The effect of CeO2–ZrO2 structural differences on the origin and reactivity of carbon formed during methane dry reforming over NiCo/CeO2–ZrO2 catalysts studied by transient techniques

Michalis A. Vasiliades a, Petar Djinović b, Albin Pintar b, Janez Kovač c and Angelos M. Efstathiou *a
aHeterogeneous Catalysis Laboratory, Chemistry Department, University of Cyprus, 1678 Nicosia, Cyprus. E-mail: efstath@ucy.ac.cy
bDepartment of Environmental Sciences and Engineering, National Institute of Chemistry, Hajdrihova 19, 1000 Ljubljana, Slovenia
cDepartment of Surface Engineering and Optoelectronics, Jožef Stefan Institute, Jamova cesta 39, 1000 Ljubljana, Slovenia

Received 18th May 2017 , Accepted 16th August 2017

First published on 17th August 2017


Nickel (1.2 wt%) and cobalt (1.8 wt%) were dispersed over Ce0.75Zr0.25O2−δ solid solution (3NiCo EG) or over a mixture of CeO2 and ZrO2 single phases (3NiCo HT) and tested for dry reforming of methane (DRM) at 750 °C. The structural, morphological, textural and redox differences between 3NiCo EG and 3NiCo HT catalysts were probed by powder XRD, HAADF/STEM and SAED, N2 adsorption/desorption at 77 K, H2-chemisorption, H2-TPR and H2 transient isothermal reduction (H2-TIR) techniques. The origin, concentration and reactivity of “carbon” deposits formed in DRM (via the CH4 and CO2 activation routes) towards gaseous H2 and O2 but also towards the support's labile oxygen species in the prepared catalysts were analyzed by a series of various transient and SSITKA experiments (use of 18O2 and 13CO2). Regardless of the EG or HT support, the %-contribution of CH4 and CO2 to the “carbon” deposition is very similar but the amount and reactivity towards oxidation are largely different. On the other hand, the concentration of active carbon formed in the carbon pathway of the CO2 activation route is very small (θC < 0.16% after 2 h in DRM). Participation of mobile lattice oxygen species in gasification of deposited “carbon” to CO(g) does occur to a large extent on both EG- and HT-supported NiCo catalysts. The catalyst deactivation rate during the first 5 h of DRM was found to depend on the structure of the support (EG vs. HT). The faster deactivation observed with the 3NiCo EG catalyst cannot be linked to the existing differences in the rates of inactive “carbon” formation and depletion or the concentration of active carbon but rather to the different rates of encapsulation of NiCo bimetallic particles in the carbon layers formed (∼30 nm thick) as revealed by HAADF/STEM.


1. Introduction

Syngas, which is a mixture of H2 and CO in a broad range of compositions, represents the backbone of modern chemical industry and is also expected to maintain its central role in the near future. Syngas is currently the platform for methane and coal upgrading to synfuels via Fischer–Tropsch synthesis,1 as well as the main source of H2 and CO for hydrocarbon upgrading via hydrogenation and carbonylation reactions.2 In the near future, hydrogen is expected to play a central role as fuel in PEM fuel cells.3

Considering the limited reserves of fossil fuels and their pollution related impact on the environment, renewable resources, like biomethane and waste gas streams, especially those that are CO2-rich, represent a possible surrogate feedstock for syngas production. As a result, the methane dry reforming (DRM) reaction (CH4 + CO2 ↔ 2H2 + 2CO) represents a possible utilization pathway for simultaneous CH4 and CO2 valorization to syngas. The methane dry reforming reaction can be performed over supported noble metal (Rh, Pd, Ru and Pt), as well as transition metal (Ni, Co, Fe) catalysts.4–8 Among transition metals, which represent the most economically viable solution, Ni is the most extensively investigated as it exhibits the highest activity. However, due to the fact that nickel-based catalysts are prone to fast carbon accumulation and deactivation, they need to be appropriately designed in order to minimize this severe drawback. For example, the metal component can be alloyed or passivated to modulate the activity, or its most active coordinatively unsaturated sites could be selectively poisoned.9 It is also known that carbon accumulation is strongly related to the size of metallic clusters,10 which can be controlled by the use of ordered mesoporous supports with narrow pore size distribution,11 as well as by the chemical nature of the support.12 The use of basic promoters (MgO, CaO),13,14 and especially redox, mainly as CeO2-based supports15,16 has proven to be a viable solution. The latter approach relies on the high mobility of labile oxygen within the CeO2 lattice at DRM reaction temperatures, which can be further enhanced by doping ceria with Zr, Pr or Sm.15,17,18 In particular, carbon deposition on the 5 wt% Ni/Ce0.8Pr0.2O2−δ catalyst with an average Ni particle size larger than 50 nm was reduced by more than two orders of magnitude compared to the case of the 5 wt% Ni/CeO2 catalyst.19 Oxygen species originating from the doped-CeO2 lattice can migrate to the metallic clusters, where they react with carbon precursors and gasify them mainly to carbon monoxide.19,20

Despite the noticeable and undisputed positive role of doped-CeO2 supports towards minimization of carbon accumulation during the DRM reaction, dedicated experimental studies which would correlate the morphological differences of doped-CeO2 solid solutions and their consequences on the oxygen mobility, the nature and the origin of carbon deposits, are very limited.19,20 To extend the knowledge in this field and enable in-depth understanding of the actual redox promotional role of the support, two structurally different NiCo/CeZrO2 catalysts were subjected to a variety of transient experiments coupled with the use of 18O2 and 13CO2 isotopes, enabling detailed analysis of the carbon pathway as well as oxygen pool participation under relevant DRM reaction conditions, for the first time over supported NiCo bimetallic catalysts. The obtained information along with that related to the deposited carbon morphology and location and NiCo surface composition changes with TOS in DRM (HRTEM and XPS studies) was used to explain the dry reforming of methane activity behaviour of the NiCo/CeZrO2 catalysts with TOS under the reaction conditions applied.

2. Experimental section

2.1. Materials

CeO2–ZrO2 supports (Ce/Zr = 2.87) were prepared by two different methods. For the hydrothermal method (coded HT), 4.04 g of Ce(NO3)3·6H2O (Fluka, p.a.) and 1.1 g ZrO(NO3)2·6H2O (Sigma Aldrich, >99% purity) were dissolved in 100 ml of distilled water. The solution was added dropwise to 250 ml of 25% NH4OH (Merck, p.a.) under stirring. The produced suspension was transferred to a PTFE-lined autoclave and aged for 6 h at 120 °C. The aged product was filtered, washed with distilled water and ethanol and dried overnight in a laboratory drier at 70 °C.

For the ethylene glycol method (coded EG), 4.04 g of Ce(NO3)3·6H2O and 1.1 g of ZrO(NO3)2·6H2O were dissolved in 5 ml of ultrapure water and mixed with 5 ml of propionic acid (Merck, >99% purity) and 154 ml of ethylene glycol (Merck, >99% purity). The solution was transferred to a PTFE-clad autoclave and aged for 200 min at 180 °C. The precipitated solid was separated from the solution by 15 min centrifugation at 8950 rpm, washed with distilled water and ethanol and dried overnight in a laboratory drier at 70 °C. After drying, the material was calcined at 400 °C for 4 h in air with a heating ramp of 5 °C min−1.

A total amount of 3 wt% cobalt and nickel metals (1.8 wt% Co, 1.2 wt% Ni) was deposited over both supports in an identical manner, namely: 0.129 g of Ni(NO3)2·6H2O (Merck, p.a.), 0.193 g of Co(NO3)2·6H2O (Merck, p.a.), 2.1 g of CeZrO2 powder and 1.68 g urea (Merck, p.a.) were dispersed in 100 ml of distilled water. One drop of concentrated HNO3 was added in order to decrease the initial pH value below 3. The suspension was heated in a slow and controlled manner from room temperature to 90 °C and maintained under reflux for 22 h. Afterwards, the suspension was filtered, washed with ethanol and water and dried overnight in air at 70 °C. The catalysts were calcined at 750 °C for 4 h in air. For simplicity, the two prepared catalysts are named 3NiCo HT and 3NiCo EG, respectively, the supports of which were prepared by the hydrothermal and ethylene glycol sol–gel methods.

2.2. Catalyst characterization

XRD analyses were performed on a PANalytical X'pert PRO diffractometer with Cu Kα radiation (λ = 0.15406 nm). The Scherrer equation and the (111) diffraction peak of the CeZrO2 solid support (Ce0.75Zr0.25O2–δ solid solution and CeO2 phases) were used for the calculation of the average scattering domain size.

The BET specific surface area (m2 g−1), total pore volume (cm3 g−1) and average pore size (dp, nm) were determined using N2 adsorption/desorption isotherms at 77 K (Micromeritics, model TriStar II 3020). The samples were degassed before measurements using a SmartPrep degasser (Micromeritics) in a N2 stream.

