Syngas production by CO₂ reforming of methane using LnFeNi(Ru)O₃ perovskites as precursors of robust catalysts

Abstract

LnFeNi(Ru)O₃ perovskites synthesized via a modified Pechini method have been studied as catalyst precursors in CO₂ reforming of methane. The structural and redox properties along with catalytic performance were found to depend upon Ln nature and Ru content in the perovskite. Active and stable catalysts have been formed from perovskite precursors in reaction media at high temperatures due to Fe–Ni–(Ru) alloy particles segregation from the perovskite lattice and their stabilization on the oxide surface. Coking stability of these catalysts in dry reforming of real natural gas containing admixtures of higher hydrocarbons is explained by combined effects of Ni dilution by Ru in alloy nanoparticles and high mobility and reactivity of surface/lattice oxygen in the disordered Ln ferrite support.

---

1. Introduction

The carbon dioxide reforming of methane has recently attracted considerable attention as a process generating synthesis gas due to both environmental and commercial reasons.^1,2 Dry reforming transforms cheap undesirable greenhouse gases and is particularly important in the case of biogas or gas fields containing a significant amount of methane and carbon dioxide. The synthesis gas produced in this reaction having the H₂/CO ratio close to unity is a favorable feedstock for further chemical processes such as the Fischer–Tropsch synthesis and production of oxygenates followed by the liquid fuel synthesis.^3–7

The most active catalysts for methane dry reforming (MDR) are based on the noble metals (Rh, Ru, Ir, Pt) or nickel.^3–12 The major problem of MDR hindering its industrial application is the coking of catalysts and, as a result, their deactivation in the case of Ni-containing catalysts.^3–9 Nevertheless, due to much lower cost of nickel compared with noble metals, the development of Ni-based catalysts resistant to coking is profitable from the commercial point of view. Different approaches have been considered up to now to minimize the coke deposition on catalysts.^{3–9,13–24} Among them an attractive option is the use of perovskite-like oxides ABO₃ as the catalyst precursors when formation of highly dispersed active metal fixed in the oxide matrix occurs under MDR reaction conditions as a result of perovskite reduction.^13–24 The structural and thermal stability as well as catalytic activity of such systems can be adjusted by the substitution with various cations on A and/or B sites. For Ni-containing perovskite-type oxides of general formula (A, A′ – La, Ca, Sr, Ce, Pr; B′ – Fe, Co, Ru; x, y – the substitution degree) as precursors of MDR catalysts, a high dispersion of Ni depending on the nature of A, A′ and/or B′ cations as well as participation of mobile oxygen of the oxide matrix in the reaction thus preventing the catalyst coking and increasing its activity and stability have been demonstrated.^13–24 For example, in the case of La_1−xA_xNiO₃ (A = Ce, Pr) perovskites, La_0.9Pr_0.1NiO₃ demonstrated the highest catalytic activity and resistance to coking in MDR explained by a higher dispersion of Ni in the reduced catalyst as well as redox properties of praseodymium oxide.¹⁶ For Ln_1−xCa_xRu_0.8Ni_0.2O₃ (Ln = La, Sm, Nd) catalysts, change in the nature and composition of cations in the A-site was shown to strongly affect stability and selectivity of catalysts.²² A partial substitution of Ni with Co,^17,18 Fe^19–21 and/or Ru^22–26 leading to formation of Ni–Me alloys under reducing MDR reaction conditions improves catalyst activity and stability. Thus, for the LaNi_1−xRu_xO₃ (x = 0, 0.1, 0.2, 1) catalysts, partial substitution of Ni with Ru hampers coking.²⁴

In our previous study of LnFe_0.7Ni_0.3O_3−δ (Ln = La, Pr, Sm) perovskites²⁷ under realistic conditions of MDR, their transformation into composites comprised of Ni–Fe alloy particles and LnO_x epitaxially bound with remaining Ln–Fe–O perovskite particles was demonstrated. This microstructure plays an important role in MDR through activation of CO₂ on the oxide sites and transfer of active oxygen-containing species to alloy particles where they interact with CH_x fragments producing syngas. The nature of Ln cation affects both the composition of the Ni–Fe alloy and oxygen mobility/reactivity in perovskites, PrFe_0.7Ni_0.3O_3−δ being the most active and stable catalyst.

This work presents results on synthesis and characterization of structural and catalytic properties in dry reforming of LnFe_1−xRu_xNi_0.3O_3−δ (Ln = La, Pr, x = 0–0.1) perovskites as catalyst precursors. The impact of the Ln cation nature and partial substitution of Fe with Ru on the real structure of perovskites, their reduction features and catalytic activity and stability in MDR has been elucidated.

2. Experimental

Perovskites were synthesized by a Pechini route using metal (Me) nitrates, Ru(OH)Cl₃, citric acid (CA), ethylene glycol (EG) and ethylenediamine (ED) as reagents.^28,29 The molar ratio of CA : EG : ED : Me was 3.75 : 11.25 : 3.75 : 1. CA and metal nitrates were dissolved in ethylene glycol at 80 °C and in distilled water at room temperature, respectively. The prepared solutions were mixed together at room temperature under stirring followed by addition of ED. The mixed solution was stirred for 60 min and then heated at 70 °C for 24 h allowing for the gel formation. The gel was calcined at 600 °C for 2 h, then the obtained solid was ground and annealed at 900 °C.

The samples of perovskites were characterized by XRD, BET, TEM with EDX, H₂ and CH₄ temperature-programmed reduction. X-ray diffraction patterns were obtained with an ARLX′TRA diffractometer (Thermo, Switzerland) using CuKα radiation in the 2θ scanning range of 15–80° with a step of 0.05°. The structural refinement has been carried out using PCW software (version 2.4). Qualitative phase analysis has been carried out by using PDF-2–ICDD files and the ICSD/retrieve database. BET specific surface area (SSA, m² g⁻¹) was determined from the data on Ar thermodesorption. The TEM micrographs were obtained with a JEM-2010 instrument (lattice resolution 1.4 Å and acceleration voltage 200 kV). Local elemental analysis was performed with an EDX method (a Phoenix Spectrometer). The X-ray photoelectron spectra (XPS) were recorded on a photoelectron spectrometer, KRATOS ES 300, with Mg K_α radiation (1253.6 eV) at pressure 1 × 10⁻⁸ Torr. Preliminarily all samples were ground in an agate mortar and deposited on conductive carbon tape.

