Introduction
Several studies have clearly demonstrated the connection between the greenhouse gases (GHGs) emissions and the climate system [1]. Carbon dioxide is estimated to account for 77% of GHGs emissions and is thus considered as the main cause of climate change. The major contribution to such a massive emission of CO2 can be associated with the use of carbon-based fuels that currently provide about 80%–85% of the global total primary energy supply [2,3]. CO2 concentration is also expected to increase in the next decades, especially because of the robust economic growth of emerging countries, such as China and India. The energy structure of these countries is dominated by coal, which is the most environmentally impacting raw material, since its combustion produces the highest CO2 amount per unit of heat compared to the other carbon-based fuels [4,5]. However, contrary to both oil and natural gas, whose reserves are expected to be depleted in the next few decades, coal availability is estimated to last for more than 150 years. In light of this, the production of synthetic natural gas (SNG) through coal gasification into syngas and its subsequent transformation into methane has attracted increasing attention. In addition, syngas can also be derived from gasification of biomass, which is a renewable source. Thus, the production of CH4 through CO or CO2 hydrogenation, known as the “Sabatier” or “methanation” reaction (Scheme 1), can be considered a promising alternative for the reduction of GHGs emissions as well as for a sustainable exploitation of coal reserves, especially if hydrogen is produced using renewable resources and energy. Moreover, carbon oxides methanation allows H2 to be transformed into a substance compatible with the current technologies, since methane could be injected directly into the already existing pipeline networks or storage infrastructures.
The methanation reactions of CO [6,7] and CO2 [6,8] have been extensively studied on supported noble (e.g., Ru, Rh) and non-noble (Ni, Co, Fe) metals, using a wide variety of metal oxides carriers prepared with different procedures, also in the presence of promoters. Among the non-noble metals, Ni has been mainly investigated, due to its low cost, high activity and high methane selectivity. On the other hand, thanks to its peculiar oxygen storage and redox properties and to the presence of oxygen vacancies in a wide range of conditions, the use of cerium oxide as a support or promoter has also been widely reported in the literature. Several papers can be found in which CeO2-supported Ni catalysts have been used for separately converting either CO [9–13] or CO2 [13–18] into methane.
CO methanation in the presence of different amounts of CO2 is a well-known process for purifying H2 streams from low CO contents (as, for example, purification of H2 for NH3 production or for Fuel Cells applications). Several publications regarding the selective methanation of CO in CO2-rich reformate gas streams over NiO-CeO2 systems also exist [19–24]. However, to the best of the present authors knowledge, only few papers [25–36] about CO and CO2 co-methanation for SNG production have been published so far, among which five deal with catalysts based on Ni supported on CeO2-containing carriers [29,30,32,35,36].
In this work, the co-methanation of CO and CO2 was studied on soft-templated NiO-CeO2 mixed oxides catalysts with different Ni/Ce molar ratios (0.3–4.0). The synthesized samples were characterized by different techniques as to their chemical composition, structure, morphology, and texture. On the catalysts submitted to the same reduction pretreatment adopted for the activity tests the surface basic properties and specific metal surface area were also determined. The reaction was performed in a fixed-bed continuous-flow quartz-glass microreactor, at atmospheric pressure and 300 °C, with CO/CO2/H2 molar compositions equal to 1/1/7 or 1/1/5, and space velocities (SV) of 72000 or 450000 cm3∙h–1∙gcat–1. Furthermore, on a selected sample, the effect of reaction temperature and the influence of SV, as well as the stability, were also investigated.
Experimental
Materials
Cetyltrimethyl-ammonium bromide (CTAB,≥98%), Ni(NO3)2·6H2O (99.999%), Ce(NO3)3∙6H2O (99%), and NaOH (pellets, 97%) were supplied by Aldrich. Ethanol (96%) was supplied by Fluka. HCl (37%) was provided by Merck. The gaseous feed mixture with known concentration of the components was supplied by SAPIO.
Synthesis of catalysts
NiO-CeO2 mixed oxides with Ni/Ce molar ratios in the range 0.3–4.0 were synthesized through the soft template method [37,38], using Ce(NO3)3·6H2O and Ni(NO3)2·6H2O as cerium and nickel oxides precursors, CTAB as the template, and NaOH as the precipitating agent. For the synthesis, appropriate amounts of the nitrate precursors and the template (CTAB/precursors= 0.62 mol∙mol–1) were dissolved at room temperature in 100 cm3 of distilled water under stirring. After 30 min, a 0.17 mol∙L–1 solution of NaOH was added dropwise until a pH value of 13 was reached; the mixture was then stirred for 15 h. After digestion at 90 °C for 3 h, the solid formed was recovered by filtration and washed with hot distilled water (70 °C) until the formation of AgBr precipitate in the filtrate by reaction of bromides with AgNO3 was no longer observed. Then, it was dried at 110 °C for 6 h and finally calcined at 450 °C for 4 h. The materials obtained were named xNiCe, were x represents the nominal value of the Ni/Ce molar ratio.
Characterization of catalysts
Transmission electron microscopy (TEM) images were obtained with a JEOL 200CX instrument. A JEOL JEM 2010 UHR microscope, equipped with a Gatan imaging filter with a 15 eV window and a 794 slow scan CCD camera, was instead used to collect high resolution TEM (HRTEM) micrographs. Both instruments operated at 200 kV. Finely ground samples were dispersed in n-octane in an ultrasonic bath. The suspension was then dropped onto a copper grid covered with a carbon thin film for the observation.
Raman spectra were collected at room temperature in back-scattering geometry with an inVia Renishaw micro-Raman spectrometer equipped with an air-cooled charge coupled device detector and edge filters. A 488.0 nm emission line from an Ar ion laser was focused on the sample under a Leica DLML microscope, using 20× or 5× objectives. The power of the incident beam was about 5 mW. Repeated (10 or 20 s) accumulations were generally acquired for each sample. The resolution was 2 cm–1 and spectra were calibrated using the 520.5 cm–1 line of a silicon wafer. Spectra processing included baseline removal and curve fitting using a Gauss-Lorentz cross-product function by Peakfit 4.12 software (Jandel, AISN Software).
X-ray photoelectron spectroscopy (XPS) analysis was performed using a PHI 5000 VersaProbe scanning instrument equipped with a monochromatic Al-Kα X-ray source (1486.6 eV, 15 kV, 1 mA) and a combined electron and argon ion gun neutralizer system to reduce charging effects. C1s, O1s, Ce3d and Ni2p regions were acquired in FAT mode (23.5 eV), setting C1s at 284.7 eV. Data analysis involved smoothing, non-linear Shirley-type background subtraction, curve fitting (mixed Gaussian-Lorentzian function by a least-square method), and peak area determination by integration of the appropriate signal after data analysis (Esca Tools 4.2 software, Surface Interface Inc., Mountain View, CA). The shapes of the Ce3d and Ni2p signals were analyzed by a curve fitting procedure with Ce3d and Ni2p doublets endowed with fixed spectroscopic parameters, but using variable position, full width at half maximum, and intensity. The surface composition was determined from the peak area ratios using the empirically derived atomic sensitivity factors reported by Wagner et al. [39].
