Comparative Evaluation of Various ABO₃ Perovskites (A = La, Ca, Sr; B = Mn, Fe) as Oxygen Carrier Materials in Chemical Looping Hydrogen Production

Abstract

This study comparatively evaluates the performance of ABO₃ perovskite materials (A = La, Ca, Sr; B = Mn, Fe) as oxygen carriers in three-step Chemical Looping Hydrogen (CLH) technology, focusing on redox behavior, oxygen transport capacity, hydrogen production, and selectivity under controlled pulse-mode conditions. The redox behavior of the materials is analyzed in relation to their defect chemistry. Perovskites such as (La_1−xCa_x)MnO₃, (La_1−xSr_x)MnO₃, and (La_0.6Ca_0.4)(Mn_1−xFe_x)O₃ were synthesized via wet chemical methods and tested in chemical looping cycles. Doping A-site cations with Ca or Sr enhanced oxygen delivery capacity by more than 100% upon reduction with CH₄ when dopant content (x) increased from 0 to 0.5. However, H₂ selectivity decreased from 52% to 2.5% for (La_1−xCa_x)MnO₃ and from 46% to 14% for (La_1−xSr_x)MnO₃ under the same conditions. In contrast, substituting Mn with Fe significantly improved hydrogen production, particularly in LaFeO₃, which exhibited the highest hydrogen selectivity and yield. At 1000 °C, LaFeO₃ produced nearly 10 mmol H₂ g⁻¹, with 80% generated during the reduction step at 99.9% selectivity and the remaining 20% during the water-splitting step at 100% selectivity. These results are linked to the extent of B-site cation reduction reactions (i) B⁴⁺ → B³⁺, which facilitates complete fuel oxidation and (ii) B³⁺ → B²⁺, which leads to partial fuel oxidation. The reverse of (ii) also contributes to H₂ production during water splitting. Additionally, the study assesses the materials’ microstructure and stability over prolonged cycles. The findings highlight Fe-based perovskites, particularly LaFeO₃, as promising candidates for CLH applications, emphasizing the need for structural and compositional optimization to enhance hydrogen production efficiency.

1. Introduction

Chemical looping processes play a crucial role in advancing sustainable energy and industrial applications by enabling efficient combustion, hydrogen production, and carbon capture [1,2,3]. These processes utilize single or mixed metal oxides as oxygen carriers to facilitate oxidation and reduction reactions without direct contact between fuel and air, significantly reducing greenhouse gas emissions. In energy production, chemical looping combustion (CLC) inherently separates CO₂, making carbon capture more cost-effective and environmentally friendly. Additionally, chemical looping hydrogen technology (CLH) supports hydrogen generation and syngas production, contributing to cleaner fuels and decarbonization efforts.

Among various chemical looping technologies, CLH has attracted considerable research interest [4,5,6] due to its ability to produce hydrogen while inherently separating carbon dioxide. The process involves metal oxides as oxygen carriers, facilitating fuel oxidation in a cyclical reaction. It typically consists of three main steps: (1) Reduction of the metal oxide with a carbonaceous fuel (e.g., natural gas or biomass) producing a reduced or oxygen-deficient oxide and CO₂, (2) oxidation of the reduced material with steam to generate high-purity hydrogen, and (3) regeneration of the metal oxide using air to complete the cycle.

In the CLH process, the oxygen carrier is a key component that directly influences hydrogen yield, reaction kinetics and process stability. An ideal oxygen carrier should exhibit high oxygen capacity, thermal and chemical stability, and resistance to deactivation over multiple cycles. Optimizing its composition and properties is essential for improving energy efficiency, reducing emissions, and ensuring the long-term viability of CLH.

Numerous materials have been investigated as oxygen carriers in CLH, each exhibiting unique properties under different operating conditions. Commonly used oxygen carriers include: Iron-based carriers (i.e., Fe₂O₃/Fe₃O₄ [7]), ferrite-spinel materials A_0.25Fe_2.75O₄ (where A = Co, Cu, Ni, Zn or Mn) [8]), Fe₂O₃-Al₂O₃-based carriers [9]), Nickel-based carriers (i.e., NiO/Ni [10], NiFe₂O₄ [11], Copper-based carriers (i.e., CuO/Cu₂O [12]), or Manganese-based carriers (i.e., Mn₂O₃/Mn₃O₄ or MnFe₂O₄ [13,14]). Additionally, perovskite materials have gained significant attention in CLH due to their excellent redox properties, structural stability, and tunable oxygen transport capacity [4,6,15]. Their unique crystal structure enables efficient oxygen exchange, enhancing hydrogen production while enabling inherent CO₂ separation. Furthermore, perovskites offer high thermal and chemical stability, making them more durable over multiple reaction cycles compared to conventional metal oxides. By optimizing their composition and properties, perovskites have the potential to enhance the efficiency and sustainability of the CLH process, positioning them as promising candidates for next-generation energy technologies.

