(La_1−xCa_x)MnO_3−δ (x = 0, 0.2, 0.3, 0.4) Perovskites as Redox Catalysts in Chemical Looping Hydrogen Production Process: The Relation between Defect Chemistry and Redox Performance

Abstract

The interaction between point defects in (La_1−xCa_x)MnO_3−δ (x = 0, 0.2, 0.3, 0.4) perovskites and their redox catalytic properties in a three-reactor chemical looping hydrogen production process is investigated. During the reduction step with CH₄, the behavior of the materials is extrinsically determined and strongly depends on the Ca content. At small oxygen deficiencies, CH₄ becomes oxidized to CO₂. As the deficiency increases, partial oxidation to CO and H₂ at a molar ratio of approximately 2 is favored. During the water-splitting step, the dependency on the Ca content is much weaker since it is intrinsically determined by the Mn²⁺→Mn³⁺ oxidation with simultaneous annihilation of oxygen vacancies that are not required to compensate for the extra negative charge of the Ca dopant. Hydrogen productivities in the order of 13 cm³ (STP) H₂/g solid could be achieved during the water-splitting step at 1000 °C. The materials exhibited reproducible catalytic behavior during 10 cycles of the complete three-step process and were found to retain their perovskite structure.

1. Introduction

Τhere is no doubt that one of the major concerns of humanity is related to the continuously rising emissions of CO₂ originating from fossil resources [1]. These anthropogenic emissions are considered responsible for the global temperature rise and consequent climate changes [2]. On the other hand, since fossil fuels will most likely remain the primary energy source for many years, an efficient reduction of CO₂ emissions could be accomplished by the development of fuel combustion technologies with integrated CO₂ capture or storage. Based on these arguments, Chemical Looping Combustion (CLC) technologies have been developed and extensively investigated [3,4]. The basic idea of this technology is the utilization of a solid oxygen carrier (OC), which is reduced by supplying oxygen for the combustion of fuel in the fuel reactor (FR). This omits the utilization of air as an oxygen supply and thus enables the fuel conversion to a pure stream of CO₂ and H₂O that can be easily separated by condensation. The oxygen-deficient OC is then transferred to the air reactor (AR), where it is reoxidized by air to its initial state, and the loop cycle is repeated (Figure 1a).

Quite soon after, the basic chemical looping concept was extended in order to cover other industrial needs besides energy production [5,6]. Chemical Looping Reforming (CLR) is a process very similar to CLC (Figure 1a), with the main difference being that the fuel (which is mainly CH₄) is partially oxidized to CO and H₂ [7,8]. It is attractive for CO₂ capture and H₂ generation through the reforming process, and when combined with high efficiencies and low H₂ production costs, it may provide a good alternative to traditional H₂ production technologies. Other chemical looping variations include Syngas Chemical Looping (SCL), which uses syngas (CO+H₂) as a fuel to produce H₂ and CO₂ [9,10], or Coal Direct Chemical Looping (CDCL), which uses coal as a solid fuel in a three-reactor configuration [11].

An interesting variant of the chemical looping concept is the so-called chemical looping hydrogen (CLH) or Chemical Looping Water-Splitting Process (CLWS) (Figure 1b) [12,13]. The fuel (usually CH₄) is oxidized in the fuel reactor with the OC, which becomes reduced. The reduced and oxygen-deficient OC is then transferred to the steam reactor, where it is partially oxidized through the water-splitting reaction. Finally, the partially oxidized OC is transferred to the third air reactor, where it is fully oxidized to its initial state, and the loop closes. An advantage of the CLH process is that, next to the hydrogen produced in the FR, very pure H₂ is produced at the exit of the steam reactor by simply cooling and condensing the H₂/H₂O mixture. An important technological challenge is the circulation of solids through the three reactors.

The selection of the proper OC is one of the most important technological issues for all chemical looping technologies and plays a decisive role in the techno-economical evaluation of the process. High oxygen exchange capacity, favorable oxidation or reduction thermodynamics or structural stability are among the requirements. Extensive investigations have been performed to evaluate a large number of OCs for chemical looping processes. Summarized results have been published in several reviews [14,15].

