Hydrogen-Rich Gas Production by Steam Reforming and Oxidative Steam Reforming of Methanol over La0.6Sr0.4CoO3−δ: Effects of Preparation, Operation Conditions, and Redox Cycles

La0.6Sr0.4CoO3−δ (LSC) perovskite, as a potential catalyst precursor for hydrogen (H2)-rich production by steam reforming of methanol (SRM) and oxidative steam reforming of methanol (OSRM), was investigated. For this purpose, LSC was synthesized by the citrate sol–gel method and characterized by complementary analytical techniques. The catalytic activity was studied for the as-prepared and prereduced LSC and compared with the undoped LaCoO3−δ (LCO) at several feed gas compositions. Furthermore, the degradation and regeneration of LSC under repeated redox cycles were studied. The results evidenced that the increase in the water/methanol ratio under SRM, and the O2 addition under OSRM, increased the CO2 formation and decreased both the H2 selectivity and catalyst deactivation caused by carbon deposition. Methanol conversion of the prereduced LSC was significantly enhanced at a lower temperature than that of as-prepared LSC and undoped LCO. This was attributed to the performance of metallic cobalt nanoparticles highly dispersed under reducing atmospheres. The reoxidation program in repetitive redox cycles played a crucial role in the regeneration of catalysts, which could be regenerated to the initial perovskite structure under a specific thermal treatment, minimizing the degradation of the catalytic activity and surface.


INTRODUCTION
Catalytic steam reforming of methanol (SRM) presents a great potential to supply hydrogen (H 2 ) for fuel cells. 1 Although methanol (CH 3 OH) is highly toxic and miscible in water, it possesses several advantages, such as a high hydrogen-tocarbon ratio, liquid state at atmospheric conditions, and biodegradability. 2 Compared to other fuels like hydrogen, synthesis gas, or methane, the hydrogen volumetric density in methanol is higher than those of either compressed fuel gas or liquid H 2 . It is easier to distribute with the existing infrastructure and safer to handle than compressed H 2 . 3,4 Like other fuels, methanol can be a carbon-neutral renewable feedstock because it is easily obtained as a synthetic fuel from renewable sources. 5,6 On the other hand, the operating temperatures for SRM (200−300°C) are lower compared to other reforming processes such as ethanol steam reforming around 400°C 7,8 or methane steam reforming above 500°C. 9 The SRM process is mainly described by three main reactions: direct SRM (eq 1), methanol decomposition (MD) (eq 2), and reverse water−gas shift (RWGS) (eq 3). 10 MD (eq 2) and RWGS (eq 3) occur simultaneously with SRM. Because the overall process is remarkably endothermic, it consumes so much energy that it has to be supplied from an external source. The addition of O 2 to the SRM process is a common practice to combine the steam reforming and partial oxidation reactions that could approximate an autothermal oxidative steam reforming of methanol (OSRM; eq 4): Autothermal SRM can be reached at around a = 0.1−0.2. 12 Higher O 2 compositions favor the formation of CO 2 and H 2 O, Given the works above, Sr-substituted lanthanum−cobalt oxides (La 1−x Sr x CoO 3−δ ) may be another interesting perovskite-type oxide for SRM and OSRM, particularly when a high H 2 /CO ratio is not required. Then, the absence of Pd in the catalysts could be an advantage in decreasing their price. Furthermore, the strontium oxide in LaXCoO 3 (X = Mg, Ca, Sr, Ce) could enhance the activity and stability, such as for steam reforming of ethanol, showing a better catalytic performance than the other compositions. 27 In this regard, La 1−x Sr x CoO 3 is known to be interesting because of its excellent performance for multiple catalytic applications: oxidation of CO, propane, and methane 28,29 soot oxidation activity, 30 toluene combustion, 31 Li−O 2 battery, 32 and solar photocatalyst. 