Microcalorimetric measurements were performed with a Tian-Calvet heat flow microcalorimeter (Setaram), equipped with a volumetric vacuum line. Each sample (0.1 g, 40–80 mesh), previously reduced under flowing pure H2 (15 cm3∙min–1) at 400 °C for 1 h, was thermally pretreated at 250 °C for 12 h under vacuum (5×10–3 Pa). Adsorption was carried out by admitting successive doses of the probe gas (CO2) at 80 °C in order to limit physisorption. The equilibrium pressure relative to each adsorbed amount was measured by means of a differential pressure gauge (Datameritics) and the thermal effect was recorded. The run was stopped at a final equilibrium pressure of about 133 Pa. After outgassing overnight at the same temperature, a second adsorption run was carried out in order to calculate the amount of the probe gas irreversibly adsorbed.
Fourier transform infrared (FTIR) spectra were collected on an EQUINOX 55 spectrometer (Bruker), equipped with a MCT cryodetector working at 2 cm–1 resolution. Prior to FTIR measurements, thin self-supporting wafers were outgassed at 400 °C in dynamic vacuum (residual pres-sure<1×10–1 Pa) and then exposed to H2 (1.5×104 Pa) at the same temperature for 1 h. After cooling to room temperature under Ar atmosphere and outgassing to 1×10–1 Pa, successive doses of CO2 were admitted and a first FTIR spectrum was collected after the equilibrium pressure of 5×103 Pa was reached; then, a second spectrum was recorded after outgassing for 30 min at the same temperature.
The specific surface areas of the Ni0 metal particles were calculated by means of hydrogen temperature programmed desorption (H2-TPD) measurements performed after H2-pulse chemisorption runs, both carried out in a TPD/R/O 1100 apparatus (ThermoQuest). Before analysis, the sample (0.050 g) was reduced under H2 flow (15 cm3∙min–1) at 400 °C for 1 h. Then, N2 (20 cm3∙min–1) was used for purging (1 h) and cooling to 50 °C. Pulses of H2 (5 vol-% in N2) were then admitted into the reactor until the area of the peaks remained constant. Finally, TPD analysis was carried out under flowing N2 (20 cm3∙min–1) while heating (10 °C∙min–1) from 50 °C to 500 °C (held for 30 min). The specific metal surface area was then automatically computed by the instrument software from the desorption profile assuming a H/Ni stoichiometric factor equal to 1. Calibration of the H2 amount evolved was previously carried out by using a copper oxide standard.
CO-TPD analyses were performed after CO-pulse chemisorption runs by means of the TPD/R/O 1100 instrument. After the sample (0.050 g) reduction under pure H2 (15 cm3∙min–1) at 400 °C for 1 h, pulses of CO (5 vol-% in He) were sent until the peaks area remained constant. Then, CO-TPD analyses were carried out under flowing He (20 cm3∙min–1) while heating (10 °C∙min–1) from 50 °C to 800 °C. Desorption profiles were collected by a mass spectrometer detector, through which CO and CO2 were monitored.
Elemental analysis was carried out with an EA 1110 CHNS-O analyzer (CE Instruments) by total combustion of the samples in a mixture of oxygen and helium.
Thermogravimetric analyses (TGA) were carried out on a Perkin-Elmer STA6000 thermal analyser. The samples were placed in an alumina pan and heated under oxygen flow (20 cm3∙min–1) from 25 °C to 850 °C (heating rate, 10 °C∙min–1).
CO2 and CO co-methanation
Typical CO and CO2 co-methanation tests were carried out for 6 h at atmospheric pressure and 300 °C in a tubular (fint = 0.8 cm) quartz-glass fixed-bed continuous-flow microreactor. The reactor is provided with an external jacket containing a fluidized bed of silicon carbide, which allows keeping the system in isothermal conditions once the reaction temperature has been reached upon heating in an electric furnace. The temperature was monitored by a K-type thermocouple located in the thermostatic jacket and connected to a temperature-programmer/controller. A second K-type thermocouple was positioned inside the reactor to check the temperature of the catalytic bed (differences between the setpoint and the detected temperature values were never observed). Prior to the reaction, the catalysts were pretreated in air (30 cm3∙min–1) at 400 °C overnight; after purging in He (30 cm3∙min–1), they were subsequently reduced under H2 flow (15 cm3∙min–1) at the same temperature for 1 h and then cooled in He (30 cm3∙min–1) to the reaction temperature. CO and CO2 co-methanation tests were performed with 0.050 g of catalyst by using a CO/CO2/H2 reactant gas mixture with a molar composition equal to 1/1/7 (CO, 9.5 mol-%; CO2, 9.8 mol-%; H2, 73.9 mol-%; N2, used as the internal standard, 6.8 mol-%) or 1/1/5 (CO, 12.7 mol-%; CO2, 13.2 mol-%; H2, 64.9 mol-%; N2, used as the internal standard, 9.2 mol-%). Two SVs, 72000 and 450000 cm3∙h–1∙gcat–1, were investigated, corresponding to a total flow rate of 60 and 150 cm3∙min–1, respectively. After 1 h on stream, online analysis of the reactor effluent was performed every hour on a 6890 (Agilent) gas chromatograph (GC), equipped with a Carboxen 1010 PLOT capillary column and a thermal conductivity detector. Water was removed from the reaction mixture through an ice trap and a molecular sieves trap placed between the reactor outlet and the GC injection valve. The results of the quantitative analysis of the carbon-containing components were used for checking the carbon mass balance and for calculating COx conversion and products selectivity. On the 1.0NiCe sample, the influence of the reaction temperature (from 200 °C to 400 °C) and of a wide range of space velocity (from 48000 to 450000 cm3∙h–1∙gcat–1) was investigated. Long-time runs were also performed at different temperature ranges to study the catalytic stability under thermal stress. In addition, on the 0.3NiCe and 1.0NiCe samples, CO methanation tests at 300 °C, H2/CO = 3, and SV = 72000 or 450000 cm3∙h–1∙gcat–1 were carried out in order to get information on the occurrence of side-reactions possibly responsible for carbon residues formation.
Results and discussion
Characterization
The compositional, textural, and structural characterization of the xNiCe samples has already been reported and discussed in a previous work [17]. However, for the sake of clarity, the results are summarized in Table 1 and the X-ray diffraction (XRD) patterns of the present samples collected before and after the H2-pretreatment are also shown in Fig. S1 (cf. Electronic Supplementary Material, ESM).