The amount of reducible and labile oxygen species in the CeZrO2 mixed metal oxide of the support and its redox characteristics were determined using the H2-TPR technique (AutoChem II 2920 apparatus). During the experiment, 100 mg of sample was positioned inside the quartz tube and pretreated in 10% O2/He at 500 °C for 20 min. After cooling the sample to 50 °C, it was degassed in Ar for 30 min. TPR analysis was performed between 50 and 750 °C, with 25 mL min−1 of 5% H2/Ar and a heating rate of 10 °C min−1. The extent of CeO2 (or Ce0.75Zr0.25O2–δ) reduction in the CeZrO2 mixed metal oxide was calculated based on the amount of consumed H2 during the H2-TPR experiment. Ni and Co were assumed to be completely reduced and that ZrO2 did not contribute to the observed H2 consumption.

Transient isothermal reduction by hydrogen (H2-TIR) was applied at 750 °C (DRM reaction T) to study in more detail the kinetic features of the redox process of the solid by recording the transient evolution rate of Ce4+ to Ce3+ reduction and the amount of labile oxygen removed. For this, the gas switch He → 1% H2/1% Kr/Ar/He (750 °C, t) was conducted following treatment of the solid (25 mg) at 800 °C for 2 h in 20% O2/He and He purge. The transient response curves of Kr and H2 were continuously monitored by online mass spectrometry (MS). The transient rate of reduction as a function of time was estimated on the basis of an appropriate material balance.

Metal dispersion was evaluated using the H2-TPD technique. The catalyst was first reduced in a 5% H2/Ar stream for 30 min at 750 °C. The flow was switched to Ar and the temperature increased to 780 °C and maintained for 20 min to completely desorb H2. The sample was cooled to 30 °C in an Ar gas flow, saturated in a 5% H2/Ar gas stream for 30 min and degassed in an Ar gas flow at the same temperature for 20 min, followed by temperature ramping to 800 °C. The low-temperature H2 desorption peak (35 °C < T < 200 °C) was used for active metal dispersion calculation, where a M–H stoichiometry of 1 was assumed (M = Ni, Co).

The morphology and chemical composition of NiCo bimetallic particles as well as the morphology of the inactive “carbon” formed after the DRM reaction were examined by HRTEM–EDX (HAADF/STEM, SAED) using a JEOL ARM 200 CF scanning transmission electron microscope, equipped with a cold field-emission gun, a probe spherical aberration corrector (CESCOR unit from CEOS, Germany) and a JEOL Centurio EDXS system with a 100 mm2 SDD detector. Specimens were ultrasonically dispersed in ethanol and deposited on a copper grid covered with a carbon film prior to analysis. The morphology of inactive “carbon” formed on the 3NiCo EG catalyst sample after the DRM reaction (5, 10 and 20 h on TOS) was also analyzed by scanning electron microscopy. SEM images were taken on a field emission electron microscope (Carl Zeiss, model FE-SEM SUPRA 35VP). The sample was deposited on conductive carbon tape prior to measurements.

X-ray photoelectron spectroscopy (XPS) analyses were carried out on a PHI-TFA XPS spectrometer (Physical Electronics Inc.). Samples were mounted on the metallic sample holder and introduced into the ultra-high vacuum spectrometer (∼10−9 mbar). The analyzed area was 0.4 mm in diameter and the sample surface was excited by X-ray radiation from a monochromatic Al source at 1486.6 eV. The survey wide-energy spectra were taken over the energy range of 0–1400 eV with an analyzer pass energy of 187 eV in order to perform surface elemental quantitative analysis. High-energy resolution spectra were acquired with the energy analyzer operating at a resolution of ∼0.6 eV and pass energy of 29 eV. During data processing, the spectra were aligned by setting the C 1s peak at 284.8 eV. The accuracy of binding energies was ±0.3 eV. Quantification of the surface composition (at%) was performed from XPS peak intensities taking into account relative sensitivity factors provided by the instrument's manufacturer.21 XPS spectra were analyzed by the Multipak software (version 8.1, Physical Electronics Inc.). The relative error of calculated surface concentrations is ∼15% of the reported values.

2.3. Characterization of deposited “carbon” after CH4/He reaction

Transient isothermal hydrogenation (TIH) and temperature-programmed hydrogenation (TPH). TIH and TPH were performed following 30 min treatment of the fresh catalyst sample (25 mg) with a 20% CH4/He gas mixture at 750 °C according to the following procedure. The catalyst was first reduced in H2 (1 bar) at 750 °C for 1 h and the feed was then switched to He for 20 min, where the H2, CH4, CO and CO2 MS signals reached their respective background values. The catalyst was then exposed to 20% CH4/He (100 NmL min−1) for 30 min, followed by 10 min He purge (no CH4 signal was recorded in the MS after He treatment). The catalyst was then cooled in a He flow to the temperature (T = 600, 645, 670 and 700 °C) at which isothermal hydrogenation was performed and the feed was subsequently changed to 20% H2/He (100 NmL min−1). In the case of TPH, the catalyst sample after the 30 min CH4/He treatment at 750 °C was cooled in a He flow to 100 °C and the feed was switched to 20% H2/He (100 NmL min−1), while the temperature was increased to 750 °C at a rate of 30 °C min−1; 10 min isothermal hydrogenation at 750 °C was also followed. During TIH and TPH, CH4 (m/z = 15) was monitored in the mass spectrometer and quantification of the “carbon” hydrogenated to methane was performed using a certified methane gas mixture (ca. 0.998% CH4/He). During the transient CH4 decomposition step (30 min CH4/He treatment), H2 (m/z = 2), CO (m/z = 28) and CO2 (m/z = 44) MS signals were continuously recorded and their conversion to mol% composition was made using standard calibration gas mixtures: 2% H2/He, 2% CO/He and 985 ppm CO2/He, respectively. The signal contribution of CO2 (m/z = 44) to the m/z = 28 signal was considered in order to quantitatively estimate the concentration of CO (based on m/z = 28).

2.4. Contribution of CO2 and CH4 activation routes to “carbon” deposition

Temperature-programmed oxidation (TPO), following dry reforming of methane with 13C-labelled carbon dioxide (12CH4/13CO2/He feed stream), had been conducted in order to estimate the relative contribution of the CO2 and CH4 activation routes to the formation of “carbon”. The experimental procedure was as follows. After 30 min of dry reforming at 750 °C (5% 13CO2/5% 12CH4/45% Ar/45% He; 50 NmL min−1), the reactor was purged in a He flow for 10 min until the CO and CO2 MS signals reached their respective background values. The catalyst was then cooled in a He flow to 100 °C and the feed was switched to 10% O2/He (50 N mL min−1), while at the same time the temperature was increased to 750 °C (TPO run, β = 30 °C min−1). The 12CO (m/z = 28), 13CO (m/z = 29), 12CO2 (m/z = 44) and 13CO2 (m/z = 45) gases were continuously monitored in the MS. The amount (μmol g−1) of “carbon” derived from the CO2 activation route was estimated based on the 13CO and 13CO2 TPO traces, while the “carbon” derived from the CH4 activation route was estimated based on the 12CO and 12CO2 TPO traces. It should be mentioned that the “carbon” estimated in the TPO experiment includes both the inactive and active “carbon” formed during DRM, the latter estimated independently by the SSITKA experiment but only for the CO2 activation route (see section 2.6).

2.5. Probing the participation of support lattice oxygen in the carbon pathway of DRM

The participation of support lattice oxygen in the carbon pathway of the DRM reaction was probed after performing partial exchange of lattice 16O with 18O at 750 °C followed by dry reforming of methane. Formation of C18O(g) might be an indication of the participation of lattice 18O present in the catalyst via the reaction of surface “carbon”, “C-s”, the latter derived from both the CH4 and CO2 activation routes:
 
C-s + 18O-L → C18O(g) + s + Vo(1)
where, s is a catalytic site at the metal–support interface, the support or both, 18O-L is a lattice 18O and Vo is an oxygen vacancy of support. The chemical structure of “C-s” could vary depending on the reaction conditions. However, the possibility of C16O2(g) (present in the DRM feed stream) for exchange of its oxygen(s) with surface 18O-L forming C16O18O(g) and C18O2(g) and the formation of C18O by the following reaction steps (2)–(3) should be examined:
 
C16O18O(g) + C-s ↔ C16O(g) + C18O(g) + s(2)
 
C18O18O(g) + C-s ↔ 2 C18O(g) + s(3)

Formation of C18O(g) might also proceed via the dissociation of C16O18O(g) or C18O2(g) on an oxygen vacant site (Vo) of the support:22,23

 
C16O18O(g) + Vo → C16O(g) (or C18O(g)) + 18O-L(or16O-L)(4)

In order to correctly interpret the transient evolution of C18O(g) upon the switch to the DRM reaction gas mixture (20% CH4/20% CO2/He) at 750 °C, the following gas switches after the 16O/18O isotopic exchange at 750 °C were performed: (i) 20% CO2/He and (ii) x% CO/He; x is the value of CO concentration formed in 20% CH4/20% CO2/He gas treatment of the catalyst at the end of the transient. On the basis of several features of the transient response curves of C18O(g), C16O18O(g) and C18O2(g), such as time delay, peak maximum appearance (tmax, s) and its shape, and estimation of the amount of 18O (mol g−1) incorporated in these species, it is possible to justify whether some C18O(g) is produced viaeqn (1) as will be illustrated in the Results and discussion section.