Manganate perovskites have been extensively investigated for their redox behavior in Chemical Looping Technologies. However, reports on their behavior in the three-step Chemical Looping with Hydrogen (CLH) process, including water splitting, are scarce in the literature [16]. CaMnO₃ has been reported to be a suitable material for Chemical Looping Combustion (CLC) because it exhibits the oxygen uncoupling effect [17], while its reactivity and resistance seem to increase with the addition of certain dopants such as Ti or Mg [18,19]. LaMnO₃ perovskites are considered suitable for the Chemical Looping Reforming (CLR) process. Partial substitution of La³⁺ with divalent ions such as Ca²⁺ and Sr²⁺ increases the oxygen delivery capacity and promotes total oxidation of the fuel [20,21,22]. Similarly, partial substitution of Mn with elements such as Al, Fe, or Co enhances hydrogen production [23,24]. Regarding the three-step CLH process, the most extensively investigated materials so far are those based on iron, such as Fe₂O₃ [7,25] and Fe-based perovskites [26]. In particular, LaFeO₃ has emerged as a very promising material. It is generally accepted that CH₄ oxidation over LaFeO₃ proceeds via a Mars–van Krevelen mechanism, utilizing lattice oxygen and regenerating it during the oxidation cycle [27]. Current research directions focus on optimizing LaFeO₃’s structural performance by developing advanced morphologies, such as synthesizing nanoscale-structured LaFeO₃ [28] or fabricating three-dimensionally ordered macroporous structures [29] to increase surface area, porosity, and oxygen accessibility. In addition, significant research effort is being devoted to optimizing its chemical performance through the incorporation of dopants (e.g., Ni, Sr, Co, or Ir) to modify the A- or B-site in the perovskite, thereby improving oxygen vacancy formation [30,31,32,33].

While extensive research has been conducted, directly comparing these materials remains challenging due to variations in reactor design, temperature, pressure, fuel type, and reaction conditions. Differences in experimental setups and evaluation criteria further complicate performance assessments. Moreover, the role of oxygen vacancies—responsible for the total or partial reduction with the fuel and subsequent oxidation with H₂O—is often described in general terms without being explicitly linked to the specific defect chemistry characteristics of the materials.

This study focuses on perovskite materials of the general formula [La_1−x(Ca/Sr/)_x](Mn_1−yFe_y)O₃, evaluating their performance as oxygen carriers in CLH. The research examines their oxygen transport capacity, product selectivity and hydrogen production under identical operation conditions. Experiments were conducted in pulse mode to enable precise and controlled investigation of the materials’ redox behavior. The study also explores how the redox behavior correlates with the defect chemistry of the materials.

2. Materials and Methods

All perovskites with the general formula (A_1−xM_x)(Mn_1−yFe_y)O₃ A = La, Ca, Sr were wet chemically prepared using the citrate method [34,35]. The nitrate salts of the metals used were La(NO₃)₃⦁6H₂O (Alfa Aesar; 99.9%), Ca(NO₃)₂⦁4H₂O (Sigma Aldrich, Burlington, MA, USA; 99.0%), Sr(NO₃)₃⦁6H₂O (Sigma Aldrich, Burlington, MA, USA; 99.0%), Mn(NO₃)₂⦁4H₂O (Merck, Rahway, NJ, USA; 98.5%), and Fe(NO₃)₃⦁9H₂O (Carlo Erba, Cornaredo, Italy; 98.0%). Stoichiometric amounts of the precursors of the corresponding metals were dissolved in deionized water. After the addition of an aqueous citric acid C₆H₈O₇⦁H₂O (Chemsolute, Renningen, Germany; 99.5%) solution, 10% in excess, the solution was stirred for 15 min at 25 °C. The water was evaporated at 70 °C, and the obtained solid was dried in an oven at 250 °C overnight. Finally, the solids were calcined at 1000 °C in air for 6 h. X-ray diffraction (XRD) was performed using a Bruker D8-Advance diffractometer (Bruker, Karlsruhe, Germany) with CuKα radiation (λ = 0.154 nm) to examine the crystalline structure of the samples at various stages of the CLH process. Morphological examinations were performed with a Scanning Electron Microscope (SEM, JEOL JSM-IT500, Tokyo, Japan) on gold-sputtered specimens.