33 Furthermore, La 1−x Sr x CoO 3−δ (x = 0.2−0.5) is typically used, showing excellent performances as an oxygen electrode and a current collector for SOFCs 34−36 and solid oxide electrolyzer cells (SOECs) 37−39 and also as a cathode for single-chamber SOFCs. 40,41 Among the studied compositions of La 1−x Sr x CoO 3−δ for catalytic applications, La 0.6 Sr 0.4 CoO 3−δ (LSC) showed the best catalytic activity, attributed to the highest oxygen chemisorption and desorption capacity. 28−30 In this regard, Morales et al. 42−44 reported the performance and stability of La 1−x Sr x CoO 3−δ (x = 0.4 and 0.5) perovskites, as a catalyst precursor, for the production of H 2 and syngas by the partial oxidation of methane and steam reforming of ethanol. The results evidenced a remarkable catalytic activity due to the stability of Co particles, which were highly dispersed in their reduced state. Similarly, Chen et al. 45 recently reported a series of A-site Sr-doped La 1−x Sr x CeO 3−δ (x = 0.2, 0.4, 0.6, 0.8) perovskite catalysts for H 2 production by SRM, showing that La 0.6 Sr 0.4 CeO 3−δ exhibited the best performance at optimal reaction conditions. In summary, Sr-substituted lanthanum− cobalt oxide perovskites may be potential catalytic precursors for the SRM and OSRM processes to supply H 2 -rich gas for SOFC applications. However, few studies reported the SRM and OSRM over LSC perovskites, and information on how the preparation, operation conditions, and degradation under redox cycles affected the catalytic activity is still scarce. Meanwhile, these issues also have a great interest in the SOFC development using LSC perovskites because their degradation is primarily due to Sr segregation on the electrode surfaces at operating temperatures higher than 500°C. 46,47 Recently, several strategies to control Sr segregation have been reported on the modification of surfaces with more and less reducible cations, the control and change of the oxygen partial pressure and applied voltage, the substitution of isovalent A-or B-site cations with different ionic radii, or the lowering of the operating temperature to deactivate Sr segregation thermodynamically and kinetically, among others. 48−50 Despite many efforts, completely effective practical solutions have not yet been proposed to inhibit Sr segregation, and, therefore, the use of thermal treatments and HCl etching to remove the Sr-rich particles at the surface of the used LSC have been proposed. 50 Here, we propose to study the LSC under multiple SRM and OSRM conditions because understanding the preparation, operation conditions, and redox cycles of SRM and OSRM perovskites is critical for finding new strategies, optimizing their performance, and enabling their practical application in catalysis but also in other systems such as SOECs for renewable energy conversion and storage. In the present work, LSC was studied as a precursor catalyst for SRM and OSRM. For this purpose, LSC was synthesized by the sol−gel citrate method. The prepared material was deeply charac-terized by different complementary analysis techniques, and, subsequently, the catalytic activity toward SRM and OSRM was investigated. Particular attention was paid to studying the effect of LSC's reductive thermal pretreatment and feed gas composition on the catalytic performance. Furthermore, the degradation and regeneration of LSC under several reaction conditions and repeated redox cycles were investigated. , homogenized in an agate mortar, and calcined at 500°C to promote decomposition of the organic fraction. Finally, the precursor was calcined at 900°C in synthetic air for 12 h and at 300°C in pure O 2 for 72 h.

Catalyst Characterization.