Concerning the CLH process, the research work is almost exclusively concerned with iron oxide-based OCs [12,13,16]. The reduced iron oxide (FeO) in the FR is partially oxidized (Fe₃O₄) in the SR and totally oxidized (Fe₂O₃) in the AR. In order to improve the redox performance or the mechanical properties, several additives are added, such as ZrO₂ [17], CeO₂ [18], CaO [19] and Al₂O₃ [20].

Perovskites comprise an interesting family of materials known for their high thermal stability and sintering resistance or their high oxygen mobility and ability to accommodate relatively large deviations from stoichiometry. As redox catalysts, they have been reported to have promising properties in CLR applications since they facilitate the partial oxidation of methane to CO and H₂ [21,22]. A large number of perovskite materials has been investigated, including La_1−xSr_xMO₃ (M = Mn, Ni) [23,24], La_1−xSr_xFeO₃ [25], LaFe_1−xNi_xO₃ [26], LaFe_1−xCo_xO₃ [27] and La_1−xCa_xMnO₃ [28].

The investigations of perovskites for the three-reactor CLH process are rather limited. In certain cases, perovskites have been employed to complement iron oxide-based catalysts in order to improve the redox performance or lifetime [29,30,31]. LaFeO₃ perovskites showed quite promising results that could be enhanced with partial substitution of Fe with Co (in order to increase the $V_O^{\bullet \bullet }$ concentration) in the modified compositions LaFe_1−xCO_xFeO₃. An optimum performance could be achieved at x = 0.3 [27,32]. La_0.8Ca_0.2MO_3±δ (M = Co, Ni, Fe, Cu) perovskites have been investigated for the two-step thermochemical water-splitting reaction; the material reduction occurred only upon equilibration under a nitrogen atmosphere (PO₂~10⁻⁵ atm). The pure hydrogen produced during the water-splitting step approaches 5 cm³ (STP)/g solid with OC reduction and oxidation temperatures of 800 °C [33,34].

In this article, we report on the performance of (La_1−xCa_x)MnO_3−δ perovskites as OCs in the three-reactor CLH process. Τhese materials are considered of interest since, due to their high melting point, ionic and electronic conductivity or mechanical stability under oxidizing or reducing conditions, they have been reported to be the preferred materials for several applications, such as solid oxide fuel cells or mixed conducting membranes, [35,36] and are being investigated for various catalytic processes [37,38,39]. In addition, their defect chemistry is well known [40] and shows good stability in chemical looping processes [24]. Attention has been paid to the connection between the properties of interest, such as the products of the various oxidation-reduction reactions, and the nonstoichiometry and defect structure of these compounds, which is described based on well-accepted models published in the literature. These models are mainly constructed based on macroscopic electrical behavior evaluation, although currently, they can be verified using more powerful physical or chemical techniques [41]. Although perovskite materials are well known to exhibit intrinsic or extrinsic point defect concentrations that depend strongly on oxidative or reductive environments and affect their catalytic properties [42], attempts toward understanding the interaction between redox performance and defect chemistry are quite limited in the chemical looping literature.

2. Results

The XRD patterns of the as-prepared samples are shown in Figure 2. From the overall spectrum (shown in the left part of the figure), it is concluded that all samples have the perovskite structure. Detailed examination of the patterns revealed the presence of traces of phases identified as La₂O₃ or La₂CaO_x. Due to the low concentrations, these are considered of no importance for the general behavior of the materials in the CLH process. From the detailed spectrum of the peak at 2θ between 32 and 33° (shown in the right part of the figure), it is indicated the substitution of Ca for La transforms the rhombohedral structure to orthorhombic and decreases the unit cell size. This result is in agreement with the results reported and explained in the literature [28,43].

In Figure 3, the evolution of the oxygen stoichiometric parameter δ is shown as the material passes from the reduction step to the water-splitting step and finally to the total oxidation step.