A schematic flow diagram of the experimental setup is shown in Figure 1. A U-type microreactor, with an outer diameter of 6 mm and an inner diameter of 4 mm, was used for the reactivity experiments, conducted in pulse mode at a constant temperature of 1000 °C, with some experiments also performed at 900 °C. Pulse mode operation allows for the stepwise introduction of reactants, providing detailed insights into oxygen release, fuel conversion, and material cyclic stability. This approach minimizes the influence of external factors such as gas flow dynamics and reactor design, enabling a more accurate evaluation of intrinsic material properties. The duration of each 100 μL pulse of CH₄ or O₂ was 2 min. The time required for the loop valve to fill with the reactant was 1 min. The carrier gas was He, flowing at 50 cm³ (STP) min ⁻¹. All gases were of were of high purity (>99.999%).

Approximately 100 mg of material was placed in the reactor, heated to the experimental temperature, and equilibrated under a helium atmosphere. Typically, 25 pulses of undiluted CH₄ were introduced into the reactor via a 100 μL loop valve. In some cases, reduction steps involving 50 or 95 CH₄ pulses were conducted. The reduction step was followed by water injections until no further variations in the material’s oxygen content were observed. Oxidation was completed using 100 μL O₂ pulses until the material returned to a fully oxidized state, completing one cycle. Throughout this article, the terms “reduction” and “oxidation” refer to the oxygen carrier rather than the other reactants.

The reactor outlet stream was directed to a quadrupole mass spectrometer (Pfeiffer Vacuum Omnistar GSD 350, Pfeiffer Vacuum GmbH, Asslar, Germany.) for quantitative analysis. The mass fractions corresponding to H₂O (m/e: 18, 17, 16), H₂ (m/e: 2), O₂ (m/e: 32, 16), CO (m/e: 28, 12), and CO₂ (m/e: 44, 28, 12) were monitored after each pulse throughout the three-step cycle. The mass spectrometer response to each reactant and reaction product was calibrated using pulse injections of pure substances or calibration mixtures. Prior to evaluating the perovskites, control experiments were conducted using an empty reactor or a reactor containing inactive α-Al₂O₃ powder, ensuring no external activity influenced the results. In all experiments, only CO, CO₂, H₂ and H₂O were detected as reaction products.

The conversion of CH₄ per pulse $\left(X_{{CH}_4}\right)$ was defined as the percentage of injected CH₄ that was consumed (converted to CO and CO₂):.

$$X_{{CH}_4}={\displaystyle \frac{n_{{CH}_{4, cons.}}}{n_{{CH}_{4, inj.}}}}\times 100$$

(1)

where $n_{{CH}_4, cons.}$ and $n_{{CH}_4, inj.}$ represent the consumed and injected moles of CH₄, respectively. The exact amount of injected CH₄ was determined based on the sum of unreacted CH₄, CO and CO₂. This value was cross-verified using the sum of unreacted CH₄, H₂ and H₂O. No discrepancies larger than 5% were observed, which falls within the experimental error, accounting for uncertainties in integrating the water peak. These results strongly indicate that carbon deposition, if any, was negligible and was therefore not considered. Furthermore, during the subsequent water splitting step or oxidation steps, no detectable traces of CO or CO₂ were observed in the reaction products, as confirmed by mass spectrometry.

The hydrogen selectivity $\left(S_{H_2}\right)$ and yield $\left(Y_{H_2}\right)$ per pulse, were defined as follows:

$$S_{H_2}={\displaystyle \frac{n_{H_2}}{2\times n_{{CH}_{4, cons.}} }}\times 100$$

(2)

$$Y_{H_2}={\displaystyle \frac{n_{H_2}}{2\times n_{{CH}_{4, inj.}} }}\times 100$$

(3)

where $n_{H_2}$ the moles of the produced H₂ per pulse. During the water-splitting step, were represented, and H₂ was the only detected product. For reaction parameters referring to the entire reduction step involving multiple pulses, the corresponding mole numbers were determined cumulatively as the sum over all pulses, indicated by a bar over the symbols.

$$\overline{X_{{CH}_4}}={\displaystyle \frac{\sum _1^v{n}_{{CH}_{4, cons.}}}{\sum _1^v{n}_{{CH}_{4, inj.}}}}$$

(4)

$$\overline{S_{H_2}}={\displaystyle \frac{{\sum }_1^v{n}_{H_2}}{2\times {\sum }_1^v{n}_{{CH}_{4, cons.}}}}$$

(5)

$$\overline{Y_{H_2}}={\displaystyle \frac{{\sum }_1^v{n}_{H_2}}{2\times {\sum }_1^v{n}_{{CH}_{4, inj.}}}}$$

(6)

$\sum _1^{\nu }{n}_{H_2}$, ${\sum }_1^v{n}_{{CH}_{4, cons.}}$, and ${\sum }_1^v{n}_{{CH}_{4, inj.}}$ represents the moles of produced H₂ and the consumed and injected CH₄, respectively, summed over a number of ν pulses during the reduction step.

