Redox Performance and Optimization of the Chemical Composition of Lanthanum–Strontium–Manganese-Based Perovskite Oxide for Two-Step Thermochemical CO₂ Splitting

Abstract

The effects of substitution at the A- and B sites on the redox performance of a series of lanthanum–strontium–manganese (LSM)-based perovskite oxides (Z = Ni, Co, and Mg) were studied for application in a two-step thermochemical CO₂ splitting cycle to produce liquid fuel from synthesis gas using concentrated solar radiation as the proposed energy source and CO₂ recovered from the atmosphere as the prospective chemical source. The redox reactivity, stoichiometry of oxygen/CO production, and optimum chemical composition of Ni-, Co-, and Mg-substituted LSM perovskites were investigated to enhance oxygen/CO productivity. Furthermore, the long-term thermal stabilities and thermochemical repeatabilities of the oxides were evaluated and compared with previous data. The valence changes in the constituent ionic species of the perovskite oxides were studied and evaluated by X-ray photoelectron spectroscopy (XPS) for each step of the thermochemical cycle. From the perspectives of high redox reactivity, stoichiometric oxygen/CO production, and thermally stable repeatability in long-term thermochemical cycling, Ni_0.20-, Co_0.35-, and Mg_0.125-substituted La_0.7Sr_0.3Mn perovskite oxides are the most promising materials among the LSM perovskite oxides for two-step thermochemical CO₂ splitting, showing CO productivities of 387–533 μmol/g and time-averaged CO productivities of 12.9–18.0 μmol/(min·g) compared with those of LSM perovskites reported in the literature.

1. Introduction

Thermochemical cycles employ two or more endothermic and exothermic reactions to dissociate water and/or CO₂, requiring lower temperatures compared to the thermolysis processes (H₂O → H₂ + 1/2O₂, CO₂ → CO + 1/2O₂). The two-step thermochemical cycle generally proceeds via redox reactions:

MO_ox + heat → MO_red + 1/2O₂
(endothermic thermal reduction (TR) step)

(1)

MO_red + CO₂ (H₂O) → MO_ox + CO (H₂) + heat
(exothermic CO₂ splitting (CS) or H₂O splitting (WS))

(2)

where MO_ox and MO_red correspond to the oxidative and reductive forms of reactive metal oxides [1]. In the first step, the metal oxide is thermally reduced at the high temperatures generated by concentrated solar radiation, releasing oxygen (Equation (1)). In the second step, the reductive form of the metal oxide (MO_red) reacts with CO₂ (or H₂O) to produce CO (or H₂), regenerating the initial material (MO_ox) in Equation (2). Concentrated solar radiation is used as a sustainable thermal energy source to drive the thermochemical cycle, whereas water and CO₂ are used as chemical sources. The combination of a thermochemical splitting cycle and renewable solar radiation is a promising solution for producing storable and transportable liquid chemical fuels (e.g., ethanol) using an industrially practical Fischer–Tropsch (FT) process. The two-step thermochemical cycle is simple; however, the temperature required for this process is extremely high. A significant challenge for this cycle is the development of reactive materials that can dissociate CO₂ (or H₂O) with a high rate and productivity while affording thermally stable performance during repeated cycling. The reactive materials for use in solar-driven thermochemical cycles must satisfy the following conditions: decrease the TR temperature (T < 1400 °C); afford high production of CO (or H₂) during the CS (or WS) step; promote fast kinetics of both steps; exhibit superior thermal stability and repeatability with low cost and safety to the environment and humans; and must be prepared from readily available and abundant resources.

All possible candidate materials for use in the two-step thermochemical cycle can be categorized as follows: volatile systems (ZnO/Zn, CdO/Cd, and SnO₂/SnO) [2,3,4], non-volatile systems (Fe₃O₄/FeO, Mn₃O₄/MnO, and mixed metal oxide) [3,5,6,7,8], nonstoichiometric ceria systems (fluorite-based structures of CeO₂/CeO_2-δ and substituted -CeO₂) [9,10,11,12,13,14], and nonstoichiometric perovskite systems (ABO₃/ABO_3-δ). In this research field, nonstoichiometric ceria is recognized as a current state-of-the-art redox material, and solar fuel production over 500 cycles has been demonstrated using nonstoichiometric ceria in a solar cavity receiver reactor [15,16]. To date, nonstoichiometric ceria systems are one of the most promising materials for the two-step thermochemical cycle because of their favorable properties, such as long-term thermal stability, high oxygen storage capacity, resistance to sintering, fast kinetics in both steps of the thermochemical cycle, and superior redox reactivity [17,18,19,20]. Other potential candidates for the two-step thermochemical cycle are nonstoichiometric perovskite (ABO₃/ABO_3-δ) systems with a wide range of oxygen nonstoichiometry (δ). δ governs fuel production through an extensive range of variable nonstoichiometry, which induces the formation of lattice defects (oxygen vacancies) in the crystal structure over various operating conditions. Both cationic sites (A- and B sites) in the crystal structure of perovskite oxides can be substituted to control the oxygen nonstoichiometry, kinetics, and thermodynamic properties over a wide range of chemical compositions. Unlike nonstoichiometric ceria, most perovskite oxides require long duration of the splitting step owing to their slow kinetics compared to the TR step [21]. Thus, either a temperature swing between the TR and splitting steps or an excess CO₂ (or H₂O) gas flow is generally required to enhance the extent and rate of CO (or H₂) production [22].

Exploratory research on perovskite oxides is generally conducted to enhance the O₂/CO productivity, optimize the range of reduction and oxidation, and reduce the temperature of the TR step and temperature swings between the steps while improving the kinetics. To improve reaction kinetics of perovskite materials, several studies on material molding to increase the specific surface area of the materials were performed [23,24]. Recently, owing to the improvement in the redox reactivity of both steps, as well as their high thermal durability and reliability, the possibility of ionic substitution in the crystal structure and nonstoichiometric perovskite systems have drawn the attention of the scientific community. Previous studies have extensively studied the effects of ion substitution in several perovskites with La or Sr at the A site and Mn at the B site for the CO₂ (or H₂O) thermochemical cycle [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47]. The results of test conditions and O₂ and CO (or H₂) productivities for the perovskites related to the present study are summarized in

Table 1. Comparison of redox performances for the perovskites related to the present study. O₂ and CO productivities, reaction conditions, and equipment are summarized.

