Boosting methane partial oxidation on ceria through exsolution of robust Ru nanoparticles

PAPER J. M. Serra et al . Boosting methane partial oxidation on ceria through exsolution of robust Ru nanoparticles Materials Advances rsc.li/materials-advances ISSN 2633-5409 Volume 2 Number 9 7 May 2021 Pages 2763–3122 2924 | Mater. Adv., 2021, 2, 2924–2934 © 2021 The Author(s). Published by the Royal Society of Chemistry Cite this: Mater. Adv., 2021, 2, 2924 Boosting methane partial oxidation on ceria through exsolution of robust Ru nanoparticles† A. J. Carrillo, L. Navarrete, M. Laqdiem, M. Balaguer and J. M. Serra * Finding sustainable routes for the transformation of CO 2 into fuels and added-value chemicals is key for mitigating greenhouse gas emission. In this respect, chemical-looping reforming coupled with CO 2 splitting emerges as a promising technology to produce syngas, using waste or solar heat as an energy source. It relies on metal oxides that act as redox intermediates and, thus, the stability and catalytic activity of the oxides are crucial. For that purpose, ceria has been widely used due to its superior multicyclic stability and fast CO 2 splitting kinetics. However, it also presents low capacity for oxygen exchange or supply compared with other oxides and slow methane partial oxidation kinetics, which is normally improved by cationic doping or catalytic surface activation via metal impregnation. The high temperatures (900 1C) required for these reactions lead to catalyst deactivation over time due to sintering of metallic clusters. In order to circumvent this issue, in this work we have utilized the exsolution method to create uniformly dispersed Ru nanoparticles (ca. 5 nm) that remain anchored to the cerium oxide backbone, guaranteeing its microstructural stability and catalytic activity over prolonged cycling. We provide evidence for metallic Ru exsolution and further demonstrate the outstanding benefits of exsolved nanoparticles in the partial oxidation of methane following a chemicalloop reforming scheme, especially in the temperature range in which industrial waste heat could be used as an energy source to drive the reaction. Remarkably, at 700 1C surface functionalization with exsolved Ru nanoparticles enables high CO selectivity (99% versus 62% for CeO 2 ) and about 2 orders of magnitude faster H 2 production rates. The dispersion and size of the exsolved Ru nanoparticles were maintained after a durability test of 20 chemical loops at 900 1C, indicating their robustness. Overall, the results presented here point towards the unique characteristics of nanoparticle exsolution for preventing agglomeration, which could find application in other catalytic or electrochemical processes for target hydrocarbon production. 1. Introduction The intensification of chemical processes through the implementation of membrane reactors or chemical-looping schemes 1 is an attractive alternative for the sustainable production of fuels with reduced CO 2 footprint and increased exergy efficiency. 2,3 In particular, chemical-looping reforming of methane (e.g. natural gas or biogas) coupled with CO 2 splitting represents a unique niche for production of fuels utilizing two greenhouse gasesasreagents. 4 In the first step, CH 4 is partially oxidized with the lattice oxygen of a metal oxide, CeO 2 in this case (eqn (1); Fig. 1). This reaction results in the production of H 2 and CO, known as syngas, in a 2 :1 molar ratio, which is ideal for further processing into liquid fuels via the Fischer–Tropsch process. In the second step, the reduced metal oxide is reoxidized with CO 2 , producing CO (eqn (2)). Importantly, the first step, which is an endothermic reaction occurring at high temperatures (800–1000 1C), could be driven with waste 5 or concentrated solar heat 6–8 CH4þ1=dCeO2!CO þ2H2 ðÞþ1=dCeO2d(1) CO2þ1=dCeO2d!1=dCeO2þCO (2) Net reaction: CH4þCO2!2CO þ2H2(3) This 2-step cyclic process presents several operational advantages. First, the main side reaction in thermocatalytic dry reforming, reverse water gas shift, is avoided. Secondly, carbon depositions caused by methane cracking in the first step could be transformed into CO during the second step by the Boudouard reaction. 9 To guarantee the feasibility of the Instituto de Tecnologı ´a Quı ´mica, Universitat Polite `cnica de Vale `ncia-Consejo Superior de Investigaciones Cientı ´ficias, Avenida Los Naranjos s/n, 46022 Valencia, Spain. E-mail: [email protected] †Electronic supplementary information (ESI) available. See DOI: 10.1039/ d1ma00044f Received 18th January 2021, Accepted 23rd February 2021 DOI: 10.1039/d1ma00044f rsc.li/materials-advances Materials Advances PAPER Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online View Journal | View Issue © 2021 The Author(s). Published by the Royal Society of Chemistry Mater. Adv., 2021, 2, 2924–2934 | 2925 process, it is key to find stable redox materials that could withstand repeated high-temperature thermochemical cycling. In this respect, ceria has been widely studied due to its high structural stability at high temperatures and fast CO 2 splitting kinetics, 6,10 based on its performance tested in solar-driven reactors. 