Production of pectic extracts from sugar beet pulp with antiproliferative activity on a breast cancer cell line

Abstract

In the last years, sugar beet pectins have been the subject of several investigations involving extraction methodologies, chemical composition and functional properties. The structure of pectins, which depends on the extraction method, is decisive in their capacity to induce apoptosis on several cancer cell lines like colon, prostate and breast. In this work, sugar beet pectin extraction was performed in the following steps: lipid extraction with hexane, removal of soluble complex carbohydrates and proteins, and enzymatic treatment with amyloglucosidase, protease, and pectinase. The enzymatic treatment was carried out with Rohapect DA6L under the following conditions: 50°C, pH 4.0, 2% enzyme/substrate (E/S) ratio, 15 h, and a solid to liquid ratio of 1 ∶ 10. The pectic extract showed a degree of polymerization (DP) profile of 55.8% with DP≥7; 4.9% with DP6; 5.8% between DP2 and DP6 ; 4.7% with DP2; and 28.8% with DP1. The pectic extract was examined for its antiproliferative activity on the MCF-7 breast cancer cell line. At a concentration range of 12.5–25 mg/mL the pectic extract killed 80.6% of the cells, exhibiting a higher antiproliferative activity than 4-hydroxytamoxifen (4-OHT), a classical anticancer drug, which killed 56.5% of the cells.

Keywords

pectic extracts / antiproliferative activity / breast cancer / enzymatic treatment

1 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.

Fig.1 Scheme 1 Stoichiometric equations for CO and CO2 hydrogenation to methane.

Full size|PPT slide

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∙g_cat^–1. Furthermore, on a selected sample, the effect of reaction temperature and the influence of SV, as well as the stability, were also investigated.

2 Experimental

2.1 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.

2.2 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.

2.3 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).

2.4 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 (f_int = 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∙g_cat^–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 CO_x 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∙g_cat^–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∙g_cat^–1 were carried out in order to get information on the occurrence of side-reactions possibly responsible for carbon residues formation.

3 Results and discussion

3.1 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/${\mathrm{g}}_{{\mathrm{C}\mathrm{e}\mathrm{O}}_2}$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.

Fig.2 (a) TEM and (b–d) HRTEM images of the 1.0NiCe sample (b) in the original oxidized form, (c) after the H2-pretreatment at 400 °C for 1 h and (d) after reaction.

Full size|PPT slide

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 (F_2g 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 (F_2g 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 F_2g 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 F_2g 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.

Fig.3 Raman spectra of CeO2 and xNiCe samples.

Full size|PPT slide

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 F_2g 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 F_2g 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 (F_2g 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 F_2g 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.

Fig.4 Raman spectra of (a) CeO2, (b) 0.3NiCe and (c) 4.0NiCe: before H2-pretreatment (grey curve); after H2-pretreatment (black curve); spectral difference between the curves after and before the H2-pretreatment (dashed curve).

Full size|PPT slide

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 3d_5/2 and 3d_3/2 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 2p_3/2 and 2p_1/2 spin-orbit components (spin-orbit splitting, 17.2 eV) and associated satellite (separated by 6.1 eV, about 50% of the total intensity), the 2p_3/2 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 Ce3d_5/2 and Ni2p_3/2 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.

Fig.5 XPS spectra of CeO2 and xNiCe samples.

Full size|PPT slide

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 Ni2p_3/2 component to overcome the difficulty due to the partial overlap with the Ce3d region and to the complexity of the Ni2p spectroscopic features. The Ni2p_3/2 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 Ni2p_3/2 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.

Fig.6 XPS spectra of the 0.3NiCe sample: (a) Ce3d, (b) Ni2p3/2 and (c) O1s regions.