Tab.1 Chemical composition and structural and textural features of the xNiCe samples and of the CeO2 and NiO pure oxides (from ref. 17) |
| Sample | Ni/Ce a) molar ratio | gNi/gCeO2a)/wt-% | Average crystallite size b)/nm | SBET c)/(m2∙g−1) | Vp c)/(cm3∙g–1) | ||
|---|---|---|---|---|---|---|---|
| CeO2 | NiO | Ni0 | |||||
| CeO2 | – | – | 5 | – | – | 191 | 0.36 |
| 0.3NiCe | 0.29 | 9.8 | 3 | 4 | n.d. | 174 | 0.17 |
| 1.0NiCe | 0.96 | 32.6 | 3 | 4 | 6 | 206 | 0.29 |
| 2.5NiCe | 2.50 | 85.2 | 3 | 4 | 8 | 208 | 0.37 |
| 4.0NiCe | 3.84 | 131.0 | 3 | 4 | 8 | 209 | 0.38 |
| NiO | – | – | – | 3 | 7 | 209 | 0.31 |
a) determined by inductively coupled plasma-atomic emission spectrometry (ICP-AES) analyses; b) determined by applying the Scherrer Equation [40] to the XRD data (Fig. S1) of the samples before (CeO2 and NiO) and after (Ni0) the H2-pretreatment; c) determined by the N2 adsorption/desorption isotherms. Surface area (SBET) values were calculated by using the Brunner-Emmett-Teller (BET) method [41]. |
TEM analysis was performed on the 1.0NiCe sample before and after the H2-pretreatment as well as after the catalytic reaction and the results are reported in Fig. 1. TEM micrographs of cerium and nickel pure oxides are shown in Figs. S2 and S3 (cf. ESM) for comparison. In agreement with previous results on NiCe samples with different Ni/Ce molar ratios [17], the TEM micrograph of the starting 1.0NiCe sample (Fig. 1(a)) shows the presence of CeO2 (black arrows), in the form either of small, roughly spheroidal and hexagonal, nanoparticles of 3 to 5 nm in size or of thin nanofilaments (3–4 nm in diameter and up to 200 nm in length) (see also Fig. S2). Elongated nanoparticles (white arrows) are also visible, which are ascribable to NiO in the form of nanorods of 3–4 nm in diameter and 10–40 nm in length (see also Fig. S3). The comparison of these values with the average size of crystallites estimated by XRD data (i.e., around 3 and 4 nm for CeO2 and NiO, respectively) suggests that both the nanofilaments and the nanorods could be composed of chains of small spheroidal nanocrystals, possibly because a self-organizing mechanism is established during the synthesis in the presence of the ionic surfactant (CTAB). The presence of CeO2 and NiO in close contact is clearly evidenced by the HRTEM image in Fig. 1(b) (see also Fig. S4 (cf. ESM)), where d-spacings of 3.12 and 2.41 Å, corresponding to the (111) planes of CeO2 and NiO, respectively, were measured on the rounded and the elongated nanocrystals. As expected, in the HRTEM image of the sample after the H2-pretreatment (Fig. 1(c)) CeO2 and Ni0 nanocrystals are present (for the latter the d-spacing of the (111) plane is 2.03 Å). Noteworthy, though not observable in the relevant XRD pattern (Fig. S1(b)), the existence of residual NiO on the reduced catalyst is indicated by the detection of crystallites with d-spacing of 2.41 Å. The presence of NiO was also evidenced in the XRD patterns of 2.5NiCe and 4.0NiCe by the appearance of weak and broad signals at 2q values of 37.3°, 43.3° and 62.9° (Fig. S1(b)), supporting the view of an incomplete reduction of the starting NiO crystallites, at least for Ni/Ce molar ratios ≥1. The micrograph of the post-reaction 1.0NiCe catalyst (Fig. 1(d)) is very similar to that of the fresh one, suggesting that both the nature and the size of the Ni species are not modified during the co-methanation reaction.
The Raman spectra of the xNiCe samples are compared in Fig. 2, where the spectrum of a soft-templated pure cerium oxide is also reported. Pure CeO2 shows a predominant band centered at 460 cm–1 (F2g mode of the CeO2 fluorite phase) and weak bands at 258, 595 and 1179 cm–1, due to second-order transverse acoustic (2TA) mode, defect-induced (D) mode, and second-order longitudinal optical (2LO) mode, respectively [42–44]. All the xNiCe samples show Raman spectra with main features at about 440 cm–1 (F2g mode), 578 cm–1 (D mode), and in the range 1080–1180 cm–1 in which both the 2LO mode of CeO2 and the 2TO mode (2P) of NiO [45] are expected. Compared to the CeO2 spectral shape, the F2g mode of the xNiCe system shifts to lower wavenumbers at increasing Ni amounts, and both the features at about 578 cm–1 and in the range 1080–1180 cm–1 increase in intensity. The shift of the F2g mode indicates a strong interaction between the NiO and CeO2 species, with a consequent weakening of the Ce–O bond. The increase in intensity along with the Ni content of the band at about 578 cm–1, which encloses contributions at about 560 cm–1, due to oxygen vacancies (D1), and at about 600 cm–1, due to cation substitution in the lattice (D2), supports the hypothesis of the formation of a solid solution [44,46]. The presence of NiO nanocrystals is clearly evidenced by the band at about 1090 cm–1, which partially overlaps the 2TA mode of CeO2 at about 1179 cm–1 and becomes a prominent feature with increasing Ni content.
The reduction treatment in H2 slightly downshifts the position of the CeO2 bands (Fig. 3(a), black curve) with respect to the oxidized sample (Fig. 3(a), grey curve) and causes a little broadening of the F2g band and a small variation in the intensity of the D band. These small changes, evidenced by the spectral profile obtained after subtracting the spectrum before H2-pretratment from that after H2-pretratment (Fig. 3(a), dashed curve), indicate a small reduction extent of CeO2 [43,47], which influences the Ce–O bond strength and the defectivity. As for the 0.3NiCe sample, after the H2-pretreatment (Fig. 3(b), black curve) both the F2g and the D bands result a little broadened and red-shifted compared to those of the starting material, which are positioned at about 455 and 580 cm–1, respectively (Fig. 3(b), grey curve). The difference spectrum (Fig. 3(b), dashed curve) shows the appearance of components at about 440 and 550 cm–1 due to a decrease in particle size (F2g mode) and increase in defects (D band) because of the reduction process. Conversely, the H2-treatment significantly modifies the Raman spectrum of the 4.0NiCe sample (Fig. 3(c)). The spectral trace resulting from the difference between spectra after and before H2-pretreatment (Fig. 3(c), dashed curve) shows a strong decrease in the intensity of the signals at about 422 and 1080 cm–1 and a parallel increase in the intensity of those located at about 453 and 605 cm–1. The strong decrease in intensity of the band at about 1080 cm–1, associated with the 2TO mode (2P) of NiO, clearly indicates the reduction of segregated NiO species to Ni0. The positive shift of the F2g mode to about 453 cm–1 together with the increased D contribution at about 605 cm–1 suggest a structural rearrangement caused by the reduction process involving both CeO2 and NiO species. Since the defect-induced mode (D) encloses contributions due to oxygen vacancies (at about 560 cm–1) and to cation substitution in the lattice (at about 600 cm–1), the higher concentration of defects could result from both the loss of oxygen and the vacancies left by the reduction of incorporated nickel ions.