9,11–13 However, the limited oxygen-exchange capacity of ceria could hamper its further applicability. In order to overcome this problem, doping with other cations, such as Zr 14 or La, 5 has been employed in order to increase the reducibility of cerium oxides and, subsequently, the fuel yields obtained by H 2 O and/or CO 2 splitting. Morphological modifications 15 or combinations in the form of composites with other metal oxides such as perovksites 16 or binary oxides 17,18 have been also successfully tested. Additionally, during the methane partial oxidation (MPO) step, most literature reports show syngas selectivity far from 100%, 6 which results from the combustion of CH 4 with readily available oxygen anions at the CeO 2 surface producing CO 2 and H 2 O. In order to solve this problem, surface promotion with metallic nanoparticles is commonly used in chemical looping reforming. 2 In the case of CeO 2 , noble metals such as Pt or Rh 19 or Ni in Ti-doped CeO 220 have been employed to suppress the formation of oxygenated molecules and increase H 2 and CO selectivity. In the latter case, Ni decorated Ti-doped CeO 2 resulted in CO selectivity (S CO ) values of ca. 80%. 20 These metallic catalysts are commonly incorporated in the oxide surface via impregnation methods; however, at the high temperatures needed to drive the reduction of the metal oxide with methane, metal nanoparticles can suffer sintering with neighboring particles, which affects its catalytic activity during prolonged operation. 21 In this sense, the exsolution method emerges as a promising alternative to obtain more durable metal nanoparticles with high catalytic activity. Nanoparticle exsolution consists of migration, under reductive atmospheres, of metal cations contained in the oxide lattice to the oxide surface, forming nanoparticles that remain anchored to the oxide backbone. 22 The fact that the nanoparticles are socketed into the oxide surface imparts beneficial effects in terms of stability, avoiding nanoparticle sintering. 23 For this reason, exsolution has gained considerable attention in the field of solid oxide fuel cells and electrolyzers, 24,25 and more recently its application has been found to be promising in methane reforming reactions. 26–29 For instance, Kousi and co-workers obtained exsolved and submerged Ni nanoparticles on La 0.8 Ce 0.1 Ni 0.4 Ti 0.6 O 3 which synergistically enhanced the oxygen supply capacity. 26 Additionally, Co incorporation into the B-site enables the activation of methane at temperatures as low as 450 1C. 30 For simultaneous methane partial oxidation coupled with CO 2 splitting, Dimitrakopoulos et al. used a membrane reactor in which the exsolved nanoparticles were formed in situ during the reaction. 31 More recently, Carrillo et al. demonstrated the beneficial effects of Co exsolution in the La 0.6 Sr 0.4 Cr 0.8 Co 0.2 O 3 perovskite for syngas production, leading to faster MPO and CO 2 splitting rates, higher syngas selectivity and stable nanoparticle size over 28 chemical loops. 32 Regarding ceria and other fluorite type structures, the exsolution method has been scarcely explored. Recently, Naeem et al. demonstrated that Ru exsolution could boost the catalytic stability of Sm 2 Ru x Ce 2x O 7 materials for methane dry reforming, especially when compared with Sm–Ce pyrochlores with impregnated Ru particles, denoting the beneficial effects of nanoparticle anchoring via exsolution. 27 Based on these recent discoveries, in this work, we have applied the exsolution method to create robust Ru nanoparticles that could boost syngas production via chemical looping reforming of methane coupled with CO 2 splitting, which to date has not been tested for ceria-based materials. The results reported herein denote the benefits of nanoparticle exsolution and could open the path for designing more stable cerium catalysts for other thermocatalytic applications or solid oxide electrochemical reactors and electrolyzers. 