In the reduction step in the fuel reactor, the deficiency initially increases at almost the same rate per pulse while gradually flattening toward some constant value. The pulse number at which deviation from the initial linearity and flattening start to occur increases with increasing Ca content. The same also holds for the deficiency level toward the end of the reduction step. When H₂O is subsequently fed to the reactor, water splitting occurs, and the oxygen deficiency starts to decrease toward a constant value. The reduced materials become partly reoxidized. The deficiency difference between the start and the end of the water-splitting test does not seem to strongly depend on the Ca content, at least not so strongly as the deficiency evolution during the reduction step. The deficiency order after the water-splitting step is the same as that at the end of the reduction step. Finally, during the final oxidation, the deficiency further decreases down to a constant value of δ ~ 0.05, which is independent of the Ca content and differs slightly from the deficiency of the initial “as prepared” materials (δ = δ_ref ≈ 0). In the insert of Figure 3, the deficiency evolution is shown for identical experiments without the intermediate water-splitting test. All reduced samples are fully oxidized down to almost their initial state.

In Figure 4, the CH₄ conversion per pulse $\left(X_{{CH}_4}\right)$ and the cumulative CH₄ consumption are shown as a function of the pulse number during the reduction step.

Initially, almost complete CH₄ conversion per pulse is achieved, which decreases as the pulse number increases. The higher the Ca content, the higher the CH₄ conversion per pulse. Toward the end of the reduction step, in all cases, a steady-state conversion of approximately 20% is achieved, independent of the Ca content. Moreover, the higher the Ca content, the higher the cumulative CH₄ consumption. In agreement with the identical steady-state conversions, all cumulative lines end in parallel linear regions that correspond to a constant CH₄ consumption of ~8 μmoles CH₄ per g solid per pulse.

The cumulative distributions of products CO, CO₂, H₂ and H₂O during the reduction step are shown in Figure 5a,b.

Initially, at small oxygen deficiencies, CO₂ and H₂O are mainly produced. Subsequently, as the oxygen deficiency increases (with increasing pulse number), the production of CO₂ and H₂O is almost entirely substituted by the production of CO and H₂. The higher the Ca content, the higher the deficiency at which CO and H₂ appear and finally dominate the reaction products. At the end of the reduction step, an almost constant production of 5.5–6 μmoles CO per g solid per pulse and 11–12 μmoles H₂ per g solid per pulse are produced, independently of the Ca content of the materials. The H₂ to CO molar ratio is very close to 2. By comparing Figure 3, Figure 4 and Figure 5, it is observed that toward the end of the 25 pulses of the reduction step, a situation is achieved, in all cases, with a constant CH₄ conversion of ~20% toward the production of CO and H₂, without significantly affecting the oxygen deficiency of the solids. An explanation will be provided in a subsequent paragraph.

In Figure 6, the cumulative H₂ production as a function of the pulse number is shown only during the water-splitting step.

The cumulative H₂ production increases as the Ca content increases from 147 μmoles H₂/g solid (~3.3 cm³ (STP)/g solid) for the LaMnO₃ material up to 255 μmoles H₂/g solid (~5.7 cm³ (STP)/g solid) for the (La_0.6Ca_0.4)MnO₃ material.

In Figure 7, the effect of temperature on the evolution of the oxygen deficiency during the reduction and the water-splitting step is shown for the (La_1−xCa_x)MnO₃ materials with x = 0.3 and 0.4. Similar results are also achieved by the other tested compositions. Except for the standard temperature of 900 °C, CLH experiments are also performed at 800 °C and 1000 °C. As indicated in Figure 7, the temperature increase does not only cause an increase in the oxygen deficiency evolution during the reduction step but also increases the deficiency reduction (i.e., the H₂ production) during the water-splitting step. This is a strong indication that the reactions involved in these two steps are endothermic. It is worthwhile to notice that at 1000 °C, the oxygen deficiency evolution did not reach a steady state, and apparently, in this case, more than 25 pulses were necessary.

Corresponding CH₄ conversions and cumulative H₂ productions are shown in Figure 8. The increased CH₄ conversions with increased reduction temperature are also associated with an increase in the almost constant conversion toward the end of the reduction step. The cumulative H₂ production, either during the reduction step or during the water-splitting step, is also increased with increased process temperature. The total H₂ production of the (La_0.7Ca_0.3)MnO₃ material is 927 μmoles H₂/g solid (~21 cm³ (STP)/g solid), from which 352 μmoles H₂/g solid (~8 cm³ (STP)/g solid) are produced as syngas during the reduction step and 575 μmoles H₂/g solid (~13 cm³ (STP)/g solid) as pure H₂ during the water-splitting step.