The XPS analysis allowed the determination of the surface composition of the samples and a better understanding of the nature of the interactions between Ni species and cerium oxide by following the changes in the Ce3d, Ni2p and O1s features. The Ce3d spectrum (Fig. S5(a), cf. ESM)) shows a complex feature due to a satellite structure resulting from hybridization with O2p orbitals and partial occupancy of the 4f levels. According to Burroughs et al. [48] the 3d5/2 and 3d3/2 spin-orbit components (spin-orbit splitting, 18.5 eV) are labeled as v and u, respectively. The v, vʹʹ'', vʹʹʹ''', and u, uʹʹ'' and u'''ʹʹʹ components refer to Ce4+ final states, while the v0, vʹ' and u0, uʹ' components refer to Ce3+ final states. The well-defined peak uʹʹʹ''' at the highest binding energy (BE) is peculiar of Ce4+ and can be used as a measure of the Ce4+ amount [49]. In fully oxidized CeO2 the area of the u'''ʹʹʹ component should constitute around 14% of the area of the entire Ce3d region [50]. The Ce3+ concentration in reduced CeO2 can instead be more accurately determined by the ratio Ce3+/(Ce3+ + Ce4+) where the numerator is the sum of the Ce3+ signals areas and the denominator is the total area of the Ce3d (Ce3+ + Ce4+) signals [51–53]. Concerning NiO, the Ni2p region consists of 2p3/2 and 2p1/2 spin-orbit components (spin-orbit splitting, 17.2 eV) and associated satellite (separated by 6.1 eV, about 50% of the total intensity), the 2p3/2 peak being at 854.5 eV [54]. The identification of Ni species in doped systems, originating from defective sites on the surface (Ni3+ surface species) or in the bulk due to the formation of a solid solution, has long been debated, leading to the assignment of BE values in the range 855.8–856.4 eV [55,56]. As for oxygen (Fig. S5(b)), the O1s BE for CeO2 can vary between 529.6 and 530.3 eV, corresponding to Ce4+–O and Ce3+–O bonds, respectively. Surface OH groups are expected to contribute a component at a BE of 531.6 eV [57]. Similar values are reported in the literature for NiO, with O1s peaks at 529.1 and 530.9 eV assigned to the lattice oxygen and to the defective sites on the NiO surface, respectively [55].
The XPS spectra collected on CeO2 and on the xNiCe samples in their oxidized form are illustrated in Fig. 4. The partial superimposition of the Ce3d5/2 and Ni2p3/2 regions originated complex features which were resolved by curve fitting. It can be noted that the Ni2p components become predominant with increasing Ni content and Ni species as NiO and Ni(OH)2 are detected.
For the xNiCe samples, the curve fitting procedure yielded distinct features arising from Ce4+ and Ce3+ species as well as from Ni2+ (NiO, Ni(OH)2) and defective Ni states (Ni3+). Similar trends were obtained for all the samples, indicating that the nature of the specific Ni and Ce surface species do not change significantly with increasing NiO content.
Since the 0.3NiCe sample presents the lowest contribution of NiO as a separate phase (as can be inferred from both the corresponding XRD pattern (Fig. S1(a)) and Raman spectrum (Fig. 2)), it was considered as a representative case to investigate the interaction between NiO and CeO2 oxides. The Ce3d region is resolved with components due to Ce4+ (v, vʹʹ'', vʹʹʹ''' and u, uʹʹ'', uʹʹʹ''') and to Ce3+ (vʹ', uʹ') species (Fig. 5(a)). The amount of Ce3+ surface species accounts for 16% of the total intensity, a value higher than that calculated for the pure CeO2 (11%). This result suggests the existence of Ce3+ species other than those due to X-ray irradiation, which reasonably arise from the interaction with Ni species, indicating that the presence of Ni atoms alters the chemical environment of the cerium atoms. The Ni2p region was analyzed for the Ni2p3/2 component to overcome the difficulty due to the partial overlap with the Ce3d region and to the complexity of the Ni2p spectroscopic features. The Ni2p3/2 signal (Fig. 5(b)) is reproduced with peaks at 854.4, 855.8 and 857.3 eV (and associated satellites), ascribable to Ni2+ as NiO, Ni3+ species, and Ni2+ as Ni(OH)2, respectively. The Ni3+ component could correspond to defective sites at the NiO-CeO2 interface, where Ni species are in strong interaction with the support [55,57], whereas the presence of Ni(OH)2 species is most likely due to exposure of the samples to the atmosphere. The O1s region (Fig. 5(c)) is fitted with a main peak at 529.3 eV, due to lattice oxygen in CeO2 and NiO, and with two less intense peaks at 530.7 and 532.3 eV, which are assigned to surface oxygen species resulting from the incorporation of Ni2+ in the CeO2 lattice and to surface hydroxyl groups, respectively [55,57]. The peak fitting procedure applied to the 1.0NiCe, 2.5NiCe and 4.0NiCe samples for the Ce3d and Ni2p3/2 regions (Figs. S6–S8, cf. ESM)) revealed the same surface species found for 0.3NiCe, i.e., Ce4+ and Ce3+; Ni2+ (NiO, Ni(OH)2) and Ni3+ (resulting from the strong interaction between NiO and CeO2 outer layers). The same oxygen species (lattice oxygen, surface oxygen due to NiO-CeO2 interaction, and hydroxyl groups) were also found in all the xNiCe samples.
By using the sensitivity factors method, the surface atomic composition was determined by integrating the XPS peaks area of the Ni and Ce species and the results are shown in Table 2 (the calculation details are reported in Tables S1 and S2 (cf. ESM) for CeO2 and xNiCe samples, respectively). The calculated surface Ni/Ce atomic ratios can be compared to those obtained for the bulk by ICP-AES (see Table 1). A Ni surface enrichment is always observed, which is parallel to the increase in the contribution of NiO as a segregate phase with increasing Ni/Ce molar ratios, evidenced by both the XRD and the Raman results. Interestingly, the values of the Ni3+/Nitotal and Ce3+/Cetotal atomic ratios remain almost constant for all the samples in the series (Table 2), indicating that the fraction of Ni species strongly interacting with CeO2 at surface and/or subsurface level is independent of the Ni content. The increase in the amount of Ce3+ surface species with respect to that obtained for pure CeO2 also suggests that the presence of Ni atoms alters the chemical environment of the cerium atoms, in line with the Raman results.