To estimate measurement accuracy, several experiments were repeated two or three times with a fresh catalyst loading in the reactor. In such cases, average values were plotted along with associated error intervals.

3. Results

3.1. (La₁₋xCax)MnO₃, x = 0, 0.1, 0.2, 0.3, 0.4, 0.5 Perovskites

All initially synthesized (La_1−xCa_x)MnO₃ (x = 0.1, 0.2, 0.3, 0.4, 0.5) perovskite samples were examined by XRD to determine their structure. The analysis confirmed the presence of a single perovskite phase in all cases. The XRD spectra of the two end members of this series are shown in Figure 2 (spectra (a) and (b)). Occasionally, barely detectable traces of minor phases were observed, but these were considered insignificant in terms of their influence on the general redox behavior of the materials. As seen in Figure 2, the substitution of La³⁺ in rhombohedral LaMnO₃ with Ca²⁺ or Sr²⁺ leads to an orthorhombic structure with a smaller unit cell size, as indicated by the shift of the diffraction peaks to larger diffraction angles. This agrees with previously reported results [21,35] and strongly suggests that the negative charge of the defect created by the dopant is compensated by the oxidation of Mn³⁺ to the smaller Mn⁴⁺, as will also be discussed in a subsequent paragraph.

All powders had comparable morphologies consisting of primary particles in the order of 2–5 μm, the majority of which form larger porous agglomerates up to 20–30 μm. Τhe BET surface areas of all synthesized perovskites were comparable, in all cases, ranging between 3 and 5 m² g ⁻¹. More detailed information on the morphological characterization of the materials has been reported elsewhere [21,22]

Figure 3 presents the overall redox performance of the (La_1−xCa_x)MnO₃ materials over 25 CH₄ pulses. As the Ca content increases, there is a notable increase in both the oxygen withdrawn from the material (Figure 3a) and the CH₄ conversion (Figure 3b). However, this enhanced oxygen transport capacity is accompanied by a decrease in H₂ selectivity and yield, which decline from $\overline{S_{H_2}}\approx 52\%,$ and $\overline{Y_{H_2}}\approx 16\%$ for LaMnO₃ to $\overline{S_{H_2}}\approx 2.5\%,$ and $\overline{Y_{H_2}}\approx 1.5\%$ for (La_0.5Ca_0.5)MnO₃. Simultaneously, the oxygen withdrawn from the material increased from 740 μmole g⁻¹ to 1940 μmole g⁻¹. The high-purity H₂ produced during the water-splitting step remained relatively unaffected and varied between 150 and 270 μmole g⁻¹, with a maximum observed at (La_0.6Ca_0.4)MnO₃.

The CH₄ conversion, H₂ yield and selectivity per pulse for the two end members of the series are shown in Figure 4. The behavior of the intermediate compositions falls between these two extremes.

Initially, CH₄ conversions exceeding 90% were obtained. Despite the higher per-pulse CH₄ conversion achieved with (La_0.5Ca_0.5)MnO₃, the appearance of hydrogen in the reaction products was significantly delayed, and the total H₂ yield remained low. It is evident that the enhanced oxygen delivery capacity of the materials, due to Ca substitution, primarily favors total oxidation of the fuel rather than H₂ production.

3.2. (La_1−xSr_x)MnO₃, x = 0, 0.3, 0.5 Perovskites

In a second experimental series, the divalent Ca²⁺ was replaced with Sr²⁺ at 0, 0.3 and 0.5 Sr²⁺ ions per chemical formula unit. A typical XRD spectrum of the end-member composition (La_0.5Sr_0.5)MnO₃ is shown in Figure 2 (spectrum (c)), confirming the presence of a single perovskite phase. These experiments were conducted at 1000 °C.

Figure 5 presents the overall redox performance during the 25-pulse reduction step. A remarkable similarity is observed between these results and those for (La_1−xCa_x)MnO₃ (Figure 3). As the Sr content increases, CH₄ conversion increases, while H₂ selectivity and yield decrease (Figure 5b). The total H₂ yield during reduction never exceeded 20%. The oxygen transport capacity also increases with increasing Sr content (Figure 5a). The H₂ produced during the water splitting step does not appear to be significantly affected by the Sr content and remains approximately 300 μmoles g⁻¹ for all materials. The somewhat higher levels of CH₄ conversion and H₂ production observed in comparison to (La_1−xCa_x)MnO₃ materials are likely due to the higher operating temperature.