Material	Synthesis Method	Reactor	Run No.	Reduction Temp. [°C]	Reduction Duration [min]	Average O2 Productivity [μmol/g]	Partial Pressure of CO2 (Steam) [ber]	CS (WS) Temp. [°C]	CS (WS) Duration [min]	Average CO (H2) Productivity [μmol/g]	CO(H2)/O2 Ratio [-]	Reference
La0.5Sr0.5MnO3	Pechini	TGA	4	1400	90	143	0.5	1200	60	248	1.73	[53]
La0.7Sr0.3Mn0.9Mg0.1O3	Pechini	TGA	4	1400	90	197	0.5	1200	60	354	1.8	[53]
La0.7Sr0.3Mn0.9Co0.1O3	Pechini	TGA	4	1400	90	184	0.5	1200	60	347	1.89	[53]
La0.7Sr0.3Mn0.9Ni0.1O3	Pechini	TGA	4	1400	90	188	0.5	1200	60	351	1.87	[53]
La0.7Sr0.3Mn0.9Cr0.1O3	Pechini	Fixed bed	3	1350	30	0	−0.84	−1200	−50	−129	−0.15	[21]
La0.7Sr0.3Mn0.9Cr0.1O3	Pechini	Fixed bed	3	1350	30	98	1	1200	50	215	2.19	[47]
La0.5Sr0.5Mn0.83Mg0.17O3	Solid-state	TGA	2	1400	45	170	0.5	1050	60	208	1.22	[31]
La0.6Sr0.4Al0.6Mn0.4O3	Pechini	TGA	1	1350	45	60	0.4	1000	60	286	-	[36]
La0.5Sr0.5Mn0.95Sc0.05O3	Pechini	TGA	3	1400	45	323	0.4	1100	45	506	1.57	[39]
La0.5Sr0.5Mn0.75Ga0.25O3	Pechini	TGA	3	1400	45	264	0.4	1100	45	460	1.74	[39]
La0.5Sr0.5Mn0.9Mg0.1O3	Pechini	TGA	2	1400	45	190	0.5	1050	60	215	1.13	[32]
La0.5Sr0.5Mn0.6Al0.4O3	Pechini	TGA	2	1400	45	129	0.5	1050	60	205	1.59	[32]
La0.6Sr0.4Co0.8Cr0.2O3	Pechini	TGA	3	1200	-	-	0.5	800	60	169	-	[40]
La0.6Sr0.4Co0.8Cr0.2O3	Pechini	TGA	3	1200	-	-	0.5	800	60	-50	-	[40]
La0.5Ca0.5Mn0.8Ga0.2O3	Pechini	TGA	2	1400	45	231	0.5	1050	60	208	0.9	[37]

. As shown in Table 1, most investigations on thermochemical CO₂ splitting have been implemented using a thermogravimetric reactor. There were some limited partial substitutions in the A- or B sites of perovskite oxides [48,49,50]. In addition, the difference in the test conditions complicates the comparison with the literature data. Recently, the chemical composition of perovskites has been explored based on computational chemistry using density functional theory [51,52]. Insight into the redox performance based on computational chemistry enables the prediction of the optimum chemical composition and redox performance of perovskites in the thermochemical cycle. However, validation of these insights using extensive experimental data is required. Extensive experimental analysis of the substitution strategy can complement and support the theoretical data for controlling redox performance and selecting the chemical composition and structure of perovskites.

Recently, the authors reported that Ni, Co, and Mg substitutions at the B site in a series of LaSrMn-based perovskites (La_0.7Sr_0.3Mn_0.9X_0.1O₃ (X = Mg, Al, Cr, Fe, Co, Ni, Cu, and Ga)) enhanced the productivity and rate of CO production and maintained stoichiometric production (CO/O₂ = 2) with high thermal durability over 50 cycles [53]. Based on the screening test, Mg, Co, and Ni substitutions were selected in the current study to improve the O₂ and CO productivities and production behavior of the respective steps in the CO₂ thermochemical cycle using LSM perovskites by identifying the optimum chemical composition by A- and B-site substitution and to evaluate the redox performance in comparison with previous literature. In terms of the productivity and repeatability of O₂ and CO release, the time-relevant production behavior of the respective steps and chemical compositions of the A- and B sites of La_1−xSr_xMn_1−yZ_yO₃ (Z = Ni, Co, or Mg) were experimentally investigated, evaluated, and optimized for the thermochemical cycle. Moreover, the optimal samples for achieving repeatability, reliability of oxygen and CO production, and thermal stability were examined using long-term redox tests. To study and evaluate the role of the substituted ions in the thermochemical cycle, the best-performing samples were analyzed by XPS and compared with nonsubstituted LSM. Finally, the CO productivities per gram of material and the duration of the CS step were compared with the literature data. This result shows that the Co35%- and Ni20%-substituted LSMs afforded the highest level of CO production among the LSM-based perovskites, which has never been previously reported.

2. Materials and Methods

2.1. Material Preparation

La_1−xSr_xMn_1−yZ_yO₃ (Z = Ni, Co, or Mg) perovskite oxides (LS_xMZ_y) were synthesized using a modified Pechini method [54]. The chemical compositions and sample names of the synthesized materials are listed in Table 1. The following reagents were used to prepare the LS_xMZ_y samples: La(NO₃)₃·6H₂O (Wako, 99.9% purity), Sr(NO₃)₂ (Wako, ≥98.0% purity), Mn(NO₃)₃·6H₂O (Wako, ≥98.0% purity), Mg(NO₃)₂·6H₂O (Wako, ≥98.0% purity), Co(NO₃)₂·6H₂O (Kanto Chemical, ≥98% purity), Ni(NO₃)₃·6H₂O (Wako, 99.9% purity), citric acid (Wako, ≥98.0% purity), and ethylene glycol (Wako, ≥99.5% purity). The metal cations, citric acid, and ethylene glycol were dissolved in distilled water in a molar ratio of 1:5:5. These solutions were stirred and heated at 80 °C for 1 h in an oil bath, and then heated at 170 °C to evaporate the water and form a gel. The obtained gel was dried at 300 °C for 5 h in an electric oven. The resulting precursor was pulverized using a mortar and pestle and calcined at 1200 °C for 8 h. Finally, powder-like LS_xMZ_y samples were prepared. For comparison, cerium oxide (High Purity Chemicals, Saitama, Japan, 99.9% purity) was prepared to evaluate its redox performance.

2.2. Characterization of Samples

Powder X-ray diffraction (XRD) analysis was performed using a Bruker D2 PHASER with Cu-K_α radiation (λ = 0.15418 nm, 30 kV–10 mA) at room temperature to identify the crystallographic phase of the as-prepared samples and the samples after the redox performance test. Powder diffraction data were collected in the 2θ range of 20°–80° with a step interval of 0.02° and a recording time of 1 s. The crystalline phases were identified by comparison with standard reference databases (Inorganic Crystal Structure Database (ICSD) and Crystallography Open Database (COD)). Rietveld refinement was performed using the whole-pattern fitting (WPF) method in the FullProf package to evaluate the space groups, crystal systems, and lattice parameters of the LS_xMZ_y samples. Scanning electron microscopy (SEM, JCM-6000, JEOL, Tokyo, Japan) with an acceleration voltage of 15 kV was used to observe the morphology and grain size of the powder samples before and after the long-term repeatability tests.