The 3NiCo HT or 3NiCo EG fresh catalyst sample (25 mg) was first reduced in 20% H2/He at 750 °C for 1 h and then purged with He until the H2 (m/z = 2), 16O2 (m/z = 32) and 16O18O (m/z = 34) MS signals reached their respective background values. The catalyst was then exposed to a 2% 18O2/1% Kr/He gas mixture at 750 °C for 10 min during which exchange of 16O-lattice oxygen of the support and oxidation of Ni/NiO to Ni18O with 18O2(g) took place. The signals of 16O2 (m/z = 32), 16O18O (m/z = 34), 18O2 (m/z = 36) and Kr (m/z = 84) were recorded continuously with on line mass spectrometer. The amount of oxygen exchanged (mol 16O g−1) was estimated from the 16O2(g) and 16O18O(g) transient response curves (up to 10 min) after calibration of the m/z = 32 and m/z = 34 MS signals and use of appropriate material balances. The 10 min treatment of the catalyst with 18O2(g) was followed by 10 min He purge at 750 °C before the switch to each of the two gas streams (i) or (ii) indicated above. Calibration of the C18O2 (m/z = 48) and C16O18O (m/z = 46) MS signals was made based on the natural abundance of 18O in a certified 54.3% CO2/He gas mixture. Furthermore, contributions of C18O2 (m/z = 48) and C16O18O (m/z = 46) to the m/z = 30 (C18O) MS signals were estimated using a standard C16O2/He gas mixture and after considering the same contribution of m/z = 44 (CO2) to m/z = 28 for the m/z = 48 (C18O2) to m/z = 30 and m/z = 46 (C16O18O) to m/z = 30. Calibration of the C18O (m/z = 30) signal to the concentration (mol%) was made after using a standard C16O/He gas mixture (use of m/z = 28 signal) and assuming same sensitivities for the C16O (m/z = 28) and C18O (m/z = 30) gases.

2.6. SSITKA studies – tracing the carbon path of CO2 activation

Steady-state isotopic transient kinetic analysis (SSITKA) experiments for the DRM reaction were conducted over a 50 mg fresh catalyst sample at 750 °C in order to trace the carbon pathway of the CO2 activation route. After 30 min and 2 h of dry reforming, the following SSITKA switches were performed: 5% 12CO2/5% 12CH4/45% Ar/45% He (30 min) → 5% 13CO2/5% 12CH4/2% Kr/43% Ar/45% He (Δt = 10 min) → 5% 12CO2/5% 12CH4/45% Ar/45% He (90 min) → 5% 13CO2/5% 12CH4/2% Kr/43% Ar/45% He (Δt = 10 min) → 5% 12CO2/5% 12CH4/45% Ar/45% He (t). During the switches from the isotopic to the non-isotopic DRM gas mixture, the decay of 13CO (m/z = 29), 13CO2 (m/z = 45) and Kr (m/z = 84) signals were continuously monitored by on line mass spectrometer. Quantification of the concentration of active carbon-containing reaction intermediates, NC (μmol g−1), present in the carbon pathway from the CO2 reactant to the CO gas product was estimated based on the following material balance eqn (5):19,20
 
image file: c7cy01009e-t1.tif(5)
where FT is the total molar flow rate (mol s−1) of the 13CO2/12CH4/Kr/Ar/He feed gas stream and y13CO (s.s.) is the mole fraction of 13CO(g) at steady-state in the 13CO2/12CH4/Kr/Ar/He feed gas stream. Z13CO and ZKr are the dimensionless concentrations of 13CO(g) and Kr tracer gas;24,25t = 0 is the time that the Kr signal at the switch 13CO2/12CH4/Kr/Ar/He → 12CO2/12CH4/Ar/He starts to decay, whereas tf is the final time at which the new steady-state under the non-isotopic feed is obtained. It should be pointed out that NC according to eqn (5) includes any adsorbed CO2-s that truly participates in the carbon pathway of CO formation via the CO2 activation route.20

According to the SSITKA theory and its interconnected pools formalism reported elsewhere,19,24 it is expected that the gaseous CO2 pool (reactant) forming first an active CO2-s pool which feeds with “carbon” the following pools (in series) of active reaction intermediates, responsible for the formation of the CO(g) product, should produce a transient 13CO(g) response upon the switch 13CO2/12CH412CO2/12CH4 lagging behind that of 13CO2(g). In the opposite case, when the 13CO2(g) transient curve lags behind that of 13CO(g), this difference is likely to be associated with the fact that another pool of reversibly adsorbed CO2-s exists on the catalyst surface which does not participate in the formation of CO(g). The concentration (μmol g−1) of such inactive reversibly adsorbed CO2 formed on the catalyst surface under dry reforming can be estimated as reported elsewhere.19

3. Results and discussion

3.1. Catalyst characterization studies

3.1.1. Structural properties. X-ray diffraction analysis (Fig. S1 in the ESI) revealed that the unit cell size (α, nm) of a 3NiCo HT solid is dimensionally very similar to that of pure CeO2 (0.539 and 0.540 nm, respectively), suggesting a very limited extent of Ce1−xZrxO2−δ solid solution formation. In addition, the X-ray diffraction pattern of the 3NiCo HT solid (Fig. S1) shows noticeable asymmetry in the diffraction peaks, which is a result of peak overlapping due to the presence of a separate tetragonal t-ZrO2 phase. This suggests that the material is comprised of preferentially separated fcc CeO2 and t-ZrO2 crystalline phases. In contrast, the support's unit cell size in the 3NiCo EG solid corresponds closely to that of Ce0.75Zr0.25O2−δ solid solution (α = 0.5340 and 0.5349 nm, respectively) and no contribution of a separate t-ZrO2 phase is visible (Fig. S1). An average scattering domain size of 8 nm was estimated to be compared with 25 nm (CeO2 phase) in the 3NiCo HT catalyst. This manifests itself also in the obtained higher specific surface area of EG (51 m2 g−1) compared to that of the HT sample (41 m2 g−1). Based on the XRD analysis results, it is concluded that despite the identical nominal chemical composition of the support, structurally the CeO2–ZrO2 EG and HT supports are considerably different. No diffraction peaks related to Ni- and Co-containing crystalline solid phases could be identified using XRD. This is a consequence of the low metal loading used. H2 chemisorption followed by TPD analysis resulted in an average bimetallic NiCo cluster size of 8 and 11 nm for the 3NiCo EG and 3NiCo HT catalysts, respectively.

HAADF/STEM, SAED and HRTEM analyses (Fig. S2–S4) confirm that Ni and Co are present in both 3NiCo EG and 3NiCo HT catalysts as alloyed bimetallic particles dispersed with a non-uniform size over the corresponding EG and HT ceria–zirconia support.

3.1.2. Redox properties. Quantification of the redox properties of the present solid supports is expressed through the amount of mobile oxygen removed from the CeZrO2 structure during H2-TPR and thus the equivalent degree of reduction (Ce4+ → Ce3+). The latter was found to be 52 and 33% for the EG and HT samples, respectively. This result shows that formation of Ce0.75Zr0.25O2−δ solid solution (EG sample) favors a higher amount of removable oxygen compared to the case of the HT sample for which CeO2 and ZrO2 exist as separate phases (Fig. S1).

The dynamics of oxygen removal from the NiCo oxides and the lattice of EG and HT supports upon interaction with hydrogen at 750 °C was evaluated using transient isothermal reduction, H2-TIR (Fig. 1). The transient reduction rate is very similar in the first 40 s for 3NiCo EG and 3NiCo HT catalysts and this is related to the elimination of oxygen from the surface of NiCo oxides, CeO2 and Ce0.75Zr0.25O2−δ solid phases. At longer times, the reduction is governed by oxygen diffusion through bulk oxygen sub-lattice and consequently, its rate decreases considerably. This reduction rate is noticeably higher for the 3NiCo EG catalyst (Fig. 1a). It should be noted that bulk oxygen diffusion coefficients at 700 °C in doped-CeO2 materials were found to be approximately 10-fold smaller compared to the values obtained for surface/grain boundary oxygen diffusion.26 Thus, the observed differences in the transient reduction rates for longer times of reduction are related to the presence of Ce0.75Zr0.25O2−δ solid solution, where oxygen diffusion is faster compared to pure CeO2.27


image file: c7cy01009e-f1.tif
Fig. 1 Transient rate of solid reduction by hydrogen (μmol g−1 s−1) as a function of time at 750 °C over 3NiCo EG (a) and 3NiCo HT (b) catalysts. Wcat = 25 mg.

