Preferential oxidation of CO on a La-Co-Ru perovskite-type oxide catalyst

1 Preferential oxidation of CO on a La-Co-Ru perovskite-type oxide catalyst Rosa Pereñíguez,1,2 Alfonso Caballero,1 and Davide Ferri2,3 1 Instituto de Ciencia de Materiales de Sevilla (CSIC-Universidad de Sevilla) and Dpto. Química Inorgánica (Universidad de Sevilla), Av. Américo Vespucio, 49, E-41092 Sevilla, Spain 2 Swiss Federal Laboratories for Materials Science and Technology, Empa, Ueberlandstrasse 129, CH-8600 Dübendorf, Switzerland 3 Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland This document is the accepted manuscript version of the following article: Pereñíguez, R., Caballero, A., & Ferri, D. (2017). Preferential oxidation of CO on a La-Co-Ru perovskitetype oxide catalyst. Catalysis Communications, 92, 75-79. https://doi.org/10.1016/j.catcom.2016.12.020 https://www.elsevier.com/about/open-science/open-access/journal-embargo-finder Für Elsevier Artikel: Creative Commons: Attribution-NonCommercial-NoDerivs 2 Abstract A Ru-containing perovskite-type oxide La(Co,Ru)O3 of nominal composition LaCo0.8Ru0.2O3 was prepared by ultrasonic spray combustion and tested for the preferential oxidation of CO (PROX). EXAFS indicated that Ru adopted the coordination environment of Co in LaCoO3 while Co was present as LaCoO3 and Co3O4. PROX activity was replaced by CO hydrogenation activity above 250 ºC. Short oxidation at 500 ºC between temperature programmed reaction ramps did not restore the initial La(Co,Ru)O3 structure but generated a catalyst with improved PROX activity compared to the initial La(Co,Ru)O3. Under reductive PROX conditions the material experienced structural changes that improved its overall catalytic activity only if the catalyst was oxidized after each temperature programmed ramp. Keywords: Ru; LaCoO3; PROX; CO methanation; reduction; structural regeneration 3 Introduction Perovskite-type oxides (ABO3) are acknowledged as efficient catalysts for oxidation reactions [1-5]. It was recognized early that when doped with transition and precious metals this class of mixed oxides exerts a stabilizing function towards the dopant to avoid formation of volatile oxides and catalyst degradation. To this end, ruthenium was the major focus of studies already in the early 70’s as it became a potential candidate for the removal of NOx from treatment of exhaust gases of vehicles [6-8]. Moreover, the possibility to produce perovskite-type oxides with catalytically relevant B-site doping elements was shown to protect the metals from particle growth in automotive catalysis as a consequence of reversible segregation-incorporation within the perovskite-type structure [9]. Recently, the interest for Ru-doped perovskite-type oxides has been placed rather on high temperature applications such as fuel reforming [10-18] because of the thermal stability of perovskite-type oxides and the intrinsic activity of ruthenium. Fierro and co-workers [11-15] have shown that addition of Ru to LaCoO3 improves especially the stability of the catalyst to high temperature reaction conditions. Because of the reducing environment of these reactions, the reducible LaCoO3 can be used as an ideal catalyst precursor. During reaction or pretreatment, La2O3 or La2O2CO3 supported Co and Ru particles are generated and constitute the final catalyst. The improved catalytic activity of Ru was attributed to the enhanced reducibility of LaCoO3 by Ru and to the consequent improved formation of Co, Ru and possible Co-Ru metallic phases. This strategy could be exploited to generate suitable materials for catalytic processes operating at low and moderate temperatures. Therefore, in this work we use Ru-doped LaCoO3 prepared by ultrasonic spray combustion [19, 20] as a possible catalyst for preferential oxidation of CO in excess hydrogen (PROX) in a broad temperature range. Ru is an active metal for this reaction [21] but it also catalyzes CO hydrogenation to CH4 occurring at higher