2.3. Redox Reactivity of Samples

The redox reactivity of the LS_xMZ_y samples was evaluated by analyzing the oxygen (O₂)/carbon monoxide (CO) productivity and production rates of O₂ and CO for the two-step thermochemical CO₂ splitting cycling. Reactivity was examined using a thermogravimetric (TG) reactor (NETZSCH STA2500 Regulus, weight resolution of 0.03 μg, temperature resolution of 0.3 K) equipped with a type S thermocouple (temperature resolution of ±0.0025 × |t| °C) for the TR and subsequent exothermic CS steps of the sample. The influences (general drift and amount of sample) of the thermogravimetric reactor used in this study on the redox reactivity were described and discussed in detail in our previous study [53]. The redox performance of the samples was evaluated using the following procedure: approximately 50 mg of the sample was packed into a platinum pan and placed on the balance of the reactor. First, 99.999% high-purity nitrogen gas at a flow rate of 100 cm³/min at standard state (unit of sccm) was passed through the reactor for 75 min at a flow rate of 100 cm³/min to purge residual gases in the reactor. To eliminate adsorbed gaseous species on the surface of the sample, the sample was preliminarily heated at 300 °C for 0.5 h at a rate of 50 K/min while purging with nitrogen gas. The sample was subjected to the TR step at 1400 °C for 1.5 h to thermally release oxygen from the sample. The sample temperature was monitored using an S-type thermocouple in contact with the bottom of the platinum pan. The sample was cooled to 1000–1200 °C at a rate of 20 K/min under N₂ flow. The purge gas was switched to a gas mixture with an N₂ flow rate of 50 sccm and a CO₂ flow rate of 50 sccm to perform the CS step. The samples were maintained at the temperature for 0.5 h in the reactor to complete the CS step. The TR and CS steps of the thermochemical cycle were repeated thrice. The weight changes of the samples were recorded as a function of time to observe the fractional extent of the TR and CS steps during the thermochemical cycle.

The long-term thermal stability and repeatability of the samples were evaluated using the following procedures. The TR step was performed at 1400 °C at an N₂ flow rate of 100 sccm for 10 min, and the subsequent CS step was conducted at 1200 °C at a gas mixture flow rate of 100 sccm (CO₂ 50% and N₂ 50%) for 15 min. The TR and CS steps were repeated 50 times to evaluate the repeatability of the O₂ release and CO production. The O₂ and CO productivities per unit sample weight (n_O2 and n_CO [mol/g]) were calculated using the molar weight of the oxygen atom (M_O [g/mol]), the weight change of the sample (${∆m}_{TR}$, ${∆m}_{CS}$ [g]) during the TR and CS steps, and initial weight of the sample (m_s [g]). The weight changes of the samples during the TR and CS steps were calculated from the weight decreases and increases, respectively. The O₂ and CO production rates per unit weight (r_O2 and r_CO [mol/(min·g)]) were calculated from the time derivative ($\raisebox{1ex}{$dm$}\!\left/ \!\raisebox{-1ex}{$dt$}\right.$) of weight $m$ [g] at time t = t, as shown in Equations (3)–(6).

$$n_{O2}=\frac{{∆m}_{TR}}{2\times M_O\times m_s}$$

(3)

$$n_{CO}=\frac{{∆m}_{CS}}{M_O\times m_s}$$

(4)

$$r_{O2}=-\frac{\raisebox{1ex}{$dm$}\!\left/ \!\raisebox{-1ex}{$dt$}\right.}{2\times M_O\times m_s}$$

(5)

$$r_{CO}=\frac{\raisebox{1ex}{$dm$}\!\left/ \!\raisebox{-1ex}{$dt$}\right.}{M_O\times m_s}$$

(6)

X-ray photoelectron spectroscopy (XPS) was performed using a Quantum 2000 instrument (ULVAC-PHI. Inc., Kanagawa, Japan) with an Al anode (Al-Kα, 25 W (15 kV, 1.67 mA), hν = 1486.6 eV) to evaluate the valence states of the elements. For energy calibration of the XPS spectrum, the peaks were normalized relative to the C1s peak of adventitious carbon at a binding energy of 284.8 eV. The Gaussian function for peak fitting, Shirley method [55] for the baseline, and Savitzky–Golay method [56] for data smoothing were used to analyze the XPS data.

CO₂-to-CO conversion ratio [%] was calculated as a function of the duration of the CS step as follows:

$$X_{CO2}=\frac{{\int }_0^t{r}_{CO}dt·m_s}{{\int }_0^t{F}_{CO2}dt}$$

(7)

where $F_{CO2}$ is the flow rate of CO₂ during the CS step. The results for the best samples were compared with those of cerium oxide.

3. Results and Discussion

3.1. Thermochemical Redox Performance of La_0.7Sr_0.3 Mn_1−yZ_yO₃ (Z = Ni, Co, and Mg)

The chemical compositions, lattice parameters, space groups, and crystal systems of all prepared samples of La_1−xSr_xMn_1−yZ_yO₃ (LS_xMZ_y) are listed in

Table 2. Abbreviation, chemical composition, lattice parameter, space group, and crystal system of refined structures of samples prepared in this study.

Name	Chemical Composition	Lattice Parameters [Å]			Space Group	Crystal System
		a	b	c
LSM	La0.7Sr0.3MnO3	5.508(2)	5.508(2)	13.368(2)	R-3c(167)	trigonal
LSMMg0.05	La0.67Sr0.28Mn0.95Mg0.05O3	5.490(9)	5.490(9)	13.379(5)	R-3c(167)	trigonal
	La0.5Sr1.5MnO4	3.865(7)	3.865(7)	12.348(3)	I4/mmm(139)	tetragonal
	La2O3	3.932(6)	3.932(6)	6.122(3)	P63/mmc(194)	hexagonal
LSMMg0.10	La0.7Sr0.3Mn0.90Mg0.10O3	5.499(7)	5.499(7)	13.345(3)	R-3c(167)	trigonal
LSMMg0.125	La0.7Sr0.3Mn0.875Mg0.125O3	5.502(5)	5.502(5)	13.358(6)	R-3c(167)	trigonal
LSMMg0.15	La0.7Sr0.3Mn0.85Mg0.15O3	5.525(2)	5.525(2)	13.347(9)	R-3c(167)	trigonal
LSMCo0.05	La0.7Sr0.3Mn0.95Co0.05O3	5.507(0)	5.507(0)	13.359(2)	R-3c(167)	trigonal
LSMCo0.10	La0.7Sr0.3Mn0.90Co0.10O3	5.504(4)	5.504(4)	13.348(3)	R-3c(167)	trigonal
LSMCo0.15	La0.7Sr0.3Mn0.85Co0.15O3	5.501(6)	5.501(6)	13.339(3)	R-3c(167)	trigonal
LSMCo0.20	La0.7Sr0.3Mn0.80Co0.20O3	5.494(7)	5.494(7)	13.323(3)	R-3c(167)	trigonal
LSMCo0.25	La0.7Sr0.3Mn0.75Co0.25O3	5.499(5)	5.499(5)	13.334(4)	R-3c(167)	trigonal
LSMCo0.30	La0.7Sr0.3Mn0.70Co0.30O3	5.486(3)	5.486(3)	13.308(9)	R-3c(167)	trigonal
LSMCo0.35	La0.7Sr0.3Mn0.65Co0.35O3	5.479(9)	5.479(9)	13.295(7)	R-3c(167)	trigonal
LSMCo0.40	La0.7Sr0.3Mn0.60Co0.40O3	5.474(2)	5.474(2)	13.282(7)	R-3c(167)	trigonal
LSMCo0.50	La0.7Sr0.3Mn0.50Co0.50O3	5.466(3)	5.466(3)	13.267(3)	R-3c(167)	trigonal
LSMNi0.05	La0.7Sr0.3Mn0.95Ni0.05O3	5.503(3)	5.503(3)	13.347(9)	R-3c(167)	trigonal
LSMNi0.10	La0.7Sr0.3Mn0.90Ni0.10O3	5.497(6)	5.497(6)	13.337(4)	R-3c(167)	trigonal
LSMNi0.15	La0.7Sr0.3Mn0.85Ni0.15O3	5.492(8)	5.492(8)	13.325(7)	R-3c(167)	trigonal
LSMNi0.20	La0.7Sr0.3Mn0.80Ni0.20O3	5.485(0)	5.485(0)	13.307(9)	R-3c(167)	trigonal
LSMNi0.25	La0.7Sr0.3Mn0.75Ni0.25O3	5.478(9)	5.478(9)	13.299(8)	R-3c(167)	trigonal
LSMNi0.30	La0.7Sr0.3Mn0.70Ni0.30O3	5.475(5)	5.475(5)	13.289(8)	R-3c(167)	trigonal