Tab.2 Surface composition of the xNiCe samples |
| Sample | (nNi/nCe)surface | (/nNi, total)surface a) | (/nCe, total)surface b) |
|---|---|---|---|
| CeO2 | – | – | 0.11 |
| 0.3NiCe | 0.35 | 0.36 | 0.18 |
| 1.0NiCe | 1.60 | 0.32 | 0.16 |
| 2.5NiCe | 3.41 | 0.31 | 0.17 |
| 4.0NiCe | 5.50 | 0.32 | 0.17 |
a) areas ratio of Ni3+ (NiO-CeO2 interacting Ni species) component to the total area of the Ni2p3/2 region [Ni2+ (NiO, Ni(OH)2) + Ni3+]; b) areas ratio of Ce3+ (vʹ', uʹ') signals to the total area of the Ce3d region [Ce3+ (vʹ', uʹ') + Ce4+ (v, vʹʹ'', vʹʹʹ''' + u, uʹʹ'', uʹʹʹ''')]. |
The microcalorimetric results for the xNiCe catalysts after the H2-pretreatment are reported in Fig. 6 in terms of the differential heat of adsorption as a function of the CO2 uptake. The Qdiff vs. CO2 uptake curves of the reduced CeO2 and NiO soft-templated oxides are also reported for comparison. It can be observed that Qdiff continuously decreases with the increasing surface coverage, indicating the presence of energetically heterogeneous adsorbing sites. It is worthy of note that the presence of nickel up to a Ni/Ce molar ratio of 2.5 induces an increase in the basic character with respect to CeO2, all the corresponding microcalorimetric curves lying above that of the pure oxide. Noteworthy, after the marked increase in basicity by addition of the lowest content of nickel, further increasing its amount leads to a concomitant decrease in the basic character, which in the case of 4.0NiCe is even lower than that of the pure cerium oxide, as indicated by the corresponding calorimetric curve located below that of ceria. Nevertheless, it still possesses some basicity, as indicated by the presence of a not negligible number of sites with Qdiff greater than 100 kJ∙mol–1. For NiO, the surface basic properties are very scarce, as expected by considering that only the residual nickel oxide still present after the H2-pretreatment is able to adsorb the CO2 probe molecule. From the Qdiff vs. CO2 uptake curves, it is possible to estimate the strength distribution of the basic sites. A Qdiff value of 40 kJ∙mol–1, corresponding to about three times the CO2 condensation heat at 80 °C (13.7 kJ∙mol–1), has been considered as the threshold value between chemical and physical or nonspecific adsorption, sites with lower adsorption heats have hence been neglected when assessing basicity. Basic sites have been divided into weak (40≤Qdiff<90 kJ∙mol–1), medium (90≤Qdiff≤150 kJ∙mol–1), and strong (Qdiff> 150 kJ∙mol–1) sites. The obtained results are summarized in Table 3. The aforementioned trend of the basicity as a function of the Ni content is confirmed by the changes in the site strength distribution: a manifest increase in the basic sites concentration irrespective of their strength is first observed by addition of the lowest amount of nickel to the pure CeO2; then, a decrease in concentration of the weak- and medium-strength sites, as well as in that of the sites on which CO2 is irreversibly adsorbed, is observed with increasing Ni content, while the number of strong basic sites is much less affected. It has been reported in the literature that the surface basicity of ceria can be enhanced by the presence of oxygen vacancies [58–60] and that the concentration of such vacancies can be tuned by doping CeO2 with different metal oxides [61,62] owing to the size/charge mismatch between host cerium ions and dopant ions. A manifest increase in the number of oxygen vacancies was observed for Cu- and La-doped CeO2 prepared by a surfactant-template method [62]. In accordance with the literature, it can be proposed that the presence of Ni, by increasing oxygen vacancies, as confirmed by Raman analysis, enhances the basic character of the NiCe catalysts. However, as the amount of Ni increases, the exposed surface of ceria decreases, which leads to a diminution of both the accessibility and the number of sites responsible for CO2 adsorption. This may explain the decrease in CO2 uptake and basic sites strength along with the increase in the Ni content.
Tab.3 Microcalorimetric results for the xNiCe catalysts and for the pure CeO2 and NiO oxides after the H2-pretreatment at 400 °C |
| Sample | nB,w a)/(mmol∙g–1) | nB,m b)/(mmol∙g–1) | nB,s c)/(mmol∙g–1) | nB,tot d)/(mmol∙g–1) | nB,irr e)/(mmol∙g–1) |
|---|---|---|---|---|---|
| CeO2 | 35 | 60 | 20 | 115 | 26 |
| 0.3NiCe | 101 | 109 | 27 | 237 | 129 |
| 1.0NiCe | 55 | 88 | 31 | 174 | 71 |
| 2.5NiCe | 43 | 74 | 35 | 152 | 54 |
| 4.0NiCe | 39 | 32 | 22 | 93 | 31 |
| NiO | 6 | 14 | – | 20 | 6 |
a) weak basic sites, 40≤Qdiff<90 kJ∙mol–1; b) medium basic sites, 90≤Qdiff≤150 kJ∙mol–1; c) strong acid sites, Qdiff>150 kJ∙mol–1; d) total basic sites, Qdiff≥40 kJ∙mol–1; e) sites on which CO2 is irreversibly adsorbed at 133 Pa and 80 °C, calculated by the difference in the amount of CO2 adsorbed between the first and the second adsorption run. |
Surface basic properties of metal oxides can generally be ascribed to the presence of basic hydroxyl groups (OH) and basic surface oxygens (O2–), which can react directly with CO2 producing, respectively, hydrogen carbonate and carbonate species. FTIR technique provides a powerful means of investigation of the surface properties of the metal oxides [59,62–64], by studying the interaction with probe molecules. FTIR spectra in the region 2000–1190 cm–1, collected at room temperature on reduced NiO, CeO2, and selected xNiCe samples (0.3NiCe and 2.5NiCe) in contact with CO2 (CO2 pressure equal to 5×103 Pa), are shown in Fig. 7 (solid lines). The figure also reports spectra obtained after outgassing for 30 min at the same temperature (dashed lines). The FTIR spectra of 2.5NiCe, 0.3NiCe and CeO2 display broad signals characterized by the presence of distinct bands with different intensities, indicating, in agreement with the microcalorimetric measurements, the presence of a heterogeneous surface, in which sites of different nature and strength exist [63]. After CO2 adsorption, the 2.5NiCe sample shows a spectrum with at least five partially superimposed contributions. According to the literature [64], the bands centered at about 1614, 1405 and 1214 cm–1 can be mainly ascribed to the presence of hydrogen carbonates (HC). However, a contribution of weaker hidden signals due to bidentate and monodentate carbonates (MC, typically observed at about 1580 and 1360 cm–1, respectively) is suggested by the appearance of two shoulders at about 1510 (MC) and 1315 cm–1 (bidentate carbonates (BC)), due to the interaction of CO2 with basic surface O2– species. As expected, a decrease in the intensity of the bands is observed because of the outgassing procedure, though to a lesser extent in the case of the MC, for which a higher stability can be supposed. Similar features appear in the spectra of the 0.3NiCe sample, which exhibit bands very similar to those observed for 2.5NiCe, in terms of both frequencies and relative intensities. Concerning the pure CeO2 oxide, the most intense peak observed after CO2 adsorption is centered around 1594 cm–1 and this can be reasonably explained by the superimposition of the signals of HC and BC. The presence of HC is also clearly pointed out by the observation of bands at about 1409 and 1216 cm–1, the former characterized by an evident tail towards higher frequencies, possibly associated with polydentate carbonates (PC) [64]. In addition, a well-defined peak ascribable to BC can be observed around 1300 cm–1, while the little hump at about 1510 cm–1 accounts for the presence of MC. The contribution at about 1510 cm–1 becomes slightly more visible after outgassing, confirming the higher stability of the monodentate species with respect to the other carbonate types. In particular, it has been reported in the literature that the formation of monodentate or BC strictly depends on the basicity of the surface oxygen atoms, with stronger sites associated with the formation of monodentate species [63]. A closer inspection of the CeO2 spectra in comparison with those of 0.3NiCe and 2.5NiCe reveals a different distribution between monodentate and BC. The contributions of the monodentate species become more evident in the presence of Ni, in particular in the spectra recorded after outgassing, suggesting that the presence of Ni gives rise to a higher concentration of stronger sites. It is worthy of note that this observation is in agreement with the microcalorimetric results, from which an increase in the amount of medium- and high-strength basic sites with the increase in nickel content was observed up to a Ni/Ce ratio equal to 2.5. Compared to the other oxides, NiO spectra point out a lower amount of carbonate species (bands with low intensity) and the absence of MC, suggesting a poor basic character, as already revealed by microcalorimetry.