temperature than PROX [22]. Besides exploiting the reducibility of LaCoO3 4 enabling the segregation of Ru, we show preliminary data on the structural regeneration of the catalyst, under sequential temperature programmed reaction and oxidation steps. Experimental An aqueous solution of La(NO3)3·6H2O (Fluka, >99%), Co(NO3)2·6H2O (Sigma–Aldrich, >98%), RuCl3·xH2O (Aldrich >99%) and citric acid (Riedel-de Haën, >99.5%) in the necessary molar ratio to obtain LaCoO3-δ and LaCo0.8Ru0.2O3-δ (La(Co,Ru)O3) was atomized in a ultrasonic nebulizer and transported by air flow through a vertical quartz reactor heated to 600 ºC [23]. The ratio between the amount of citric acid and the total amount of metal ions was 1. The resulting oxide powders were collected on a heated membrane filter (150 ºC) and calcined at 600 ºC for 4 h. X-ray diffractograms (XRD) were recorded with a Philips X’pert Pro MPD diffractometer using Ni-filtered Cu Kα radiation (= 1.5418 Å). In situ XRD data during temperature programmed reduction (H2-TPR) were collected at steady state in 5 vol% H2/Ar up to 700 ºC (10 ºC/min) and 70 vol% H2/N2 up to 500°C (5 ºC/min) at temperature increments of 25 ºC (2θ range, 24–56º; step size, 0.05º). The BET specific surface area was measured from N2 adsorption–desorption isotherms at -196 ºC using a Micromeritics ASAP 2020 instrument. Temperature programmed reduction (H2-TPR) was performed in a setup equipped with a thermal conductivity detector (TCD) to quantify the H2 concentration and an online mass spectrometer (MS) to monitor products formation [24]. CuO was used for quantification of the H2 consumption. After heating in 5 vol% H2/Ar to 700 ºC (10 ºC/min), cooling to room temperature and purging with He, the sample was heated in 3 vol% O2/He to 700 ºC (10 ºC/min). This procedure was repeated three times and the reduction profiles were labelled as TPR-1, TPR-2 and TPR-3. Transmission extended X-ray absorption fine structure (EXAFS) 5 spectra were collected at the Co (7709 eV) and Ru (22117 eV) K-edges (BM01B beamline, ESRF, Grenoble) using Co and Ru foils for energy calibration. The preferential oxidation of CO (PROX) was carried out by heating the catalyst (50 mg) in a quartz tubular reactor to 500 ºC (1 ºC/min) in 1 vol% O2, 1 vol% CO and 70 vol% H2, 100 mL/min (He balance). After cooling to room temperature, the catalyst was heated to 500 ºC in 20 vol% O2/He followed by 5 min dwell at 500 ºC and cooling to room temperature. Five activity-oxidation cycles were performed that were followed by five activity cycles without intermediate oxidation (Figure S1). The effluent gas was analyzed by gas chromatography (Varian GC-3800; TCD; Molecular Sieve 5 Å and Porapak®-N columns). CO and O 2 conversion, CO2 yield and CH4 yield were determined as follows: X(CO) (%) = 100  (nCOin-nCOout)/nCOin X(O2) (%) = 100  (nO2in-nO2out)/nO2in Y(CO2) (%) = 100  nCO2out/nCOin Y(CH4) (%) = 100  nCH4out/nCOin where ni is the amount of every component (i) in moles. Results and Discussion The XRD pattern of La(Co,Ru)O3 (30.2 m2/g) calcined at 600 ºC exhibited only the reflections of the rhombohedral structure of LaCoO3 (ICDD, 01-084-0847, R-3c; Figure 1). The missing doublets characteristic of the rhombohedral structure reflected the low degree of criystallinity as a result of the low synthesis and calcination temperatures. The significant shift to lower 2 values of the reflections compared to LaCoO3 suggests that Ru was inserted at the B-site [12]. The edge energy (7723 eV) of the Co K-edge XANES spectrum of 6 La(Co,Ru)O3 (Figure 2) was intermediate between that of LaCoO3 (Co 3+, 7724 eV ) and Co3O4 (Co 2+/3+, 7717 eV). The whiteline presented a second peak attributable to Co3O4. Beside the Co-O coordination shell at 1.5 Å, which did not match that of Co3O4 (1.7 Å), the peak at ca. 3.2 Å that is due to multiple contributions of Co-La, Co-Co, and Co-O-Co bonds was weaker than in LaCoO3 in the FT-EXAFS spectrum (Figure 2) suggesting a large degree of disorder