Full size|PPT slide

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⁺/Ni_total and Ce³⁺/Ce_total 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	($n_{{\mathrm{N}\mathrm{i}}^{3+}}$/nNi, total)surface a)	($n_{{\mathrm{C}\mathrm{e}}^{3+}}$/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 Q_diff vs. CO₂ uptake curves of the reduced CeO₂ and NiO soft-templated oxides are also reported for comparison. It can be observed that Q_diff 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 Q_diff 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 Q_diff vs. CO₂ uptake curves, it is possible to estimate the strength distribution of the basic sites. A Q_diff 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≤Q_diff<90 kJ∙mol^–1), medium (90≤Q_diff≤150 kJ∙mol^–1), and strong (Q_diff> 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.

Fig.7 Differential heat of adsorption as a function of CO2 uptake for the CeO2 and NiO pure oxides and xNiCe catalysts, after the H2-pretreatment at 400 °C: (■) CeO2; (○) 0.3NiCe; (△) 1.0NiCe; (▿) 2.5NiCe; (◇) 4.0NiCe; (•) NiO.

Full size|PPT slide

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 (O^2–), 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 O^2– 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.

Fig.8 FTIR spectra of CO2 adsorbed at room temperature (solid lines; equilibrium pressure: 5×103 Pa) and after outgassing (dashed lines; residual pressure: 10–1 Pa) for 30 min at the same temperature on reduced NiO, CeO2, 0.3NiCe and 2.5NiCe samples.

Full size|PPT slide

The specific surface areas of the Ni⁰ metal particles, expressed per unit mass of either nickel (A_Ni) or catalyst (A_cat) (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 A_Ni and A_cat 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 A_Ni 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 A_Ni 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 A_Ni 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, A_cat, 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 A_cat 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

3.2 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∙g_cat^–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 (X_CO and $X_{{\mathrm{C}\mathrm{O}}_2}$, respectively) and selectivity to methane ($S_{{\mathrm{C}\mathrm{H}}_4}$) 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 X_CO close to 100 mol-%, while $X_{{\mathrm{C}\mathrm{O}}_2}$$X_{{\mathrm{C}\mathrm{O}}_2}$ 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 ($X_{{\mathrm{C}\mathrm{O}}_2}$) conversions and selectivity to methane ($S_{{\mathrm{C}\mathrm{H}}_4}$) of the xNiCe catalysts a)

Sample	XCO/mol-%	$X_{{\mathrm{C}\mathrm{O}}_2}$/mol-%	$S_{{\mathrm{C}\mathrm{H}}_4}$/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 (C_ads) and hydrogen (H_ads) 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 (CeO_2–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.

Fig.9 Scheme 2 Proposed mechanisms for (a) the CO and (b) CO2 hydrogenation to methane.

Full size|PPT slide

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, X_CO 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, $X_{{\mathrm{C}\mathrm{O}}_2}$ progressively rises going from 0.3NiCe to 2.5NiCe, in parallel with the increase in the A_cat 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 H_ads 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 A_cat 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 C_ads and H_ads intermediates rather than migration of the H_ads 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.

Fig.10 Scheme 3 Stoichiometric equations for: (1) reverse methane dry reforming and (2) water gas shift.

Full size|PPT slide

At a much higher space velocity (450000 cm³∙h^–1∙g_cat^–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, X_CO 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∙g_cat^–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 $S_{{\mathrm{C}\mathrm{H}}_4}$ 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.

Fig.11 CO (black symbols) and CO2 (grey symbols) conversions for 0.3NiCe (■,■), 1.0NiCe (•, •), 2.5NiCe (▲,▲), and 4.0NiCe (✶,✶). Reaction conditions: P = 1.013×105 Pa; T = 300°C; CO/CO2/H2 molar ratio equal to 1/1/7; SV = 450000 cm3∙gcat–1∙h–1.

Full size|PPT slide

It was reported in the literature that on NiCe-based catalysts the establishment of strong interactions between C_ads 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].

Fig.12 Scheme 4 Stoichiometric equation for the CO disproportionation (Boudouard) reaction.