The stability and performance reproducibility of the (La_1−xCa_x)MnO₃ materials have been tested for 10 successive reduction–partial oxidation–total oxidation cycles. The results are shown in Figure 9 for material (La_0.6Ca_0.4)MnO₃. Similar results are also obtained with the other tested compositions. As shown, the materials exhibit a stable and reproducible oxidation-reduction behavior after 10 cycles. The oxygen deficiency of the material after each process step is considered reproducible with cycle repetition.

In Figure 10, XRD results are shown for samples (La_0.7Ca_0.3)MnO₃ and (La_0.6Ca_0.4)MnO₃ after each step of the CLH process. For sample (La_0.6Ca_0.4)MnO₃, results are shown after seven reduction–total oxidation cycles and after ten reduction–partial oxidation–total oxidation cycles. Similar results are also achieved with all compositions. In all cases, the main perovskite structure is preserved, and no decomposition into low- or high-valence oxides is observed. The appearance of some minor secondary phases is not considered important for the general material behavior. The reduction step clearly leads to larger cell dimensions since lattice oxygen is consumed for the oxidation of CH₄, and oxygen vacancies are created. Both partial oxidation with H₂O and further total oxidation with O₂ restore oxygen deficiency and decrease the orthorhombic unit cell size. However, it never returns to its initial size. This result is in qualitative agreement with the fact that the oxygen deficiency after total oxidation does not return to the reference deficiency of the fresh samples (i.e., δ = δ_ref ≈ 0) and indicates the existence of some non-reversible phenomena that affect the structure, which cannot be explained at the moment. Nevertheless, the structural and lattice size similarity after 7 CH₄/O₂ cycles and 10 CH₄/H₂O/O₂ provides an indication that a structural situation is achieved that remains stable with cycle repetition.

3. Discussion

It is mentioned in the literature that an intrinsic mixed oxidation state of Mn (Mn⁴⁺/Mn³⁺) or even Mn²⁺ exists in pure LaMnO₃ perovskites through the so-called discommendation reaction [44,45]. These considerations are important when the material attains excess oxygen under oxidative conditions. In the experiments reported in this article, all materials start the reduction step of the looping cycles after being equilibrated at 900 °C in a He atmosphere, and it is not likely that they contain excess oxygen. Instead, they approach stoichiometry [46]. It is, therefore, assumed that LaMnO₃ contains only Mn in its reference valence 3⁺ $\left({\mathrm{M}\mathrm{n}}_{\mathrm{M}\mathrm{n}}^{\times }\right)$. In the following, the Kroger–Vink notation will be used for all point defects involved in the discussion [47].

When Ca²⁺ substitutes for La³⁺ in A sites, the defect ${\mathrm{C}\mathrm{a}}_{\mathrm{L}\mathrm{a}}^{\prime }$ is created. The extra negative charge of this defect is not compensated by oxygen vacancies $\left({\mathrm{V}}_{\mathrm{O}}^{\bullet \bullet }\right)$ but by the oxidation of one Mn³⁺ to Mn⁴⁺, leading to the compensating defect $\left({\mathrm{M}\mathrm{n}}_{\mathrm{M}\mathrm{n}}^{\bullet }\right)$ [43]. This is in qualitative agreement with the cell shrinkage as a function of the Ca content shown in Figure 2, as well as with the transformation of the rhombohedral structure to orthorhombic upon the introduction of Ca due to increasing Mn⁴⁺ content and reduction of Jahn–Teller distortions caused by Mn³⁺. If we assume that per chemical formula, fraction x of La³⁺ ions is substituted by Ca²⁺ ions, then the material formula could be written in the following form:

$$\left({La}_{La,1-x}^{\times }{Ca}_{La,x}^{\prime }\right)\left({Mn}_{Mn,1-x}^{\times }{Mn}_{Mn,x}^{\bullet }\right)O_3$$

(1)

The concentration of ${\mathrm{M}\mathrm{n}}_{\mathrm{M}\mathrm{n}}^{\bullet }$ is extrinsically determined by the dopant through the neutrality condition $\left[{\mathrm{C}\mathrm{a}}_{\mathrm{L}\mathrm{a}}^{\prime }\right]=\left[{\mathrm{M}\mathrm{n}}_{\mathrm{M}\mathrm{n}}^{\bullet }\right]$.