This has in the present case two origins: a) the atomic displacement of oxygen atoms, which is directly related to the oxygen diffusion rate, is higher in Ce0.75Zr0.25O2−δ compared to CeO2[thin space (1/6-em)]28 and b) the smaller CeO2-based crystallite size for EG compared to HT solid (8 vs. 25 nm), resulting in a shorter migration path of oxygen atoms towards the surface.

3.2. Transient evolution rates of CH4/He reaction and characterization of deposited “carbon”

3.2.1. Transient evolution rates of H2, CO and CO2. Catalytic methane activation in the absence of CO2 at 750 °C was investigated over previously reduced (1 atm H2, 750 °C, 2 h) 3NiCo EG and 3NiCo HT catalysts, as well as their corresponding supports. Transient rate profiles of H2 and CO formation (Fig. 2A and B, respectively) are distinctly different in the case of bare CeZrO2 EG and CeZrO2 HT supports, compared to 3NiCo EG and 3NiCo HT catalysts. Over bare supports, both H2 and CO formation rates are practically very small for 150 s of reaction. On the other hand, over 3NiCo EG and 3NiCo HT catalysts, the H2 and CO evolution rates are significantly larger and pass through a maximum, which decreases slowly to low steady-state values, which are very similar to the ones observed over pure CeZrO2 supports; 20 μmol H2 g−1 s−1 and 10 μmol CO g−1 s−1 (Fig. 2A and B at 5 min, respectively). The appearance of H2 and CO formation rate maximum is due to the fact that methane alone is activated over the reduced or partially reduced bimetallic NiCo particles, which was recently confirmed during CH4/CO2 reforming over Ni (111).29
image file: c7cy01009e-f2.tif
Fig. 2 Transient rates (μmol g−1 s−1) of H2 (A), CO (B) and CO2 (C) as a function of time obtained during the 20% CH4/He gas treatment at 750 °C for 30 min over 3NiCo EG (a), 3NiCo HT (b), CeZrO2 EG (c) and CeZrO2 HT (d) catalysts. Wcat = 25 mg.

With increasing reaction time in CH4/He, the active oxygen pool of the catalyst, dominated by that present in the CeZrO2 support, is becoming progressively depleted. The adsorbed CHx fragments derived from methane decomposition are more slowly removed as CO(g) since available mobile oxygen atoms need to diffuse from the bulk to the surface and, therefore, the H2 and CO rates decrease concomitantly.

The products of the methane decomposition reaction quantified over the first 5 min of reaction are shown in Table 1. The amount of H2 formed was about six times higher when Ni and Co with a total loading of 3 wt% were deposited over the CeZrO2 support, indicating that methane dehydrogenation takes place predominantly over (bi)metallic sites. The amounts (mmol g−1) of H2 and CO produced were found to be higher over 3NiCo EG compared to the 3NiCo HT catalyst, highlighting its higher activity for methane dissociation. This is in line with previous research results reported by Liu et al.30 over Ni/CeO2, where oxygen vacant sites in CeO2 were identified as CH4 activation sites in addition to those of Ni, which contribute in the DRM reaction. A higher degree of support reduction was confirmed during H2-TPR and H2-TIR (see section 3.1.2) for 3NiCo EG, suggesting a higher abundance of active sites for methane activation and CO formation via the participation of very mobile support lattice oxygen. The same trend was observed over pure supports; the CeZrO2 EG sample produced 47% and 27% higher quantities of H2 and CO, respectively, compared to CeZrO2 HT (Table 1). The H2/CO stoichiometric ratio produced by the 3NiCo EG and 3NiCo HT catalysts (assuming participation of the lattice oxygen in the reaction to generate CO) should be 2.0 in the absence of any side reactions. The values obtained are larger, 3.3 and 2.8, respectively, suggesting that a considerable amount of “carbon” species remains on the catalyst surface. Thus, not all CHx-s species derived from CH4 dissociation get gasified to COx by support lattice oxygen. The H2/CO ratio over pure HT and EG supports is close to the expected value of 2.0 (1.9 and 1.6, respectively), which suggests that “carbon” accumulation over 3NiCo EG and 3NiCo HT is related mainly to the presence of NiCo metal particles.

Table 1 Amounts of H2, CO and CO2 (mmol g−1) produced after 5 min of methane decomposition (20% CH4/He reaction) at 750 °C
Catalyst H2 mmol g−1 CO mmol g−1 CO2 mmol g−1
3NiCo EG 17.04 5.10 0.104
3NiCo HT 13.17 4.72 0.130
CeZrO2 EG 2.98 1.60 0.097
CeZrO2 HT 2.02 1.26 0.035


The CO2 formation rates (Fig. 2C) show distinctly different behaviors over the analyzed materials compared to H2 or CO. Carbon dioxide can originate either from the WGS reaction (CO + H2O ↔ CO2 + H2) or deep oxidation of CHx-s intermediates with the support's lattice oxygen. The total amount of CO2 formed is 30–100 times lower compared to H2 and CO, indicating that the mentioned CO2 formation pathways over the studied materials under the employed CH4 decomposition reaction conditions negligibly influence the distribution of the main methane decomposition reaction products. The carbon mass balance considered on the basis of the methane dehydrogenation products listed in Table 1 suggests accumulation of carbon on the catalysts either as active, which can react with H2 or CO2 and participate in chemical conversion over the catalyst, or inactive, which blocks the surface and causes catalyst deactivation, a subject that is described below.

3.2.2. Characterization of deposited “carbon” via temperature-programmed (TPH) and transient isothermal (TIH) hydrogenation. Temperature-programmed hydrogenation (TPH) was performed after CH4 dehydrogenation at 750 °C for 30 min (Fig. 2) aiming at the characterization of deposited “carbon” over the 3NiCo EG and 3NiCo HT catalysts. The TPH analysis (Fig. 3) shows intrinsically different reactivities and amounts of “carbon” formed on both materials, namely: 4.91 and 0.83 mmol C g−1 (equivalent to CH4 formation) for the 3NiCo EG and 3NiCo HT catalysts, respectively.
image file: c7cy01009e-f3.tif
Fig. 3 TPH trace of “carbon” to CH4 obtained following 30 min treatment in 20% CH4/He at 750 °C over 3NiCo EG (a) and 3NiCo HT (b) catalysts.

Carbon hydrogenation is initiated at ∼500 °C over the 3NiCo EG catalyst, which is ∼100 °C lower compared to the 3NiCo HT catalyst. H2 dissociation, which is a prerequisite for hydrogenation and gasification of carbon deposits, occurs with the lowest activation energy over Ni and Co metal sites.

In addition, H2 can also be dissociated over oxygen vacancies and coordinatively unsaturated edge sites of CeZrO2 crystallites, which are related to the physicochemical properties of the catalysts, specifically their crystallite size.31 On the other hand, pure t-ZrO2, which is present as a separate phase in 3NiCo HT, is much less active for H2 dissociation. The considerably smaller Ce0.75Zr0.25O2−δ crystallite size (8 nm) compared to CeO2 (25 nm), higher abundance of oxygen vacancies (section 3.1.2) and 39% higher exposed active metal surface area of the 3NiCo EG solid all act favorably for providing a larger concentration of active sites for hydrogen activation and an enhanced methanation rate for the deposited “carbon”.

The transient isothermal hydrogenation (TIH) curves (Fig. 4) show clear distinctive contribution of two very different carbon species present in the 3NiCo EG catalyst. The first type of carbon is hydrogenated to methane as a sharp peak in t < 0.5 min, while in the second type, the methane formation rate reaches a maximum at about t = 4 min and decays slowly. In the case of the 3NiCo HT catalyst, the profile of carbon hydrogenation to CH4 is significantly different. CH4 originating from highly reactive “carbon” species is visible as a sharp peak but is much smaller and the CH4 trace at t > 1 min becomes flat, indicating that carbon hydrogenation is performed with a practically constant rate at least for 10 min in a hydrogen stream. With increasing temperature from 645 to 700 °C, “carbon” hydrogenation traces maintain their shapes (Fig. 4A, B) and the amount of “carbon” removed increases (Fig. 4C).


image file: c7cy01009e-f4.tif
Fig. 4 Transient response curves of CH4 concentration at 645 °C (A) and 700 °C (B) obtained during transient isothermal hydrogenation (TIH). (C) The total amount (μmol g−1) of “carbon” formed after 30 min treatment of the catalyst in 20% CH4/He at 750 °C. Wcat = 25 mg; a: 3NiCo EG, b: 3NiCo HT.