The redox properties of perovskites were studied using temperature-programmed reduction (TPR) by H₂ (10% H₂ in Ar, feed rate 2.5 L h⁻¹, temperature ramp from 25 to 900 °C at 10 °C min⁻¹) or CH₄ (1% CH₄ in He, feed rate 10 L h⁻¹, temperature ramp from 25 to 880 °C at 5 °C min⁻¹). The experiments were carried out in kinetic installations equipped with GC Tcvet-500 and a PEM-2M analyzer (IR absorbance and electrochemical gas sensors). In the case of repeated red–ox cycles, samples were reoxidized in the flow of O₂ for 1 h at 500 °C.

Methane dry reforming (MDR) was studied in the temperature-programmed mode (1% CH₄ + 1% CO₂ in He, contact time 0.005 s) as well as in steady-state conditions (10% CH₄ + 10% CO₂ in He, contact time 0.01–0.015 s) at temperatures up to 800–850 °C in a flow installation. The long-term testing was performed in the reaction mixture 45% CO₂ + 45% natural gas +10% N₂ at 800 °C and contact time 0.24 s. Prior to testing, the catalysts were activated in the reaction media at 850 °C for 2 hours. The concentrations of reactants and products were analyzed by using GC Tcvet-500. Temperature-programmed oxidation (TPO) in the feed of 1% O₂ in He was used to evaluate the amount of carbon formed over catalysts during long-term testing.

3. Results and discussion

3.1. Structural and textural features

The X-ray diffraction patterns for the initial samples are shown in Fig. 1 and 2. While samples without Ru (PrFe_0.7Ni_0.3O_3−δ and LaFe_0.7Ni_0.3O_3−δ) are single-phase orthorhombic perovskites,²⁷ the partial substitution of Fe with Ru in LnFe_0.7−xNi_0.3Ru_xO_3−δ (x = 0.05, 0.1) leads to appearance of NiO and the praseodymium or lanthanum oxychloride [JCPDF 88-0064] phase and distortion of perovskite structure (Table 1). The formation of oxychlorides is caused by using Ru(OH)Cl₃ as a source of Ru, which is confirmed by the increase in the oxychloride reflections intensity with increase in Ru concentration, being more pronounced for La-containing samples. Simultaneously, especially in the case of LaFe_0.7−xNi_0.3Ru_xO_3−δ samples, the noticeable growth of NiO reflections is observed (Fig. 1). The shift of all perovskite reflections towards lower angles shows the unit cell expansion due to partial segregation of nickel from perovskite and insertion of larger Ru³⁺ cations into the structure (0.068 nm versus 0.06 and 0.0645 nm for Ni³⁺and Fe³⁺³⁰).

Fig. 1 XRD patterns of LaFe_0.7−xRu_xNi_0.3O_3−δ: as prepared (1–3); reduced in H₂ (4–6); x = 0 (1, 4); x = 0.05 (2, 5); x = 0.1 (3, 6). NiO (), LaOCl (▼), La₂O₃ (), Ni–Fe alloy (◆), Ni–Fe–Ru alloy (*), Ru (+), Ru–Ni alloy (●). Reduction conditions: 5% H₂ in He, 800 °C, 2 h.

Fig. 2 XRD patterns of PrFe_0.7Ni_0.3O_3−δ (1, 3, 5) and PrFe_0.6Ru_0.1Ni_0.3O_3−δ (2, 4, 6): as prepared (1, 2), reduced in H₂ (3, 4) and used in MDR (5, 6). NiO (), PrOCl (▼), Pr₂O₃ (v), Pr_xO_y (#), Ni–Fe alloy (◆). Reduction conditions: 5% H₂ in He, 800 °C, 2 h. MDR conditions: 800 °C in the feed 50% CO₂ + 50% natural gas, contact time 0.2 s.

Table 1 Unit cell parameters^a, secondary phases^a and BET^b for perovskites LnFe_0.7−xNi_0.3Ru_xO_3−δ (Ln = La, Pr, x = 0.05, 0.1), calcined at 900 °C

Perovskite	a (Å)	b (Å)	c (Å)	V (Å^3)	Secondary phase	SSA, m^2 g^−1
a The structural refinement has been carried out using PCW software (version 2.4). Qualitative phase analysis has been carried out by using PDF-2–ICDD files and the ICSD/retrieve database. b From the data on Ar thermodesorption.
LaFN	5.511	7.808	5.537	238.25	—	7
LaFNRu0.05	5.529	7.817	5.547	239.74	LaOCl, NiO	7
PrFN	5.522	7.742	5.462	233.54	—	4
PrFNRu0.05	5.574	7.779	5.480	237.61	PrOCl, NiO	7
PrFNRu0.1	5.584	7.777	5.478	237.89	PrOCl, NiO	4