The catalyst was characterized in different oxidation states: as-prepared LSC, after a reductive thermal treatment in 5% H 2 /Ar at 650°C for 1 h, after SRM and OSRM tests, and after a specific thermal treatment of regeneration. To investigate in situ regeneration of the LSC precursor under the reaction conditions, redox cyclic tests were carried out, reoxidizing the catalyst by decreasing the temperature from reaction temperature (600°C) to room temperature (RT) in several cooling programs: a quenching in N 2 , a quenching under SRM and OSRM, and a slow cooling under SRM, OSRM, and air. Regeneration experiments of the LSC precursor were also carried out under synthetic air at 900°C for 12 h and subsequently at 300°C for 72 h, similar to that described previously. 43,44 The specific surface area of the catalysts was determined by the Brunauer−Emmett−Teller method using a Micromeritics model Tristar 3000, measuring the nitrogen adsorption at 77 K. Temperature-programmed reduction (TPR) experiments were performed in a reactor equipped with a thermal conductivity detector (TCD). A gas mixture of 5% H 2 /Ar (30 mL min −1 ) was used to reduce 50 mg of LSC by heating from 100 to 800°C at 10°C min −1 . Phase analysis of the synthesized and thermally treated powders was performed by Xray diffraction (XRD). The XRD patterns were obtained by using a Siemens 2000 diffractometer with Cu Kα radiation (40 kV, 40 mA, λ = 0.154 nm). The XRD patterns were collected in the range 2θ = 20− 80°, and the crystalline phases were identified using the JCPDS database. The crystallite sizes were estimated through the line broadening of the XRD peaks using the Scherrer equation. 51 The surface properties related to the chemical states and surface compositions of La, Sr, Co, and O in the LSC samples were determined by X-ray photoelectron spectroscopy (XPS). XPS analyses were conducted in an ultrahigh-vacuum multichamber system by SPECS with a PHOIBOS 150 EP hemispherical energy analyzer and an MCD-9 detector XR-50. It possesses an X-ray source with a twin anode (Al and Mg) and a high-pressure and high-temperature chamber for gas treatments of the samples. The samples were compensated for charging with a low-energy electron beam, and the peak of C 1s (binding energy = 284.4 eV) was used to correct for sample charging effects. 52 SpecsLab Prodigy, an experiment control software package, was used for data acquisition and CasaXPS for spectral analysis. The catalyst microstructures were observed by transmission electron microscopy (TEM; Titan G2 60-300, USA) and scanning electron microscopy (SEM; Carl Zeiss Merlin, Germany) equipped with energy-dispersive spectroscopy (EDS; Oxford Instruments INCA-350 system, United Kingdom). The samples were coated with carbon to minimize the electrical charge on the surface.
2.3. Catalytic Tests. SRM and OSRM experiments were carried out in a fixed-bed quartz tubular reactor (5 mm inner diameter) at atmospheric pressure. The catalytic tests were performed using 50 mg of LSC packed on a bed of quartz wool in the reactor. The reactor was kept in a horizontal tubular furnace. Two K-type thermocouples were used: one outside the reactor to control the furnace temperature using a Eurotherm PID controller and another in contact with the catalyst to measure the bed temperature. Before catalyst tests, several blanks (without catalyst) were analyzed to confirm the reactor's absence of direct oxidation reactions. Before the SRM reaction, the catalyst precursor was reduced at 650°C for 1 h in 5 vol % H 2 /Ar (30 mL min −1 ) and then cooled under N 2 (30 mL min −1 ) to the initial testing temperature of 150°C. In addition, several tests were performed with the as-prepared catalyst (without reductive pretreatment) to study the effect of the previous oxidation state of the catalyst on the catalytic activity. Afterward, the reaction mixture was introduced to the reactor to analyze the impact of the temperature, varying the feed composition of the O 2 /H 2 O/CH 3 OH molar ratio on the methanol conversion. Finally, the selectivities of the products were investigated under different SRM and OSRM conditions.