The amount of reacted “carbon” in TIH in the case of the 3NiCo EG catalyst was found to be 110, 160, 215 and 337 μmol g−1 at 600, 645, 670 and 700 °C, respectively. Significantly lower amounts of deposited “carbon” were found in the case of 3NiCo HT, namely: 47, 68, 92 and 140 μmol g−1 at 600, 645, 670 and 700 °C, respectively. These quantities are in agreement with those of TPH (Fig. 3), where higher amounts of “carbon” were identified over the 3NiCo EG catalyst.

3.3. Transient evolution rates of CH4/CO2/He (DRM) reaction and characterization of “carbon” deposition

Fig. 5A and B present the evolution of the transient CO and H2 formation rates, respectively, after the step gas switch He → 20% CH4/20% CO2/He (DRM) at 750 °C was made over pre-reduced 3NiCo EG and 3NiCo HT catalysts. This particular transient experiment reveals the initial methane dry reforming behavior of materials, where “carbon” deposition is limited, and which is much related to the structural differences exhibited by their supports. The rate of CO formation reaches a pseudo steady-state after ∼30 s at values of 495 and 400 μmol g−1 s−1 for 3NiCo EG and 3NiCo HT catalysts, respectively. On the other hand, the H2 transient rate evolution is markedly different. It reaches a pseudo steady-state after ∼50 s at a value of 573 μmol H2 g−1 s−1 for 3NiCo EG, whereas in the case of the 3NiCo HT catalyst, a clear delay (t ∼ 5 s) is apparent with respect to the H2 trace observed for 3NiCo EG and only after ∼120 s did it reach a steady-state value of 464 μmol H2 g−1 s−1.
image file: c7cy01009e-f5.tif
Fig. 5 Transient rates (μmol g−1 s−1) of CO (A) and H2 (B) formation as a function of time obtained during dry reforming of methane at 750 °C over 3NiCo EG (a) and 3NiCo HT (b) catalysts. Wcat = 25 mg.

At this very short time of reaction (150 s), 3NiCo EG exhibits higher methane reforming activity and the H2/CO ratio obtained is 1.15. The same order of activity (3NiCo EG > 3NiCo HT) was observed also during methane dehydrogenation (Fig. 2). The time delay in the appearance of H2 response (Fig. 5B) with respect to that of CO (Fig. 5A) is the result of the different kinetic rates with respect to CH4 dissociation and CO and H2 formation routes on an initially reduced and clean metallic surface. In addition, the delay in the appearance of H2 response compared to CO might be influenced by the fact that nickel and nickel containing rare-earth oxides are known for their hydrogen storage ability.32,33

The more pronounced H2 delay observed on 3NiCo HT might also be partly related to the fact that the separate tetragonal ZrO2 phase, which is present only in this sample, can store H2 in the form of thermally stable OH and hydride species at adjacent O and Zr3+ surface sites.34

3.4. Quantifying the origin of inactive deposited “carbon” (CH4vs. CO2 activation routes)

The contribution of CO2 and CH4 towards carbon formation during the DRM reaction over the 3NiCo HT and 3NiCo EG catalysts was investigated by using isotopically labeled 13CO2 and monitoring both 12CO2 and 13CO2 during a following TPO experiment (Fig. 6). It is observed that the 13CO2 and 12CO2 traces over individual catalysts are practically identical, showing no distinction in the reactivity of “carbon” deposited during DRM regardless of its origin (12CH4 or 13CO2). This suggests that upon the activation of each reactant over the catalyst's surface, the “C-s” derived forms a carbon pool of very similar size and reactivity towards gasification with molecular oxygen. The characteristic features of the TPO traces, however, differ when Ce0.75Zr0.25O2−δ solid solution (3NiCo EG) or separate CeO2 and ZrO2 phases (3NiCo HT) are used to support NiCo bimetallic particles. Carbon oxidation is in both cases initiated at about 300 °C forming a small peak extended to T < 450 °C. At higher temperatures, two additional carbon oxidation peaks emerge: the first between 500–650 °C and the second between 650–760 °C. The total amount of carbon formed during DRM is found to be higher on 3NiCo HT compared to 3NiCo EG (Table 2), resulting in a substantial increase of the peak shown in the 500–650 °C range.
image file: c7cy01009e-f6.tif
Fig. 6 TPO of “carbon” formed during DRM with labeled 13CO2 at 750 °C for 3NiCo EG (A) and 3NiCo HT (B) catalysts. Wcat = 50 mg.
Table 2 12CO2 and 13CO2 (μmol g−1) and 12CO2/13CO2 ratio obtained during temperature-programmed oxidation (TPO) of carbon species formed during 30 min reforming in 5% 13CO2/5% 12CH4/He at 750 °C over the 3NiCo EG and 3NiCo HT catalysts
Catalyst 12CO2 mmol g−1 13CO2 mmol g−1 12CO2/13CO2 Total “carbon” mmol g−1
3NiCo EG 0.442 0.395 1.12 0.837 (1.01 wt%)
3NiCo HT 0.703 0.695 1.01 1.398 (1.68 wt%)


By considering that the carbon nano-filament is the predominant carbon morphology during extensive catalyst coking, to be presented below, one can assume that the middle TPO peak, which is the most pronounced over the 3NiCo HT sample, is related to this kind of carbon morphology.

The larger extent of catalyst coking on 3NiCo HT during DRM originates from the fact that the amount and rate of supply of mobile lattice oxygen present in the support, as analyzed by H2-TPR and transient isothermal reduction (Fig. 1), are lower on the 3NiCo HT catalyst, and as such, less active oxygen species for carbon gasification are available during DRM, leading to faster carbon buildup.12 Considering the fact that some of NiCo bimetallic clusters are located on ZrO2, the oxygen spillover from the adjacent CeO2 phase for prompt carbon gasification is substantially hindered. Also, metallic particles can be extracted from the support by the growth of carbon nano-filaments, resulting in loss of contact with the support.

It is also interesting to emphasize at this point that “carbon” deposition during CH4 decomposition alone was found to be larger on the EG-supported than the HT-supported NiCo catalyst, which is the opposite for the case of the DRM reaction (Table 2).

3.5. Probing the participation of support lattice oxygen in DRM

The ability of the 3NiCo EG and 3NiCo HT catalysts to provide active oxygen species originating from the support, which enable CO formation during methane decomposition, was confirmed in the previous section 3.2.1 (Fig. 2). However, during the DRM reaction, the oxygen originating from the CO2 reactant or the support lattice oxygen should be differentiated. For this, the participation of the support's lattice oxygen was investigated after the support's 16O was partially replaced by 18O before DRM. Fig. S5 describes this 16O/18O isotopic exchange according to section 2.5. The 16O2 formation proceeds with equal rate in both materials in the initial transient after the He → 18O2/He switch (Fig. S5A), which is related to the recombination and desorption of surface 16O atoms.

The surface oxygen desorption rate appears to be slightly influenced by the Zr incorporation into the CeO2 lattice, as was observed also in the H2-TIR experiment (Fig. 1). At longer times, the observed 16O2 generation rate is governed by bulk 16O diffusion, which appears to be slightly faster in the 3NiCo EG catalyst. Again, this correlates with the redox and oxygen mobility differences between the present Ce0.75Zr0.25O2−δ and CeO2 solids.

Contrary to the 16O2 rate maximum which appears at ∼0.5 min, the 16O18O rate maximum is observed at significantly larger times (between 4–4.5 min), indicating that bulk oxygen diffusion is slower within the crystal lattice of CeZrO2 compared to surface/grain boundary diffusion.35 We can see again that the 16O18O formation rate is slightly faster in EG compared to the 3NiCo HT solid.

The catalyst's oxygen pool which participates in the DRM reaction was monitored through the 18O-labeled reaction products as shown in Fig. 7. Prior to the CH4/CO2 gas switch, the catalyst was treated with 18O2 for 10 min at 750 °C allowing for partial lattice 16O/18O exchange (see Fig. S5). During the following DRM reaction, practically the same amount (14.6 mmol 18O/g) of the total 18O was found to be incorporated in the observed 18O-labelled CO2's and CO gaseous products for the two catalysts. It should be noted that the equivalent 18O stored in Ni18O and Co318O4 during catalyst treatment at 750 °C with 18O2 gas is only 1.6 mmol g−1.


image file: c7cy01009e-f7.tif
Fig. 7 Transient concentration (mol%) response curves of C16O18O (A), C18O2 (B) and C18O (C) obtained during dry reforming of methane at 750 °C, after treatment with 2% 18O2/He at 750 °C for 10 min, performed over 3NiCo EG (a) and 3NiCo HT (b) catalysts. Wcat = 25 mg.