33–36 2. Results and discussion 2.1. Physicochemical insights into Ru nanoparticle exsolution on CeO 2 Ruthenium doped cerium oxide, Ru 0.01 Ce 0.99 O 2 (Ru–CeO 2 ), was prepared via a modified version of the Pechini method (see the Experimental section). For benchmarking, CeO 2 was also synthesized following the same method. According to the X-ray diffractograms depicted in Fig. 1, pristine Ru–CeO 2 and CeO 2 presented solely the reflections of fluorite CeO 2 , space group Fm3m. Interestingly, incorporation of Ru into the fluorite crystal induced a lattice expansion according to the shift of the main reflection to lower 2yvalues, pointing to the presence of a solid solution. The cubic lattice parameter, a, was calculated for both materials and was found to be 5.405 and 5.421 Å for CeO 2 and Ru–CeO 2 , respectively. This result is somehow counterintuitive since the ionic radius for Ce 4+ in 8-fold coordination is Fig. 1 Schematic of the chemical looping reforming of methane process studied in this work. In the first step, CH 4 reacts with the lattice oxygen from CeO 2 , resulting in the formation of CO and H 2 . In the second step, CO 2 reacts with the reduced oxide (CeO 2d ), replenishing the oxygen vacancies formed in the first step, and, subsequently, forming CO. Paper Materials Advances Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online 2926 | Mater. Adv., 2021, 2, 2924–2934 © 2021 The Author(s). Published by the Royal Society of Chemistry 0.97 Å, whereas for Ru 4+ (in 6-fold coordination) it is 0.64 Å, 37 which would suggest lattice shrinkage. However, this result is in accordance with previous works that reported a lattice expansion with Ru doping. 38,39 In order to confirm the lattice expansion upon Ru substitution we analyzed the materials with Raman spectroscopy (Fig. 2b). Due to the high symmetry of the cubic fluorite structure, CeO 2 has a unique Raman band ascribed to a one triply degenerate oxygen breathing mode of F 2g symmetry. 40 This mode normally occurs at 465 cm 1 ; however, under the instrumental conditions used here, a red laser of 785 cm 1 wavelength, the F 2g peak for undoped CeO 2 is at 463.7 cm 1 . Upon Ru incorporation, the mode slightly shifted towards lower wavenumbers, 463.3 cm 1 , which corroborates the lattice expansion observed by XRD. Interestingly, no bands associated with oxygen vacancies, the region between 550 and 600 cm 1 , 27 were observed for Ru–CeO 2 . Fig. 2a also shows the XRD diffraction patterns for Ru–CeO 2 material after the exsolution process performed at 900 1C for 2 h under a 5% H 2 atmosphere, revealing that the fluorite crystal phase prevailed after the reductive treatment without the appearance of any secondary phase. However, when scrutinizing the materials’ morphology before and after exsolution, important differences were observed. Fig. 3 depicts the SEM images of Ru–CeO 2 before and after the exsolution treatment. Fig. 3a reveals that Ru–CeO 2 is formed by aggregates of necked micron-sized particles. Fig. S1 (ESI†) shows the EDX elemental analysis of the pristine material. However, after 2 h reduction in 5% H 2 , well-dispersed nanoparticles emerged over the oxide surface indicative of nanoparticle exsolution (Fig. 3b). Turning back to the XRD diffraction, the exsolution treated Ru–CeO 2 sample (exRuCeO 2 ) experienced a shift towards higher 2y values, denoting lattice shrinkage. That is to say, exRuCeO 2 presented a cubic lattice parameter, a, of 5.413 Å, whereas for the pristine sample it was 5.421 Å. This slight lattice contraction was also observed in the Raman spectra depicted in Fig. 2b. The F 2g peak position for exRuCeO 2 is at 464.5 cm 1 , which is one wavenumber higher compared to that for Ru–CeO 2 . In addition, exRuCeO 2 exhibited the Raman modes ascribed to oxygen vacancies at around 550–600 cm 1 . 41 This indicates that the exsolution treatment in the reductive atmosphere triggers the creation of oxygen vacancies in addition to the exsolved nanoparticles. Although both XRD and Raman analyses point toward lattice contraction upon nanoparticle exsolution, the origin of this phenomenon is not clear. Previous reports on reduction of Ru–CeO 2 at 500 1CinaH 2 atmosphere showed a slight lattice expansion upon reduction; 38 however, in that work the presence of exsolved Ru nanoparticles was not inferred. Under the present reductive atmosphere, the lattice oxygen reacts with H 2 giving rise to electrons and oxygen vacancies, as demonstrated by Raman (Fig. 2b), which in the Kroger–Vink notation can be written as H2gðÞþOX O!V Oþ2eþH2OðgÞ(4) Subsequently, the oxidation state of cerium cations will be lowered according to Ce 4+ +e  -Ce 3+ . 42 Commonly, it has Fig. 2 X-ray diffraction patterns (a) and Raman spectra (b) for Ru–CeO 2 before and after exsolution performed at 900 1C for 2 h under 5% H 2 flow (exRuCeO 2 ) and for CeO 2 . Raman spectra were collected with a 785 cm 1 laser. The inner sketch in (b) depicts the vibration of oxygen anions around the Ce cation associated with the F 2g Raman mode. Fig. 3 SEM images of Ru-doped CeO 2 samples: (a) pristine and after exsolution treatment for (b) 2 h and (c) 5 h at 900 1C under a 5% H 2 atmosphere. (d) Schematic depicting the morphology transformations at each stage. Materials Advances Paper Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online © 2021 The Author(s). Published by the Royal Society of Chemistry Mater. Adv., 2021, 2, 2924–2934 | 2927 been believed that lattice expansion of CeO 2 upon reduction has been ascribed to the larger ionic radii of Ce 3+ (1.143 Å) compared to that of Ce 4+ (0.97 Å) cations. However, Muhich demonstrated that Ce 3+ –O 2 elongation can be compensated by Ce 4+ –O contraction, with the lattice expansion being driven by non-counterpoised forces. 