*2.4. CO2 Decomposition*

To the best of our knowledge, only two studies have reported CO2 decomposition in a continuous gas-flow reactor before our previous report [17]. One [11] provided TGA measurement results, and the other [15] provided extremely limited information. We have used several metal oxides for CO2 decomposition experiments in a continuous-flow reactor. In our previous report, we reported CO2 decomposition with SrFeCo0.5Ox using data obtained under nonisothermal conditions. Even if nonisothermal data are insufficient to cover all practical applications, they can serve as a cornerstone to determine the most economically efficient temperature region for CO2 decomposition. Furthermore, temperature fluctuates during both activation and decomposition processes. Considering these applications, we performed nonisothermal tests, isothermal tests, and cyclic experiments for CO2

decomposition with a noncobalt metal oxide, SrFeO3−δ. A comparison of nonisothermal data for SrFeCo0.5Ox and SrFeO3−<sup>δ</sup> is presented below.

Nonisothermal CO2 decomposition: Figure 2 shows a comparison of the results of nonisothermal CO2 decomposition using SrFeO3−<sup>δ</sup> and SrFeCo0.5Ox for temperatures ranging between 25 and 800 ◦C. Data for SrFeCo0.5Ox were extracted from our previous report [17], and the same experimental conditions were applied. Initially, we started CO2 decomposition with NiFe2O4 as an oxygen-deficient ferrite; however, it decomposed only up to 20% of CO2 in the continuous gas-flow system. We obtained a ~90% efficiency of CO2 decomposition using SrFeCo0.5Ox selected based on our proposed mechanism. Here, we demonstrated several enhanced CO2 decomposition results obtained by using SrFeO3−δ. First, SrFeO3−<sup>δ</sup> could be activated at a much lower temperature and for shorter duration. SrFeO3−<sup>δ</sup> was primarily activated at 280 ≤ T ≤ 600 ◦C, as shown in Figure 2a. The H2 concentration during reduction decreased rapidly up to ≈460 ◦C, which indicated the phase changes from perovskite to brownmillerite. This behavior can be confirmed from the in-situ XRD data shown in Figure 1. It is believed that oxygen vacancies are created the most at these temperatures. Second, the amount of CO2 decomposed using SrFeO3−<sup>δ</sup> is approximately 2.2 times higher than that decomposed using SrFeCo0.5Ox based on the calculated result of ≥50% CO2 decomposition, as shown in Figure 2b. As the temperature increased, the CO2 decomposition efficiencies of both metal oxides increased by up to ~90%. After reaching 800 ◦C, the CO2 decomposition efficiency of SrFeCo0.5Ox decreased, whereas the high decomposition efficiency of SrFeO3−<sup>δ</sup> was maintained over 100 min. This indicates that SrFeO3−<sup>δ</sup> might be a more appropriate material for CO2 decomposition than SrFeCo0.5Ox. The amount of CO produced using SrFeO3−<sup>δ</sup> was also slightly higher than that produced using SrFeCo0.5Ox.

**Figure 2.** CO2 decomposition results: (**a**) Consumed H2 concentration and (**b**) CO2 and CO concentrations during sample activation and CO2 decomposition tests, respectively. The black lines indicate the results of SrFeCo0.5Ox extracted from our previous work [17], and the straight dotted lines indicate temperature profiles (i.e., 3 ◦C/min).

The CO2 decomposition ability could be expressed in units of millimoles of decomposed CO2 and generated CO per gram of sample loaded (i.e., mmol g−1). This calculation was made using several assumptions. For example, the final decomposition was determined to be the point at which CO2 decomposition ceased, resulting in the revelation of the initial CO2 concentration. The final CO2 decomposition time was determined during the point at which SrFeO3−<sup>δ</sup> started to decompose at 200 ◦C and continued for over 4 h, even reaching 800 ◦C. This calculation was made using the decomposition time limit that ranged between 54 and 500 min (see Figure 2b). Secondly, in spite of nonisothermal CO2 decomposition, the ideal gas law was used to calculate the decomposed amount (i.e., mmol g<sup>−</sup>1) of input CO2. The exact same conditions were applied for the NiFe2O3−<sup>δ</sup> and SrFeCo0.5Ox samples. The results are summarized in Table 1. Both CO2 decomposition and CO generation using a perovskite (SrFeO3−<sup>δ</sup>) demonstrated enhanced performance compared to those using a spinel (NiFe2O3−δ) or a nonperovskite (SrFeCo0.5Ox). In addition, SrFeO3−<sup>δ</sup> is a cobalt-free compound that is economical and environmentally friendly. Generally, cobalt-containing metal oxides display good catalytic behavior but have several shortcomings, such as structural instability even at intermediate operating temperatures (500 to 800 ◦C) in a long-term test [34]. In the case of NiFe2O3−δ, other shortcomings were observed, such as too long an oxidation time.