Transient formation of C18O2 goes through a maximum at 20 s and decays to zero, 75 s after the switch to the DRM feed stream (Fig. 7B). The total C18O2 amount formed is found to be slightly higher in 3NiCo EG compared to the 3NiCo HT catalyst (2.70 vs. 2.38 mmol g−1) and is related to the reaction of surface and sub-surface lattice 18O with the reactive “carbon” pool. Formation of C16O18O is related to the reaction between lattice 18O, 16O originating from CO2 and reactive “carbon” formed over the catalyst during the DRM reaction. The possible exchange of surface lattice 18O with CO2 is presented below.

The C16O18O formation rate goes through a maximum at 25 s and tails strongly until decaying to zero at about 150 s. This is consistent with longer times required for diffusion of bulk 18O to the surface of the catalyst, where 18O can take part in catalytic reaction steps with surface adsorbed species.

Contrary to the C16O18O and C18O2 traces, which light off quickly after the He → 20% CH4/20% CO2/He gas switch, the C18O signal (CO is a direct product of the DRM reaction) lights off with a delay of ∼25 s. This may suggest that the initially oxidized 3NiCo EG and 3NiCo HT solids are unable to catalyze the DRM reaction but favor only full methane oxidation.27 The slow decay of the C18O transient response lasts about 250 s, which by far outlives the formation of any 18O labeled CO2. This indicates a gradual transition from the initial total oxidation to the predominant DRM reaction pathway with increasing time on stream, where lattice 18O diffuses to the surface and takes part in the DRM reaction.

The amounts of 18O-labeled CO and CO2 produced by 3NiCo EG and 3NiCo HT catalysts are listed in Table 3. The EG catalyst produces a substantially higher quantity of 18O-containing CO2, whereas HT is more selective towards 18O-containing CO. As mentioned before, during the course of the DRM reaction (Fig. 7), the catalyst becomes progressively more reduced. If the oxygen diffusion is slow (3NiCo HT), a larger dynamic oxygen concentration gradient between the surface (predominantly reduced) and bulk (predominantly oxidized) is established, favoring the DRM reaction and CO production. On the other hand, if bulk oxygen diffusion is fast (3NiCo EG), the gradient is smaller, favoring total oxidation.

Table 3 Amounts (mmol g−1) of C16O18O, C18O2 and C18O formed during the various gas switches applied over the 3NiCo EG and 3NiCo HT catalysts after 10 min 18O2/He treatment at 750 °C
Catalyst C16O18O mmol g−1 C18O2 mmol g−1 C18O mmol g−1 Total 18O mmol g−1
a After 20%CH4/20%CO2/He treatment. b After 20%CO2/He treatment. c After x% CO/He treatment; x = 0.65% for 3NiCo HT and 0.85% for the 3NiCo EG catalyst.
3NiCo EG 3.04a 2.70 6.19 14.63
2.04b 2.38 0.20 7.0
0.79c 0.25 4.60 5.89
3NiCo HT 2.28 2.38 7.55 14.59
2.64 1.68 0.1 6.1
0.59 0.21 4.08 5.09


As mentioned in section 2.5, the formation of 18O-labelled CO2 during the DRM gas switch (Fig. 7A and B) on a partially pre-covered catalyst surface by 18O might also be considered as the result of exchange of C16O2 (DRM feed stream) with surface lattice 18O, according to reaction steps (6)–(7), but also to the reversibility of the Boudouard reaction given in steps (2)–(3).

Fig. S6A and B presents the transient evolution of 18O-labelled CO2 and that of C18O during the step-gas switch He → 20% CO2/He (t) at 750 °C over 3NiCo EG and 3NiCo HT, respectively, conducted under exactly the same experimental conditions as the experiment described in Fig. 7. Formation of C16O18O and C18O2 takes place according to the following reaction steps:

 
C16O16O(g) + 18O-s → C16O18O(g) + 16O-s(6)
 
C16O18O(g) + 18O-s → C18O18O(g) + 16O-s(7)

The most important result from this experiment is the fact that the C18O transient response obtained is different in shape and quantity from the corresponding one shown in Fig. 7C. The amount of C18O formed was only 0.20 and 0.10 mmol g−1 for the EG-supported and HT-supported NiCo, respectively. These results exclude the possibility that the main source of C18O formed during DRM (Fig. 7C) is the dissociation of 18O-labelled CO2 on the metal surface or via the participation of the support's oxygen vacant sites (step (4)).

The step gas switch He → x% CO/He (t) at 750 °C over the catalyst surface partially pre-covered with 18O-s resulted also in C18O transient responses that are different in shape and quantity (Fig. S7A and B), namely, 4.60 and 4.08 mmol g−1 for the EG-supported and HT-supported NiCo catalysts, respectively. On the basis of the shape, position and amount (Table 3) of C18O transient responses recorded under the CH4/CO2 (Fig. 7C), CO2/He (Fig. S6) and CO/He (Fig. S7) gas switches, it is concluded that large quantities of highly mobile oxygen of the support react more with “C-s” derived from CH4 decomposition, the latter occurring on the NiCo surface, than “C-s” derived from the Boudouard reaction, towards gasification to CO(g).

3.6. Quantifying the active “carbon” present in the CO2 activation reaction path via SSITKA

Quantification of the active carbon formed in the CO2 activation pathway was performed by employing the SSITKA technique. The SSITKA response (Fig. 8) after 30 min of DRM shows that the 13CO response follows closely that of the Kr tracer gas, suggesting that the concentration of active carbon-containing intermediates leading to the CO gas product is very small. This was found to be ∼0.01 μmol g−1 after 30 min of dry reforming over both catalysts. This amount was found to increase after 2 h of reaction, namely, 0.08 and 0.03 μmol g−1 for the 3NiCo EG and 3NiCo HT, respectively.
image file: c7cy01009e-f8.tif
Fig. 8 SSITKA dimensionless concentration response curves (Z) of Kr, 13CO and 13CO2 obtained after 30 min of DRM at 750 °C performed over the 3NiCo HT catalyst. Wcat = 50 mg; FT = 100 NmL min−1. SSITKA switches: 5% 12CO2/5% 12CH4/He (30 min) → 5% 13CO2/5% 12CH4/He (3 min) → 5% 12CO2/5% 12CH4/He (t).

In contrast, the 13CO2 trace lags behind that of 13CO, and this has been reported to be the result of the presence of a pool of reversibly chemisorbed carbon dioxide which does not participate in the carbon pathway of the CO2 activation route (spectator species).19,20 The (NCO2) amount of such inactive carbonate-type species was found to accumulate with reaction time, where after 30 min in the DRM reaction it was found to be 0.12 and 2.5 μmol g−1, respectively, for the 3NiCo EG and 3NiCo HT catalysts, and 1.3 and 6.0 μmol g−1, respectively, after 2 h of dry reforming.

3.7. Relationship between active and inactive carbon formation and long-term catalyst stability in DRM

Fig. 9 presents the DRM activity behavior of the two catalytic systems at 750 °C, in terms of CH4-conversion (%), for long time-on-stream (TOS). The feed gas composition used was 44.2% CO2, 44.3% CH4 and 11.5% N2, and the flow rate was 56.5 NmL min−1. During the first 1 h of reaction, the activity of 3NiCo EG is higher than that of 3NiCo HT, in agreement with the initial rates reported in Fig. 5. However, during the first ∼5 h on TOS, the EG-supported NiCo catalyst is deactivated faster than the HT-supported one. While the latter catalyst presents stable activity after ∼5 h on TOS, the EG-supported one still shows further deactivation but less than that obtained during the first 5 h of reaction (Fig. 9).
image file: c7cy01009e-f9.tif
Fig. 9 CH4-conversion (%) as a function of time on stream (TOS) for DRM at 750 °C on the 3 NiCo EG (a) and 3NiCo HT (b) catalysts. Feed gas composition: 44.2% CH4, 44.2% CO2, 11.5% N2; Wcat = 50 mg; FT = 56.5 NmL min−1.

A similar behavior was observed for the CO2-conversion (%) and H2/CO product ratio (Fig. S8). It should be noted here that the trend of activity behavior with TOS was very similar to that shown in Fig. 9 after using the 20% CH4/20% CO2/He feed gas stream (Fig. 5).

“Carbon” deposits (0.51 wt% as quantified using CHNS analysis) over the 3NiCo EG catalyst were characterized after 20 h in the DRM reaction using SEM and HAADF/STEM techniques. Carbon in the form of nano-filaments was observed by SEM (Fig. 10), which represents the only identifiable carbon morphology. Additional information on the “carbon” attachment to the catalyst surface was obtained by HAADF/STEM. It can be seen in Fig. S9 that in the case of carbon nano-filament formation, the bimetallic cluster was extracted from the CeZrO2 support and is located embedded with carbon (layer about 30 nm thick) at the tip of the filament. Also, several smaller NiCo bimetallic clusters were found to be embedded with carbon in the vicinity of the CeZrO2 support (Fig. S10). Based on these SEM and HAADF/STEM analyses, it can be concluded that the NiCo bimetallic clusters covered with a carbon layer are no longer in contact with the CeZrO2 support.


image file: c7cy01009e-f10.tif
Fig. 10 SEM image of the spent (20 h in DRM) 3NiCo EG catalyst, where the growth of carbon nano-filaments is clearly shown.