43 Anyhow, the nucleation of exsolved metallic nanoparticles compensates to a certain extent the expansion with contraction, 42 which for this particular case, Ru 4+ +4e  -Ru 0 , could be higher explaining the slight lattice contraction upon metallic exsolution. In order to shed more light on this matter, we analyzed these three materials with Temperature Programmed Reduction (TPR). The TPR results (Fig. S2, ESI†) showed a common peak at ca. 800 1C peak maximum, ascribed to the reduction of the Ce 4+ bulk species. 44 Interestingly, for both CeO 2 and RuCeO 2 this peak is located at 800 1C. Normally, partial substitution of Ce 4+ in doped ceria materials leads to a shift of this peak toward lower temperatures due to the creation of oxygen vacancies that facilitate oxygen diffusion. 45,46 However, in this case Ru-doping did not result in the same effect, which might corroborate that Ru cations are located in interstitial positions, rather than partially substituting the Ce 4+ cations, which results in the lattice expansion. After the exsolution treatment (5% H 2 ,9001C, 2 h) of the exRuCeO 2 material, the main peak shifted to 769 1C. This shift to lower temperatures is ascribed to the presence of oxygen vacancies, as corroborated by Raman spectra. This fact is in principle beneficial for the MPO step, indicating that exRuCeO 2 could outperform CeO 2 at lower temperatures. Additionally, with the TPR analyses it was possible to determine the oxygen exchange/ supply capacity of the materials (see the Experimental section for more details). Our results indicate that the oxygen supply capacity of exRuCeO 2 was equal to 0.18 mol O mol oxide1 , whereas for CeO 2 it was 0.09 mol O mol oxide1 . Despite the increased value for exRuCeO 2 with respect to CeO 2 ,the oxygen supply capacity is still below the values reported for perovskites, for instance ca. 0.4 mol O mol oxide1 for La 0.8 Ce 0.1 Ni 0.4 Ti 0.6 O 3d . 30 SEM analysis (Fig. 3b) of the sample reduced for 2 h in H 2 at 900 1C reveals the formation of exsolved nanoparticles (dB2–5 nm), well dispersed over all the facets of the oxide. Interestingly, when increasing the exsolution treatment up to 5 h (Fig. 3c) the nanoparticle size increases, as previously observed in Ni exsolution in perovskites, 47,48 and grow mainly in the edges of the oxide particles, which progressively change from a rounded-like shape into a more faceted morphology caused by a longer exposure to the reductive environment. The changes in the exsolved nanoparticle habitus depending on time are reflected in the schematic depicted in Fig. 3d. We now turn to TEM to further analyze the size and level of anchorage of the exsolved Ru nanoparticles. Fig. 4a depicts some grains of exRuCeO 2 and the presence of exsolved nanoparticles over the oxide backbone. At higher magnifications (Fig. 4b), it is possible to observe that the nanoparticle is embedded in the oxide backbone, confirming the level of anchorage characteristic from metal nanoparticle exsolution. 49 Fig. 4c shows the HAADF-STEM image of exRuCeO 2 ,inwhichit is possible to observe the high and uniform dispersion of exsolved nanoparticles over the oxide backbone. EDX was utilized to further explore the nature of the exsolved nanoparticles (Fig. 4d), by analyzing two different zones in Fig. 4c, one pointing to an exsolved particle (point 1) and the other pointing to the oxide surface (point 2). By comparison, one can observe the higher concentration of Ru on the exsolved nanoparticles. Although Ru exsolution on Sm 2 Ce 2 O 7 pyrochlores has been recently reported, 27 to the best of our knowledge this work represents the first demonstration of Ru exsolution from the CeO 2 fluorite host. To further explore the nature of the exsolved nanoparticles we performed X-ray photoemission spectroscopy (XPS; Fig. 5). In order to analyze the oxidation state of Ru, we checked the Ru3p core shell, since Ru5d overlaps with C1s. 34,38,39 Fig. 5a compares the Ru3p spectra for Ru–CeO 2 before and after exsolution (i.e., for exRuCeO 2 ). The Ru3p 3/250 binding energy for Ru–CeO 2 was observed at 463.26 eV, whereas after exsolution, it shifted to 461.64 eV, indicative of the Ru metal. 