For SRM tests, a mixture of vaporized water−methanol, with H 2 O/ CH 3 OH molar ratios of 1.3:1, 2:1, and 4:1, and N 2 carrier gas was cofed to the reactor using a mass flow controller (Bronkhorst High-Tech). The reaction gas mixture was composed of 80 vol % N 2 and 20 vol % (CH 3 OH + H 2 O). The water−methanol solution was continuously dosed by a peristaltic pump. The gas hourly space velocity resulting from the total gas mixture was 40000 h −1 . For OSRM tests, O 2 was added to the reaction mixture SRM, using a mass flow controller (Bronkhorst High-Tech), with O 2 /H 2 O/CH 3 OH molar ratios of 0.1:1.3:1, 0.2:1.3:1, and 0.3:1.3:1. The other conditions were similar to those of SRM. The effluent gases were analyzed by online gas chromatography (3000A micro GC, Agilent Technologies, USA) equipped with three channels, using a molecular sieve, a Poraplot, QV-1 columns, and a TCD. The chromatograph was calibrated using known flow rates of pure gases (CO, CO 2 , and H 2 ) with N 2 as the internal standard. Other byproducts, such as methanol or ethylene, were not detected significantly. Methanol conversion and selectivity to products were determined as follows: where n(CH 3 OH) in is the flow rate (mol min −1 ) at the reactor inlet and n(CO) out , n(CO 2 ) out , and n(H 2 ) out are the flow rates (mol min −1 ) of CO, CO 2 , and H 2 , respectively, at the reactor outlet. Figure 1 shows the TPR results of LSC and LaCoO 3−δ (LCO). Co reduction in both catalyst precursors occurred in two steps. In the case of LCO, the initial reduction of Co 3+ to Co 2+ resulted between 300 and 420°C with a peak centered at ∼380°C, and the complete reduction of Co 2+ to Co 0 occurred at temperatures of 440−660°C (maximum at ∼570°C). Similar to the LCO perovskite, the Co of LSC was first reduced between 200 and 400°C due to a one-electron reaction, which took place with two peaks at ∼280 and ∼380°C. The second reduction step was a twoelectron process in the 430−680°C range, with two maxima centered at ∼440 and ∼640°C. Compared to LCO, the TPR peaks of LSC were shifted toward lower temperatures. The Sr 2+ doping in LSC led to a less stable structure, increasing the average oxidation state of Co 3+ to Co 4+ and/or the generation of oxygen vacancies. 53,54 These TPR data suggested that the presence of Sr 2+ in LCO-based perovskites should promote the ACS Applied Energy Materials www.acsaem.org Article reduction process and, consequently, the SRM of LSC at temperatures lower than those of LCO. Figure 2 shows the XRD patterns and SEM images of LSC as-prepared and after a reductive thermal treatment at 650°C for 1 h in 5 vol % H 2 /Ar. XRD analysis evidenced that asprepared LSC presents the cubic perovskite structure, the only phase with a highly crystalline symmetry. After the reductive thermal treatment, XRD exhibited the formation of a mixture of phases composed of cubic Co, hexagonal La 2 O 3 , and cubic SrO. Therefore, due to the reduction reaction, the total decomposition of LSC generated the following process: La 0.6 Sr 0.4 CoO 3−δ → Co/La 2 O 3 /SrO. These results were in good agreement with the TPR data. The crystallite sizes were about 20−30 nm, as determined from the XRD data using Scherrer's formula. This suggested that Co nanoparticles were highly dispersed on a La 2 O 3 and SrO particles matrix. SEM and TEM micrographs of LSC showed that much larger particle dispersion and smaller particle size were achieved in the reduced state (Figures 2 and 3). The specific surface areas of the precursor increased from 6 m 2 g −1 (as-prepared) to 56 m 2 g −1 after the reduction process, which is in good agreement with the TEM images and XRD patterns. This strong increase in the specific surface area of the reduced precursor is expected due to the calcination conditions at temperatures as high as 900°C to form the perovskite phase and the capacity of perovskite-type oxides for yielding an excellent metallic dispersion formed by one or several metals in a matrix of metal oxides. 18,55 Figure 4 shows the XPS spectra of La 3d, Sr 3d, Co 2p, and O 1s corresponding to as-prepared LSC and after the reduction of LSC, which were obtained in an ultrahighvacuum chamber. Additionally, surface compositions of the samples derived from XPS measurements are listed in Table 1. The La 3d level was characterized by a double peak for each spin−orbit component (Figure 4a). The positions and shapes of the La 3d peaks for both samples were similar to those reported in the literature for LSC. 30,56 In the La 3d signal, the double contribution of each spin−orbit is characteristic of La 3+ and so is a fingerprint of the oxidation state. The samples exhibited different splittings (around 0.4 eV) in the La XPS peaks, 3.5 and 4.0 eV for the as-prepared and reduced samples, respectively. This is related to the chemical states present in each sample because the strong electron orbital−spin angular momentum interactions cause a splitting of the La XPS peaks. In the as-prepared sample, the splitting is close to that of La 2 (CO) 3 , in concordance with the Sr XPS results. After reduction, the splitting increased toward the value for lanthanum oxide, indicating that the carbonate amount decreased, but it is not pure lanthanum oxide bonding. The results are consistent with a mixture of La 2 O 3 , La 2 (CO) 3 , and La(OH) 3 . Figure 4b exhibits the Sr 3d 5/2 region (∼132 eV) characteristic of Sr 2+ in the perovskite phase (LSC lattice). In addition, the Sr 3d 3/2 peak (∼133−134 eV) indicates the formation of SrCO 3 56 (∼133 eV), and the contribution at higher binding energy (∼134 eV) suggested the presence of SrO. 26 The Sr 3d spectral comparison for the as-prepared and reduced LSC evidenced the significant intensity of the Sr 3d peak in SrCO 3 after LSC reduction, which was associated with the Sr tendency to interact with the atmospheric CO 2 . 