The results of the present work are important in order to better understand the intrinsic reason(s) for the stability behavior shown in Fig. 9. According to the results of the present work, the concentration of active carbon formed on the 3NiCo EG catalyst increases and that of inactive carbonate-type species decreases, a result opposite to that found for the 3NiCo HT catalyst (Fig. 8). It should be noted that the latter species could be linked to the poisoning of active labile oxygen of the support that facilitates “carbon” gasification (see section 3.5). Furthermore, the reactivity of “carbon” deposits from methane decomposition towards hydrogen is larger for the EG-supported than the HT-supported NiCo catalyst (Fig. 2 and 3), and less “carbon” is deposited after DRM at 750 °C on the EG-supported than the HT-supported catalyst (Table 3). The transient 16O/18O isotopic exchange studies also indicate that lattice oxygen mobility of EG is larger than that of the HT support (Fig. S4). It should also be pointed out that very low amounts of “carbon” (0.37 wt% C) were measured after 20 h of TOS for the DRM reaction conditions reported in Fig. 9 for the 3NiCo HT sample.

Based on the above offered discussion, the faster deactivation observed in 3NiCo EG compared to 3NiCo HT in the first 5 h of DRM cannot be explained by the higher rates of “carbon” deposition, the lower rates of “carbon” gasification by lattice oxygen shown by 3NiCo HT, or by the diminishing concentration of the active “carbon” intermediates that form CO(g).

In order to investigate possible oxidation of NiCo particles during DRM arising from strongly bound O-s species,36 XPS analysis was conducted over a fresh (reduced) 3NiCo EG catalyst sample and those treated for 5, 10 and 20 h under DRM reaction conditions (the same as in Fig. 9). ICP-OES analysis revealed 1.06 and 1.51 wt% Ni and Co present in the fresh 3NiCo EG catalyst, respectively, which is in good agreement with the nominal value. The presence of surface nickel in the fresh reduced 3NiCo EG catalyst could not be confirmed using the XPS technique due to its low concentration. On the other hand, the presence of a low amount of surface cobalt (0.57 at%) was detected. After 5 h and 10 h on TOS, the concentration of surface Co decreased to 0.19 and 0.20 at%, respectively, whereas after 20 h of TOS surface Co was below the detection limit (Fig. S11). Based on these data, due to the very low concentration of surface Co and Ni, it was not possible to accurately determine the oxidation states of Co and Ni with TOS in the DRM reaction. As a result, changes related to the CeZrO2 support are presented and discussed as follows.

The surface Ce/Zr atom ratio was found to change in the following order: 1.1, 0.88, 1.08 and 0.98 for the fresh reduced and 5, 10 and 20 h in TOS catalyst samples, respectively; the bulk Ce/Zr nominal value is 2.87. The binding energy of the Zr 3d5/2 photoelectron peak in the fresh reduced sample is 181.8 eV (the 3d5/2–3d3/2 spin–orbit splitting of 2.5 eV; Fig. S12). The peak position corresponds to that of CeZrO2 solid solution37 and is lower compared to the characteristic value of Zr4+ in ZrO2 (182.3 eV).21 With increasing TOS, the Zr 3d5/2 photoelectron peak shows a progressive shift toward a higher binding energy, reaching 182.6 eV after 20 h TOS. This value corresponds to Zr4+ in ZrO221 and indicates the formation of a surface ZrO2 phase during the DRM reaction. The O 1s photoelectron peak BE (529.7 eV) of the fresh reduced catalyst is associated with lattice oxygen anions in the CeZrO2 solid solution (Fig. S13). With increasing TOS, a progressive shift of the O 1 s peak to higher BEs (530.3 eV after 20 h TOS) is observed, and this corresponds to the individual oxides of CeO2 and ZrO2. This is in line with the evolution in the Zr 3d spectra. A shoulder for the high BE (532 eV) is also observed for the fresh sample, which might be attributed to surface hydroxyls and/or water molecules.37 The same shoulder appears also in the sample treated in DRM for 20 h, where an increase of carbon C 1s signal (not shown here) was also observed.

Another difference observed by the XPS analyses was a change in the Ce 3d spectra with TOS in DRM shown in Fig. S14. It consists of spectral peaks related to the Ce (4+) (v and v′′ peaks at 884 eV and 889 eV, respectively) and Ce (3+) (v′ and vo peaks at 885 eV and 881 eV, respectively) oxidation states. It can be observed (Fig. S14) that the fresh sample has the highest Ce (3+) concentration. The 5 h and 10 h treated samples in DRM resulted in similar concentrations of Ce (3+), whereas the 20 h treated sample has the lowest Ce (3+) concentration. This indicates a gradual oxidation of the CeZrO2−x solid support as a function of TOS in the DRM reaction. A progressive decrease of surface cobalt concentration seen by XPS (electron Inelastic Mean Free Path (IMFP) in CeO2 is 2.1 nm at 1.3 keV kinetic energy38) may suggest its diffusion into the bulk of CeZrO2 (forming a solid solution)30 during DRM (reductive atmosphere). This progressively results in a decrease in the surface abundance of active Co, and likely Ni metal clusters (active sites for methane activation), manifesting itself in the progressive deactivation of the catalyst, as observed in Fig. 9.

On the basis of the above described XPS results, there is experimental evidence for the decrease of the surface Ce/Zr ratio by ∼10% (compared to the fresh reduced sample) and at the same time an enrichment in the support surface with the ZrO2-like structure with increasing TOS in DRM. On the other hand, surface Co seen by XPS decreased by about a factor of three after 10 h on TOS and is hardly seen by XPS after 20 h on TOS. The latter is in good agreement with the HRTEM results (Fig. S9 and S10) previously discussed, where NiCo bimetallic clusters were found embedded in a carbon layer of about 30 nm thick. It could therefore be concluded that progressive encapsulation of NiCo bimetallic clusters in the carbon layer appears as the main reason for the significant deactivation of the 3NiCo EG catalyst after 20 h on TOS in DRM (Fig. 9). Unfortunately, due to the low surface Ni and Co present in the examined 3NiCo EG catalyst, oxidation of the remaining NiCo alloy clusters being another likely reason to explain the reduced catalytic activity with TOS cannot be discussed.

A very recent work36 that utilized in-operando XAS combined with kinetic and DFT studies, reported the importance of the oxidation state of Co and Ni in the NiCo bimetallic or alloy supported on ZrO2 under working DRM reaction conditions at 750 °C, on catalyst performance. The important conclusion derived from that work is that Co sites promote the formation of strongly bound O-s species (via CO2 and CO dissociation), hindering C–H bond activation of CH4. As TOS increases, oxidation gradually occurs to the bulk of the metal particles, as a consequence of the unbalanced kinetics of CH4 and CO2 activation. On Ni surface sites, the oxyphilicity of nickel is insufficient for O-s accumulation, thus a reduced state is apparent. However, low coverage of O-s on Ni leads to a slower removal of “carbon” species derived from CH4 and CO dissociation. In the case of the NiCo alloy (solid solution), the oxyphilicity becomes moderate, and O-s formation competes with CH4 activation. More importantly, these O-s species do not cause bulk oxidation of the NiCo alloy.

4. Conclusions

The following conclusions can be derived from the results of the present work:

(i) The initial DRM catalytic activity of 3 wt% NiCo (1.8 wt% Co and 1.2 wt% Ni) alloy particles supported on a ceria–zirconia (Ce/Zr = 2.87) mixed metal oxide was found to depend on the support's synthesis method (hydrothermal vs. sol–gel). The latter tunes the structural and redox properties of the ceria–zirconia mixed metal oxide and the resulting metal support interactions.

(ii) The rate of “carbon” deposition and the kinds and reactivity of “carbon” formed in CH4 decomposition depend on the structure of the ceria–zirconia support. The Ce0.75Zr0.25O2−δ (solid solution) support (EG) promotes the rate of CH4 decomposition and at the same time “carbon” removal, via the participation of mobile lattice oxygen, to a greater extent than the CeO2–ZrO2 (HT) support.

(iii) The %-contribution of CH4 and CO2 activation routes to “carbon” deposition during DRM over the supported NiCo alloy particles is not influenced by the structure of the ceria–zirconia support. The opposite is true for the amount, types and reactivity of deposited “carbon”.

(iv) Very small quantities (0.01 μmol g−1) of active “carbon” are formed in both EG- and HT-supported NiCo, whereas significantly larger quantities of inactive “carbon” (0.84–1.4 mmol g−1) are formed after 30 min of DRM (5%CH4/5%CO2/He) at 750 °C.

