fit is shown in Figure 11).

The refinements gave Fe/B occupancies in good agreement with those expected (

Table 4. Lattice parameters and Fe/B site occupancies obtained from Rietveld analysis using X-ray diffraction data for SrFe_1−xB_xO_3-δ. The structure of SrFeO_3-δ was refined using a tetragonal space group (P4/mmm). The structures of the doped samples were refined using a cubic space group (Pm$\overline{3}$m).

SrFe1−xBxO3-δ
B (x)	0	0.05	0.1
a (Å)	3.8648(1)	3.8593(1)	3.8561(1)
c (Å)	3.8487(1)	-	-
V (Å3)	57.486(4)	57.483(2)	57.336(4)
Rwp (%)	1.84	3.23	3.67
Rexp (%)	0.92	2.30	2.52
Fe occupancy	1	0.92(1)	0.89(1)
B occupancy	-	0.08(1)	0.11(1)
Fe/B Uiso	0.003(1)	0.015(1)	0.022(1)

). Furthermore, in this case, a decrease in the unit cell was observed on borate doping in agreement with the smaller size of B³⁺ compared to Fe^3+/4+, and the fact that unlike the situation for sulfate doping there is no additional oxygen associated with the dopant (i.e., we are effectively replacing Fe(III)O_1.5 with BO_1.5).

2.2.2. Stability under N₂

Heat treatment of the SrFe_1−xB_xO_3-δ samples at 950 °C under N₂ showed similar results to those observed on sulfate doping, with the borate-containing samples remaining cubic after this heat treatment in nitrogen (Figure 12). In line with the reduction of the Fe oxidation state to Fe³⁺, there was a shift in the peak positions to lower angles, indicating a larger unit cell.

2.2.3. Thermogravimetric Analysis

These samples were then analysed by TGA (heat treatment under N₂). In contrast to the air synthesised SrFe_1−xS_xO_3-δ samples, there was no mass loss due to CO₂ observed in these SrFe_1−xB_xO_3-δ systems, thus indicating no carbonate present.

2.2.4. Conductivity Data

For the SrFe_1−xB_xO_3-δ samples (x = 0.05, 0.1), the conductivity data showed significantly lower conductivities at lower temperatures compared with the undoped system (Figure 13). However, above 600 °C, the conductivities were comparable with the x = 0.1 sample, in particular showing a small improvement in the conductivity compared with undoped SrFeO_3-δ. Notably, these temperatures are in the range where operation as a cathode in a solid oxide fuel cell would be.

Conductivity values (at 700 °C) for the borate- and sulfate-doped samples are shown in

Table 5. Conductivity data in air at 700 °C for SrFe_1−xM_xO_3-δ where M = Si [23], S and B.

	Si (x)				S (x)				B (x)
	0	0.05	0.1	0.15	0	0.025	0.05	0.075	0.05	0.1
Conductivity 700 °C (S cm−1)	26	21	35	18	26	25	30	25	30	32

, and compared to the equivalent data for silicate-doped SrFeO_3-δ (from reference [23]). The data show similar values for all samples at this typical solid oxide fuel cell operating temperature.

3. Experimental

High-purity SrCO₃, Fe₂O₃, (NH₄)₂SO₄, and H₃BO₃ were used to prepare SrFe_1−xS/B_xO_3-δ samples. Stoichiometric mixtures of the powders were intimately ground and initially heated to 900 °C (4 °C/min) for 12 h. Samples were then ballmilled (350 rpm for 1 h, Fritsch Pulverisette 7 planetary Mill) and reheated to 1000, 1050 and 1100 °C for 12 h with ballmilling of samples between heat treatments. For the SrFe_1−xB_xO_3-δ (x = 0.05, 0.1) samples, a higher temperature (1200 °C) heat treatment was required to achieve single phase samples. In order to ensure maximum oxygen content, all samples underwent a final heat treatment at 350 °C for 12 h in air.

Additionally, portions of the SrFe_1−xS_xO_3-δ (x = 0, 0.025, 0.05, 0.075) samples were heated to 900 °C under oxygen for 12 h with slow cooling at 50 °C /h to 350 °C, with the samples then maintained at this temperature for 12 h followed by cooling at 50 °C/h to room temperature. To test the stability under low p(O₂) conditions, both sulfate and borate-doped samples were heated under N₂ to 950 °C for 12 h.

Powder X-ray diffraction data were used in order to determine lattice parameters and phase purity. For SrFe_1−xS_xO_3-δ samples heated in air, X-ray diffraction data were collected on a Panalytical Empyrean diffractometer equipped with a Pixcel 2D detector (Cu K$\alpha$ radiation). For the remaining SrFe_1−xS_xO_3-δ and SrFe_1−xB_xO_3-δ samples, a Bruker D8 diffractometer with Cu K${\alpha }_1$ radiation was used.

Samples were also analysed using thermogravimetric analysis (Netzch STA 449 F1 Jupiter Thermal Analyser with mass spectrometry attachment). Samples were heated to 950 °C in N₂ (10 °C/min) and held for 30 min to reduce the iron oxidation state to Fe³⁺.

Pellets for conductivity measurements were prepared as follows: powders of SrFe_1−xS_xO_3-δ and SrFe_1−xB_xO_3-δ heated in air were initially ball milled (350 rpm for 1 h), before pressing into compacts and sintering at 1100 °C (SrFe_1−xS_xO_3-δ) and 1200 °C (SrFe_1−xB_xO_3-δ) for 12 h in air. Four Pt electrodes were attached with Pt paste and the samples were heated at 900 °C for 1 h in air. Samples were then furnace cooled to 350 °C and held at this temperature for 12 h to ensure full oxygenation. Conductivities were measured using the four probe dc method.

4. Conclusions

The results presented in this paper demonstrate the first reports of successful incorporation of sulfate and borate into SrFeO_3-δ. This doping strategy results in a change from a tetragonal to a cubic cell, which is maintained even after heating under N₂, where undoped SrFeO_3-δ transforms to the oxygen vacancy ordered brownmillerite structure. Conductivity data in air show that the borate/sulfate-doped samples have comparable conductivities (Table 5) to undoped SrFeO_3-δ at solid oxide fuel cell operating temperatures. Given these initial promising results, further studies are warranted to investigate the performance of these doped systems as solid oxide fuel cell cathodes.