Upon the introduction of CH₄, the initial overall reaction seems to be

$${CH}_4+2O_2\to {CO}_2+2H_{2}O$$

(2)

It is believed that the oxygen that takes place in this reaction is lattice oxygen that leaves the material, while Mn⁴⁺ is reduced back to Mn³⁺, with the simultaneous creation of oxygen vacancies, which is described by the following equation:

$$O_O^{\times }+2{Mn}_{Mn}^{\bullet }\iff V_O^{\bullet \bullet }+2{Mn}_{Mn}^{\times }+\frac{1}{2}{O}_2$$

(3)

Assuming that fraction δ of oxygen vacancies are created per chemical formula, then the material formula in (1) can be modified to the following:

$$\left({La}_{La,1-x}^{\times }{Ca}_{La,x}^{\prime }\right)\left({Mn}_{Mn,1-x-2\delta }^{\times }{Mn}_{Mn,x-2\delta }^{\bullet }\right)O_{3-\delta }{V}_{O,\delta }^{\bullet \bullet }$$

(4)

And by a combination of (2) and (3), the final CH₄ reaction with the material can be written as follows:

$${CH}_4+4O_O^{\times }+8{Mn}_{Mn}^{\bullet }\to {4V}_O^{\bullet \bullet }+8{Mn}_{Mn}^{\times }+{CO}_2+2H_{2}O$$

(5)

Therefore, the higher the Ca content, the higher the Mn⁴⁺ content, and, consequently, the higher the oxygen deficiency, causing reaction (2) to proceed, which is in agreement with the results shown in Figure 3. It is believed that the initial linear part of the oxygen deficiency evolution per pulse reflects the exact reaction described by (5). This linear part is extended to higher deficiencies as the Ca content increases. The dependency of CH₄ conversion as a function of Ca content, as shown in Figure 4, can also be explained. The higher the Mn⁴⁺ concentration, the more oxygen may become available and, consequently, the higher the conversion. Finally, the results shown in Figure 5 indicate that this initial CH₄ reaction with lattice oxygen is directed to CO₂ and H₂O, in agreement with (2).

As the reduction step proceeds and Mn⁴⁺ is depleted, then another reaction starts to gradually become of importance, which leads to the production of H₂ and CO. In Figure 3, this is indicated by the gradual deviation of the oxygen deficiency evolution from the initial linear part and in Figure 5 by the gradual appearance of CO and H₂ in the products with parallel disappearance of CO₂ and H₂O. The Mn⁴⁺ is gradually depleted so that at the end, the material formula given by (4) can be modified to

$$\left({La}_{La,1-x}^{\times }{Ca}_{La,x}^{\prime }\right)\left({Mn}_{Mn}^{\times }\right)O_{3-\frac{x}{2}}{V}_{O,\frac{x}{2}}^{\bullet \bullet }$$

(6)

The charge of the dopant is no longer compensated by Mn⁴⁺ but by oxygen vacancies, and the electrical neutrality is given by $\left[{\mathrm{C}\mathrm{a}}_{\mathrm{L}\mathrm{a}}^{\prime }\right]=2\left[{\mathrm{V}}_{\mathrm{O}}^{\bullet \bullet }\right]$.