. Almost all the prepared samples comprised a single solid phase indexed to the trigonal unit cell (space group R-3c (167)), whereas the LSMMg0.05 sample was composed of three solid phases, including a main phase with space group R-3c (167). Figure 1 shows the XRD patterns of (a) Mg-, (b) Co-, and (c) Ni-substituted LSMs at room temperature in the evaluated 2θ range. As shown in Figure 1a, the XRD pattern of LSMMg0.125 is identical to that of the crystalline phase, and the peaks are indexed to the reference La_0.7Sr_0.3Mn_0.85Mg_0.15O₃ (ICSD code: 258933). This result indicates that Mg ions were incorporated into the B site of the crystal structure and formed a solid solution without byproducts. However, the pattern of LSMMg0.05 showed that some Mg ions were substituted at the A site of the crystal structure (La_0.67Sr_0.28Mg_0.05MnO₃ (ICSD code:254217)), resulting in the observation of the La₂O₃ phase (COD data number:2002286) and La_0.5Sr_1.5MnO₄ (COD data number:1008251). As shown in Figure 1b,c, the patterns of the Co and Ni-substituted LSMs are in good agreement with those of the nonsubstituted LSM. These results indicate that Co and Ni ions were substituted into the B site of the crystal structure as a solid solution within the substitution fraction range. To verify the formation of a solid solution in the substituted LSM samples, the cell parameters of all prepared samples were determined by Rietveld refinement, and the cell volume of the samples was calculated based on the crystal structure.

Figure 1d shows the relationship between the estimated cell volume and the substitution fraction (y) of La_0.7Sr_0.3Mn_1-yZ_yO₃ (Z = Mg, Ni, and Co). In the series of Ni-substituted LSMs, the cell volume decreased as the substitution fraction increased, resulting in an almost linear downward trend. One primary reason for this downtrend is the intermediate value of the ionic radius of trivalent Mn (Mn³⁺:0.65 Å, six coordination by O²⁻ ion at the high spin) between the ionic radius of divalent and trivalent Ni (Ni²⁺:0.70 Å, Ni³⁺:0.60 Å, six coordination by O²⁻ ion in the high spin state) [57] in the B site of the crystal structure. These results indicate that trivalent Ni ions, together with bivalent Ni ions, were incorporated into the B site of the Ni-substituted LSMs as the substitution fraction increased. Another possibility is that the average valence of the Mn ions in the B site increased as the fraction of Ni substitution increased, inducing a decrease in the cell volume. For the Co-substituted LSMs series (Figure 1d), the cell volume of the LSMCo0.05 sample was higher than that of the nonsubstituted LSM. These results indicated that divalent Co ions (ionic radius of Co²: 0.735 Å, six coordinated by O²⁻ ions in the high-spin state) [57] were incorporated into the B site of the Co-substituted LSMs. As the fraction of Co substitution increased, the cell volume decreased in the fraction range of y = 0.1–0.5. The resulting downtrend indicates that divalent, trivalent, or tetravalent Co ions (ionic radius of Co³⁺: 0.61 Å, Co⁴⁺: 0.53 Å, six coordination by O²⁻ ion in the high spin state) [57] are incorporated into the B site instead of Mn ions. The different tendency at y = 0.3 may be owing to the irregular change in the fraction of multivalent ions (Co²⁺/Co³/Co⁴) in the LSMCo0.30 sample. For the Mg-substituted LSMs series (Figure 1d), the cell volume of the LSMMg0.10 sample was smaller than that of the nonsubstituted LSM. Because the ionic radius of divalent Mg ions (Mg²⁺: 0.720 Å, six coordination by O²⁻ ion) [57] is larger than that of Mn³⁺ ions, the average valence of the Mn ions at the B site increases at y = 0.10. When the substitution fraction in the Mg-substituted LSM series (y = 0.10–0.15) was increased, the cell volume monotonically increased. The results indicate that the average valence of the Mn ions decreased at the B site of the Mg-substituted LSMs. Therefore, maintaining charge valence in the crystal structure is difficult with Mg substitution at y > 0.15. These results indicate that all the samples, except LSMMg0.05, were successfully prepared as a solid solution of the single-crystalline phase of La_0.7Sr_0.3Mn_1−yZ_yO₃ without impurities.

Figure 2 shows the average O₂ and CO productivities of Ni-, Co-, and Mg-substituted LSMs. Each productivity value was estimated based on the extent of the weight change. The average productivity was calculated from the second and third runs of the two-step thermochemical CO₂ splitting cycling. The value from the first run was excluded from the calculation because the data may have been affected by the sample preparation process. For comparison, nonsubstituted LSM was also evaluated to verify the impact of partial substitution at the B site. As shown in Figure 2a, the O₂ productivity of the Mg-substituted LSMs increased with increasing substitution fraction (y) and reached a maximum value of 44.3 mmol/mol-material at y = 0.10 (LSMMg0.10). The CO productivity also increased with the substitution fraction and was maximal (86.1 mmol/mol-material) at y = 0.125 (LSMMg0.125). These results indicate that the partial substitution of Mg at the B site of LSM enhances the O₂ and CO productivities. The CO/O₂ (CO/O₂ = 2.04) value at y = 0.125 is approximately equal to the stoichiometric ratio. The results in Figure 2a indicate that, from the viewpoint of productivity, LSMMg0.125 can provide stable and reproducible O₂ and CO production with maximal redox performance. The results obtained for the Co-substituted LSM perovskites are shown in Figure 2b. Maximum O₂ and CO productivities were obtained at y = 0.350 and CO/O₂ = 1.93. For the Ni-substituted LSM perovskites (Figure 2c), the O₂ and CO productivities were maximized at y = 0.200, with CO/O₂ = 1.89. These results indicate that the partial substitution of the B site for all ion species (Mg, Co, and Ni) improved the O₂ and CO productivities while maintaining stoichiometric reactivity. The most suitable chemical composition depends on the substitution species used.