The specific surface areas of the Ni0 metal particles, expressed per unit mass of either nickel (ANi) or catalyst (Acat) (Table 4), were determined by means of H2-TPD analyses performed on the reduced catalysts. It is worth noting that, although H2 chemisorption and desorption techniques are often used for determining metal surface area on Ni-based materials [65–67], the possible occurrence of spillover phenomena [66,68] in the presence of ceria could cause an overestimation of the ANi and Acat values. In the light of this, the calculated specific surface areas should be considered as relative data, which anyway allow the evaluation of the capacity of the xNiCe catalysts to adsorb H2 (either on the metal surface or as a consequence of spillover). From the results, a decrease in the ANi value can be observed with the increase in the Ni/Ce molar ratio, which appears particularly evident passing from 0.3NiCe to 1.0NiCe. Unfortunately, due to the possible simultaneous reduction of both NiO and CeO2, and by considering that NiO reduction seems to be incomplete (as indicated by XRD in Fig. S1(b) for 2.5NiCe and 4.0NiCe and by HRTEM for 1.0NiCe), the ANi values cannot be directly related to the dispersion of the Ni0 metal particles, even more so if a contribution of spillover exists. Nevertheless, the high ANi value of the 0.3NiCe sample seems to be in agreement with the absence of Ni0 reflections in its XRD pattern (Fig. S1(b)), which could depend on the high dispersion of the Ni0 nanocrystals. As expected, the metal surface area per unit mass of catalyst, Acat, increases along with the nickel content up to 2.5NiCe, whereas very similar values are instead observed for the 2.5NiCe and 4.0NiCe samples. Taking into account the comparable Ni0 crystallites size (ca. 8 nm), such similar Acat values could be explained by the presence of a comparable amount of Ni0 nanoparticles, possibly due to a lower degree of reduction of the original NiO phase on the latter catalyst.
Tab.4 Specific Ni0 metal area expressed per nickel mass unit (ANi) and catalyst mass unit (Acat) |
| Sample | ANi/(m2∙gNi–1) | Acat/(m2∙gcat–1) |
|---|---|---|
| 0.3NiCe | 65.7 | 5.7 |
| 1.0NiCe | 36.5 | 8.4 |
| 2.5NiCe | 25.8 | 10.6 |
| 4.0NiCe | 21.8 | 10.7 |
Catalytic results
The results of the catalytic tests carried out on the xNiCe catalysts at 300 °C, CO/CO2/H2 molar ratios equal to 1/1/7, and SV = 72000 cm3∙h–1∙gcat–1 are summarized in Table 5. Pure ceria was also tested for comparison and it showed to be inactive in the co-methanation reaction. For all the xNiCe samples, catalytic activity was found to be stable within 6 h on stream (t.o.s), therefore average values for CO and CO2 conversions (XCO and , respectively) and selectivity to methane () are reported. Irrespective of the catalyst composition, a CH4 selectivity higher than 99 mol-% is always observed (traces of ethane were detected in some cases), suggesting that the occurrence of reactions leading to the formation of carbon residues can be ruled out. Moreover, all the catalysts exhibit XCO close to 100 mol-%, while increases up to ca. 80 mol-% by increasing the Ni/Ce molar ratio up to 2.5 and then remains constant. Hydrogenation of CO has already been reported to be favored in comparison with that of CO2 [30,35,36]. For a CO/CO2/H2 molar composition of 1/1/7, equilibrium conversions of about 100 and 93 mol-% for CO and CO2, respectively, were calculated at 300 °C by Gao et al. by means of the total Gibbs free energy minimization method [27]. The present results indicate that, unlike that of CO, which is very close to 100 mol-%, CO2 conversion does not reach the equilibrium value, suggesting that kinetic factors (possibly related to the different reaction mechanisms by which CO and CO2 methanation occurs) rather than thermodynamics are responsible for these observations.
Tab.5 Average (6 h) CO (XCO) and CO2 () conversions and selectivity to methane () of the xNiCe catalysts a) |
| Sample | XCO/mol-% | /mol-% | /mol-% |
|---|---|---|---|
| 0.3NiCe | 98 | 57 | >99 |
| 1.0NiCe | >99 | 74 | >99 |
| 2.5NiCe | >99 | 81 | >99 |
| 4.0NiCe | >99 | 80 | >99 |
a) Reaction conditions: T = 300 °C; P = 1.013×105 Pa; mcat = 0.050 g; CO/CO2/H2 = 1/1/7; SV = 72000 cm3∙h–1∙gcat–1. |
In CO hydrogenation, metal Ni0 species are considered responsible for the activation of CO and H2, leading to adsorbed carbon (Cads) and hydrogen (Hads) species, which are reported as the intermediates in the production of methane [9,69]. Accordingly, CeO2 can be considered not directly involved in the previous reaction, although, in its partially reduced form (CeO2–x), it is reported to be beneficial in favoring an easier CO dissociation through its ability to accept oxygen [11,70]. On the other hand, the presence of ceria and, in particular, the occurrence of strong Ni-Ce interactions are reported to be crucial for obtaining high catalytic performances in the hydrogenation of CO2 to methane, whose main reaction pathway does not involve CO as a reaction intermediate [16,17,71], at variance with the mechanism proposed for Ni/SiO2 [71] and Ni/Al2O3 [72] catalysts. The proposed reaction mechanisms for the CO and CO2 hydrogenation reactions are depicted in Scheme 2.
