*2.5. Stability Tests*

Irrespective of how high the efficiency is, samples should have reproducibility and long-term stability, especially in severe redox reactions. This is the most important criterion for practical applications. In this report, we performed CO2 decomposition in five reproducibility tests. We also performed cyclic CO2 decomposition tests with partially activated SrFeO3−δ; however, we did not include the data, owing to their overlapped content. After these stability tests, the changes in the structural behavior and surface morphology of the used SrFeO3−<sup>δ</sup> were investigated.

Figure 4 shows the results of the five cyclic reproducibility tests for CO2 decomposition using SrFeO3−<sup>δ</sup> at 700 ◦C. The sample was activated for 800 min in each cycle. The CO2 decomposition of each cycle was subtracted from the amount of the blank test. The measurement data of the last cycle (i.e., the fifth cycle) did not perfectly match those of the first. However, they were reasonably close, and the difference could be attributed to the annealing effect of lasting temperatures. Average amounts of decomposed CO2 and generated CO were determined to be 1.38 and 1.02 mmol per gram of catalyst, respectively. This result indicated a certain level of coke generation or possible adsorption of part of the carbon dioxide instead of decomposition. Although SrFeO3−<sup>δ</sup> demonstrated good reproducibility even after repeating the redox experiment several times, it still took too long to activate the sample. In these days, we are using coke oven gas from the steel industry for sample activation. As it contains 55% to 60% hydrogen, the activation time will be much faster. In addition, SrFeO3−<sup>δ</sup> impregnated with a small amount of precious metals such as Ru and Rh is tested for low-temperature CO2 decomposition. Table 2 summarizes the test results.

**Figure 4.** Five cyclic reproducibility tests for CO2 decomposition using SrFeO3−<sup>δ</sup> at 700 ◦C: (**a**) Decomposed CO2 concentration and (**b**) produced CO concentration.



To investigate the structural changes, XRD measurements were performed with the tested SrFeO3−<sup>δ</sup> powders. Figure 5a shows the sample used for the nonisothermal CO2 decomposition experiment at up to 800 ◦C. Figure 5b,d illustrate the powder patterns of SrFeO3−<sup>δ</sup> that underwent cyclic tests at 700 and 650 ◦C, respectively. The brownmillerite phase remained, and it was hard to find impurities, including even traces of Fe-metal peaks. This indicates that the redox reaction is reversible and that SrFeO3−<sup>δ</sup> can presumably serve as an excellent reactant. For comparison, the in-situ XRD result extracted from Figure 1 was added in Figure 5c. Three oxidized XRD patterns (i.e., Figure 5a,b,d) shifted to a higher 2θ angle; Figure 5c shows the pattern of the reduced sample. When the sample reduced, the oxygen vacancy concentration increased, resulting in its increased volume. The unit cell volume of reduced SrFeO3−<sup>δ</sup> at 700 ◦C was 504.1 Å3, and that of the oxidized sample decreased to 489.1 Å3. Table <sup>3</sup> lists the fully indexed unit cell parameters for SrFeO3−<sup>δ</sup> tested at 700 ◦C.

**Figure 5.** XRD powder patterns of SrFeO3−<sup>δ</sup> tested for CO2 decomposition measurements: (**a**) After nonisothermal test, (**b**) after five cycles at 700 ◦C, (**c**) in-situ XRD at 700 ◦C, and (**d**) after six cycles at 650 ◦C. The symbols indicate perovskite (p) and brownmillerite (b).

**Table 3.** Experimental conditions of nonisothermal and isothermal tests with SrFeO3−δ.


It should be noted that the powder pattern of the sample tested at 800 ◦C shows a mixed phase, that is, perovskite and brownmillerite, whereas the sample tested at ≤700 ◦C primarily shows a brownmillerite phase. This result is relevant to the increased amounts of decomposed CO2 at 800 ◦C (see Figure 3a). For a perovskite phase to exist at 800 ◦C, more oxygen vacancies need to be filled. This condition is induced by CO2 decomposition. It should also be noted that three oxidized samples were slightly reduced with N2 because 1 vol% CO2/He gas was switched with N2 after decomposition tests when cooling down to room temperature.

The microstructure of SrFeO3−<sup>δ</sup> was examined using SEM before and after CO2 decomposition experiments. Figure 6a shows a secondary electron image of pristine SrFeO3−δ, indicating that many small-sized (≤100 nm) particles are attached and dispersed on bigger ones. These small particles grew twice as large after the redox tests at 650 ◦C, and some large particles appeared to aggregate to each other (see Figure 6b). Even larger agglomerates developed and were observed in the sample tested at 700 ◦C, as shown in Figure 6c. Furthermore, the agglomerated particles (even ≥1 μm) displayed distinct grain boundaries. As the activity of samples is generally believed to depend on their surface area, a detailed microstructural investigation will be conducted in a separate study.

**Figure 6.** SEM images of SrFeO3−δ: (**a**) Before the redox test, (**b**) after six cycles of redox measurements at 650 ◦C, and (**c**) after five cycles of reproducibility tests at 700 ◦C.