Figure 3 shows the O₂ production profiles of the Ni, Co, and Mg-substituted LSMs during the TR step in the first to third runs. As a reference, the results obtained with the non-substituted LSM are shown in the figure. In all runs, the production rate of O₂ reached a maximum value when the sample was heated at a temperature of 1400 °C. Figure 3a shows that LSMMg0.125 (green curve) afforded stable and repeatable O₂ production without degradation during thermochemical cycling, unlike the other Mg-substituted samples. The O₂ production of LSMMg0.150 (red curve) was poor and lower than that of the nonsubstituted LSM. Figure 3b shows the O₂ production profiles of LSMCo0.20 (blue curve), LSMCo0.35 (green curve), and LSMCo0.50 (red curve). LSMCo0.35 exhibited the most stable and highest O₂ production among the Co-substituted LSMs, whereas, with LSMCo0.50, the peak rate of O₂ production declined with an increase in the number of runs of thermochemical cycling. The O₂ production profiles of the Ni-substituted LSM are shown in Figure 3c. LSMNi 0.20 (green curve) demonstrated a stable and reproducible O₂ production profile throughout the second and third runs and exhibited the highest peak rates of O₂ production.

Figure 4a shows the CO production profiles of the Mg-substituted LSMs series during the subsequent CS steps in the first to third runs. For all the Mg-substituted samples, CO production was immediately initiated at the target temperature under a gas mixture of N₂ and CO₂ flow. The peak rate of CO production for LSMMg0.05 was higher in all runs compared to that of the nonsubstituted LSM but decreased for LSMMg0.15. The results for the Co-substituted LSMs series are shown in Figure 4b. The peak rate of CO production for LSMCo0.35 reached the maximum value but decreased for the samples with 40–50% substitution. One main reason for the improvement in the CO productivity with Co-substitution (Figure 2b) is that the duration of CO production is extended for all substitution levels compared to the case with nonsubstitution. For the Ni-substituted LSMs series (Figure 4c), the peak CO production rate was higher with 5–20% substitution but decreased with 25–35% substitution. Prolonged duration of CO production was observed for all the Ni-substituted LSMs series. These results indicate that the Co- and Ni-substituted LSMs demonstrate higher O₂ and CO productivities and superior gas releasing profiles compared to the nonsubstituted LSM. Notably, from the viewpoint of the release behavior and productivity of O₂ and CO in the thermochemical cycle, the LSMCo0.35 and LSMNi0.20 samples are promising materials among the substituted LSMs series investigated in this study.

The superior CO release behavior of Co- and Ni-substituted LSMs enables an increase in solar-to-fuel efficiency under the same reaction conditions due to the enhancement of cycle-averaged and time-averaged CO productivity. As another approach for improvement in solar-to-fuel efficiency, a reticulated porous ceramic structure [23] or nanostructuring methods [24] with higher specific surface area were investigated in the literature.

3.2. Repeatability and Thermal Durability of LSMCo0.35 and LSMNi0.20 in Continuous Multiple Cycling of Thermochemical Two-Step CO₂ Splitting

In this section, the repeatability and thermal durability of the promising LSMCo0.35 and LSMNi0.20 samples are examined and evaluated. Figure 5 shows the O₂ and CO productivities in the long-term continuous test for the thermochemical two-step CO₂ splitting cycling using LSMCo0.35 and LSMNi0.20. To monitor and evaluate the stoichiometric reactivity of the samples in each run, the CO/O₂ ratio was calculated and plotted in the figure. The results of the first run may include nonreproducible effects of the sample and reactor preparations (adsorption of moisture and gaseous species on the surface of the sample and crucible in the thermogravimetric reactor, and volatile/combustible contaminations). Thus, the results of the second–fiftieth runs were evaluated to estimate thermal durability and stability. As shown in Figure 5a, the decline fraction of CO productivity during the continuous thermochemical cycling was 1.0%. Thus, O₂ and CO production were maintained at the same level without degradation, significant effusion, or vaporization of the sample over 50 continuous runs. These results indicate that LSMCo0.35 is thermally durable during long-term thermochemical cycling at high temperatures and undergoes reproducible redox reactions. In the case of LSMNi0.20 (Figure 5b), the decline fraction of CO productivity was estimated to be 3.7%. The LSMNi0.20 and LSMCo0.35 samples showed stable and reproducible O₂ and CO productivity over multiple continuous cycles while maintaining stoichiometric reactivity. The production rates of O₂ and CO were determined for the LSMCo0.35 (Figure 5c) and LSMNi0.20 (Figure 5d) samples. Both the samples exhibited stable and reproducible O₂ and CO release profiles during continuous thermochemical cycling. The crystallographic phases of LSMCo0.35 and LSMNi0.20 samples were observed and compared to analyze the crystal structure and morphology. Figure 6 shows the XRD patterns of the samples obtained before and after continuous testing. The peak positions of the sample after the test were the same as those of the as-prepared sample and corresponded to the same crystalline phases (red vertical bars below the XRD pattern). No secondary phases or phase separation were observed in either pattern, indicating that the LSMCo0.35 and LSMNi0.20 samples underwent repeatable and thermally durable thermochemical cycling without crystal structure degradation. Figure 7 shows secondary electron image (SEI) micrographs of LSMCo0.35 (Figure 7a–e) and LSMNi0.20 (Figure 7f–i). For the as-prepared LSMCo0.35 sample (Figure 7a–c), the particle size was 0.1–0.2 µm, and agglomerations of the non-porous fine particles appeared. After the long-term test (Figure 7d,e), a coagulated mass composed of coarse particles with sizes of 2–10 µm was observed. Thus, the as-prepared samples were severely sintered at high temperatures during the continuous testing. For the as-prepared LSMNi0.20 sample (Figure 7f,g), the particle size was 0.2–0.5 µm, and agglomerations of particles were apparent. After the long-term test (Figure 7h,i), the observed coagulated mass was smaller than that of LSMCo0.35. These results indicate that Ni substitution may reduce the agglomeration of the LSM particles subjected to high temperatures during continuous testing. Both LSMCo0.35 and LSMNi0.20 were sintered at high temperatures during the continuous testing. However, the decline fractions remained at low values, and a multivalence state of Co and Ni existed in the crystal lattice of LSMCo0.35 and LSMNi0.20. Thus, it may be caused by the enhancement of oxygen transfer in the crystal lattice by Co and Ni substitution.