56 For the as-prepared LSC in Figure 4c, the high intensities of Co 2p 1/2 and Co 2p 3/2 at binding energies of ∼779 and ∼783 eV and ∼794 and ∼795 eV, respectively, indicated that the Co was mainly in the states of Co 3+ /Co 4+ and Co 2+ , respectively. 57,58 As observed by other researchers for LSC, 59 we note that Co 3+ and Co 4+ were indistinguishable and could not be quantified from the main peaks of the Co 2p XPS spectra, while both states are expected to exist on/in LSC at high temperatures. After the reduction treatment, both Co 2p peaks were wider than the as-prepared LSC due to the significant presence of metallic Co at ∼778 and ∼793 eV, respectively, and the decrease of the Co 3+ /Co 4+ peak. In addition, the surface amount of Co after the reduction catalyst became higher than that of the as-prepared one, evidencing that the Co species got out on the surface. Figure 4d shows the O 1s region with four peaks at ∼528, ∼529, ∼531−532, and    (Table 1). In addition, a certain segregation of the A site cation, especially in Sr, on the surface was observed in as-prepared and reduced precursors, which is in good agreement with previous works. 33,46,61 The characterization results suggest that a reductive pretreatment of LSC in 5 vol % H 2 /Ar at 650°C for 1 h was beneficial to achieving highly dispersed metallic Co nanoparticles on a metal oxide matrix. Consequently, it was proposed to start the catalytic tests under SRM and OSRM with a totally reduced LSC catalyst precursor.

Catalyst Characterization.
3.2. Catalytic Performance. 3.2.1. Effect of the Oxidation State of the Catalyst. Because the catalytic properties depend on the initial oxidation state of the catalyst, the effect of reductive thermal pretreatment of the catalyst precursor on the CH 3 OH conversion and product selectivity was investigated. Figure 5 shows the CH 3 OH conversion and CO 2 and H 2 selectivity under the SRM reaction, for an asprepared LSC and after a reductive pretreatment in 5 vol % H 2 /Ar at 650°C for 1 h. In addition, the catalytic properties obtained for an as-prepared LCO were also reported for comparison with the as-prepared LSC. As shown in Figure 5a, the significant CH 3 OH conversion of as-prepared LSC started at ∼300°C and gradually enhanced with an increase of the temperature, and complete conversion was reached around 400°C. In contrast, the catalytic activity of LSC after a reductive pretreatment at 650°C remarkably enhanced at a lower temperature, as the CH 3 OH conversion started around 250°C, and the total conversion was completed at ∼350°C. In the temperature range of 275−425°C, the CO 2 and H 2 selectivity progressively increased to about 65% and 90%, respectively (Figure 5b). Above 425°C, the selectivity of CO 2 and H 2 dropped with an increase in the temperature, exhibiting a similar tendency for both as-prepared and pretreated LSC. These results confirmed that the previous state of the catalyst governed the catalytic activity at low temperature. The low CO 2 and H 2 selectivity values at low temperatures indicated the importance of the MD reaction, which decreased with an increase in the temperature to reach a maximum value at ∼425°C . Above ∼425°C, RWGS contributed to diminishing the CO 2 and H 2 selectivity, evidencing the critical role of thermodynamics in the SRM reaction over this catalyst. At this point, as reported in previous works, 18 surface segregation of Co as Co 0 induced by the reductive environmental conditions is a fundamental step for active species activation. The role of reductive thermal pretreatment of the catalyst precursor on the activation of active species at the microscopic level will be discussed in the next section on catalyst characterization under SRM and OSRM.  On the other hand, the Sr influence on the SRM performance of LSC was compared to that of undoped LCO. As shown in Figure 5a, the CH 3 OH conversion of asprepared LCO started above ∼350°C, and the total conversion was completed at ∼550°C. Thus, the partial substitution of La by Sr reduced the temperature at which the catalyst started the SRM. These results were in good agreement with the TPR experiments, in which the TPR peaks of LSC were shifted toward lower temperatures concerning those of LCO. The values and trends of the CO 2 and H 2 selectivity for both LCO and LSC were close because the active phase exhibited similar behavior for both catalyst precursors (Figure 5b,c).