It is believed that the oxygen required for this second reaction is lattice oxygen delivered by the material though a Mn³⁺ to Mn²⁺ reduction, with simultaneous creation of oxygen vacancies as well, which is described by the following equation:

$$O_O^{\times }+2{Mn}_{Mn}^x\iff V_O^{\bullet \bullet }+2{Mn}_{Mn}^{\prime }+\frac{1}{2}{O}_2$$

(7)

Assuming the creation of a fraction n oxygen vacancies per chemical formula though the reduction described by (7), then Formula (6) can be modified to

$$\left({La}_{La,1-x}^{\times }{Ca}_{La,x}^{\prime }\right)\left({Mn}_{Mn,1-2n}^{\times }{Mn}_{Mn,2n}^{\prime }\right)O_{3-\frac{x}{2}-n}{V}_{O,\frac{x}{2}+n}^{\bullet \bullet }$$

(8)

It is, therefore, stated that the Mn³⁺→Mn²⁺ consumes lattice oxygen and produces Mn²⁺ and oxygen vacancies. This statement, however, does not explain the fact that a constant CH₄ conversion level is achieved corresponding to constant CO and H₂ production per pulse. Consumption of all available oxygen should either lead to total conversion elimination or to structural deterioration; neither of the two is observed. The total conversion elimination indeed occurred when, in separate experiments, the reactants were totally free of oxygen. On the other hand, the stability of the pulse number of all products and reactants, including material oxygen deficiency, has also been observed in experiments with more than 25 CH₄ pulses. It is worthwhile to mention that during the course of the entire cycle, no carbon deposition is observed, and all mass balances can be closed with very high accuracy. All previous observations point to the possibility that certain oxygen traces might exist in the He feed bottle, and these might have been responsible for the final constant CH₄ conversion.

The total reactions that take place in this phase can be described by the following equations:

$${CH}_4+\frac{3}{2}{O}_2\to CO+2H_{2}O$$

(9)

$${CH}_4+\frac{1}{2}{O}_2\to CO+2H_2$$

(10)

It is, however, not probable that H₂ coexists even with traces of oxygen at 900 °C, and it is, therefore, believed that reaction (9) dominates the fuel oxidation at high deficiencies.

Reaction (9) in combination with (7) leads to the following equation:

$${CH}_4+3O_O^{\times }+6{Mn}_{Mn}^x\to {3V}_O^{\bullet \bullet }+6{Mn}_{Mn}^{\prime }+CO+2H_{2}O$$

(11)

As indicated in Figure 5, no water is observed in the products. When Mn³⁺→Mn²⁺ reduction occurs, the material becomes active for water splitting. Consequently, the H₂O produced through (9) is dissociated to H₂ and O₂ through the reverse of Equation (7), which, combined with the water dissociation reaction, leads to the following equation:

$$H_{2}O+V_O^{\bullet \bullet }+2{Mn}_{Mn}^{\prime }\to O_O^{\times }+2{Mn}_{Mn}^x+H_2$$

(12)

As reduction through Equation (11) proceeds, vacancies are created, which are consequently annihilated through Equation (12). At a certain oxygen deficiency, the rate of vacancy creation becomes equal to the rate of vacancy annihilation, and a certain CH₄ conversion is achieved without significantly affecting the oxygen deficiency of the material. During the previously mentioned equilibrium, the concentrations of all reaction products were constant. Assuming that all μmoles of H₂ are produced from the water dissociation reaction, a H₂O conversion can be calculated. The results are shown in Figure 11. H₂O conversions as high as 100% could be achieved at oxygen deficiency levels at which the material becomes, through Equation (11), active for water splitting. The deficiency level at which this occurs increases as the Ca content increases.

During the water-splitting step, reaction (11) takes place. The oxygen vacancies are annihilated through the Mn²⁺→Mn³⁺ oxidation. Only the n fraction of vacancies may participate in this reaction, and as long as they are depleted, this partial oxidation of the material occurs through water-splitting stops. Further oxidation (i.e., through Mn³⁺→Mn⁴⁺ oxidation, as described by the reverse of Equation (3)) cannot proceed due to thermodynamic reasons. This n fraction of vacancies is intrinsically determined by the thermodynamics of the Mn³⁺→Mn²⁺ reduction and does not depend on the Ca content. In Figure 12, the dependence of the oxygen deficiency (δ) achieved at the end of the reduction step is shown as a function of the Ca content. In comparison, the oxygen recovery during the water-splitting step is shown, which is determined as the difference between the initial and final oxygen deficiency during this step (noted as ${\mathsf{\Delta }\mathsf{\delta }}_{{\mathsf{H}}_2\mathsf{O}}$ in the figure). The dependency of the extrinsically determined δ on the Ca content is much stronger than the weak dependency of the intrinsically determined ${\mathsf{\Delta }\mathsf{\delta }}_{{\mathsf{H}}_2\mathsf{O}}$. The small differences obtained in ${\mathsf{\Delta }\mathsf{\delta }}_{{\mathsf{H}}_2\mathsf{O}}$ as a function of the Ca content can be attributed to the fact that the thermodynamic equilibrium constant of reaction (7) might depend on the Ca content. Such effects are found for La_1−xSr_xMnO₃ that exhibit defect chemistry similar to La_1−xCa_xMnO₃ [40].