---

XRD patterns of perovskites reduced in H₂ at 800 °C (Fig. 1 and 2) show the reflections in the range of 2θ ∼ 44–44.5° corresponding to the Ni–Fe(Ru)-alloy²⁷ and metal Ru,²³ as well as those of orthorhombic perovskite, rare-earth oxides (La₂O₃ [JCPDF 83-1344] or Pr₂O₃, Pr_xO_y) and lanthanum or praseodymium oxychloride. The position and intensity of perovskite reflections are practically the same as in the initial and reduced samples showing, in agreement with literature data,^22–24 that reduction of Ru-containing perovskites is hampered as compared with Ru-free samples.²⁷ However, the presence of La₂O₃(Pr₂O₃, Pr_xO_y) reflections indicates some decomposition of perovskites considering the invariance of La(OH)Cl₂ reflections and only some decrease in Pr(OH)Cl₂ reflection intensity after reduction. The reflections of the metal phase are broad and of low intensity, which makes their analysis difficult. However, some information on the composition of the metal phase could be deduced. The reflections of the Ni–Fe alloy shift to higher angles with the increase in Ru concentration due to its incorporation into the alloy. Thus, in the case of the LaFe_0.65Ni_0.3Ru_0.05O_3−δ sample, the position of alloy reflection is rather close to that of the Ru-free Ni–Fe alloy (2θ = 44°) showing only a small amount of Ru in the alloy. In the XRD pattern of the LaFe_0.6Ni_0.3Ru_0.1O_3−δ sample, the reflection at 2θ = 44.2° indicates the presence of the Ni–Fe alloy enriched with Ru, the reflection at 44.35° corresponds to pure metal Ru and the reflection at 44.5° corresponds to the Ni–Ru alloy. The presence of metal Ru in this sample implies that a part of ruthenium in the oxidized sample is not incorporated into the perovskite structure, though in the XRD pattern of the initial LaFe_0.6Ni_0.3Ru_0.1O_3−δ perovskite no reflections of individual ruthenium oxide are detected.

3.2. H₂ TPR

H₂ TPR patterns for reference samples (LaNiO₃ and NiO) and perovskites are presented in Fig. 3 and 4. An Asymmetric Gauss Decomposition method (AGDM) has been applied for the analysis of complex H₂-TPR profiles of perovskites. A single peak with a maximum at 400 °C is observed in the H₂ TPR pattern of NiO whereas LaNiO₃ shows two peaks: the first one centered near 337 °C is assigned to the reduction of Ni³⁺ to Ni²⁺ and the second peak at 495 °C corresponds to the reduction of Ni²⁺ to metallic nickel (Fig. 3).^17,21,24 For LaFe_0.7Ni_0.3O_3−δ, in agreement with known data,^19–21,27 the first peak shifts to lower temperatures and the second one (corresponding to a partial reduction of perovskite) shifts to higher temperatures (Fig. 3). Furthermore, the data of AGDM show the presence of a broad low-intensity peak in the range of 300–600° which could be assigned to the reduction of NiO traces not detected by XRD and the partial reduction of Fe⁴⁺ to Fe³⁺.^20,21,27 For LaFe_0.7−xNi_0.3Ru_xO_3−δ samples (Fig. 3), in the low-temperature region, three peaks of hydrogen consumption are observed: the low intensity peak at 260–270 °C and two overlapping peaks with the maximum at 315–370 °C and 400–410 °C. The first peak and the peak at 315–370 °C could be assigned to the reduction of Ni³⁺ to Ni²⁺ and Ru³⁺ to Ru²⁺ in perovskite.^24,31 The peak at 400–410 °C corresponds to the reduction of NiO: the increase in its intensity at x = 0.1 agrees with XRD data evidencing a higher NiO content in this sample (Fig. 1). The low-intensity peaks in the 500–700 °C region could be assigned to the reduction of Fe⁴⁺ cations to Fe³⁺ within the perovskite lattice.^20,21,27 An unresolved, intense, high-temperature peak above 700 °C is attributed to the partial destruction of perovskite structure with the formation of a metallic phase and La₂O₃ as confirmed by XRD data (Fig. 1). In the case of PrFe_0.7−xNi_0.3Ru_xO_3−δ series, the TPR profiles are similar (Fig. 4), however, as compared with La-based perovskites, the low-temperature peaks are somewhat shifted to higher temperatures while the high-temperature peaks are shifted to lower temperatures. The higher temperature of the partial reduction of Ni³⁺ (Ru³⁺) cations in Pr-based perovskites agrees with the data for Ru-free systems.^27,32,33 The shift of the high-temperature peak to lower temperatures is determined by the lower lattice stability of Pr-containing perovskites.³³

Fig. 3 H₂ TPR profiles for: 1, NiO; 2, LaNiO₃; 3, LaFe_0.7Ni_0.3O_3−δ; 4, LaFe_0.65Ru_0.05 Ni_0.3O_3−δ; 5, LaFe_0.6Ru_0.1Ni_0.3O_3−δ.

Fig. 4 H₂ TPR profiles for PrFe_0.7−xRu_xNi_0.3O_3−δ: 1, x = 0; 2, x = 0.05; 3, x = 0.1.

In general, when ruthenium is incorporated into perovskite, the peak of the partial reduction of Ni³⁺ (Ru³⁺) is noticeably shifted to higher temperatures (from 315 to 370 °C and from 354 to 380 °C for LaFe_0.7−xNi_0.3Ru_xO_3−δ and PrFe_0.7−xNi_0.3Ru_xO_3−δ, respectively (Fig. 3 and 4)). Higher stability of both nickel and ruthenium in the perovskite lattice suggests strong inter-cations interaction which agrees with published data.²⁴ Some poisoning effect of Cl⁻ anions in the surface layer is possible as well.