induced by Co substitution [25]. The contribution at 2.5 Å resembled the coordination environment of cobalt in Co3O4. These observations suggest the presence of a mixed local environment of Co and the coexistence of LaCoO3 and Co3O4 phases in La(Co,Ru)O3. Its dispersion and amorphous character, reflecting the structural disorder observed in the FT-EXAFS of La(Co,Ru)O3 explain the absence of Co 3O4 in the XRD (Figure 1). The Ru K-edge XANES spectrum of La(Co,Ru)O3 (Figure 2) did not match the spectra of reference materials with nominal Ru oxidation state +3 (RuCl3) and +4 (RuO2). It was composed of two major peaks due to 1s→5px and 1s→5py(z) transitions in distorted RuO6 octahedra in perovskites [26], implying that Ru adopted the oxidation and coordination states of the B-site. By comparison of the position of the maximum of the first derivative of La(Co,Ru)O3 with that of compounds with oxidation state +3 (RuCl3), +4 (RuO2), +4.5 (Pb2Ru2O6.5) and +5 (Ca2YRuO6) (Figure 2)[27], we assign Ru an oxidation state close to +5 and octahedral coordination. This oxidation state could be justified by the similar ionic radius of Ru5+ (0.705 Å) in octahedral coordination and Co3+ (0.685 Å) compared to lower oxidation states of Ru (Ru4+: 0.76 Å or Ru3+: 0.82 Å) [28]. The slightly larger radius of Ru5+ induces the shift to lower diffraction angles in the XRD. Because no Ru K-edge spectrum was reported, we could not compare our spectra with those measured by Mota et al. [13], who proposed Ru+4 in LaCo0.8Ru0.2O3. The high oxidation state of Ru is consistent with the large degree of disorder evident from the FT spectrum of the Co K-edge and the partial reduction 7 of Co3+ to Co2+ related to the presence of Co3O4. Charge compensation/redistribution and the unit cell distortion can suggest the Co segregation if all Ru adopted the coordination state of Co and was all integrated in the perovskite lattice. Assuming that Co was present only in oxidation state +3 in the perovskite phase, while Ru was in the oxidation state +5 and all exchanged in the perovskite phase, the stoichiometry of the B-site should be Co3+0.67Ru5+0.2 (overall oxidation state of B-site, +3) rather than the theoretical Co3+0.8Ru5+0.2 (+3.4). The difference of Co concentration in the perovskite is thus compensated by the formation of Co3O4. The H2-TPR profile of La(Co,Ru)O3 exhibited three major peaks (TPR-1, Figure 3) and Ru improved the reducibility of LaCoO3 in agreement with previous observations [12]. According to the H2 consumption values and after deconvolution of the different contributions of TPR-1 (Table S1), we assume that the reduction of the cobalt phase took place in two steps, i.e. Co2.95+ → Co2+ (375 ºC) and Co2+ → Co0 (>500 ºC). The first peak at 295 ºC is ascribed to the reduction of ruthenium. The values of theoretical (ΔnT+= 1) and experimental (ΔnE+= 0.81) extent of electron exchange for this step suggest that reduction of ruthenium may have been partial and could have been completed together with the subsequent step, the reduction of cobalt (375 ºC). For the peak at 375 ºC we measured an excess of H2 consumption (ΔnE+= 1.13 against ΔnT+= 0.76). Segregated Ru0 promoted likely the first reduction peak of cobalt (250-450 ºC) by hydrogen spillover. The second (TPR-2) and third (TPR-3) H2-TPR profiles were different compared to TPR-1. The two H2 consumption peaks below 400 ºC of TPR-1 merged in a sharp peak centered at ca. 350 ºC. Similarly, the clear event at 550 ºC became a shoulder of the peak at 605 ºC, which shifted to ca. 590 ºC. It is evident that TPR-2 and TPR-3 are very similar suggesting the structural reversibility of the material over redox cycles after the first reduction segment. Therefore, the 8 short oxidation at 700 ºC did not restore completely the initial structure obtained after synthesis and calcination at 600 ºC. In order to explain these differences, in situ XRD was carried out during H2-TPR in 5 vol% H2 (Figure 4). Ideally the