34,38,39 This shift indicates that for the Ru–CeO 2 sample Ru is in the 4+ oxidation state, and after the 2 h reduction in 5% H 2 at 900 1C, all the Ru at the surface is in the metal state, corroborating that the exsolved nanoparticles observed by SEM (Fig. 3) and TEM (Fig. 4) are solely composed of the Ru metal. XPS analysis of Ce3d core-level spectra are depicted in Fig. 5b. The 3d 5/2 and 3d 3/2 components correspond to v and u, respectively, whereas v00,v 000 and u00,u000 are the satellites, respectively. Additionally, the doublets v 0 ,u 0 and v0,u 0correspond to Ce 3+ . 51–53 A detailed description of the deconvolution of each of these components can be found in the ESI†(Tables S1–S3,). According to the spectra depicted in Fig. 5b, it can be inferred that there is little presence of Ce 3+ in the surface of Ru–CeO 2 , which slightly increases after exsolution, as revealed for exRuCeO 2 . The O1s spectra depicted in Fig. 5c indicate that for the three materials the major component is Fig. 4 (a and b) TEM images of Ru-doped CeO 2 after exsolution treatment for (b) 2 h at 900 1C in a 5% H 2 atmosphere. (c) HAADF-STEM image for the same material and (d) EDX point analysis of an exsolved nanoparticle (point 1) and the bare oxide surface (point 2). Paper Materials Advances Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online 2928 | Mater. Adv., 2021, 2, 2924–2934 © 2021 The Author(s). Published by the Royal Society of Chemistry located at binding energies around 529 eV, which is generally ascribed to the lattice oxygen in oxides. 51 In addition, for Ru–CeO 2 and exRuCeO 2 there is a feature around 531 eV representative of oxygen anions in the vicinity of oxygen vacancies, indicatingthepresenceofoxygenvacancysitesinthesurfaceof these materials. For the case of CeO 2 , the components around 533 and534eVcouldbeascribedtoadsorbedspecies. In this section, we have presented the thorough physicochemical characterization of the materials before and after exsolution, confirming the presence of anchored metallic Ru nanoparticles over cerium-based oxides. In the following, the performance of exRuCeO 2 for chemical looping methane reforming coupled with CO 2 splitting will be interrogated, paying special attention to the reaction rate improvement and nanoparticle stability after prolonged cycling. 2.2. Improving methane partial oxidation with Ru exsolution Chemical looping methane reforming coupled with CO 2 splitting reactions was performed in a fixed-bed tubular quartz reactor and the gas production was monitored via mass spectrometry. First, we focus on determining the impact of Ru nanoparticles on syngas production, and for that, we benchmark exRuCeO 2 against CeO 2 . As commented in the introduction, one of the main drawbacks of CeO 2 for chemical looping reforming is that it presents slow syngas production during the partial oxidation of methane, which is the first step of this redox cyclic process (Fig. 1). In this first set of experiments, chemical looping reforming coupled with CO 2 splitting was assessed at three different temperatures (900, 800, 700 1C) for both materials, i.e.,exRuCeO 2 and CeO 2 . Here, it should be noted that the exsolution process was carried out inside the reactor, prior to the chemical looping reforming tests. We first assessed the influence of temperature on the chemical looping reforming performance of CeO 2 . Fig. 6a depicts the reactions occurring at 900 1C, which start with the injection of a stream of 5 vol% of CH 4 , resulting in the formation of H 2 and CO, which are the products of the partial oxidation of methane, and CO 2 ,which results from the complete oxidation of CH 4 with the lattice oxygen. For further information, Fig. S3 (ESI†) also shows the CH 4 curve in order to show the methane consumption, as well as the O 2 production curve, which clearly illustrates the absence of this gas during the experiments. The formation of CO 2 occurred at the beginning of the reaction, which is indicative of CH 4 reaction with readily available oxygen species at the surface of the ceria. 54 The CO 2 production in the first stage of the reaction lasted for ca. 1 min. Simultaneously, the formation of H 2 and CO occurred steadily until CH 4 injection was suppressed. At 900 1C, CO selectivity, S CO , was equal to 80% because of the prominent formation of CO 2 ; see Fig. 7 for the selectivity data at the different temperatures tested. The peak gas production rate was 2.9 and 1.3 mL min 1 g 1 for H 2 and CO, respectively, resulting in a H 2 :CO ratio slightly higher than the stoichiometric 2:1 proportion, indicative of CH 4 decomposition into solid C and additional H 2 formation. Once CH 4 injection was complete, Ar was flushed to purge the reactor for 2 min, followed by the injection of CO 2 (5 vol%). As mentioned in the introduction (Fig. 1), the methane partial oxidation step generates oxygen vacancies in the cerium oxide, since the lattice oxygen reacts with CH 4 . Thus, in the oxidation step, CO 2 reacts with the reduced cerium oxide (CeO 2d ), replenishing