Repeated reduction–reoxidation experiments revealed some evolution of reactivity of perovskites apparently reflecting variation in their real structure/microstructure. In the case of perovskites without the Ru reoxidized after the 1st TPR run, the first peak at ∼315 °C remains only as a shoulder of a new intense asymmetric peak with the main maximum at ∼420 °C. Since the amount of hydrogen consumed in these low-temperature peaks corresponds to complete reduction of all Ni in perovskite to the Ni⁰ state, this clearly indicates the increase in sample reactivity after reoxidation (Fig. 5a). Nearly identical H₂ TPR spectra for the second and the third runs demonstrate reproducibility of sample reactivity after reoxidation. Since reduction of both NiO and Ni-containing perovskites to the metallic phase is a topochemical process including generation of Ni⁰ nuclei followed by their growth,³⁵ observed variation in perovskites reactivity after red–ox cycles can be explained by microheterogeneity of reoxidized perovskites caused by preferential location of Ni cations in the surface layer and within domain boundaries of defect La ferrite. For the LaFe_0.6Ni_0.3Ru_0.1O_3−δ sample, TPR spectra change from run to run reflecting the increase in reactivity (Fig. 5b). In the third run, spectra correspond to practically complete reduction of all nickel and ruthenium cations contained in the sample to the metal state. Since the position of the main peak at ∼400 °C is the same as that for the LaFe_0.7Ni_0.3O_3−δ sample, it clearly corresponds to the topochemical process of NiO_x species reduction. Peaks at 180–230 °C are due to reduction of RuO_x species formed as a result of ruthenium segregation from the perovskite during the first reduction,^23,24 while peaks at 340–390 °C can be assigned to reduction of mixed Ni–Ru oxide species.³⁴ The driving force for the increase of Ru-doped sample reactivity in red–ox cycles is clearly segregation of Ni and Ru in the surface layers as well as Cl removal from the surface.

Fig. 5 Repeated H₂ TPR after oxidation for LaFe_0.7Ni_0.3O_3−δ (a) and LaFe_0.6Ru_0.1Ni_0.3O_3−δ (b).

3.3 CH₄ TPR

Temperature-programmed reduction of perovskites by methane usually occurs at higher temperatures as compared with the reduction by H₂ due to a strong C–H bond in CH₄ molecules.^27,35 The profiles of CH₄ and products concentration during TPR for studied perovskites are presented in Fig. 6. The results show that incorporation of Ru in the perovskite strongly affects dynamics of its reduction. Thus, for LaFe_0.7Ni_0.3O_3−δ (Fig. 6a) and PrFe_0.7Ni_0.3O_3−δ²⁷ reduction starts by formation of deep oxidation products (H₂O or/and CO₂) at 480–540 °C. CO and H₂ evolution begins at 620–700 °C and is completed during TPR run resulting in segregation of the Fe–Ni alloy and the corresponding Ln oxide within the remaining perovskite.²⁷

Fig. 6 CH₄ TPR for perovskites LnFe_0.7−xRu_xNi_0.3O_3−δ: (a) Ln = La, x = 0; (b) Ln = La, x = 0.05; (c) Ln = Pr, x = 0.05.

For all Ru-containing perovskites, the start of reduction product evolution shifts to 730–810 °C, and concentration of CO and H₂ passed through the maximum and slowly declined with time (Fig. 6b and c). In agreement with H₂ TPR data, this evidences hampering of their reduction as compared with Ru-free perovskites. This could be caused by two reasons: inter-cation interaction stabilizing both nickel and ruthenium cations in the perovskite lattice²⁴ and the presence of oxychlorides which could block the surface of perovskite hampering the active metal formation required for methane activation. Further, the dynamics of perovskite reduction and composition of products are affected by the nature of Ln cations. Thus, for PrFe_0.65Ru_0.05Ni_0.3O_3−δ, though the reduction products appear at higher temperature as compared with La-based perovskite (775 and 730 °C, correspondingly), the maximum rate of CO and H₂ formation is achieved at somewhat lower temperature (820 °C vs. 830 °C), which could be due to a lower stability of Pr(OH)Cl₂ basic salt due to a lower basicity of Pr cations. In addition, in the case of Pr-based perovskite, along with CO and H₂, products of deep oxidation which are absent for La-containing perovskite are observed. This could be due to combustion of activated CH_x species by highly reactive oxygen of PrO_x.^16,27

3.4. Methane dry reforming

Activation of the catalysts. Since methane reforming reactions are well known to proceed on Ni⁰ or Ru⁰ metal sites, the activation of perovskite precursors must be done either via reductive pretreatment or directly in the reaction media.^19,27,36 The peculiarities of the catalyst activation in the course of MDR and their activity have been firstly studied using temperature-programmed reaction mode with the dilute feed (1% CH₄ + 1% CO₂ in He). For Ru-containing perovskites, the profiles of products concentration during MDR temperature-programmed reaction presented in Fig. 7 are, in general, similar to those for Ru-free samples. Higher temperatures of the products appearance agree with the CH₄-TPR data evidencing hampering of their reduction. Indeed, for Ru-free samples the reaction starts at 600–650 °C, whereas a partial substitution of Fe by Ru increases reaction starting temperature up to 680–700 °C (Fig. 8). In all cases, the consumption of CH₄ and CO₂ is followed by the formation of CO while hydrogen appears at higher temperatures (Fig. 7). Such a sequence of products formation could be explained by a higher share of reverse water gas shift reaction (RWGS) in the overall process at lower temperatures. The reagents consumption and products formation depend on the Ln nature: for PrFe_0.65Ru_0.05Ni_0.3O_3−δ, the steady-state concentrations are established at 880 °C in 30–40 min whereas, in the case of LaFe_0.65Ru_0.05Ni_0.3O_3−δ, the steady state is not attained even after 70 min.

Fig. 7 Typical curves for reagents (a, b) and products (c, d) concentration variation in the temperature-programmed reaction. The feed: 1% CH₄ + 1% CO₂ in He, contact time 0.001 s.

Fig. 8 Temperature-programmed MDR: reaction starting temperature. Reaction conditions: 1% CH₄ + 1% CO₂ in He, contact time 0.001 s.

Along with Ln nature, the activation process depends on the Ru content in the perovskites. The variation in methane and CO₂ conversion during activation of PrFe_1−xRu_xNi_0.3O_3−δ at 850 °C in the reaction mixture 10% CH₄ + 10% CO₂ in He (Fig. 9) evidences that it is retarding with increasing Ru content in perovskite from 0.05 to 0.1% mol. This is conditioned by a bigger amount of oxychloride in the samples with higher Ru concentration as revealed by XRD data (Fig. 1). Thus, dynamics of catalyst activation during MDR reaction is clearly controlled by the same factors controlling their reduction with methane, namely, nickel and ruthenium inter-cation coupling in the perovskite lattice as well as the amount and stability of corresponding oxychloride.