Full size|PPT slide

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 C_ads 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 H_ads 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∙g_cat^–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 ($S_{{\mathrm{C}\mathrm{O}}_2}$ 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∙g_cat^–1, while 1.0NiCe still exhibited a constant catalytic performance within 24 h on stream (X_CO = 84 mol-%, $S_{{\mathrm{C}\mathrm{H}}_4}$ = 84 mol-%, $S_{{\mathrm{C}\mathrm{O}}_2}$ = 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∙g_cat^–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/CO_x molar ratio, X_CO still results close to 100 mol-%, with a very slight increase passing from 0.3NiCe to 1.0NiCe. On the other hand, $X_{{\mathrm{C}\mathrm{O}}_2}$ 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, $X_{{\mathrm{C}\mathrm{O}}_2}$ 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 ($X_{{\mathrm{C}\mathrm{O}}_2}$) conversions and selectivity to methane ($S_{{\mathrm{C}\mathrm{H}}_4}$) of the xNiCe catalysts a)

Sample	XCO/mol-%	$X_{{\mathrm{C}\mathrm{O}}_2}$/mol-%	$S_{{\mathrm{C}\mathrm{H}}_4}$/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, X_CO and $S_{{\mathrm{C}\mathrm{H}}_4}$ 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∙g_cat^–1 (Fig. 10). With the increase in SV, X_CO and $S_{{\mathrm{C}\mathrm{H}}_4}$ 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/Ce_0.50Zr_0.50O₂ 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∙g_cat^–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∙g_cat^–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.

Full size|PPT slide

Fig.14 CO (black curve) and CO2 (grey curve) conversions as a function of space velocity.

Full size|PPT slide

4 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∙g_cat^–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 (X_CO≥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 H_ads species, which depends on the metal surface area (A_cat). 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 C_ads and H_ads intermediates is favored in comparison with migration of H_ads 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∙g_cat^–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 CO_x 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.

References

[1] Oosterveld A, Beldman G, Voragen A G J. Enzymatic modification of pectic polysaccharides obtained from sugar beet pulp. Carbohydrate Polymers, 2002, 48(1): 73-81

[2] Oosterveld A, Beldman G, Schols H, Voragen A. Characterization of arabinose and ferulic acid rich pectic polysaccharides and hemicelluloses from sugar beet pulp. Carbohydrate Research, 2000, 328(2): 185-197

[3] Sun R, Hughes S. Fractional isolation and physico-chemical characterization of alkali-soluble polysaccharides from sugar beet pulp. Carbohydrate Polymers, 1999, 38(3): 273-281

[4] Schols H, Voragen G, Colquhoun I. Isolation and characterization of rhamnogalacturonan oligomers, liberated during degradation of pectic hairy regions by rhamnogalacturonase. Carbohydrate Research, 1994, 256(1): 97-111

[5] Schols H, Vierhuis E, Bakx E, Voragen A. Different populations of pectic hairy regions occur in apple cell walls. Carbohydrate Research, 1995, 275(2): 343-360

[6] Yapo B, Robert C, Etienne I, Wathelet B, Paquot M. Effect of extraction conditions on the yield, purity and surface properties of sugar beet pulp pectin extracts. Food Chemistry, 2007, 100(4): 1356-1364

[7] Yapo B. Pectin quantity, composition and physicochemical behaviour as influenced by the purification process. Food Research International, 2009, 42(8): 1197-1202

[8] Yapo B. Pectic substances: From simple pectic polysaccharides to complex pectins—A new hypothetical model. Carbohydrate Polymers, 2011, 86(8): 373-385

[9] Yeoh S, Shi J, Langrish T. Comparisons between different techniques for water-based extraction of pectin from orange peels. Desalination, 2008, 218(1-3): 229-237

[10] Willats W, Knox J, Mikkelsen J. Pectin: New insights into an old polymer are starting to gel. Trends in Food Science & Technology, 2006, 17(3): 97-104

[11] Micard V, Renard C M G C, Thibault J F. Enzymatic saccharification of sugar-beet pulp. Enzyme and Microbial Technology, 1996, 19(3): 162-170

[12] Levigne S, Ralet M C, Thibault J F. Characterisation of pectins extracted from fresh sugar beet under different conditions using an experimental design. Carbohydrate Polymers, 2002, 49(2): 145-153