most of the oxygen vacancies with oxide anions, and subsequently forming CO, as observed in the second step in Fig. 6a. Once CO 2 is injected inside the reactor, a fast CO production is observed (peak production rate of 18.2 mL min 1 g 1 ), that lasted for 2 min until CO 2 splitting reaction reached equilibrium. This fact indicates that this CeO 2d -reoxidation step could be performed with shorter CO 2 injections, which will eventually increase the CO 2 conversion per cycle, denoting the high extent of operational tunability and room for optimization of this technology. The influence of temperature on the chemical looping reforming activity of CeO 2 is depicted in Fig. 6b (reaction at 800 1C) and Fig. 6c (reaction at 700 1C). Here, we would like to note that lowering the process temperature is beneficial to leverage industrial waste heat to drive the endothermic methane partial oxidation reaction. 5 However, by lowering the reaction temperature, the driving force for the formation of oxygen vacancies in CeO 2 and the activation of CH 4 are also decreased since, thermodynamically, oxygen vacancy formation Fig. 5 X-ray photoemission spectra of (a) Ru3p, (b) Ce3d and (c) O1s core levels of CeO 2 and Ru-doped CeO 2 before (Ru–CeO 2 ) and after exsolution of Ru nanoparticles (exRuCeO 2 ). Materials Advances Paper Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online © 2021 The Author(s). Published by the Royal Society of Chemistry Mater. Adv., 2021, 2, 2924–2934 | 2929 in ceria is favored at higher temperatures. 54 This fact implies that lower reduction extents, d, are achieved during the methane partial oxidation, which directly affects the CO production extent (yield and formation rate) in the subsequent CO 2 splitting step (eqn (1) and (2)). This effect is illustrated in Fig. 6b and c, i.e. lowering the reaction temperature drastically diminished the amount of H 2 and CO produced in the first step, and hence the CH 4 conversion, with peak H 2 production rates of 0.76 and 0.18 mL min 1 g 1 , denoting one order of magnitudedecreaseinthesyngasproductionofCeO 2 by lowering 200 1C. This temperature decrease also affected S CO ,which decreased to 74% and 62% at 800 and 700 1C, respectively. Fig. S4 (ESI†)(azoom-outofFig.6c)revealsthattheCO 2 productioncurveisevenhigherthantheH 2 and CO ones. In addition, the CO 2 splitting rate and CO yield notably decreased at lower temperatures, as an effect of the lower amount of Fig. 7 Comparison of (a) selectivity, (b) CH 4 conversion during the MPO step, and (c) CO 2 conversion during the CO 2 splitting step for the two materials tested. Fig. 6 Chemical looping gas production curves for CeO 2 at (a) 900 1C, (b) 800 1C and (c) 700 1C, and exRuCeO 2 at (d) 900 1C, (e) 800 1C and (f) 700 1C. Each chemical loop consisted of 5 min methane partial oxidation (injection of 5% CH 4 , total flow 100 mL min 1 ) and 5 min CO 2 splitting (injection of 5% CO 2 , total flow 100 mL min 1 ). The exsolution of exRuCeO 2 was performed inside a quartz reactor. For that purpose, Ru–CeO 2 was reduced under 5% H 2 for 2 h at 900 1C. Paper Materials Advances Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online 2930 | Mater. Adv., 2021, 2, 2924–2934 © 2021 The Author(s). Published by the Royal Society of Chemistry vacancies created in the methane activation step, which directly correlates with the amount of CO produced in the second step. The CO yields obtained were 14.1, 3.5 and 0.5 mL g 1 for 900, 800 and 700 1C, respectively. In addition, for the reaction at 900 1C, there is a fraction of CO that was generated through the Boudouard reaction, in which deposited coke is gasified with CO 2 ,increasingCOyield.Insummary,theseresultsprovide evidence about the main challenges of CeO 2 in chemical looping reforming, i.e. (i) low activity for methane partial oxidation; and (ii) syngas selectivity far from 100%, both being drastically decreased with lowering the process temperature. Fig. 6d–f show the results of catalytic tests using exRuCeO 2 as a redox material at 900, 800 and 700 1C, respectively. By comparing Fig. 6a and d, the benefits of surface functionalization by exsolved Ru nanoparticles become noticeable. At 900 1CtheH 2 production rate for exRuCeO 2 is 33.1 mL min 1 g 1 with a H 2 production of 92 mL g 1 . In contrast, the values achieved for CeO 2 are 2.9 mL min 1 g 1 and 13.70 mL g 1 ,whichareca. one order of magnitude lower. In addition, the gas production curve when using exRuCeO 2 starts with a pronounced peak that progressively decays, which totally differs from the flattened profile exhibited for syngas production with CeO 2 . The amount of CO produced is 46.6 mL g 1 ,resultinginaH 2 :CO ratio of 1.98 close to the stoichiometric 2: 1. In comparison CeO 2 showed a H 2 :COratioof 2.17 at 900 1C, due to CH 4 cracking into H 2 (g) and C(s). This is indicative of a higher resistance to carbon depositions ascribed to the presence of exsolved nanoparticles, as previously reported in the literature. 