Fig. 9 Time dependence of CH₄ and CO₂ conversion in MDR over oxidized PrFe_0.7−xRu_xNi_0.3O_3−δ (1, x = 0.05, 2, x = 0.1) at 850 °C, the feed 10% CH₄ + 10% CO₂, He – balance, contact time 0.015 s.

Steady state activity. The typical temperature dependence of methane and CO₂ conversion and H₂/CO ratio along with equilibrium values are presented in Fig. 10 for the case of PrFe_0.65Ru_0.05Ni_0.3O_3−δ sample activated and tested in the reaction mixture containing 10% CH₄ + 10% CO₂ at contact time 0.015 s. The reagents conversions increase with the temperature remaining below equilibrium values due to a short contact time. Thus, a gradual decrease in both CH₄ and CO₂ conversions with the increase in space velocity (decrease in contact time) was observed for Ni and Ru-containing catalysts.^37,38 A higher conversion of CO₂ as compared to CH₄ in the whole temperature region is due to RWGS reaction occurring in parallel with MDR. CO₂ reacts with H₂ in the course of RWGS resulting in H₂/CO ratios lesser than 1.^24,39

Fig. 10 Temperature dependence of methane (squares) and CO₂ (circles) conversion (a) and H₂/CO ratio (b). Open symbols – equilibrium values, solid symbols – the catalyst PrFe_0.65Ni_0.3Ru_0.05O_3−δ activated in the reaction mixture at 850 °C. The feed 10% CH₄ + 10% CO₂, He – balance, contact time 0.015 s.

To analyze the effect of perovskites chemical composition and their pretreatment on the catalytic activity the specific effective first-order rate constant of the methane transformation in MDR was estimated using the model of a plug-flow reactor.⁴⁰ The values of specific rate constants of MDR at 800 °C for all catalysts activated in the reaction mixture and in the flow of H₂ at 800 °C are shown in Fig. 11. These data show that, as in the case of Ru-free samples,²⁷ the highest performance has been demonstrated by Pr-based catalysts, moreover, the activity of the catalysts increases with Ru concentration independently of the activation mode. However, the activity of catalysts activated in the reaction mixture is slightly lower due to the presence of PrOCl as detected by XRD (Fig. 2). The pretreatment of catalysts in hydrogen leads to segregation of active metal alloy particles and partial decomposition of PrOCl blocking active centers that increases the catalyst activity. Note that the values of specific rate constants for the samples activated in the reaction mixture and reduced samples are close due to comparatively easy decomposition of PrOCl which is completely destroyed during long term testing in the real reaction mixture (Fig. 2).

Fig. 11 Specific rate constant of CH₄ dry reforming over perovskites reduced at 800 °C in H₂ or activated in the reaction mixture. The feed 10% CH₄ + 10% CO₂, He – balance, 800 °C, contact time 0.01 s.

In contrast to Pr-containing samples, for LaFe_1−xRu_xNi_0.3O_3−δ catalysts, the activity order strongly depends on the activation mode. Thus, if the catalysts are activated in the reaction media, the next row of activity is observed: LFN > LFNRu_0.05 > LFNRu_0.1 (Fig. 11). In this case, the decrease of activity with increasing Ru concentration is caused by inter-cation coupling stabilizing both nickel and ruthenium cations in the perovskite lattice²⁴ as well as the rising content of stable LaOCl as shown by XRD data (Fig. 1). The particles of LaOCl blocking the perovskite precursor surface hinder the formation of active metal centers. The results evidencing the poor activation of Ru-containing LaFe_1−xRu_xNi_0.3O_3−δ samples in the reaction feed 10% CH₄ + 10% CO₂ in He agree with CH₄-TPR and MDR-TPR results (Fig. 6 and 7). The reversed activity order of catalysts is observed after the reductive activation in hydrogen. According to XRD data (Fig. 1), the perovskite reduction in H₂ at 800 °C leads to segregation of metal alloy particles which are the active centers for methane activation, and the activity of catalysts increases with Ru content.

Long-term testing and post-reaction analysis. The time-on-stream dependence of methane conversion and H₂/CO ratio during long-term stability testing of catalysts in the real feed is presented in Fig. 12. For Pr-based samples with and without Ru, a high methane conversion close to the equilibrium value is observed in the first 30 minutes after 2 hours pretreatment in the same feed. In the case of LaFe_0.65Ru_0.05Ni_0.3O₃, a high methane conversion is achieved only after 1–1.5 hours due to slow reduction of this sample (Fig. 6 and 7). Ru-doped catalysts demonstrate better performance stability, which can be explained by prevention of coking due to formation of Ni–Fe–Ru alloys. Moreover, formation of such alloys could also help to suppress coarsening of Ni clusters. A higher level of activity was obtained here for Pr-containing catalysts as well, which suggests the importance of these redox cations for efficient activation of CO₂ as well as providing higher oxygen mobility in the oxide support.

Fig. 12 Methane conversion and H₂/CO ratio in long-term tests of PrFe_0.7Ni_0.3O₃ (■), PrFe_0.65Ru_0.05Ni_0.3O₃ (▲) and LaFe_0.65Ru_0.05Ni_0.3O₃ (O). Dot line – equilibrium value. Reaction conditions: the feed 45% CO₂ + 45% natural gas + 10% N₂ at 800 °C, pretreatment in reaction media 2 h at 850 °C, contact time 0.24 s.

The amount of carbon deposits on Ru-containing catalysts during long-term testing has been determined using TPO. The results (not presented for brevity) show no CO₂ evolution up to 880 °C evidencing no coke formation over the catalysts during the reaction. The absence of any type of coke on the used catalysts has been confirmed by TEM study of PrFe_1−xRu_xNi_0.3O₃ samples (Fig. 13).