(v) The concentration behavior of active and inactive “carbon” and that of inactive carbonate-type species formed with TOS in the DRM reaction over EG- and HT-supported NiCo, cannot explain their observed deactivation behavior. HAADF/STEM studies revealed that NiCo alloy particles encapsulated in a carbon layer of ∼30 nm thick are found at the tip of the carbon nano-filaments. This loss of NiCo particles is considered to be the main cause of severe catalyst deactivation with TOS (Fig. 9). This phenomenon appears to depend on the structure of the CeZrO2 support.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

AME and MAV acknowledge the Research Committee of the University of Cyprus for financial support. PD and AP acknowledge financial support through Research program P2-0150 and research grant Z2-5463 provided by the Slovenian Research Agency. Dr. Sašo Šturm is kindly acknowledged for TEM-EDX and SAED analyses.

References

  1. A. de Klerk, Fischer Tropsch Refining, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2011 Search PubMed .
  2. G. A. Olah and A. Molnar, Hydrocarbon chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, 2nd ed., 2003 Search PubMed .
  3. F. Barbir, PEM Fuel Cells: Theory and Practice, Elsevier B.V., Amsterdam, 2nd ed., 2012 Search PubMed .
  4. A. T. Ashcroft, A. K. Cheetham, M. L. H. Green and P. D. F. Vernon, Nature, 1991, 352, 225–226 CrossRef CAS .
  5. S.-J. Alonso, J. Juan-Juan, M. J. Illán-Gómez and M. C. Román-Martínez, Appl. Catal., A, 2009, 371, 54–59 CrossRef .
  6. K. Takanabe, K. Nagaoka, K. Nairai and K. Aika, J. Catal., 2005, 232, 268–275 CrossRef CAS .
  7. R. Benrabaa, A. Löfberg, A. Rubbens, E. Bordes-Richard, R. N. Vannier and A. Barama, Catal. Today, 2013, 203, 188–195 CrossRef CAS .
  8. K. Sutthiumporn, T. Maneerung, Y. Kathiraser and S. Kawi, Int. J. Hydrogen Energy, 2012, 37, 11195–11207 CrossRef CAS .
  9. J. W. Han, J. S. Park, M. S. Choi and H. Lee, Appl. Catal., B, 2017, 203, 625–632 CrossRef CAS .
  10. J.-H. Kim, D. J. Suh, T.-J. Park and K.-L. Kim, Appl. Catal., A, 2000, 197, 191–200 CrossRef CAS .
  11. M. N. Kaydouh, N. El Hassan, A. Davidson, S. Casale, H. El Zakhem and P. Massiani, Microporous Mesoporous Mater., 2016, 220, 99–109 CrossRef CAS .
  12. P. Djinović and A. Pintar, Appl. Catal., B, 2017, 206, 675–682 CrossRef .
  13. Z. Hou, O. Yokota, T. Tanaka and T. Yashima, Appl. Catal., A, 2003, 253, 381–387 CrossRef CAS .
  14. A. I. Tsyganok, T. Tsunoda, S. Hamakawa, K. Suzuki, K. Takehira and T. Hayakawa, J. Catal., 2003, 213, 191–203 CrossRef CAS .
  15. P. Djinović, I. G. Osojnik Črnivec, B. Erjavec and A. Pintar, Appl. Catal., B, 2012, 125, 259–270 CrossRef .
  16. N. Wang, W. Chu, T. Zhang and X. S. Zhao, Int. J. Hydrogen Energy, 2012, 37, 19–30 CrossRef CAS .
  17. M. M. Makri, M. A. Vasiliades, K. C. Petallidou and A. M. Efstathiou, Catal. Today, 2016, 259, 150–164 CrossRef CAS .
  18. L. N. Bobrova, A. S. Bobin, N. V. Mezentseva, V. A. Sadykov, J. W. Thybaut and G. B. Marin, Appl. Catal., B, 2016, 182, 513–524 CrossRef CAS .
  19. M. A. Vasiliades, M. M. Makri, P. Djinović, B. Erjavec, A. Pintar and A. M. Efstathiou, Appl. Catal., B, 2016, 197, 168–183 CrossRef CAS .
  20. M. A. Vasiliades, P. Djinović, L. F. Davlyatova, A. Pintar and A. M. Efstathiou, Catal. Today, 2017 DOI:10.1016/jcattod.2017.03.057  , in press (available on line 4 April 2017).
  21. J. F. Moulder, W. F. Stickle and P. E. Sobol, Handbook of X-Ray Photoelectron Spectroscopy, ed. J. Chastain, Physical Electronics Inc., Eden Prairie, USA, 1995 Search PubMed .
  22. N. Kumari, M. A. Haider, M. Agarwal, N. Sinha and S. Basu, J. Phys. Chem. C, 2016, 120, 16626–16635 CAS .
  23. V. A. Sadykov, E. L. Gubanova, N. N. Sazonova, S. A. Pokrovskaya, N. A. Chumakova, N. V. Mezentseva, A. S. Bobin, R. V. Gulyaev, A. V. Ishchenko, T. A. Krieger and C. Mirodatos, Catal. Today, 2011, 171, 140–149 CrossRef CAS .
  24. A. M. Efstathiou, J. T. Gleaves and G. S. Yablonsky, Characterisation of Solid Materials: From Structure to Surface Reactivity, ed. M. Che and J. C. Vedrine, Wiley-VCH, 2012, ch. 22, pp. 1013–1073 Search PubMed .
  25. A. M. Efstathiou and X. E. Verykios, Appl. Catal., A, 1997, 151, 109–166 CrossRef CAS .
  26. A. S. Bobin, V. A. Sadykov, V. A. Rogov, N. V. Mezentseva, G. M. Alikina, E. M. Sadovskaya, T. S. Glazneva, N. N. Sazonova, M. Y. Smirnova and S. A. Veniaminov, Top. Catal., 2013, 56, 958–968 CrossRef CAS .
  27. P. Fornasiero, J. Kašpar and M. Graziani, Appl. Catal., B, 1999, 22, 11–14 CrossRef .
  28. M. Yashima, in Catalysis by Ceria and related materials, ed. A. Trovarelli and P. Fornasiero, Imperial College Press, 2nd edn, 2013, pp. 1–40 Search PubMed .
  29. K. Yuan, J.-Q. Zhong, X. Zhou, L. Xu, S. L. Bergman, K. Wu, G. Q. Xu, S. L. Bernasek, H. X. Li and W. Chen, ACS Catal., 2016, 6, 4330–4339 CrossRef CAS .
  30. Z. Liu, D. C. Grinter, P. G. Lustemberg, T.-D. Nguyen-Phan, Y. Zhou, S. Luo, I. Waluyo, E. J. Crumlin, D. J. Stacchiola, J. Zhou, J. Carrasco, H. F. Busnengo, M. V. Ganduglia-Pirovano, S. D. Senanayake and J. A. Rodriguez, Angew. Chem., Int. Ed., 2016, 55(26), 7455–7459 CrossRef CAS PubMed .
  31. S. M. Schimming, G. S. Foo, O. D. LaMont, A. K. Rogers, M. M. Yung, A. D. D'Amico and C. Sievers, J. Catal., 2015, 329, 335–347 CrossRef CAS .
  32. L. Jalowiecki-Duhamel, J. Carpentier and A. Ponchel, Int. J. Hydrogen Energy, 2007, 32, 2439–2444 CrossRef CAS .
  33. S. Velu and S. Gangwal, Solid State Ionics, 2006, 177, 803–811 CrossRef CAS .
  34. A. Trunschke, D. L. Hoang and H. Lieske, J. Chem. Soc., Faraday Trans., 1995, 91, 4441 RSC .
  35. E. M. Sadovskaya, Y. A. Ivanova, L. G. Pinaeva, G. Grasso, T. G. Kuznetsova, A. van Veen, V. A. Sadykov and C. Mirodatos, J. Phys. Chem. A, 2007, 111, 4498–4505 CrossRef CAS PubMed .
  36. B. AlSabban, L. Falivene, S. M. Kozlov, A. Aguilar-Tapia, S. Ould-Chikh, J.-L. Hazemann, L. Cavallo, J.-M. Basset and K. Takanabe, Appl. Catal., B, 2017, 213, 177–189 CrossRef CAS .
  37. A. Galtayries, R. Sporken, J. Riga, G. Blanchard and R. Caudano, J. Electron Spectrosc. Relat. Phenom., 1998, 88–91, 951–956 CrossRef CAS .
  38. S. Kato, M. Ammann, T. Huthwelker, C. Paun, M. Lampimäki, M.-T. Lee, M. Rothensteiner and J. A. van Bokhoven, Phys. Chem. Chem. Phys., 2015, 17(7), 5078–5083 RSC .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7cy01009e

This journal is © The Royal Society of Chemistry 2017