[13] Robert C, Emaga T, Wathelet B, Paquot M. Effect of variety and harvest date on pectin extracted from chicory roots (Cichorium intybus L.). Food Chemistry, 2008, 108(3): 1008-1018

[14] Schieber A, Hilt P, Streker P, Endreß H U, Rentschler C, Carle R. A new process for the combined recovery of pectin and phenolic compounds from apple pomace. Innovative Food Science & Emerging Technologies, 2003, 4(1): 99-107

[15] Westereng B, Michaelsen T, Samuelsen A, Knutsen S. Effects of extraction conditions on the chemical structure and biological activity of white cabbage pectin. Carbohydrate Polymers, 2008, 72(1): 32-42

[16] Koubala B, Kansci G, Mbome L, Crépeau M J, Thibault J F, Ralet M C. Effect of extraction conditions on some physicochemical characteristics of pectins from “Améliorée” and “Mango” mango peels. Food Hydrocolloids, 2008, 22(7): 1345-1351

[17] Koubala B, Mbome L, Kansci G, Tchouanguep Mbiapo F, Crepeau M J, Thibault J F, Ralet M C. Physicochemical properties of pectins from ambarella peels (Spondias cytherea) obtained using different extraction conditions. Food Chemistry, 2008, 106(3): 1202-1207

[18] Liu Y, Shi J, Langrish T. Water-based extraction of pectin from flavedo and albedo of orange peels. Chemical Engineering Journal, 2006, 120(3): 203-209

[19] Rombouts F, Thibault J. Enzymic and chemical degradation and the fine structure of pectins from sugar-beet pulp. Carbohydrate Research, 1986, 154(1): 189-203

[20] Rombouts F, Thibault J. Feruloylated pectic substances from sugar-beet pulp. Carbohydrate Research, 1986, 154(1): 177-187

[21] Happi Emaga T, Ronkart S, Robert C, Wathelet B, Paquot M. Characterisation of pectins extracted from banana peels (Musa AAA) under different conditions using an experimental design. Food Chemistry, 2008, 108(2): 463-471

[22] Concha J, Zúñiga M. Enzymatic depolymerization of sugar beet pulp: Production and characterization of pectin and pectic-oligosaccharides as a potential source for functional carbohydrates. Chemical Engineering Journal, 2012, 192: 29-36

[23] Claye S, Idouraine A, Weber C. Extraction and fractionation of insoluble fiber from five fiber sources. Food Chemistry, 1996, 57(2): 305-310

[24] Deytieux-Belleau C, Vallet A, Donèche B, Geny L. Pectin methylesterase and polygalacturonase in the developing grape skin. Plant Physiology and Biochemistry, 2008, 46(7): 638-646

[25] Iglesias M, Lozano J E. Extraction and characterization of sunflower pectin. Journal of Food Engineering, 2004, 62(3): 215-223

[26] Kulkarni S, Vijayanand P.Effect of extraction conditions on the quality characteristics of pectin from passion fruit peel (Passiflora edulis f. flavicarpa L.). LWT-Food Science and Technology, 2010, 43(7): 1026-1031

[27] Liu L, Cao J, Huang J, Cai Y, Yao J. Extraction of pectins with different degrees of esterification from mulberry branch bark. Bioresource Technology, 2010, 101(9): 3268-3273

[28] Marry M, McCann M, Kolpak F, White A R, Stacey N J, Roberts K. Extraction of pectic polysaccharides from sugar-beet cell walls. Journal of the Science of Food and Agriculture, 2000, 80(1): 17-28

[29] Sahari M, Akbarian M A, Hamedi M. Effect of variety and acid washing method on extraction yield and quality of sunflower head pectin. Food Chemistry, 2003, 83(1): 43-47

[30] Singthong J, Ningsanond S, Cui S, Goff H. Extraction and physicochemical characterization of Krueo Ma Noy pectin. Food Hydrocolloids, 2005, 19(5): 793-801