structural changes should coincide with the reduction events observed in Figure 3. At ca. 400 ºC the rhombohedral LaCoO3 structure vanished in favor of the oxygen deficient La2Co2O5. La2O3 appeared already at 500 ºC; at 700 ºC there was no remaining reflection of a perovskite-type oxide. However, no Ru-phase and Co 0 were detected indicating that Ru and Co were well dispersed on La2O3. Therefore, the results confirm that the first H2-TPR peak corresponds to the reduction of Ru5+ and of Co 3+ (Ru/LaCoO3) to Co2+ (Ru/La2Co2O5), while the peak at higher temperature belongs to the complete reduction of cobalt and loss of the perovskite phase. Figure S2a and Figure S2b show the diffractograms of calcined La(Co,Ru)O3 and of the sample after one in situ XRD redox cycle, respectively. The diminished intensity of the reflections indicated that the short oxidation at 700 ºC and the subsequent cooling in the same environment generated a more disordered perovskite structure. However, without any other experimental evidence, we speculate that the same 2θ position of the diffraction peaks suggests that Ru reverted in the perovskite structure. La(Co,Ru)O3 was subjected to a sequence of consecutive temperature programmed ramps in the preferential CO oxidation (PROX) feed alternated to temperature programmed oxidation ramps (Figure S1). Under similar reducing conditions of PROX, an in situ XRD measurement with 70 vol% H2 (Figure S3) demonstrated the same phase transitions observed with 5 vol% H2 but at lower temperature: La2Co2O5 was observed at 200 ºC and reduction to La2O3 occurred at ca. 500 ºC. Figure 5 shows the CO conversion data for the preferential CO oxidation (PROX), while Figure S4 displays the O2 conversion and CO2 yield data. In the first PROX profile (R1) the 9 reaction was initiated above 200 ºC and 90% CO conversion was attained at 350 ºC. The relatively broad temperature window in which activity was measured allowed capturing a change of reaction regime from the preferential CO oxidation to CO methanation above 300 ºC [29]. The activity profiles obtained after the first oxidation including heating and cooling segments to 500 ºC (R2-R5) demonstrated that the temperature regime for PROX shifted to lower temperature by ca. 100 ºC (CO conversion, Figure 5; O2 conversion, Figure S4). Full CO conversion was reached at 300 ºC just before activity loss. The kink in the CO conversion at 200 ºC corresponded to the maximum CO2 yield. Comparison of Figure 5 and Figure S4 suggests that between 50 and 200°C PROX dominated over H2 combustion though the stoichiometry of the reactions indicates that the latter was occurring parallel to PROX to some extent. CO methanation replaced CO oxidation above 250 ºC after the first activity run (Figure S4). The loss of activity above 350 ºC mirrored the decrease of CH4 yield. The activity profiles remained substantially unchanged after R2 confirming that the structural changes did not proceed further. This observation is in agreement with the H2-TPR data of Figure 4 showing that consecutive TPR profiles after the first and reoxidation (700 ºC) were very similar. In absence of further in situ characterization data, the observed activation can be explained as follows. The reduction of the perovskite-type structure during the temperature programmed reaction under reducing conditions (70 vol% H2) to 500 ºC caused the reduction and segregation of Ru0 and the formation of a Ru0-Co0/La2O3 catalyst. We cannot conclude whether a Ru-Co alloy was formed. After oxidation at 500 ºC, that is milder than the temperatures of synthesis and initial calcination, part of the material was restored into A- and B-site deficient LaCoO3 and this was likely not sufficient to revert completely to the initial oxidation and coordination environment, however this treatment results in a better catalyst. The subsequent H2-TPR experiments (Figure 4) then could correspond predominantly to 16 Figure 2 17 Figure 3 18 Figure 4 19 Figure 5