Fig. 13 TEM with EDX of used samples: (a) PrFe_0.7Ru_0.1Ni_0.3O₃, (b) PrFe_0.65Ru_0.05Ni_0.3O₃.

In agreement with XRD, TEM data show that the PrFe_0.6Ru_0.1Ni_0.3O₃ catalyst is comprised of Ni–Ru [JCPDF 06-663] nanoparticles (marked by 1 in Fig. 13a, lattice spacing 2.08 Å) strongly interacting with the surface of Pr ferrite PrFeO₃ [JCPDF 47-0065] domains (marked by 2 in Fig. 13a, (021) lattice spacing 2.66 Å). In the case of PrFe_0.65Ru_0.05Ni_0.3O₃, small alloy clusters are observed with a good epitaxy with the oxide particle containing both Pr and Fe (Fig. 13b), thus corresponding to the perovskite phase ((020) lattice spacing 2.79 Å). This certainly provides strong metal–support interaction required for preventing sintering and coking. Indeed, no carbon deposits or fibers were detected by TEM in catalysts discharged after reaction.

The XPS study of PrFe_0.7−xRu_xNi_0.3O_3−δ perovskites shows that the surface composition of catalysts does not vary appreciably after contact with the reaction feed (Table 2). The initial perovskite contains oxidized forms of ruthenium (III, IV) according to the well-defined Ru3d_5/2 peak observed at 282.6 eV (Fig. 14). The low-intense component at 280.5 eV could be assigned to very small (<2 nm) clusters characterized by noticeable relaxation processes in the photoemission that leads to shift of Ru3d peaks. The samples tested in MDR have mainly metal Ru in the surface as indicated by the well-defined Ru3d_5/2 peak observed at 280.0 and 280.3 eV for PrRu0.1 and PrRu0.05 samples, respectively. Mathematical analysis with Doniach–Sunjic and Gauss–Lorentz function revealed the presence of a low intensity component at 282.6 eV, which can be assigned to the Ru (III, IV) state. Since samples were transferred through air after discharging from the reactor, this feature can be explained by their partial oxidation due to this manipulation.

Table 2 The bulk and surface composition of initial and used PrFe_0.7−xRu_xNi_0.3O_3−δ

x	Bulk composition^a		Surface composition^b
	Ni/Pr	Ru/Pr	Ni/Pr	Ru/Pr
a Nominal values from preparation. b From XPS data.
0.1 (initial)	0.30	0.10	0.12	0.041
0.1 (used)	0.30	0.10	0.11	0.044
0.05 (used)	0.30	0.05	0.15	0.031

---

Fig. 14 The profile of Ru 3d core-level spectra of PrFe_0.6Ru_0.1Ni_0.3O_3−δ perovskites: initial (a) and used (b).

The oxidation state of Fe is identified by the Fe2p_3/2 core-level peak as corresponding to Fe(III). It remained almost unchanged after treatment in the reaction media followed by transfer through air. The state of Ni in the initial perovskite is characterized mainly by Ni2p_3/2 peaks at 854.1 and 856.1 eV corresponding to Ni(II) and Ni(III) states (Fig. 15). The additional component at 852.5–852.7 eV, corresponding to the Ni⁰ state, appears in spectra of samples after MDR reaction. With due regard to the presence of mainly metal Ru as well as XRD and TEM data, this Ni metal state can correspond to the Ni–Ru alloy. In addition, the main part of Ni is present in Ni(II) and Ni(III) states as indicated by components at 854.5 and 856.3 eV respectively. This feature can be again caused by the partial reoxidation of Ni metal under contact of the discharged catalyst with air, pure Ni⁰ particles (generated by reduction of the NiO admixture) being oxidized much easier as compared with Ni–Ru alloy nanoparticles.

Fig. 15 The profile of Ni 2p_3/2 core-level spectra of PrFe_0.6Ru_0.1Ni_0.3O_3−δ perovskites: initial (a) and used (b).

Table 3 presents methane conversion, H₂/CO ratio and amount of carbon deposits obtained at different GHSV and temperatures during MDR using an undiluted feed with CH₄/CO₂ = 1 for Pr(La)Fe_0.65Ru_0.05Ni_0.3O₃ catalysts and various catalysts reported in the literature.^{16,23,24,41–46} Note that the different conditions of catalysts testing make difficult the comparison of the catalysts activity, however, some conclusions could be drawn. The data show that performance of only Ni-based catalysts depends on the support nature thus the amount of coking changes from 1.2 (4.5%Ni/SiO₂)⁴¹ to 67% (6.6%Ni/Al₂O₃)⁴² while it is 0.06% in the case of 1.5%Ru/CeO₂–ZrO₂.⁴⁵ It is known that activity of Ni-containing catalysts and carbon formation over them are linked with the ensemble effect when, for reforming reaction, the ensemble of Ni atoms is smaller than that for nucleation of carbon whiskers,⁴⁷ thus, active and stable catalysts for MDR must contain nickel particles of high dispersion.^3–6,42 However, for catalysts based on pure nickel supported on different substrates sintering is a main problem which could be avoided by using alloys of nickel with other metals (e.g. Fe, Co, Ru).^3–12 Highly dispersed active metal alloy species of various composition fixed in the oxide matrix are formed under MDR reaction conditions as a result of perovskite reduction.^13–25 LaNi_1−xRu_x (x = 0–0.2),²⁴ Ln_1−xCa_xRu_0.8Ni_0.2O₃ (Ln = La, Pr, Sm)^23,25 have been used as precursors for bimetallic Ni–Ru catalysts containing metal particles distributed in the matrix of corresponding oxides (e.g. CaO–La₂O₃) which are formed after reductive treatment. Though such systems are sufficiently active and stable in MDR, the formation of filamentous coke is observed in the case of LaNi_1−xRu_x catalysts with high Ni concentrations.²⁴ As concerning Ln_1−xCa_xRu_0.8Ni_0.2O₃, the high concentration of Ru makes this catalyst rather expensive.