All reactions that take place during the reduction and the water-splitting step are endothermic and are, therefore, favored by increasing temperature. This explains the increased oxygen deficiencies during the reduction step and the increased ${\mathsf{\Delta }\mathsf{\delta }}_{{\mathsf{H}}_2\mathsf{O}}$ during the water splitting (Figure 7), as well as the increased CH₄ conversions toward an also increased steady-state conversion at the end of the reduction step (Figure 8).

Pulse experiments with CH₄ on Ruddlesden–Popper-type perovskites La_2−xM_xNiO₄ (M = Ca, Sr) have been reported in the literature [48]. Although the CH₄ conversion per pulse is not characterized by a clearly decreasing trend, the obtained CO, CO₂, H₂ and H₂O product patterns are very similar to those reported in this article. It is stated that surface oxygen is responsible for the total CH₄ oxidation to CO₂ and H₂O, while lattice oxygen favors partial oxidation to CO and H₂. This is different from the results presented in this article, where it is stated that total oxidation also consumes lattice oxygen. Besides the very systematic dependency of H₂O and CO₂ appearance and evolution on the composition (which is not expected for oxygen chemisorbed on the surface), intermediate material examinations by XRD after 5 or 10 CH₄ pulses already indicated a unit cell expansion that implies the removal of lattice oxygen and the creation of oxygen vacancies. Stated in a different way, the increased oxygen deficiency and the associated changes in the valence of manganese lead to an increase in the binding energy of oxygen and, finally, to a change in selectivity. Quite recently, it has been reported in the literature that XRD peak shifts in catalysts in operation that involve oxygen exchange with the environment might be due to the appearance of strains caused by phase segregation phenomena [49]. Although such effects cannot be excluded, the systematic relation of the shift with the extent of reduction or oxidation, the very good reproducibility and the very small amounts of detected secondary phases indicate that such phenomena, if present, are of minor importance in this study. For the LaMnO₃ samples, a transformation from rhombohedral to orthorhombic took place.

La_0.8N_0.2MO₃ (N = Al, Ca, M = Co, Ni, Fe, Co) perovskites have also been extensively evaluated for the two-step thermochemical water-splitting reaction [34,50]. Structural stability could be achieved by performing the reduction step at temperatures below 1000 °C. Under isothermal operation at 800 °C, water splitting produced about 5 cm³ (STP) H₂/g solid. This value is quite comparable with the 13 cm³ (STP) H₂/g solid reported in this article at 1000 °C if one takes into account that the material reduction is forced with CH₄.

4. Experimental Section

La_1−xCa_xMnO₃ (x = 0, 0.2, 0.3, 0.4) are chemically prepared by the citrate method, which is described elsewhere [51]. After synthesis, all powder samples are calcined to 1000 °C for 6 h under airflow. X-ray diffraction (XRD) is performed on a Bruker D8-Advance diffractometer with CuKα radiation (λ = 0.154 nm) for the examination of the crystalline structure of the samples during various stages of the CLH process. All powder had comparable morphologies consisting of primary particles in the order of 2–3 μm, the majority of which form larger porous agglomerates up to 20–30 μm [28]. The specific surface area of all specimens was between 4 and 5 m² g⁻¹.