The evolution of the CH₄ conversion, H₂ yield and selectivity during the 25-pulse reduction step is shown in Figure 6. for the two end members of the series. Similar results were obtained for the intermediate compositions.

A striking qualitative similarity is observed between Figure 4 and Figure 6. The addition of Sr favors total oxidation and delays the appearance of H₂ produced by the partial oxidation of CH₄. It appears that higher CH₄ conversion promotes total oxidation rather than H₂ production.

3.3. (La_0.6Ca_0.4)(Mn_1−xFe_x)O₃, x = 0, 0.3, 0.5/LaFeO₃ Perovskites

In the previous experimental series, the substitution occurred at the A-sites of the perovskite structure. In the current series using (La_0.6Ca_0.4)MnO₃ at the base material, Mn on the B-sites was gradually substituted by Fe. The XRD spectra of the fresh materials (La_0.6Ca_0.4)MnO₃, (La_0.6Ca_0.4)FeO₃ and LaFeO₃ are shown in Figure 2 (spectra (d) and (e)). No secondary phases were observed. The overall redox behavior of the materials is presented in Figure 7.

A notable difference is observed in this case. Unlike A-site substitutions, the total oxygen withdrawn from the solid decreases as Fe content on the B-sites increases (Figure 7a). The H₂ produced during the reduction step initially increases, then decreases, reaching a minimum at a Fe content of approximately 0.8 Fe ions per formula unit. A similar trend was observed for the H₂ yield during reduction. The H₂ selectivity during the reduction step initially increases up to 0.4 Fe ions per formula unit, beyond which it remains approximately constant up to 0.8 Fe ions per unit.

However, when Fe entirely replaces Mn, a significant increase in H₂ production, yield and selectivity is observed. Specifically, over 25 pulses, 1070 μmole H₂ g⁻¹ was produced during the reduction step, while 550 μmole H₂ g ⁻¹ was generated during the water-splitting step. The corresponding H₂ selectivity and yield during reduction were $\overline{S_{H_2}}\approx 76\%,$ and $\overline{Y_{H_2}}\approx 51\%$, respectively. Furthermore, the H₂ produced during the water-splitting step increased from an average value of 340 μmole H₂ g⁻¹ to 550 μmole H₂ g⁻¹. These results improved even further when Ca substitution on A-sites was removed, resulting in pure LaFeO₃ perovskite. In this case, H₂ selectivity and yield reached 94% and 89%, respectively, while 700 μmole H₂ g⁻¹ was produced during the subsequent water-splitting step.

In Figure 8a, the redox behavior per pulse is shown for (La_0.6Ca_0.4)MnO₃ and (La_0.6Ca_0.4)(Mn_0.7Fe_0.3)O₃, illustrating the effect of Fe substitution on the B-sites. Initially (up to the first ten pulses), both materials primarily favor total CH₄ oxidation with negligible H₂ production. However, as the reaction progresses, the effect of Fe becomes pronounced, leading to significantly higher H₂ yields and selectivities. By the end of the reduction step, H₂ selectivities up to 90% were achieved for (La_0.6Ca_0.4)(Mn_0.7Fe_0.3)O_3.

A significant shift occurs when Fe becomes the sole B-site component (Figure 8b). After an initial period of two pulses, a high and stable hydrogen selectivity of 80% is achieved, along with a constant CH₄ conversion of 66% and an H₂ yield exceeding 50%.

These results further improve when Ca is entirely removed from the A-sites, resulting in pure LaFeO₃. During the reduction step, the reaction is almost entirely directed toward partial oxidation of CH₄, leading to syngas production at very high CH₄ conversions (>90%), selectivities (~100%) and yields (>95%). Additionally, during the subsequent water-splitting step, LaFeO₃ demonstrated a high capacity for high-purity H₂ production. Overall, regarding H₂ production, LaFeO₃ emerges as the best-performing material among all tested perovskites.

The comparative material evaluation was limited to 25 pulses during the reduction step, followed by subsequent water splitting until no further hydrogen production was observed. However, for the best-performing LaFeO₃, it was deemed worthwhile to examine its behavior under prolonged reduction beyond 25 pulses. Figure 9 presents the oxygen removed during reduction or incorporated into the material during subsequent water splitting for reduction steps involving 25, 50 and 95 pulses. It is expressed as the stoichiometric parameter δ in LaFeO_3−δ, where δ = 0 was arbitrarily set for the fresh material.

The amounts of produced hydrogen, along with the overall H₂ yields and selectivity, are shown in

Table 1. Hydrogen production, H₂ yields and selectivities of LaFeO₃ during the reduction and water splitting steps for reduction steps including 25, 50 and 95 CH₄ pulses.