Table 3 Methane conversion, H₂/CO ratio and amount of carbon deposits during MDR using an undiluted feed with CH₄/CO₂ = 1 for Pr(La)Fe_0.65Ru_0.05Ni_0.3 O₃ catalysts and various catalysts reported in the literature

Sample	Methane conversion, %	H2/CO ratio	Carbon deposits, wt%	GHSV, L g^−1 h^−1	T, °C	Ref.
a Carbon balance.
6.6%Ni/Al2O3	68	—	67	22	700	41
4.5%Ni/SiO2	50	—	1.2	21	750	42
10%Ni/3% CeO2/3%La2O3/γ-Al2O3	77	1.09	17.5	12	800	43
10%Ni/8%Y2O3–ZrO2	70	0.7	Coke	36	800	44
1.5%Ru/CeO2–ZrO2	80	0.97	0.06	12	800	45
La2NiO4/ZSM-5	63	—	20	48	700	46
La0.9Pr0.1NiO3	48	0.81	0.0	60	700	16
LaNi0.9Ru0.1O3	85	0.62	20.3	72	750	24
La3.5Ru4O3	76	0.62	0.938	72	750	24
LaRu0.8Ni0.2O3	79	0.94	98.2^a	48	700	23
PrFe0.65Ru0.05Ni0.3 O3	90	0.76	0.0	22	800	This work
LaFe0.65Ru0.05Ni0.3 O3	87	0.72	0.0	22	800

---

The use of LnFe_1−xNi_0.3Ru_xO_3−δ (Ln = La, Pr, x = 0–0.1) precursors with low nickel and ruthenium concentrations provides a high dispersion of Ni–Fe–Ru alloy species epitaxially bound with the oxide matrix and, hence, high activity and stability to coking and sintering. Furthermore, if activation of methane is determined by the metallic phase, activation of CO₂ occurs with the participation of the oxide matrix and could be controlled by its red–ox properties.^27,35 In contrast to ref. 23–25, where the perovskite phase completely decomposes under reaction conditions, the nickelate-ferrite phase used as a precursor in this work is only partly destroyed even after prolonged contact of catalysts with a concentrated feed at high temperatures. A stable perovskite phase along with alloy clusters appear to be the most important factors providing a high and stable performance. This can be explained by a high reactivity and mobility of the surface/lattice oxygen of these ferrite-type perovskites disordered due to reductive segregation of Ni and Ru.^27,35 This helps to provide an efficient route of CO₂ activation on the surface of perovskites and fast transfer of oxygen atoms or oxygen-containing species (hydroxyls, hydroxocarbonates, etc.) to the interface with segregated Ni–Fe–Ru alloy particles where they consume activated CH_x fragments – coke precursors.

4. Conclusions

LnFeNi(Ru)O₃ perovskites are promising precursors of active and stable-to-coking catalysts of CO₂ reforming of methane. Under reaction conditions, perovskites are transformed into nanocomposites comprised of Ni–Fe(Ru) alloy particles and LnO_x epitaxially bound with Ln–Fe–O perovskite particles. Partial substitution of iron by ruthenium gives rise to a synergetic effect between nickel and ruthenium, which is reflected in a higher stability of mixed perovskites to reduction as well as in a higher catalytic activity and stability to coking in MDR. These catalysts provide a high syngas yield and demonstrate resistance to coking for more than 20 hours in dry reforming of the real natural gas. The study of discharged catalysts showed that neither coke deposits nor fibers have been detected.

Acknowledgements

Support by OCMOL FP7 Project, RFBR–CNRS 09-03-93112 Project and Russian Federal Innovation Agency via the program “Scientific and Educational cadres” is gratefully acknowledged.