3.3. Impact of La/Sr Substitution Fraction in a Site on the Redox Activity of Substituted LSM

In this section, we evaluate the impact of the La/Sr substitution fraction (x) in the A site of the substituted LSM on the O₂ and CO productivities during thermochemical cycling. LSMNi0.20, in which the chemical composition of the B site was fixed, was selected as the representative LSM sample. The A-site-substituted samples (La_1−xSr_xMn_0.8Ni_0.2O₃) were subjected to the TR step at 1400 °C for 90 min. Subsequently, the CS step was performed at 1000 °C for 30 min. Figure 8 shows the weight change profiles of the samples (x = 0–0.8) during the thermochemical cycling (first–fifth runs). Except for the first run, repeatable and reproducible profiles were observed for all the samples. Figure 9 shows the O₂ and CO productivities and the CO/O₂ ratios for various x values. The O₂ and CO productivities reached their highest values of 49.3 and 96.5 mmol/mol-material at x = 0.3. In addition, the O₂/CO ratio of 1.95 at x = 0.3 means that the two-step thermochemical CO₂ splitting proceeds stoichiometrically, retaining the greatest productivity among all x fractions. To optimize the chemical composition of the A site and maximize the redox reactivity, La_0.7Sr_0.3Mn_0.8Ni_0.2O₃ is a feasible chemical composition for the thermochemical two-step CO₂ splitting cycle.

The impact of the La/Sr fraction on the O₂ and CO productivities of the Ni-substituted LSM indicates the following two possibilities: (1) an initial increase in the substitution fraction (x = 0–0.4) leads to an increase in the number of lattice defects (vacancies) in the crystal structure. The results indicated that the O₂ productivity was enhanced owing to the improved oxygen mobility in the crystal structure at high temperatures in both steps. However, (2) not all vacancies induced by Sr substitution are available for the redox reaction in the thermochemical cycle (x = 0.4–0.8). The XRD patterns of the as-prepared samples (La_1−xSr_xMn_0.8Ni_0.2O₃) are shown in Figures S1–S6 in the Supplementary Information. The space group and crystal structure of the identified phases are presented in

Table 3. Space group and crystal structure of the as-prepared La_1-xSr_xMn_0.8Ni_0.2O₃.

		Main Phase		Second Phase		Third Phase
La/Sr Fractions (x)	Chemical Composition	Space Group	Crystal Structure	Space Group	Crystal Structure	Space Group	Crystal Structure
0	LaMn0.8Ni0.2O3	R-3c(167)	trigonal	-	-	-	-
0.2	La0.8Sr0.2Mn0.8Ni0.2O3	R3m(160)	trigonal	-	-	-	-
0.3	La0.7Sr0.3Mn0.8Ni0.2O3	R-3c(167)	trigonal	-	-	-	-
0.4	La0.6Sr0.4Mn0.8Ni0.2O3	R-3c(167)	trigonal	-	-	-	-
0.6	La0.4Sr0.6Mn0.8Ni0.2O3	Pm-3m(221)	cubic	-	-	-	-
0.8	La0.2Sr0.8Mn0.8Ni0.2O3	Pm-3m(221)	cubic	I4/mmm(139)	tetragonal	Fm-3m(225)	cubic

. The experimental findings revealed that the crystal structure (cubic system) of the primary phase in samples with a high degree of substitution (x = 0.6 and 0.8) was different from that of a trigonal system with less substitution (x = 0–0.4). This result indicates that the excess vacancies are accommodated in the trigonal system as x increases (x = 0–0.4). However, the phase transition stabilizes the crystal structure by accommodating excess vacancies when the value of x increases further (x = 0.6–0.8). Consequently, the acceptable number of vacancies in the cubic system may decrease compared to that in the trigonal system. The space groups and crystal structures of the samples obtained after the redox tests are listed in

Table 4. Space group and crystal structure of La_1-xSr_xMn_0.8Ni_0.2O₃ obtained after the redox reactivity test.

	Main Phase		Secondary Phase
La/Sr Fractions (x)	Space Group	Crystal Structure	Space Group	Crystal Structure
0	Pnma(62)	orthorhombic	-	-
0.2	R3m(160)	trigonal	-	-
0.3	R-3c(167)	trigonal	-	-
0.4	R-3c(167)	trigonal	-	-
0.6	R-3c(167)	trigonal	I4/mmm(139)	tetragonal
0.8	Pm-3m(221)	cubic	I4/mmm(139)	tetragonal

. Phase separation was observed at chemical compositions of x = 0.6 and 0.8. These results indicated that they were thermally or chemically unstable during thermochemical cycling. The threshold value of a stable crystal structure without phase transition and separation before and after thermochemical cycling was approximately x = 0.5.

3.4. XPS Analysis of Nonsubstituted-, Ni-, and Co-Substituted LSMs

The changes in the valence states of the constituent ionic species in the substituted LSMs were analyzed and evaluated to elucidate the functions of each ionic species in each step of the two-step thermochemical CO₂ splitting. LSMNi0.20 and LSMCo0.35 were selected as the samples with the highest reactivity among the substituted LSMs series tested in the present study. Nonsubstituted LSM was analyzed and compared. Figure 10a shows the La 3d_3/2 X-ray photoelectron spectra with peak deconvolutions for nonsubstituted LSM. Triplet peaks centered at 850.57, 853.03, and 854.99 eV are observed for the as-prepared sample. No evident changes in the peak positions and areas were observed for the samples obtained after the TR and subsequent CS steps in the two-step thermochemical cycle. The results indicate that the La valence in the nonsubstituted LSM did not vary at each step of the thermochemical cycle. Figure 10b shows the Mn 2p_3/2 spectrum with peak deconvolution for nonsubstituted LSM. For the as-prepared sample, the Mn 2p_3/2 spectrum was decomposed into triplet peaks (Mn²⁺ = 640.67 eV, Mn³⁺ = 641.80 eV, and Mn⁴⁺ 643.53 eV). The average Mn valences were estimated from the peak areas corresponding to the Mn²⁺/Mn³⁺/Mn⁴⁺ species. The estimated average valences of Mn in the nonsubstituted LSM are listed in

Table 5. Peak position, ratio of peak area, and valance state of Mn for the as-prepared non-substituted LSM, thermally reduced sample obtained after the TR step, and the oxidized sample obtained after the subsequent CS step. The values were estimated from Mn 2p_3/2 X-ray photoelectron spectra.

	Peak Position [eV]			Ratio of Peak Area [%]			Valence State of Mn
Sample	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)
As preparation	640.67	641.8	643.53	42.2	31.7	26.1	2.84
After TR	640.67	641.79	643.6	57.4	31.0	11.6	2.54
After CS	640.63	641.76	643.68	50.1	32.4	17.5	2.67

. The variation in the average valence of Mn indicates that the Mn ions at the B site of the nonsubstituted LSM are responsible for the redox reaction in the thermochemical cycle.