Effect of the Feed Gas Composition.
The feed gas composition in terms of the O 2 /H 2 O/CH 3 OH ratio was another important reaction variable that could strongly influence the catalytic activity. As shown in Figure 6, the increase of the H 2 O/CH 3 OH ratio from 1.3 to 4.0, under SRM conditions, led to a slight improvement of both the CH 3 OH conversion and selectivity of CO 2 and H 2 between 250 and 425°C. Above ∼400°C, a progressive drop of the CO 2 and H 2 selectivity with the temperature increase was observed for the H 2 O/CH 3 OH ratio of 1.3, which was associated with the RWGS effect. In contrast, the feed gas compositions with a larger excess of water exhibited a slight increase of both CO 2 and H 2 . This behavior was related to the equilibrium of the WGS, which was thermodynamically predominant above 400°C for high H 2 O/CH 3 OH ratios, thus contributing to the formation of CO 2 and H 2 at the expense of CO.
On the other hand, the effect of the H 2 O/CH 3 OH ratio was also significant to the stability of the catalyst. While no remarkable deactivation of the catalyst activity was observed using an H 2 O/CH 3 OH ratio of 4.0, a progressive activity loss with the time-on-stream reaction was detected for a ratio of 1.3. This was attributed to the carbon deposition because the carbon present in the catalyst was observed by SEM−EDS analysis. Therefore, the feed gas composition with the most water excess (H 2 O/CH 3 OH = 4.0) yielded better CH 3 OH conversion and CO 2 and H 2 selectivity and stability in the whole temperature range than that close to the stoichiometric one (H 2 O/CH 3

ACS Applied Energy Materials
www.acsaem.org Article studied. As shown in Figure 6, CH 3

Catalyst Characterization after SRM and OSRM.
In order to study the evolution of the phases formed at the catalyst under SRM and OSRM, all samples were characterized by using XRD. Figure 7 shows the XRD patterns of the asprepared catalyst and after testing under SRM and OSRM conditions at 600°C for 1 h and quenched in a N 2 atmosphere. After the SRM and OSRM reactions, the characteristic diffraction peaks of hexagonal La 2 O 3 , hexagonal La(OH) 3 , monoclinic La 2 O 2 CO 3 , cubic SrO, orthorhombic SrCO 3 , cubic Co 0 , and cubic CoO were detected, and no peaks of LSC were observed (Figure 7b,c). This evidenced that Co ions were reduced from the perovskite lattice to Co 0 nanoparticles, which were highly dispersed on a matrix of lanthanum and strontium oxides (Co/La 2 O 3 /SrO). The Co 0 crystallite sizes determined from the XRD data with Scherrer's formula were 20−30 nm. In addition, other species such as La 2 O 2 CO 3 , La(OH) 3 , SrCO 3 , and CoO were also formed under the SRM and OSRM conditions. The reactions of La 2 O 3 with H 2 O and SrO with CO 2 as reaction products generated La(OH) 3 and SrCO 3 (eqs 5 and 6, respectively).

SrO CO SrCO
As reported in previous works, 62 This mechanism described the good stability of the catalysts based on M/La 2 O 3 (M = Co, Ni, etc.) under reforming reaction conditions because active metal particles were decorated with lanthanum oxide species originating from the support. Regeneration of the active sites took place in the vicinity of lanthanum oxide particles with an intermediate formation of La 2 O 2 CO 3 . However, this mechanism in the LSC precursor showed limited efficiency in severe SRM conditions because catalyst deactivation was observed for nearly stoichiometric feed ratios (H 2 O/CH 3 OH = 1.3). Thus, the good catalyst stability exhibited under SRM for an H 2 O/ CH 3 OH ratio of 4, and also under OSRM, could be attributed to the influence of lanthanum oxycarbonate. As shown in Figure 7, the formation of lanthanum oxycarbonate was favorable in a high excess of water and an addition of O 2 , which reduced the potential catalyst deactivation caused by carbon deposition. SEM micrographs confirmed that both the dispersion and size of the particles after reduction of the catalyst precursor were more significant than those of the asprepared ones (Figure 7). Similarly, the specific surface area of the as-prepared catalyst precursor (6 m 2 g −1 ) was strongly increased to 25 and 28 m 2 g −1 after testing and quenching in N 2 under SRM and OSRM, respectively. The superior catalytic performance of OSRM compared to SRM was attributed to the higher surface area, higher metal dispersion, and smaller size of the Co 0 nanoparticles.