[31] Zykwinska A, Boiffard M, Kontkanen H, Buchert J, Thibault J F, Bonnin E. Extraction of green labeled pectins and pectic oligosaccharides from plant byproducts. Journal of Agricultural and Food Chemistry, 2008, 56(19): 8926-8935

[32] Zykwinska A, Rondeau-Mouro C, Garnier C, Thibault J, Ralet M. Alkaline extractability of pectic arabinan and galactan and their mobility in sugar beet and potato cell walls. Carbohydrate Polymers, 2006, 65(4): 510-520

[33] Zykwinska A, Gaillard C, Boiffard M, Thibault J F, Bonnin E. “Green labelled” pectins with gelling and emulsifying properties can be extracted by enzymatic way from unexploited sources. Food Hydrocolloids, 2009, 23(8): 2468-2477

[34] Sathisha U, Jayaram S, Harish Nayaka M, Dharmesh S. Inhibition of galectin-3 mediated cellular interactions by pectic polysaccharides from dietary sources. Glycoconjugate Journal, 2007, 24(8): 497-507

[35] Jackson C, Dreaden T, Theobald L, Tran N M, Beal T L, Eid M, Gao M Y, Shirley R B, Stoffel M T, Kumar M V, Mohnen D. Pectin induces apoptosis in human prostate cancer cells: correlation of apoptotic function with pectin structure. Glycobiology, 2007, 17(8): 805-819

[36] Maxwell E, Belshaw N, Waldron K, Morris V. Pectin—An emerging new bioactive food polysaccharide. Trends in Food Science & Technology, 2012, 24(2): 64-73

[37] Kidd P. A new approach to metastatic cancer prevention: Modified citrus pectin (MCP), an unique pectin that blocks cell surface lectins. Alternative Medicine Review, 1996, 1(1): 4-10

[38] Davis C, Milner J. Gastrointestinal microflora, food components and colon cancer prevention. Journal of Nutritional Biochemistry, 2009, 20(10): 743-752

[39] Glinskii O V, Huxley V H, Glinsky G V, Pienta K J, Raz A, Glinsky V V. Mechanical Entrapment is insufficient and intercellular adhesion is essential for metastasic cell arrest in distant organs. Neoplasia (New York, N.Y.), 2005, 7(5): 522-527

[40] Bergman M, Djaldetti M, Salman H, Bessler H. Effect of citrus pectin on malignant cell proliferation. Biomedicine and Pharmacotherapy, 2010, 64(1): 44-47

[41] Morra M, Cassinelli C, Cascardo G, Nagel M D, Della Volpe C, Siboni S, Maniglio D, Brugnara M, Ceccone G, Schols H A, Ulvskov P. Effects on interfacial properties and cell adhesion of surface modification by pectic hairy regions. Biomacromolecules, 2004, 5(6): 2094-2104

[42] Gulfi M, Arrigoni E, Amado R. Influence of structure on in vitro fermentability of commercial pectins and partially hydrolysed pectin preparations. Carbohydrate Polymers, 2005, 59(2): 247-255

[43] Bland E, Keshavarz T, Bucke C. The influence of small oligosaccharides on the immune system. Carbohydrate Research, 2004, 339(10): 1673-1678

[44] Garthoff J, Heemskerk S, Hempenius R. Safety evaluation of pectin-derived acidic oligosaccharides (pAOS): genotoxicity and sub-chronic studies. Regulatory toxicology and pharmacology. RTP, 2010, 57(1): 31-42

[45] Olano-Martin E, Gibson G, Rastell R. Comparison of the in vitro bifidogenic properties of pectins and pectic-oligosaccharides. Journal of Applied Microbiology, 2002, 93(3): 505-511

[46] Chauhan D, Catley L, Li G, Podar K, Hideshima T, Velankar M, Mitsiades C, Mitsiades N, Yasui H, Letai A, Ovaa H, Berkers C, Nicholson B, Chao T H, Neuteboom S T C, Richardson P, Palladino M A, Anderson K C. A novel orally active proteasome inhibitor induces apoptosis in multiple myeloma cells with mechanisms distinct from Bortezomib. Cancer Cell, 2005, 8(5): 407-419