A U-type microreactor, with outer and inner diameters of 6 and 4 mm, respectively, is used for the reactivity experiments, which have been conducted in pulse mode at a constant temperature of 900 °C and also occasionally at 800 and 1000 °C. This mode has been chosen since it allows a better understanding of the reactions involved from the products obtained per pulse. Approximately 100 mg of material are placed in the reactor, which is brought to the experiment temperature and left to equilibrate under He atmosphere (N50 grade, purity > 99.999%) at a total flow rate of 50 cm³ min⁻¹. Initially, 25 pulses of undiluted CH₄ are fed to the reactor through a 100 μL loop valve. The reduction step is subsequently followed by water injections until a point where no variations occur in the oxygen content of the material. The material oxidation is completed by ~100 μL O₂ pulses until the material returns to a fully oxidized steady state. The duration of each CH₄ or O₂ pulse was 2 min plus 1 minute needed for the valve to be filled with the reactant. For comparative reasons, experiments without the water-splitting step are also performed. The reactor outlet stream is directed to a quadrupole mass spectrometer (Pfeiffer Vacuum GmbH, Asslar Germany, Model Omnistar GSD 350), where it is quantitatively analyzed. The mass fractions corresponding to water (m/e: 18, 17, 16), hydrogen (m/e: 2), oxygen (m/e: 32, 16), carbon monoxide (m/e: 28, 12) and carbon dioxide (m/e: 44, 28, 12) are monitored after each pulse during the entire three-step cycle. The mass spectrometer response to each of the reactants and reaction products is calibrated by using pulse injections from either pure substances or calibration mixtures. Prior to the evaluation of the perovskites, it has been experimentally confirmed, either through experiments with an empty reactor or with a reactor that contained inactive α-Al₂O₃ powder, that no activity exists originating from factors other than those related to the perovskite material. Although under partial pressures of O₂, the materials attain excess oxygen and become La_1−xCa_xMnO_3+δ, upon equilibration in He, they become nearly stoichiometric (i.e., δ ≈ 0) [28]. In any case, the equilibrium in He is considered as the reference state $\left({\mathrm{L}\mathrm{a}}_{1-\mathrm{x}}{\mathrm{C}\mathrm{a}}_{\mathrm{x}}\mathrm{M}\mathrm{n}{\mathrm{O}}_{3\pm {\mathsf{\delta }}_{\mathrm{r}\mathrm{e}\mathrm{f}}}\right)$ and all stoichiometric parameters δ reported in the following paragraphs are in relation to δ_ref. Τhe stoichiometric parameter δ is calculated from the cumulative amount of atomic oxygen that left the material (as CO or CO₂) per mole of solid material. The conversion of CH₄ per pulse is defined as the percentage of injected CH₄ consumed (in CO and CO₂). The exact amount of injected CH₄ is determined by the sum of the amounts of unreacted CH₄, produced CO and produced CO₂. It is also determined by the sum of the amounts of unreacted CH₄, produced H₂, and produced H₂O. No difference larger than 5% is observed, which is considered within the experimental error if one takes into account the uncertainty in integrating the tailoring of the water peak. This result provides a strong indication that carbon deposition, if any, is negligible. In addition, the reaction products during the water-splitting step that followed the reduction step did not contain any CO or CO₂ traces detectable by the mass spectrometer. Carbon deposition is, therefore, not considered in the analysis of the results reported in this article.

5. Conclusions

The relation between nonstoichiometric of (La_1−xCa_x)MnO₃ (x = 0, 0.2, 0.3, 0.4) perovskites and their catalytic performance in a three-step LCH chemical looping process has been studied. The results were interpreted using point defect-based solid-state chemistry.

The addition of Ca in A-sites creates Mn⁴⁺, the subsequent reduction of which is associated with the creation of oxygen vacancies and with the oxidation of CH₄ to CO₂ and H₂O.

Further reduction creates Mn²⁺, as well as the simultaneous creation of oxygen vacancies and the oxidation of CH₄ to CO and H₂O. Upon this second reduction, the materials become active for water splitting so that all formed H₂O dissociates toward H₂ production and oxygen vacancy annihilation.

During the water-splitting step, the material performance does not strongly depend on the Ca content. It is stated that this process is determined intrinsically.

The H₂ production increases with increasing temperature. At 1000 °C, the water-splitting step gives rise to the production of ~13 cm³ (STP) H₂/g solid.

Under the experimental conditions reported in this article, all materials retain the perovskite structure during all steps of the process and exhibit a constant catalytic behavior after 10 cycles of the three-step CLH process.