Pulse Nr.	25	50	95
Reduction Step
H2 produced (μmole g−1)	1745	3739	7748
$\overline{{\mathrm{X}}_{{\mathrm{C}\mathrm{H}}_4}}$ (%)	94	93.5	90.5
$\overline{{\mathrm{S}}_{{\mathrm{H}}_2}}$ (%)	98	99.9%	99.9%
$\overline{{\mathrm{Y}}_{{\mathrm{H}}_2}}$ (%)	92.5	95	92
Water Splitting Step
H2 produced (μmole g−1)	700	1538	2059
$\overline{{\mathrm{S}}_{{\mathrm{H}}_2}}$ (%)	100%	100%	100%
Total H2 produced per cycle (μmole g−1)
	2445	5277	9807

.

In addition to demonstrating very good reproducibility, the results in Figure 9 indicate the material’s stability under prolonged reduction. Table 1 further shows that the reduction reaction is exclusively directed toward syngas production at CH₄ conversions exceeding 90% in all cases. Moreover, the amount of H₂ produced during water splitting increases as the degree of reduction in the material increases. Finally, with 95 CH₄ pulses, a total H₂ production of approximately 10 mmole g⁻¹ (equivalent to approximately 225 mL (STP) H₂ g⁻¹) was achieved.

In Figure 10a and b, the results are shown after three consecutive three-step cycles involving reduction, oxidation with H₂O and oxidation with O₂. The reduction duration was 25 (Figure 10a) or 50 pulses (Figure 10b). Similar results were also obtained in 95 reduction pulses. Although the amount of oxygen withdrawn from the material and, consequently, the H₂ produced during the reduction remains relatively constant, the H₂ produced during the water-splitting step gradually decreases.

In Figure 11, the XRD results are shown after the first reduction step and after the H₂O splitting step off the third cycle for cycles with reduction steps of 25 and 95 pulses. It can be seen that after reduction, the main perovskite phase is maintained; however, some La(OH)₃ and metallic Fe are also formed, with their amounts increasing as the degree of reduction increases. After H₂O splitting, these minor phases disappear, and the perovskite phases appear to be fully restored. Therefore, irreversible phase transformations cannot account for the progressively decreasing H₂ production during water splitting.

In Figure 12, typical SEM images are shown for the fresh samples (Figure 12a–d) as well as for the samples after the first cycle (Figure 12e–h). The fresh particles exhibit a relatively open structure, consisting of randomly oriented layers that provide a sufficient external surface for reactant access (Figure 12a,b). After undergoing a 25-CH₄ pulse reduction-H₂O splitting-oxidation cycle, the particle structure becomes more compact (Figure 12e,f). A more noticeable difference is observed in the microstructure within these layers. In the fresh samples, pores are visible between the primary particles (Figure 12c,d), whereas in the reduced and oxidized samples, sintering has progressed, leading to fewer pores and a denser microstructure (Figure 12g,h). This phenomenon likely intensifies with an increasing number of cycles and may be responsible for the reduced H₂ production during the H₂O splitting step.

4. Discussion

The similarities in the redox behavior of (La_1−xCa_x)MnO₃ and (La_1−xSr_x)MnO₃ perovskites can be explained by considering their defect chemistry [35,36,37,38]. The substitution of La³⁺ with Ca²⁺ or Sr²⁺ on A-sites creates negatively charged point defects, ${Ca}_{La}^{\prime }$ or ${Sr}_{La}^{\prime }$, which are compensated by the oxidation of Mn³⁺ to Mn⁴⁺ on B-sites, leading to the formation of the positively charged point defect ${Mn}_{Mn}^{⦁}$ (point defect symbols follow the Kröger-Vink notation [39].

During the reduction step with CH₄, Mn⁴⁺ is reduced to Mn³⁺, accompanied by the formation of positively charged oxygen vacancies $\left(V_O^{⦁⦁}\right)$, which take over the charge compensation for the defects introduced by the dopants. Consequently, as the dopant content (x) increases in (La_1−xCa_x)MnO₃ and (La_1−xSr_x)MnO₃, the Mn⁴⁺ concentration increases. The subsequent reduction of Mn⁴⁺ leads to greater oxygen release and a higher $V_O^{⦁⦁}$ concentration. This explains the increased oxygen release in Ca- and Sr-doped materials, as shown in Figure 3a and Figure 5a, and the enhanced overall CH₄ conversion observed in Figure 3b and Figure 5b.