Catalytic Performance in Redox Cyclic Tests.
As mentioned above, the best resistance of the catalyst to deactivation due to carbon deposition was under OSRM reaction conditions (O 2 /H 2 O/CH 3 OH = 0.1−0.3/1.3/1.0). Therefore, these reaction conditions were selected to evaluate the effect of redox cycles on the catalytic activity and in situ self-regeneration of the catalyst, thus completely repeating several times the process of reduction and oxidation for the same sample. Figure 8 shows the CH 3 OH conversion and selectivity of CO 2   operating under OSRM and reoxidized in air at 900°C and then slow cooling to RT in air (Figure 8b). In the redox cyclic tests under OSRM, both the CH 3 OH conversion and H 2 selectivity were strongly decreased with an increase in the redox cycles. In contrast, the CH 3 OH conversion and selectivity of CO 2 and H 2 during the six cyclic tests with a specific thermal treatment remained almost constant, indicating that the conditions of the reoxidation process play a remarkable role in the performance and stability of the catalyst in successive redox processes.

Catalyst Characterization after Redox Cyclic
Tests. XRD analyses were carried out to determine the evolution of phases in the LSC precursor after redox cyclic tests under OSRM and in situ self-regeneration of the catalyst ( Figure 9). As shown in Figure 9, the XRD patterns after the first cycle exhibited the formation of the tetragonal phase [T′(La 1−x Sr x ) 2 CoO 4 ] due to the reaction of La 2 O 3 , SrO, and Co favored by slow cooling during the cyclic tests. In addition, a small fraction of La, Sr, and Co remained as La 2 O 2 CO 3 , SrCO 3 , and CoO. Therefore, the self-regeneration process of the catalyst in a redox cycle was not completed under the OSRM reaction conditions. However, the XRD results evidenced that the used catalyst after one redox cycle could be regenerated entirely under a reoxidation thermal treatment in synthetic air at 900°C for 12 h and finally in pure O 2 at 300°C for 72 h (Figure 9c). Complete regeneration of the used catalyst suggests that Co nanoparticles were highly dispersed on the matrix of lanthanum and strontium carbonates and oxides under OSRM, which could be interesting for the mitigation of Co coarsening and the deactivation of the catalyst in operation. In another series of experiments, degradation and regeneration were studied in successive redox cycles ( Figure   10). After six redox cycles, the XRD patterns showed higher amounts of the same compounds ([T′(La 1−x Sr x ) 2 CoO 4 ], La 2 O 2 CO 3 , SrCO 3 , and CoO) compared to the first cycle (Figure 10b,c). In this regard, the used catalyst after six redox cycles could not be regenerated entirely under a reoxidation thermal treatment in synthetic air at 900°C for 12 h and finally in pure O 2 at 300°C for 72 h (Figure 10d). In this case, the catalyst presented small amounts of (La 1−x Sr x ) 2 CoO 4 , and the presence of SrCO 3 and CoO was also detected; therefore, the irreversible segregation of Sr-rich species took place. SEM images of the as-prepared catalyst and after redox cyclic tests and thermal treatment of regeneration are shown in Figure 11. After the first redox cycle, the used catalyst and slow cooling in OSRM presented slightly smaller particle sizes than the as-prepared and regenerated catalysts (Figure 11a,b,d). However, after the sixth cycle, it can be observed that both the used catalyst (slow cooling under OSRM) and the regenerated catalyst exhibited the grains bonded together slightly, evidencing a slight sintering after redox cycles at high temperature (Figure 11a,c,e). In addition, the morphology of the used catalysts after the first cycle (Figure 10b,d) exhibited more porosity and larger voids between particles compared to the six cycles (Figure 11c,e). The interconnected porosity and voids between the particles enable the diffusion of reactant and product gas into the solid particles. These observations in the microstructures are consistent with the values of the specific surface area because the used catalyst after the first cycle presented a higher specific surface area (12 m 2 g −1 ) than the regenerated catalyst (7 m 2 g −1 ) and the as-prepared one (6 m 2 g −1 ). In contrast, the specific surface area significantly decreased during successive redox cycles, reaching values of