The deconvoluted Co 2p_3/2 and Mn 2p_3/2 XPS profiles of the LSMCo0.35 sample are shown in Figure 11. The Co 2p_3/2 XPS profile of the as-prepared sample fitted the quadruplet peaks centered at 778.99, 779.98, 780.98, and 782.00 eV, respectively (Figure 11a). Each peak centered at 778.99 and 779.98 eV corresponded to the Co³⁺ and Co²⁺ signal, respectively. According to the literature [58], the XPS profile of Co₃O₄ shows a peak centered at 779.6 eV, assigned to trivalent cobalt, whereas the peaks centered at 780.1 (strongest), 782.1, 785.5, and 786.5 eV were attributed to bivalent cobalt. The XPS results of LiCoO₂ showed a peak centered at 779.6 eV in the Co 2p_3/2 spectrum, attributed to trivalent cobalt. This value agrees with the previously reported data [59], in which the peak centered at 781 eV in the Co 2p_3/2 spectrum of Li_0.66CoO₂ was attributed to tetravalent cobalt. This value corresponds to the current results (tetravalent cobalt). Finally, it was proposed that the other peak-centered redox reaction occurred in LSMCo0.35 during the thermochemical two-step process. In contrast, 782.00 eV corresponded to the satellite peak of the Co²⁺ signal. Figure 11b shows the deconvoluted Mn 2p_3/2 XPS profile of LSMCo0.35. The triplet peaks at 640.50, 641.87, and 643.50 eV for the as-prepared sample were attributed to Mn²⁺, Mn³⁺, and Mn⁴⁺, respectively. The binding energy peaks of the as-prepared samples agreed with those reported in the literature [60,61,62]. The valence states of the Mn and Co ions in the LSMCo0.35 sample were calculated and estimated from the peak positions and areas of the signals for the Mn and Co ions. As shown in

Table 6. Peak position, ratio of peak area, and valance state of Mn for the as-prepared LSMCo0.35, thermally reduced sample obtained after the TR step, and the oxidized sample obtained after the subsequent CS step. The values were estimated from Mn 2p_3/2 X-ray photoelectron spectra.

	Peak Position [eV]			Ratio of Peak Area [%]			Valence State of Mn
Sample	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)
As preparation	640.5	641.87	643.5	51.2	26.8	22.1	2.71
After TR	640.58	642.24	643.58	58.4	15.8	25.8	2.67
After CS	640.65	641.9	643.56	21.4	65.1	13.5	2.92

, the valence states of Mn decreased after the TR step, indicating the release of oxygen from the sample, and increased after the subsequent CS step owing to oxygen absorption via CO₂ splitting. The Mn 2p_3/2 XPS data indicate that the valence states of the Mn ions contributed to the redox reaction in LSMCo0.35 during the thermochemical two-step process. In contrast to the variation in the valence of Mn, the valence states of Co (

Table 7. Peak position, ratio of peak area, and valance state of Co for the as-prepared LSMCo0.35, thermally reduced sample obtained after the TR step, and the oxidized sample obtained after the subsequent CS step. The values were estimated from Co 2p_3/2 X-ray photoelectron spectra.

	Peak Position [eV]			Ratio of Peak Area [%]			Valence State of Co
Sample	Co 2p3/2 (2+)	Co 2p3/2 (3+)	Co 2p3/2 (4+)	Co 2p3/2 (2+)	Co 2p3/2 (3+)	Co 2p3/2 (4+)
As preparation	779.98	778.99	780.98	28.7	50.6	20.7	2.92
After TR	780.01	778.98	781.04	21.6	58.2	20.2	2.99
After CS	780.04	779.02	781.00	31.6	53.3	15.0	2.83

) either did not change in either step or changed negligibly, the change having no relation to the redox reaction. These results indicate that the Co ions at the B site of the LSMCo0.35 sample may promote the redox reaction of Mn ions in the thermochemical cycle. Compared to the valence change of the Mn ions in the nonsubstituted sample, the valence change increased in the LSMCo0.35 sample. The result of the LSMCo0.35 sample agreed well with the result of the 10%Mg-substituted sample without a valence change in Mg in a previous study [53].

Figure 12a shows the deconvoluted La 3d_3/2 and Ni 2P_3/2 X-ray photoelectron spectra of the as-prepared LSMNi0.20 after TR and after subsequent CS steps in the thermochemical cycle. For the as-prepared LSMNi0.20, the triplet peaks centered at 850.60, 852.50, and 854.98 eV can be assigned to the La 3d_3/2 main and satellite peaks. Triplet La 3d_3/2 peaks have previously been reported for LaFe_xO_3-δ (0.7 ≤ x ≤ 1.3) perovskites [63]. According to the literature, the main peak in the La 3d_3/2 spectrum appears at a binding energy of ~850 eV, whereas satellite peaks are present between 850 and 855 eV. These results are in agreement with those of the current study. Thus, the La 3d_3/2 XPS data indicate that lanthanum ions were present in trivalent form in the as-prepared LSMNi0.20. The binding energies of the main and satellite peaks in the La 3d_3/2 region were not significantly different for the as-prepared LSMNi0.20 and samples obtained after the TR and subsequent CS steps. These results indicate that the valence state of La did not change during either step of the thermochemical cycle. For the double perovskite Nd₂NiMnO₆, the Ni 2p XPS profile shows the most prominent peak in the vicinity of 850–860 eV [64]. The Ni 2p_3/2 peak appeared at lower binding energies and higher intensities than the Ni 2p_1/2 peak [65]. The Ni 2p_3/2 peaks for the La_0.8Sr_0.2Mn_1−xNi_xO₃ nanoparticles were observed together with the La 3d_3/2 peaks, and the former were separated into doublet peaks centered at ~854 and 855.5 eV. The low and high binding energies correspond to the divalent and trivalent ions. The peak position of the doublet did not change with Ni substitution levels of 20% or 40% [66]. In the case of the porous Ni (II) hydroxide nanosheets supported on carbon paper [65], doublet peaks due to divalent and trivalent Ni appeared at low and high binding energies, respectively, in the Ni 2p_3/2 spectra. As shown in Figure 12a, the Ni 2p_3/2 XPS profile of the as-prepared LSMNi0.20 shows doublet peaks centered at 853.98 and 855.60 eV. The low and high binding energies correspond to the divalent and trivalent ions. These peak positions are in good agreement with those reported previously [66]. The results in Figure 12a indicate that the as-prepared LSMNi0.20 contained trivalent and divalent Ni ionic species at the B site of the perovskite. For the samples obtained after the TR and subsequent CS steps, negligible differences (853.95–854.05 eV for Ni² and 855.54–855.60 eV for Ni³⁺) in the binding energies were observed. The corresponding Ni valences were calculated from the peak areas of the Ni²⁺ and Ni³⁺ curves. The results are summarized in

Table 8. Peak position, ratio of peak area, and valance state of Ni for the as-prepared LSMNi0.20, thermally reduced sample obtained after the TR step, and the oxidized sample obtained after the subsequent CS step. The values were estimated from Ni 2p_3/2 X-ray photoelectron spectra.