As reduction progresses and Mn⁴⁺ is depleted, further reduction of Mn³⁺ to Mn²⁺ occurs. It has been proposed that oxygen released by the Mn⁴⁺→Mn³⁺ reduction favors total fuel oxidation, whereas oxygen from the Mn³⁺→Mn²⁺ reduction leads to partial oxidation and H₂ production. This observation aligns with the product evolution trends seen in the pulse sequence during the reduction step (Figure 4 and Figure 6). Higher dopant concentrations lead to increased Mn⁴⁺ content, requiring more pulses to deplete Mn⁴⁺ and initiate H₂ production. The progressively decreasing CH₄ conversion with pulse number (Figure 4 and Figure 6) likely results from the increasing difficulty of oxygen removal as the material becomes more oxygen-deficient. The overall CH₄ conversion $\left(\overline{X_{{CH}_4}}\right)$ during reduction (Figure 3b and Figure 5b) increases with dopant content, likely due to the dominance of the initial total oxidation stage, which extends with increasing dopant levels. Generalizing these results, divalent A-site substitutions in manganates appear to favor total fuel oxidation rather than H₂ production.

During the H₂O splitting step, oxygen vacancies are eliminated, and Mn²⁺ is oxidized back to Mn³⁺. This oxidation appears to be largely independent of A-site dopant content, as does the H₂ yield from water splitting, as shown in Figure 3a and Figure 5a. It is possible that the higher Mn²⁺ content in low-dopant samples is counterbalanced by kinetic effects associated with the greater oxygen deficiency in high-dopant samples.

When Fe³⁺ partially substitutes for Mn³⁺, distinct differences in redox behavior are observed. Since the Mn³⁺→Mn⁴⁺ oxidation is thermodynamically favorable at 1000 °C ($\Delta G_f^o\approx -187 \mathrm{k}\mathrm{J} {\mathrm{mol}}^{-1}<0)$ while Fe³⁺→Fe⁴⁺ oxidation is not ($\Delta G_f^o\approx +102 \mathrm{k}\mathrm{J} {\mathrm{m}\mathrm{o}\mathrm{l}}^{-1}>0)$, it is reasonable to assume that ${Mn}_{Mn}^{⦁}$ remains the primary compensating defect for A-site dopants. However, in this case, reduction involves Mn⁴⁺→Mn³⁺ and the more challenging Fe³⁺→Fe²⁺ transition, which favors partial oxidation of the fuel. Consequently, as Fe content increases, total oxygen removal from the material decreases, while H₂ production during the reduction step initially increases (Figure 7a).

At a certain Fe concentration, Fe³⁺ becomes the majority B-site ion, and the Mn³⁺ content is insufficient to fully compensate for the ${Ca}_{La}^{\prime }$ defects. At this point Fe⁴⁺ formation appears to occur. The presence of Fe⁴⁺ in Fe-based perovskites, particularly in A-site doped materials, is well documented in the literature and has been linked to catalytic activity [40,41]. Similar to Mn-based perovskites, Fe⁴⁺ → Fe³⁺ reduction promotes total oxidation, leading to a slight decrease in H₂ production, as observed in Figure 7. When Fe fully or nearly fully replaces Mn, the amount of Fe not involved in ${Ca}_{La}^{\prime }$ compensation increases, and Fe³⁺ ↔ Fe²⁺ redox cycling significantly enhances H₂ production in both the reduction step and water-splitting steps. These observations suggest that the presence of Ca²⁺ on A-sites is detrimental to H₂ production, and its removal would be beneficial. This is supported by the high H₂ yields observed for LaFeO₃ in Figure 7 and Figure 8b during both reduction and water splitting.

LaFeO₃ also exhibits stability under extended reduction conditions, as indicated by the nearly constant rate of stoichiometric parameter (δ) variation per pulse up to 95 pulses (Figure 9). The amount of H₂ produced during water splitting increases with the degree of reduction, as more Fe²⁺ is produced during Fe³⁺ → Fe²⁺ reduction, leading to increased H₂ generation during Fe²⁺ → Fe³⁺ oxidation with H₂O.

Material repeatability was tested over three complete cycles. The reduction step was found to be reproducible even at high pulse numbers, with variations within experimental error. However, H₂ production during water splitting declined with cycle number. Since no irreversible phase transformations were detected (Figure 11), this decline is attributed to sintering and surface area loss. Notably, the formation of La(OH)₃ and metallic Fe upon reduction, as observed in this study, has not been previously reported in the literature.

Perovskites are ionic solids composed of close-packed O²⁻ ions, and their sintering rate is determined by the mobility of the slowest-moving species. In oxides, this is typically O²⁻, which diffuses via oxygen vacancies [42]. The strongly reducing conditions during the reduction step promote oxygen vacancy formation, thereby increasing O²⁻ diffusion and, consequently, the sintering rate. The resulting loss of accessible surface area is evident in the SEM images (Figure 12).