ACS Applied Energy Materials
www.acsaem.org Article 8 and 5 m 2 g −1 (after six cycles) for the slow cooling and regenerated catalysts, respectively. With the experimental observations discussed above, we can suggest how the different programs of the redox cycles (in terms of reoxidation and cooling from the operation temperature to RT) affect the regeneration of a catalyst. In the first program of the redox cycle, reoxidation of the catalyst with slow cooling under OSRM caused severe degradation of the catalytic activity in successive redox cycles. This was attributed to a remarkable segregation of phases [(La 1−x Sr x ) 2 CoO 4 , SrCO 3 , ...] with the accumulation of redox cycles in an irreversible way, of which the entire regeneration of the catalyst using a regeneration thermal treatment is difficult. As reported in previous studies of LSC for electrochemical applications such as SOFCs, 46,65 the accumulation of Sr-rich species, particularly SrCO 3 , on the catalyst surface covered the catalytically active Co surface sites, thus very effectively decreasing the catalytic activity of LSC. In the second program of the redox cycle, the use of a specific regeneration treatment in air at 900°C between consecutive cycles stopped degradation of the catalytic activity, which suggests that the regenerated LSC presented a surface quite similar to the original one.

CONCLUSIONS
SRM and OSRM over LSC perovskite, as a catalyst precursor, were studied to supply syngas enriched in H 2 for SOFC applications. The SRM performance of the catalyst precursor strongly depended on the initial oxidation state of the catalyst, feed gas composition, and reaction temperature. CH 3 OH conversion of the catalyst with a reductive pretreatment at 650°C was significantly enhanced at lower temperatures than the as-prepared one. Sr doping in LaCoO 3 -based perovskites facilitated the reduction process for the as-prepared catalyst precursor. Consequently, CH 3 OH conversion of LSC under SRM was increased at lower temperatures compared to undoped LaCoO 3 . MD and RWGS occurred simultaneously with SRM and OSRM. The SRM performance was also significantly affected by the feed gas composition. The feed composition with significant water excess (CH 3 OH/H 2 O = 4.0) yielded better CO 2 and H 2 selectivity (at temperatures of >400°C) than that close to the stoichiometric one due to the increases of the WGS reaction. In addition, this reduced the potential catalyst deactivation caused by carbon deposition. In OSRM, adding O 2 to SRM was an effective way of avoiding carbon deposition but at the expense of reducing the H 2 selectivity and increasing the CO 2 selectivity. The character-  ization results evidenced the stability of highly dispersed Co nanoparticles over a matrix composed of metal carbonates and oxides, mainly La 2 O 3 , La 2 O 2 CO 3 , SrO, and SrCO 3 . In addition, the formation of lanthanum oxycarbonate could also contribute to minimizing carbon deposition on the catalyst surface.
The identification of O loss and certain segregation of Srrich species on the reduced precursor surface by XPS evidenced a strong connection between the surface chemistry and catalytic activity. The reoxidation program in repetitive redox cycles played a crucial role in the regeneration of catalysts. A reoxidation with slow cooling under OSRM contributed to a severe degradation of the catalytic activity in successive redox cycles probably due to the accumulation of Sr-rich species (SrCO 3 ) on the catalyst surface, thus covering the catalytically active Co surface sites. In contrast, the reoxidized catalyst in air at 900°C and then slow cooling to RT between consecutive redox cycles stopped degradation of the catalytic activity. Furthermore, the catalyst precursor could be regenerated to the initial perovskite structure with a specific regeneration treatment (in air at 900°C for 12 h and in pure O 2 at 300°C for 72 h). The authors recognize that these results do not demonstrate long-term cycling stability of LSC; however, it can be an effective approach to mitigating the degradation of LSC, enabling its practical application in catalysis but also in other systems such as SOECs for renewable energy conversion and storage.