Sample	Peak Position [eV]		Ratio of Peak Area [%]		Valence State of Ni
	Ni 2p3/2 (2+)	Ni 2p3/2 (3+)	Ni 2p3/2 (2+)	Ni 2p3/2 (3+)
As preparation	853.98	855.6	30.2	69.8	2.70
After TR	853.95	855.57	49.2	50.8	2.51
After CS	854.05	855.54	60.3	39.7	2.40

. The estimated average values of Ni valence were 2.70, 2.51, and 2.40 for the as-prepared sample and samples obtained after the TR and CS steps. The variations in the Ni valences did not change for either step or remained limited without being related to the redox reaction. Figure 12b shows the Mn 2p_3/2 spectra with peak deconvolutions of the as-prepared samples and those obtained after each step. The peak positions obtained for the as-prepared samples were within the range of previously reported peak position variations and were in good agreement with the data for the La_0.8Sr_0.2Mn_1-xNi_xO₃ perovskite [66] with similar crystal structures and chemical compositions. The positions and areas of the peaks corresponding to the Mn²⁺/Mn³⁺/Mn⁴⁺ species are summarized in

Table 9. Peak position, ratio of peak area, and valance state of Mn for the as-prepared LSMNi0.20, thermally reduced sample obtained after the TR step, and the oxidized sample obtained after the subsequent CS step. The values were estimated from Mn 2p_3/2 X-ray photoelectron spectra.

Sample	Peak Position [eV]			Ratio of Peak Area [%]			Valence State of Mn
	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)	Mn 2p3/2 (2+)	Mn 2p3/2 (3+)	Mn 2p3/2 (4+)
As preparation	640.7	642.05	643.6	34.0	31.7	34.3	3.00
After TR	640.7	641.86	643.6	46.2	26.3	25.4	2.73
After CS	640.8	642.02	643.5	33.3	30.7	36.9	3.06

. The corresponding Mn valences were estimated based on the peak areas. The estimated Mn²⁺, Mn³⁺, and Mn⁴⁺ fractions for the as-prepared samples were 34.0%, 31.7%, and 34.3%, respectively, and the average Mn valence was identical to the stoichiometric value of 3.00+. As shown in Table 9, the average Mn valences of the samples obtained after the TR and CS steps were 2.73+ and 3.06+, respectively. The variations in Mn valences can be used to describe the redox reaction in two-step thermochemical cycling. Notably, the average Mn valence returns to its initial value, suggesting that the sample undergoes continuous stoichiometric cycling without degradation.

3.5. Comparison of Mg, Ni, and Co-Substituted LSMs with Previous Literature

The CO₂-to-CO conversions and CO productivities of LSMNi0.20, LSMCo0.35, and LSMMg0.125 were compared with those of cerium oxide and previously reported congeners. The weight change profiles of cerium oxide during thermochemical cycling are shown in Figure S7 of the Supplementary Information. The O₂ and CO productivities are summarized in Table S1. Figure 13a shows the time variations in CO₂-to-CO conversion ($X_{CO2}$) of the tested samples. The results for the nonsubstituted sample and cerium oxide are also shown for comparison. The peak value of $X_{CO2}$ for the LSMMg0.125 sample was ~0.17%, which was higher than that of the nonsubstituted sample but less than that of cerium oxide. However, the values of $X_{CO2}$ after 30 min of the CS step for the three samples were enhanced compared with those of the nonsubstituted sample and cerium oxide. This is owing to the extended duration of CO production. Figure 13b shows the CO productivities per gram of material and average value per duration. Previous data on the thermochemical two-step CO₂ splitting using LSM-based perovskites in a thermogravimetric reactor have been cited from the literature [29,31,32,33,36,39,40,41,45,47,53,67]. The best-performing samples in this study ranked at the highest level of both CO productivities compared to the reported substituted LSMs, where LSMCo0.35 exhibited the maximum CO productivity (533 μmol/g material) among the reported LSM perovskites. Compared to the authors’ previous results marked with a triangle (LSMMg0.10, LSMCo0.10, and LSMNi0.10) [53], obtained under the same test conditions using the same methods and devices, the CO productivity of LSMCo0.35 was ~1.5 times higher. Other perovskite materials that are not LSM series, Sm_0.6Ca_0.4Mn_0.8Al_0.20O₃ [68], Sr_0.6Ce_0.4Mn_0.8Al_0.2O₃ [69], and Sm_0.6Sr_0.4MnO₃ [70], were studied for the thermochemical two-step CO₂ splitting cycle. A comparison with previous literature data can be biased by the test equipment, test conditions (reaction time, temperature in both steps, gas flow rates, and concentrations), morphological features of the samples, configuration of the samples, passing gases (inert gas and CO₂), and whether the gas flow directly contacts the surface of the sample or indirectly reacts with the sample mounted in the crucible. The thermogravimetric reactor measures the weight change of the sample under various atmospheres during the thermochemical two-step CO₂ splitting cycle and estimates the O₂ and CO productivities. However, some reactors directly measure O₂ and CO gases using gas chromatography and gas analysis [34,45]. When the test results from the thermogravimetric reactor contain anything other than oxygen release/absorption based on the reaction mechanism (Equations (1) and (2)), the results should be evaluated and compared with literature data.

4. Conclusions

The redox performances of lanthanum/strontium–manganese-based perovskite oxides La_1−xSr_xMn_1−yZ_yO₃ (Z = Ni, Co, or Mg) were studied for a two-step thermochemical CO₂ splitting cycle to produce liquid fuel from syngas. Based on the analysis of the stoichiometry of oxygen/CO production and chemical composition at the A- and B sites, La_0.7Sr_0.3Mn_0.8Ni_0.20O₃ and La_0.7Sr_0.3Mn_0.65Co_0.35O₃ are the most promising materials for enhancing oxygen/CO productivity among the LSM perovskite oxides that have been studied for two-step thermochemical CO₂ splitting. Especially, La_0.7Sr_0.3Mn_0.65Co_0.35O₃ exhibited CO productivity of 533 μmol/g on the redox reactivity tests, and this result is 2.15 times higher than that of nonsubstituted La_0.7Sr_0.3MnO₃ (248 μmol/g) in the same reaction conditions. In addition, both samples showed stable and reproducible O₂ and CO release profiles during long-term continuous thermochemical cycling tests. Compared with the second and fiftieth run of CO productivity during the long-term cycling tests, the decline fraction for LSMCo0.35 and LSMNi0.20 was 1.0% and 3.7%, respectively. Ni and Co substitutions in the LSM perovskite enhanced the amount of O₂ released and improved fuel production by inducing large variations in the Mn valence owing to the existence of multivalences of Ni and Co during the thermochemical two-step cycle. Notably, LSMCo_0.35 exhibits the maximum CO productivity (533 μmol/g material) compared to the previously reported LSM perovskites.