23 Fig. 6a also shows that there is concomitant production of CO 2 , however, in a minor extent when compared with CO, resulting in a high S CO of 97%. These results highlight the benefits that exsolved Ru nanoparticles impart on improving the methane partial oxidation performance. However, it should be also kept in mind that exRuCeO 2 allocates oxygen vacancies prior to the reaction since during the exsolution treatment H 2 was utilized to reduce the material and drive the exsolution of Ru nanoparticles. Warren and Scheffe pointed towards surface oxygen vacancies as active sites for methane activation, 55 and recently demonstrated how the increase of oxygen vacancies in CeO 2 , in their case by not reaching full conversion in the previous CO 2 splitting oxidation step, could greatly enhance the rate of syngas production. 9 The higher syngas yield in the first step is an indicative of a higher extent of reduction for exRuCeO 2 when compared with CeO 2 , triggered by the presence of Ru. As a result, in the subsequent CO 2 splitting step, the CO yield (Fig. 6d) was also higher than for CeO 2 (Fig. 6a) with 14.1 and 66 mL g 1 , respectively. It can be also observed that for exRuCeO 2 , the CO 2 splitting reaction did not achieve equilibrium in the 5 min CO 2 injection, opposite to the behavior of CeO 2 , which reached full oxidation after 2 min exposure. Similar to what happened to CeO 2 , lowering the temperature also affected the kinetics and syngas yield, although to a minor extent (Fig. 6e and f). In particular, the performance observed at 700 1C is remarkable (Fig. 6f), which is the temperature closer to the range needed for utilization of industrial waste heat as an energy source. For instance, the peak H 2 production rate reaches 23.9 mL min 1 g 1 for exRuCeO 2 , whereas for CeO 2 it is 0.18 mL min 1 g 1 . This 2-order of magnitude increase in the syngas production rate at 700 1C illustrates the high catalytic activity of the Ru nanoparticles for the MPO into syngas. In addition, exRuCeO 2 displayed S CO B99% at 700 1C. Fig. 7b shows that methane conversions (X CH4 )B32% can be reached for exRuCeO 2 at 700 1C whereas for CeO 2 just 1% conversions are observed. Nevertheless, it should be noted here that our experiments were not targeted for reaching higher conversions of CH 4 or CO 2 , which could be improved by shortening the injection time or increasing the solid loading. For the CO 2 splitting step, just a slight decrease is observed in the peak production rate when lowering the temperature. That is to say, CO production rates were 24.9, 23.2 and 22.4 mL min 1 g 1 at 900, 800 and 700 1C, respectively. However, since by lowering the temperature the amount of oxygen vacancies generated in the previous reduction step was also decreased, it was observed that CO 2 splitting reaction reached equilibrium before the 5 min CO 2 injection was complete. That is to say, at 700 1C, CO 2 splitting reached equilibrium after 3 min of exposure to CO 2 (Fig. 6f) ascribed to a lower amount of oxygen vacancies that should be replenished. This also affected the CO 2 conversion values (X CO2 ) which increased with increasing temperature (Fig. 7c). The highest conversion was for exRuCeO 2 at 900 1Cwithca. 45%, whereas for CeO 2 it was ca. 16% at the same temperature. 2.3. Chemical activity and stability test of exsolved Ru nanoparticles In the previous section, the benefits of functionalizing the CeO 2 oxide surface with Ru nanoparticles became visible, with a remarkable reaction activity improvement when compared with bare CeO 2 . Now, we assess the stability of the reaction over prolonged cycling at a fixed temperature as well as the stability of the Ru nanoparticles in harsher conditions. For that purpose, we conducted a 20-cycle chemical looping reforming test at 900 1C. The reaction conditions were modified with respect to the previous section, viz., a 10-fold higher concentration of CO 2 (50 vol%) was used for the 2nd step. The motivation for the increase of the CO 2 partial pressure was 2-fold. The first was in order to assess the reaction in more realistic conditions in which more concentrated CO 2 stream would be used. Second, we observed in Fig. 6d that with 5 vol% the CO 2 splitting reaction did not reach equilibrium in the given 5 min CO 2 injection period. By increasing the CO 2 partial pressure faster kinetics are expected, 56 achieving full oxidation in the given time. By this way, the material will be fully oxidized in the subsequent reduction step under CH 4 , and, thus, the kinetic effect provided by the oxygen vacancies acting as active sites at the surface will be suppressed. Raman (Fig. 2b) and XPS analyses (Fig. 5c) proved that exRuCeO 2 allocates oxygen vacancies before the first MPO reaction, which Warren and Scheffe showed to be vacancymediated. 