[47] Favela-Torres E, Volke-Sepúlveda T, Viniegra-González G. Production of Hydrolytic Depolymerising Pectinases. Production, 2006, 44(2): 221-227

[48] Miller G. Use of dinitrosalicylic acid reagent for determination of reducing sugar. Analytical Chemistry, 1959, 31(3): 426-428

[49] Dubois M, Gilles K, Hamilton J, Rebers P A, Smith F. Colorimetric method for determination of sugars and related substances. Analytical Chemistry, 1956, 28(3): 350-356

[50] Skehan P, Storeng R, Scudiero D, Monks A, McMahon J, Vistica D, Warren J T, Bokesch H, Kenney S, Boyd M R. New colorimetric cytotoxicity assay for anticancer-drug screening. JNCI J Natl Cancer Inst, 1990, 82(13): 1107-1112

[51] Holck J, Hjernø K, Lorentzen A, Vigsnæs L K, Hemmingsen L, Licht T R, Mikkelsen J D, Meyer A S. Tailored enzymatic production of oligosaccharides from sugar beet pectin and evidence of differential effects of a single DP chain length difference on human faecal microbiota composition after in vitro fermentation. Process Biochemistry, 2011, 46(5): 1039-1049

[52] Gnanasambandam R. Determination of pectin degree of esterification by diffuse reflectance Fourier transforms infrared spectroscopy. Food Chemistry, 2000, 68(3): 327-332

[53] Bermudez G.Obtencion y caracterizacion de pectinas de alto y bajo metoxilo de la manzana, variedad Pachacamac. Revista-sociedad quimica del Perú, 2003, 69(3): 155-162

[54] Ovodov Y. Current views on pectin substances. Russian Journal of Bioorganic Chemistry, 2009, 35(3): 269-284

[55] Gullón B, Yáñez R, Alonso J, Parajó J. Production of oligosaccharides and sugars from rye straw: A kinetic approach. Bioresource Technology, 2010, 101(17): 6676-6684

[56] Sriamornsak P. Chemistry of pectin and its pharmaceutical uses: A review. Silpakorn University International Journal, 2003, 3(1-2): 206-228

[57] Nangia-Makker P, Hogan V, Honjo Y, Baccarini S, Tait L, Bresalier R, Raz A. Inhibition of human cancer cell growth and metastasis in nude mice by oral intake of modified citrus pectin. Journal of the National Cancer Institute, 2002, 94(24): 1854-1862

[58] Sharma B, Naresh L, Dhuldhoya N.An overview on pectins. Time processing Journals, 2006, 4: 44-51

[59] Ghazarian H, Idoni B, Oppenheimer S B. A glycobiology review: Carbohydrates, lectins and implications in cancer therapeutics. Acta Histochemica, 2011, 113(3): 236-247

[60] Glinsky V, Raz A. Modified citrus pectin anti-metastatic properties: One bullet, multiple targets. Carbohydrate Research, 2009, 344(14): 1788-1791

[61] Nangia-Makker P, Conklin J, Hogan V, Raz A. Carbohydrate-binding proteins in cancer, and their ligands as therapeutic agents. Trends in Molecular Medicine, 2002, 8(4): 187-192

[62] Nakahara S, Oka N, Raz A. Raz. On the role of galectin-3 in cancer apoptosis. Apoptosis. An International Journal on Programmed Cell Death, 2005, 10(2): 267-275

[63] Danguy A, Camby I, Kiss R. Galectins and cancer. Biochimica et Biophysica Acta, 2002, 1572(2-3): 285-293

[64] Nangia-Makker P, Balan V, Raz A. Regulation of tumor progression by extracellular galectin-3. Cancer Microenvironment. Official Journal of the International Cancer Microenvironment Society, 2008, 1(1): 43-51

Acknowledgements

Generous financial support by DIPUV GRANT 37/2009 and Fondecyt GRANT N 1121062 is gratefully acknowledged.

RIGHTS & PERMISSIONS

2014 Higher Education Press and Springer-Verlag Berlin Heidelberg