Indeed, for the present samples the existence of strongly interacting Ni and Ce species is indicated by the Raman and XPS results (which are also in agreement with previous TPR data [17]). Moreover, as highlighted by the microcalorimetric and FTIR results, CeO2 is able to effectively adsorb CO2, forming carbonates and HC, which would be subsequently hydrogenated to formates and finally to methane by hydrogen dissociatively adsorbed on Ni0 particles. In light of these mechanisms, the catalytic behavior observed along the xNiCe series might be elucidated on the basis of the different nature of the sites responsible for activating CO, H2 and CO2. Passing from 0.3NiCe to 4.0NiCe, XCO is always very close to the equilibrium value, indicating that, even for the catalyst with the lowest Ni content, the number and the efficacy of the Ni0 metal sites are such that all the CO molecules supplied can be activated and hydrogenated to methane. By converse, progressively rises going from 0.3NiCe to 2.5NiCe, in parallel with the increase in the Acat value (Table 4). Noteworthy, the increase in such parameter, related to the Ni content, is associated with a diminution in the number of basic sites on which CO2 is activated (Table 3), due to the decrease in the amount of ceria. These results hence suggest that the conversion of CO2 is principally governed by the amount of Hads species available for its hydrogenation, such amount would be limited by the competition between CO and H2 for the same Ni0 sites and should reasonably become higher with the increase in the specific metal area. The comparable catalytic activity of 2.5NiCe and 4.0NiCe (Table 5) can therefore be explained by their very similar Acat values (Table 4).
It could be hypothesized that the preferential conversion of carbon monoxide is related to its quicker hydrogenation when compared to CO2, due to the proximity of the sites on which both CO and H2 are activated, which would favor the reaction between the Cads and Hads intermediates rather than migration of the Hads species by spillover processes toward the ceria sites on which HC and carbonates (coming from CO2 activation) can be hydrogenated. However, in addition to methanation (Scheme 1), CO and CO2 could also be involved in the reverse methane dry reforming and in the water gas shift reactions (Scheme 3). Indeed, both reactions are thermodynamically favored at 300 °C (the equilibrium constant for the reaction of reverse methane dry reforming is even higher than those of the methanation reactions) [27,73]. Moreover, NiCe-based systems have been reported as active catalysts for the water gas shift [74,75] and the methane dry reforming reactions [76,77]. Notably, reverse dry reforming has been reported to be the main reaction for CO hydrogenation at low H2/CO molar ratios with La2O3-promoted Ni/Al2O3 [73] and Ni/SiO2 [78] at temperatures between 300 °C and 550 °C. Consequently, it might be possible that not only the hydrogenation of CO2 is repressed by the competitive CO methanation, but also that the occurrence of the reverse dry methane reforming and/or the water gas shift reactions, through which carbon dioxide is formed, can lead to an overall lower conversion of CO2.
At a much higher space velocity (450000 cm3∙h–1∙gcat–1), the 1.0NiCe, 2.5NiCe and 4.0NiCe catalysts are stable within the investigated 6 h on stream (Fig. 8), with methane as the only product. Though decreasing slightly with the increase in SV, XCO remains very high (96–98 mol-%) and close to the equilibrium value regardless of the Ni content, indicating that a significant portion of Ni0 sites still remains active in these severe reaction conditions. A pronounced effect of the increase in space velocity is instead observed on CO2 conversion, which decreases by about 20 mol-% for the three samples. Interestingly, 2.5NiCe and 4.0NiCe show the same catalytic behavior also in these conditions, confirming what previously observed at the lower SV value (72000 cm3∙h–1∙gcat–1).
These results highlight that the increase in space velocity further disadvantages the hydrogenation of CO2 with respect to that of CO. Compared to the other catalysts in the series, the catalytic behavior of 0.3NiCe appears significantly different (Fig. 8), besides showing a manifest decrease (which is much more pronounced for CO2, whose conversion decreases by about 47% compared to that at the lower space velocity), CO and CO2 conversion values also decline with time-on-stream. Considering that was around 88 mol-%, with only traces of ethane detected as a byproduct, the evident deactivation of this catalyst can supposedly be associated with the formation of carbonaceous deposits on its surface, which, interestingly, appears to be promoted by the low contact time. Carbon analysis, performed on fresh and post-reaction 0.3NiCe and 1.0NiCe catalysts, confirms this hypothesis. While only a slight increase in the percentage of carbon is observed in the 1.0NiCe catalyst after the reaction (values of 0.2 and 1.0 wt-% are obtained for the fresh and post-reaction samples, respectively), a significant formation of carbonaceous species results for 0.3NiCe after the reaction, whose carbon content passes from 0.1 (fresh) to 4.8 wt-% (post-reaction). It is worth noting that such a difference between 0.3NiCe and 1.0NiCe appears even more significant if the carbon content on the post-reaction samples is expressed per unit mass of nickel, i.e., the catalytic active species on which carbonaceous deposits can mainly be supposed to form. The obtained results, 4.1 and 55.4 wt-% for 1.0NiCe and 0.3NiCe, respectively, clearly show the much greater tendency of the 0.3NiCe catalyst to deactivate. To get information on the nature of the carbonaceous species, a TGA was performed, and the results are shown in Fig. S9 (cf. ESM). Weight losses of about 1.3% and 5.0% have been obtained at high temperature (~640 °C) for 1.0NiCe and 0.3NiCe, respectively (Fig. S9(a)). Such losses, well evidenced in the DTG curve (Fig. S9(b)) correspond to an exothermic process (Fig. S9(c)), which can reasonably be ascribed to the combustion of carbon-like species [79], suggesting the occurrence of CO disproportionation.
It was reported in the literature that on NiCe-based catalysts the establishment of strong interactions between Cads species and Ni0 metal sites, which are favored in the presence of small Ni0 particles, can lead to the deactivation of the catalyst due to the possible occurrence of the Boudouard reaction (Scheme 4), through which carbon is formed [9].
Considering that very small Ni0 particles are supposed to form on the 0.3NiCe sample, its deactivation can be explained accordingly: due to the low nickel content, the amount of hydrogen activated at high SV on this catalyst would be insufficient to reduce a significant portion of the Cads species formed, which therefore tend to polymerize forming carbon deposits. Taking into account the small Ni0 crystallites size (≤8 nm), a preferential adsorption and activation of CO with respect to H2 might also be hypothesized for the other xNiCe catalysts, which could explain the notable decrease in CO2 conversion at high space velocity (Fig. 8); unlike the case of 0.3NiCe, due to the higher Ni content, the available amount of Hads is however sufficient to hydrogenate CO almost completely. In order to verify the role of carbon monoxide in the deactivation of the catalyst, CO methanation tests were carried out on 0.3NiCe and 1.0NiCe at the same temperature and SV, using a H2/CO molar ratio equal to 3 (stoichiometric). At 72000 cm3∙h–1∙gcat–1, no deactivation phenomena were observed after 24 h of reaction, with CO conversions of 98 and 100 mol-% for 0.3NiCe and 1.0NiCe, respectively. Besides methane, a significant amount of carbon dioxide was detected ( equal to 16 and 11 mol-% for 0.3NiCe and 1.0NiCe, respectively), possibly formed through the reverse methane dry reforming and/or the water gas shift reactions (Scheme 3). A very different catalytic behavior was instead observed at 450000 cm3∙h–1∙gcat–1, while 1.0NiCe still exhibited a constant catalytic performance within 24 h on stream (XCO = 84 mol-%, = 84 mol-%, = 16 mol-%), 0.3NiCe was completely deactivated after 1 h of reaction. This result is not surprising and clearly confirms the involvement of carbon monoxide in the deactivation of 0.3NiCe at high SV. Proof of the possible occurrence of the Boudouard reaction was also obtained by performing CO-TPD analyses on the same samples. In the TPD profiles (Fig. S10, cf. ESM), collected by a massspectrometer detector, a clear contribution of CO does not appear, while a noticeable signal ascribable to CO2 is observed in the range 150 °C–500 °C for both catalysts. These findings highlight the ability of the catalysts to promote CO disproportionation that, as revealed by the catalytic results, can become significant even in the presence of hydrogen when the number of Ni0 sites and the reaction conditions adopted are not suitable to ensure the activation of a proper amount of H2.