Notes and references

1. C. Graves, S. D. Ebbesen, M. Mogensen and K. S. Lackner, Renewable Sustainable Energy Rev., 2011, 15, 1.
2. D. G. Vlachos and S. Caratzoulas, Chem. Eng. Sci., 2010, 65, 18.
3. M. C. J. Bradford and M. A. Vannice, Catal. Rev. Sci. Eng., 1999, 41(1), 1.
4. J. J. Spivey and T. Inui, Catalysis, The Royal Society of Chemistry, 2002, vol. 16, p. 133.
5. Y. H. Hu and E. Ruckenstein, Adv. Catal., 2004, 48, 297.
6. A. P. E. York, T. Xiao, M. L. H. Green and J. Claridge, Catal. Rev. Sci. Eng., 2007, 49, 511.
7. J. Ma, N. Sun, X. Zhang, N. Zhao, F. Xiao, W. Wei and Y. Sun, Catal. Today, 2009, 148, 221.
8. V. M. Gonzalez-Delacruz, F. Ternero, R. Pereniguez, A. Caballero and J. P. Holgado, Appl. Catal., A, 2010, 384, 1.
9. C. E. Daza, A. Kiennemann, S. Moreno and R. Molina, Appl. Catal., A, 2009, 364, 65.
10. M. M. V. M. Souza, D. A. G. Aranda and M. Schmal, J. Catal., 2001, 204, 498.
11. M. Safariamin, L. H. Tidahy, E. Abi-Aad, S. Siffert and A. Aboukais, C. R. Chim., 2009, 12, 748.
12. C. Carrara, A. Roa, L. Cornaglia, E. A. Lombardo, C. Mateos-Pedrero and P. Ruiz, Catal. Today, 2008, 133–135, 344.
13. J. Woo Nam, H. Chae, S. Ho Lee, H. Jung and K.-Y. Lee, Stud. Surf. Sci. Catal., 1998, 119, 843.
14. T. Hayakawa, S. Suzuki and J. Nakamura, Appl. Catal., A, 1999, 183, 273.
15. G. Valderrama, M. R. Goldwasser, C. Urbina de Navarro, J. M. Tatibouet, J. Barrault, C. Batiot-Dupeyrat and F. Martinez, Catal. Today, 2005, 107–108, 785.
16. G. S. Gallego, J. G. Marín, C. Batiot-Dupeyrat, J. Barrault and F. Mondragon, Appl. Catal., A, 2009, 369, 97.
17. G. S. Gallego, C. Batiot-Dupeyrat, J. Barrault, E. Florez and F. Mondragon, Appl. Catal., A, 2008, 334, 251.
18. G. Valderrama, A. Kienneman and M. R. Goldwasser, Catal. Today, 2008, 133–135, 142.
19. H. Provendier, C. Petit, C. Estournes and A. Kiennemann, Stud. Surf. Sci. Catal., 1998, 119, 741.
20. H. Provendier, C. Petit, C. Estournes and A. Kiennemann, Chemistry, 2001, 4, 57.
21. H. Falcon, J. Baranda, J. M. Campos-Martin, M. A. Pena and J. L. G. Fierro, Stud. Surf. Sci. Catal., 2000, 130, 2195.
22. M. R. Goldwasser, M. E. Rivas, E. Pietri, M. J. Perez-Zurita, M. L. Cubeiro, L. Gingembre, L. Leclercq and G. Leclercq, Appl. Catal., A, 2003, 255, 45.
23. M. R. Goldwasser, M. E. Rivas, E. Pietri, M. J. Perez-Zurita, M. L. Cubeiro, A. Grivobal-Constant and G. Leclercq, J. Mol. Catal. A: Chem., 2005, 228, 325.
24. G. C. Araujo, S. M. Lima, J. M. Assaf, M. A. Pena, J. L. Garcia Fierro and M. D. C. Rangel, Catal. Today, 2008, 133–135, 129.
25. M. E. Rivas, J. L. G. Fierro, M. R. Goldwasser, E. Pietri, M. J. Perez-Zurita, A. Griboval-Constant and G. Leclercq, Appl. Catal., A, 2008, 344, 10.
26. N. K. Labhsetwar, A. Watanabe, T. Mitsuhashi and H. Haneda, J. Mol. Catal. A: Chem., 2004, 223, 217.
27. L. Kapokova, S. Pavlova, R. Bunina, G. Alikina, T. Krieger, A. Ishchenko, V. Rogov and V. Sadykov, Catal. Today, 2011, 164, 227.
28. M. P. Pechini, US Patent, 3 330 697, 1967.
29. T. G. Kuznetsova, V. A. Sadykov, E. M. Moroz, S. N. Trukhan, E. A. Paukshtis, V. N. Kolomiichuk, E. B. Burgina, V. I. Zaikovskii, M. A. Fedotov, V. V. Lunin and E. Kemnitz, Stud. Surf. Sci. Catal., 2002, 143, 659.
30. Electronic sources http://abulafia.mt.ic.ac.uk/shannon/ptable.php.
31. N. W. Hurst, S. J. Gentry, A. Jones and B. D. McNicol, Catal. Rev. Sci. Eng., 1982, 24(2), 233.
32. P. Ciambelli, S. Cimino, S. De Rossi, L. Lisi, G. Minelli, P. Porta and G. Russo, Appl. Catal., B, 2001, 29, 239.
33. P. Porta, S. Cimino, S. De Rossi, M. Faticanti, G. Minelli and I. Pettiti, Mater. Chem. Phys., 2001, 71, 165.
34. C. Crisafulli, S. Scire, S. Minico and L. Solarino, Appl. Catal., A, 2002, 225, 1.
35. V. A. Sadykov, S. N. Pavlova, T. S. Kharlamova, V. S. Muzykantov, N. F. Uvarov, Yu. S. Okhlupin, A. V. Ishchenko, A. S. Bobin, N. V. Mezentseva, G. M. Alikina, A. I. Lukashevich, T. A. Krieger, T. V. Larina, N. N. Bulgakov, V. M. Tapilin, V. D. Belyaev, E. M. Sadovskaya, A. I. Boronin, V. A. Sobyanin, O. F. Bobrenok, A. L. Smirnova, O. L. Smorygo and J. A. Kilner, Perovskites: Structure, Properties and Uses, Nova Science Publishers Inc, 2010, p. 67.
36. J. Juan-Juan, M. C. Roman-Martinez and M. J. Illan-Gomez, Appl. Catal., A, 2009, 355, 27.
37. D. Liu, R. Lau, A. Borgna and Y. Yang, Appl. Catal., A, 2009, 358, 110–118.
38. J. Chen, C. Yao, Y. Zhao and P. Jia, Int. J. Hydrogen Energy, 2010, 35, 1630.
39. H. M. Swaan, V. C. H. Kroll, G. A. Martin and C. Mirodatos, Catal. Today, 1994, 21, 571.
40. N. N. Sazonova, V. A. Sadykov, A. S. Bobin, S. A. Pokrovskaya, E. L. Gubanova and C. Mirodatos, React. Kinet. Catal. Lett., 2009, 98, 35.
41. J. Juan-Juan, M. C. Roman-Martınez and M. J. Illan-Gomez, Appl. Catal., A, 2006, 301, 9.
42. A. Effendi, K. Hellgard, Z.-G. Zhang and T. Yoshida, Catal. Commun., 2003, 4, 203.
43. R. Yang, C. Xing, C. Lv, L. Shi and N. Tsubaki, Appl. Catal., A, 2010, 385, 92.
44. J. D. A. Bellido and E. M. Assaf, Appl. Catal., A, 2009, 352, 179.
45. J. Chen, C. Yao, Y. Zhao and P. Jia, Int. J. Hydrogen Energy, 2010, 35, 1630.
46. W. D. Zhang, B. S. Liu, C. Zhu and Y. L. Tian, Appl. Catal., A, 2005, 292, 138.
47. J. R. Rostrup Nielsen, Stud. Surf. Sci. Catal., 1991, 68, 85.

---