Although LaFeO₃ and doped LaFeO₃ have been reported to exhibit high selectivity for the partial oxidation of CH₄, achieving both CH₄ conversion and H₂ selectivity greater than 90% is relatively rare [26]. Additionally, the production of approximately 10 mmol H₂ per complete cycle per gram of oxygen carrier is among the highest reported in the literature [4,26] for chemical looping hydrogen (CLH) processes involving water splitting. Reported H₂ production per complete cycle for LaFeO₃-based oxygen carriers typically ranges between 2 and 5 mmol H₂·g⁻¹ [32,33,43].

The Fe²⁺ → F³⁺ redox transition in LaFeO₃ has been assumed based on its general redox behavior. However, this is not the only crucial factor; the Fe ion coordination number also significantly influences reaction pathways [27]. Comparative experiments using hematite (Fe₂O₃) showed that the Fe³⁺ → Fe²⁺ reduction favors total CH₄ oxidation, as evidenced by the following results after 25 CH₄ pulses: Overall CH₄ conversion $\overline{X_{{CH}_4}}=73.5\%,$ H₂ selectivity $\overline{S_{H_2}}=2.8\%$ and H₂ yield $\overline{Y_{H_2}}=2\%$, H₂ from water splitting 410 μmole g ⁻¹. Comparing these results with Table 1 for LaFeO₃ suggests that Fe₂O₃ promotes total oxidation during reduction.

Efforts to enhance LaFeO₃’s redox performance have been explored, including structural modifications to improve cycling stability [28,29,33]. A-site substitutions have been found to favor total oxidation [40,44], whereas B-site modifications could further enhance H₂ production. Ongoing laboratory experiments indicate promising results when partially substituting Fe with Co.

5. Conclusions

Perovskites, particularly those with the general formula (La_1−x(Ca/Sr/)_x)(Mn_1−yFe_y)O₃, demonstrate significant potential as oxygen carriers for chemical looping hydrogen (CLH) due to their excellent redox properties, structural stability, and tunable oxygen transport capacity. Their performance in hydrogen production is strongly dependent on their composition and defect chemistry.

Substitution of La with Ca or Sr on the A-site of the perovskite structure increases the oxygen transport capacity but also shifts the reaction towards total oxidation of CH₄, reducing hydrogen yield. By increasing x from 0 to 0.5 in (La_1−xCa_x)MnO₃, the total oxygen delivered by the solid at 900 °C after 25 CH₄ pulses increases from 740 to 1940 μmole O₂ g⁻¹ while the total H₂ selectivity decreases from 52 to 2.5%. Corresponding results for (La_1−xSr_x)MnO₃ (x = 0, 0.5) at 1000 °C were 890 and 1940 μmole O₂ g⁻¹ for oxygen delivery and 46% and 14% for H₂ selectivity, respectively. As the dopant content increases, the reduction of Mn⁴⁺ to Mn³⁺ enhances oxygen release, but the overall hydrogen production decreases due to the predominance of total oxidation reactions.

Substituting Mn with Fe at the B-site in (La_0.6Ca_0.4)MnO₃ improves hydrogen yield and selectivity. When Fe fully replaces Mn in (La_0.6Ca_0.4)FeO₃, a significant increase in hydrogen production and selectivity is observed. The H₂ production total selectivity and yield for (La_0.6Ca_0.4)FeO₃ at 1000 °C were, during the 25 CH₄ pulse reduction step, 76% and 51%, respectively. They become even higher when Ca is removed from A-sites in LaFeO₃, emerging as the best-performing material for H₂ production among the tested perovskites, with a corresponding H₂ total reduction step selectivity of 98% and yield of 92.5%. These numbers become even higher when the reduction step is prolonged to more than 25 CH₄ pulses.

LaFeO₃ shows remarkable stability under extended reduction conditions of up to 95 CH₄ pulses, maintaining a constant rate of oxygen delivery across multiple reduction cycles. This indicates that LaFeO₃ can undergo prolonged reduction without significant degradation, ensuring its potential for long-term applications in chemical looping processes.

Sintering, which leads to a denser structure and loss of surface area, may contribute to a gradual decrease in hydrogen production during the water-splitting step. This phenomenon is most evident in the perovskites after multiple reduction-oxidation cycles.

The results suggest that modifications to the B-site of perovskites, particularly using Fe, could further improve hydrogen production efficiency. Ongoing experiments with Co substitution for Fe in LaFeO₃ show promising results, indicating that further optimization could lead to even more efficient materials for CLH applications.