With the aim of evaluating the performance of the catalysts using a feed composition similar to that of a generic syngas (for which H2 is usually in defect with respect to the stoichiometric amount imposed by the CO and CO2 methanation reactions (Scheme 1)), further co-methanation tests were carried out at 300 °C, with a CO/CO2/H2 molar composition equal to 1/1/5, and a space velocity of 72000 cm3∙h–1∙gcat–1. The relevant results are reported in Table 6. The catalysts were stable during the 6-hour tests, with a CH4 selectivity higher than 99 mol-%. Despite the less favorable H2/COx molar ratio, XCO still results close to 100 mol-%, with a very slight increase passing from 0.3NiCe to 1.0NiCe. On the other hand, appears clearly limited by the low concentration of H2 in the reactant mixture, confirming that the activated hydrogen is preferentially consumed in the conversion of CO. As expected, increases with the nickel content up to a Ni/Ce molar ratio equal to 2.5, while, once again, the 2.5NiCe and 4.0NiCe samples are characterized by the same conversion values.
Tab.6 Average (6 h) CO (XCO) and CO2 () conversions and selectivity to methane () of the xNiCe catalysts a) |
| Sample | XCO/mol-% | /mol-% | /mol-% |
|---|---|---|---|
| 0.3NiCe | 97 | 29 | >99 |
| 1.0NiCe | 99 | 34 | >99 |
| 2.5NiCe | 99 | 37 | >99 |
| 4.0NiCe | 99 | 38 | >99 |
Being the catalyst on which, in the operating conditions discussed so far, the most significant effects on CO2 conversion were found, without however observing any deactivation phenomena, 1.0NiCe was chosen to perform additional catalytic tests, through which co-methanation was studied in terms of thermal stability and activity at different reaction temperatures and a wide range of SV. Noteworthy, methane was always detected as the only product, confirming the high selectivity of the present catalysts toward hydrogenation in different reaction regimes, provided that the nickel content is appropriate. The conversion values as a function of the reaction temperature in the range 200 °C–500 °C are plotted in Fig. 9(a). From 200 °C to 300 °C, XCO and rise with temperature and then diminish because of the exothermicity of both the CO and CO2 methanation reactions. Thermal stability was studied by carrying out a long-run test (50 h) in which different temperature ranges were investigated up to 500 °C, considering 300 °C as the control value. The results, summarized in Fig. 9(b), highlight that, even after exposure to the temperature of 500 °C, the catalytic activity at 300 °C is always restored, indicating excellent thermal stability of the catalyst.
Finally, the influence of space velocity was studied in the range 48000–450000 cm3∙h–1∙gcat–1 (Fig. 10). With the increase in SV, XCO and decrease by 5% and 29%, respectively, which is coherent with the influence of the contact time on H2 activation and with the preferential hydrogenation of carbon monoxide. Compared to that of CO, a steeper decrease in the conversion of CO2 during the co-methanation reaction has already been reported in the literature. For a Ni/Ce0.50Zr0.50O2 catalyst with a nickel content similar to that of the present 1.0NiCe, CO and CO2 conversions were found to diminish by 5% and 36%, respectively, in the SV range between 72000 and 450000 cm3∙h–1∙gcat–1, using a substoichiometric amount of H2 [36]. A decrease in CO2 conversion of about 23% was instead reported for a Mn-promoted Ni/Al2O3 catalyst [33] in a much narrower SV range (from 48000 to 84000 cm3∙h–1∙gcat–1) while CO conversion remained constant.
Fig.13 CO and CO2 co-methanation results for the 1.0NiCe sample. COx conversions as a function of (a) reaction temperature and of (b) t.o.s. by varying the temperature during the run: (•), CO conversion; (•), CO2 conversion; (---), equilibrium conversions. Other reaction conditions: CO/CO2/H2 = 1/1/5; SV= 72000 cm3∙h–1∙gcat–1. |
Conclusions
NiO-CeO2 mixed oxides, prepared by the soft template method with different Ni/Ce molar ratios (0.3–4.0), were used as catalysts in the CO and CO2 co-methanation reaction after reduction in H2 at mild conditions (400 °C for 1 h). At 72000 cm3∙h–1∙gcat–1, 300 °C, and CO/CO2/H2 molar composition equal to 1/1/7, a CH4 selectivity higher than 99 mol-% was observed, irrespective of the catalyst composition. Whereas CO was almost completely transformed (XCO≥98 mol-%) on all the xNiCe catalysts, the conversion of CO2 was at most 80 mol-%, considerably lower than the equilibrium value of 93 mol-%, suggesting that kinetic factors rather than thermodynamics play a role. The trend of CO2 conversion as a function of the Ni/Ce molar ratio suggested that, though the presence of ceria is important for activating CO2, its hydrogenation is mainly governed by the availability of Hads species, which depends on the metal surface area (Acat). In the presence of a lower H2 content in the feed (CO/CO2/H2 = 1/1/5), CO hydrogenation was even more favored with respect to that of CO2. The explanation proposed for the faster and definitely preferred CO hydrogenation is that, due to the proximity of the sites on which both CO and H2 are dissociatively adsorbed and activated, the reaction between the Cads and Hads intermediates is favored in comparison with migration of Hads species by spillover processes from the Ni0 metal sites toward the nearby ceria sites on which CO2 is activated. The catalytic results obtained at higher SV (450000 cm3∙h–1∙gcat–1) further supported this hypothesis, highlighting that the hydrogenation of CO2 is considerably disadvantaged by the increase in space velocity while that of CO is only slightly affected. Moreover, in addition to the competitive CO methanation, the overall conversion of CO2 might also be lowered by the occurrence of the reverse dry methane reforming and/or the water gas shift reactions. In the case of the 0.3NiCe catalyst, a manifest decrease in both CO and CO2 conversions as a function of t.o.s was observed using a stoichiometric amount of H2 at high space velocity. This deactivation was proven to be caused by deposition of carbon, deriving from the Boudouard reaction, on the Ni0 metal sites. Such result highlighted the importance of optimizing the Ni content to prevent deactivation and to obtain good catalytic performances for the COx co-methanation reaction in a wide range of operating conditions. Indeed, when the number of Ni0 sites was enough to favor the hydrogenation of CO over its disproportionation, the NiCe investigated system was very promising, being highly active and stable (up to 50 h) also when submitted to thermal stress.