**Table 1.** The rate of decomposed CO2 and generated CO in nonisothermal experiments.


Isothermal CO2 decomposition: For practical applications, data for CO2 decomposition using SrFeO3−<sup>δ</sup> at constant temperature should be determined. Figure 3 shows the isothermal results. The measurements were performed at 500, 600, 625, 650, 700, and 800 ◦C. The temperatures for sample activation and decomposition were identically controlled, and a blank test was also performed in the same reactor. GC was used to determine data points every ~4 min after switching the gas with 1 vol% CO2/He. Fresh powder samples were used in each measurement. As the temperature increased, the amounts of CO2 decomposition and CO production increased (see Figure 3a,b). This is probably attributable to the amount and high mobility of oxygen vacancies at higher temperatures. The ionic conductivities are proportional to the mobility of perovskite metal oxides (i.e., σ = n × e × μ, where σ is the specific conductivity; n, the number of charge carriers of a species; e, its charge; and μ, its mobility) and generally increases with the temperature [35].

**Figure 3.** Isothermal CO2 decomposition results obtained using SrFeO3−<sup>δ</sup> at various temperatures: (**a**) Decomposed CO2 concentration, (**b**) produced CO concentration, (**c**) CO2 conversion as a function of time, and (**d**) decomposed CO2 and produced CO area extracted from (**a**) and (**b**).

The CO2 decomposition results between 600 and 700 ◦C were noteworthy. As the operating temperature increased from 625 to 650 ◦C, the amount of CO2 decomposition doubled. We performed in-situ XRD and TGA experiments to analyze this unusual behavior in this temperature region (not shown). However, no special structural phase or weight changes were seen in the sample activation process. The amounts of consumed hydrogen for sample activation and cell parameters also demonstrated no considerable difference. The reason for the sudden increase in CO2 decomposition upon increasing temperature by only 25 ◦C remains unclear. We presume that the thermal energy at 650 ◦C might boost CO2 decomposition and the reverse Boudouard reaction (i.e., C(s) + CO2 → 2CO). The mobility increase caused by the thermal energy might be an important factor because other factors, such as the unit cell volume, oxygen ion vacancy concentration, and weight change from TGA, did not change abruptly. These issues are discussed further by comparing sample characteristics before and after performing measurements in Section 2.5.

Based on the obtained data, CO2 conversion rates were calculated using Equation (1).

$$\text{CO2 Consumption} \left(\% \right) = \frac{\text{CO2 In} - \text{CO2 Out}}{\text{CO2 In}} \times 100 \tag{1}$$

Although Figure 3c illustrates the same data in the same format as Figure 3a, the conversion degree is easier to distinguish from the CO2 conversion plot. Furthermore, ≥90% of CO2 conversion lasted for ≈65 min at 650 ◦C. This drastic change was much more evident from the area plots of the decomposed CO2 and produced CO shown in Figure 3d. We calculated these areas by subtracting those obtained in the isothermal blank tests. It should be noted that the area for CO2 (i.e., amount of CO2 decomposition) was unusually high at 650 ◦C. It was even slightly higher than that at 700 ◦C. Further, CO production

increased rapidly until the temperature was increased up to 800 ◦C. The shape of the isothermal CO2 decomposition curve at 650 ◦C also slightly differed from those of the others. We plan to analyze these behaviors using temperature-programed reduction and temperature-programed oxidation in a separate paper.