55 The syngas production curves during the 20-cycle chemical looping reforming test are shown in Fig. 8a, in which the syngas (H 2 and CO) produced during the MPO and the CO produced during the CO 2 splitting step are depicted. It can be observed that over the 20 cycles, the chemical activity of the Materials Advances Paper Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online © 2021 The Author(s). Published by the Royal Society of Chemistry Mater. Adv., 2021, 2, 2924–2934 | 2931 exRuCeO 2 material is preserved, although some differences with respect to the chemical looping test in Fig. 6d are visible. First, regarding the CO production during the CO 2 splitting step; with the increase in the vol% of CO 2 , the peak production rate is greatly enhanced. That is to say, in Fig. 6d we observe a peak production rate of 24.9 mL min 1 g 1 when using 5 vol% of CO 2 , whereas when increasing the concentration up to 50 vol%, the rate increases up to 189. 9 mL min 1 g 1 , which is about a 7-fold increase. However, this high value was not maintained during the whole cycling test, and after a small decrease in the second cycle, it shows stable values at around 156 mL min 1 g 1 from the 6th to the 20th cycle. To put these results into perspective, Ruan et al. reported 168.8 mL min 1 g 1 using 100 vol% CO 2 and Ti-doped CeO 2 impregnated with metallic Ni. 20 In addition, on increasing the vol% of CO 2 the reaction reached equilibrium in ca. 1 min (see Fig. S5 in the ESI†). This fact indicates that complete equilibration in CO 2 via re-oxidation is achieved, and, thus, in the subsequent MPO step, the effect of additional oxygen vacancies – generated upon exsolution – as active sites for the activation of methane could be ruled out. Indeed, when comparing the syngas production curves between the first methane injection and the second, this effect becomes evident. That is to say, during the first MPO reaction, the H 2 production is 34.4 mL min 1 g 1 , which is very close to the reported value in Fig. 6a wherein the same conditions of 5 vol% of CH 4 and 900 1C were employed. However, in the second cycle the peak H 2 production rate lowers down to 23.4 mL min 1 g 1 , which is ascribed to the initial lack of surface oxygen vacancies that help in activating MPO. Thus, after the first cycle, since high CO 2 concentrations are used and, hence, full oxidation achieved, the catalytic improvement for the MPO could be solely ascribed to the exsolved Ru nanoparticles. Importantly, from the 2nd to the 20th cycle the MPO performance was stable (Fig. 8a) with a H 2 production rate around 23.4 mL min 1 g 1 . Next, we assessed the microstructural stability of the exsolved Ru nanoparticles after the 20-cycle longevity test depicted in Fig. 8a. For that purpose, we analyzed with TEM the materials after reaction (Fig. 8b and c). In Fig. 8b, we provide evidence for the existence of Ru nanoparticles that remained well dispersed over the oxide surface after the multi-cycle treatment at high temperatures. Since the images were taken after the last CO 2 splitting step, this also proves that the exsolved Ru nanoparticles are not dissolved back into the lattice under these experimental conditions used in the oxidation step. This is probably due to the short exposure (5 min) to a less oxidative atmosphere (CO 2 ) when compared to air, which is the gas used in re-dissolution treatments. Thus, under these oxidative conditions there is not enough driving force to produce the dissolution of the exsolved Fig. 8 (a) H 2 and CO gas production curves obtained from a durability test carried out for exRuCeO 2 , previously exsolved in a tubular furnace at 900 1C for 2 h under 5% H 2 . The test consisted of 20 chemical loops, with each of them with first 5 vol% CH 4 injection for 5 min, followed by a 2 min purge under pure Ar, and 50 vol% CO 2 injection for 5 min. The total flow rate was always kept at 100 ml min 1 . (b and c) TEM images of exRuCeO 2 after the 20 chemical loops of methane reforming coupled with CO 2 splitting, illustrating the stability of the exsolved Ru nanoparticles. In (c) the interplanar distance of the exsolved particle after 20 cycles is illustrated. Fig. 9 (a) XRD diffractograms comparing exRuCeO 2 before and after the 20-cycle chemical looping reforming test. (b) Raman spectra comparison of exRuCeO 2 before and after 20 cycles. They grey-shaded region is zoomed out in (c) to show the shift of the F 2g peak towards lower Raman shift values. The Raman spectra were collected with a 785 cm 1 laser. Paper Materials Advances Open Access Article. Published on 02 March 2021. Downloaded on 3/30/2022 9:11:16 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online