Influence of Ca/P ratio on the catalytic performance of Ni/hydroxyapatite samples in dry reforming of methane

Influence of Ca/P ratio on the catalytic performance of Ni/hydroxyapatite samples in dry reforming of methane Zouhair Boukhaa, María Pilar Yesteb, Miguel Ángel Cauquib, Juan R. GonzálezVelascoa (a)Chemical Technologies for Environmental Sustainability Group, Department of Chemical Engineering, Faculty of Science and Technology, University of the Basque Country UPV/EHU, P.O. Box 644, E-48080 Bilbao, Spain (b)Departamento de Ciencia de los Materiales e Ingeniería Metalúrgica y Química Inorgánica, Faculty of Sciences, University of Cadiz, Campus Río San Pedro s/n, 11510, Puerto Real, Cádiz, Spain *Corresponding author: Zouhair Boukha Phone: +34 946015502 Fax: +34 946013500 E-mail address: [email protected] This is the accepted manuscript of the article that appeared in final form in Applied Catalysis A: General 580 : 34-45 (2019), which has been published in final form at https://doi.org/10.1016/j.apcata.2019.04.034. © 2019 Elsevier under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Abstract A series of Ni/hydroxyapatite samples presenting different Ca/P molar ratios were synthesised to study the influence of the hydroxyapatite support composition on their catalytic properties in the dry reforming of methane. Our results reveal that a preparation starting from a sub-stoichiometric composition (Ca/P < 1.67) followed by the impregnation of Ni results in suitable properties which give the highest catalytic performance compared with stoichiometric (Ca/P = 1.67) and over-stoichiometric (Ca/P = 1.73) compositions, respectively. The characterisation of the investigated Ni/HAP materials shows that their texture, surface chemistry (acid/base) and the Ni species distribution are mainly derived from the structural properties of the used support. The activity of the Ni/HAP samples in DRM shows that their performances follow this trend: Ni/HAP-D2 (Ca/P = 1.62) > Ni/HAP-D1 (1.57) > Ni/HAP-S (1.67) > Ni/HAP-E (1.73). The superiority of the sample with a Ca/P molar ratio of 1.62 was explained by a suitable surface chemistry consisting of an abundance of strong acid sites and basic sites. While the former act as anchoring sites for Ni species the latter serve as CO2 chemisorption sites producing intermediate species which in turn react with deposited carbon to form CO. This distribution together with its improved textural properties lead to the deposition of highly dispersed, efficient and coke resistant Ni species. Keywords: Hydroxyapatite, Ca/P molar ratio, Ni dispersion, acid/base properties, methane dry reforming. 1. Introduction The choice of an appropriate technology for the production of synthesis gas (H2 and CO) from natural gas reforming is based on its end-use. In this sense, dry reforming of methane (DRM) appears to be a suitable strategy for chemicals and fuel production which require a H2/CO ratio close to unity, as in the case of Fisher-Tropsch synthesis [1-4]. Furthermore, its combination with other reforming processes, such as steam reforming of methane (SRM), allows the adjustment of the resulting H2/CO ratio when specific values are needed [2,3]. Likewise, for environmental concerns, DRM is an attractive strategy, fulfilling one of the constant challenges, because it allows the use of greenhouse gases (CH4 and CO2) as reactants [1-4]. Nickel-based catalysts are widely used for DRM reaction owing to their activity and competitive cost compared to noble metals [1-5]. Despite their high activity they suffer from a rapid deactivation, due to carbon deposition, which is considered a major drawback [1-10]. Hence, the catalysts for methane dry reforming should exhibit adequate properties to endure the severe conditions of the reaction which is generally carried out at high temperatures (> 600 ºC). For this purpose, numerous catalytic formulations were investigated in order to improve the sintering-resistance of the Ni particles and to reduce the coke formation [1, 4-10]. The nature of the used support seems to play a crucial role in controlling the distribution and then the stability of the Ni active phases. In this sense, it would be essential to design Ni catalysts that exhibit suitable metal-support interaction in order to provide highly dispersed and resistant active phases. It is also well-known that the surface catalyst chemistry is a determinant factor in orienting the process towards either the desirable products or undesirable side reactions, such as methane cracking (CH4 2 C + CO2) [4-7]. However, a suitable distribution of acid and basic sites is still a matter of controversy. For instance, it was reported that F addition to Ni/Al2O3 catalysts induces the formation of Lewis acidic species which stabilise the metallic Ni particles from high-temperature sintering by enhancing metal-support interaction and reduces coke deposition under DRM conditions [4]. By contrast, Bang et al. [5] linked the excellent coke resistance and stability of Ni/P-Al2O3 catalysts in DRM reaction to their weak acidity. Likewise, several studies reported that increasing the basic character of the exposed surface minimises the possibility of carbon formation [3,6,9,10]. For instance, Zhang et al. [6] found an improvement of the activity and stability of their Ni/Al2O3 catalysts by introducing small amounts of La as promoter. They attributed the observed behaviour to the increased number of medium-strength basic sites. Since the pioneer work by Boukha et al. [1] which demonstrated the promising behaviour of hydroxyapatite (Ca10(PO4)6(OH)2 : HAP) supported Ni catalysts in the DRM reaction, a number of studies dealing with the use of hydroxyapatite as a DRM catalyst support have been available in the literature [11-17]. Commercial and/or synthesised materials with a composition close to that of stoichiometric hydroxyapatite (Ca/P = 1.67) are by far the most investigated for DRM application. These investigations specially concern the effect of the nature of the active phase, metal loading, promoter effect, preparation method and the experimental conditions of the reaction, among others. However, no special attention has been devoted to the effect of the stoichiometry of the HAP support on the catalytic performance of the supported active phases in DRM. According to many reports, varying the Ca/P molar ratio leads to the formation of additional phases and induces significant changes in the textural and acid-base properties of the hydroxyapatite [18,19]. Silvester et al. [18] reported that as a result of structural rearrangements Ca-deficient HAP (Ca/P < 1.67) tends to decompose to stoichiometric HAP and Ca3(PO4)2 whereas for a Ca/P ratio higher than 1.67 the material should decompose to stoichiometric apatite and CaO. Tsuchida et al. [19] have shown that the highly selective synthesis of 1-butanol and 1,3-btadiene from ethanol could be merely They found that product selectivity bore a strong correlation with the acid and basic properties tuned by varying Ca/P ratio of the HAP catalysts. The present study is dealing with the investigation on the influence of the Ca/P molar ratio on the textural, structural, surface chemistry (acid/base) and catalytic behaviour in DRM of a series of hydroxyapatite supported nickel samples (Ni/HAP). For this purpose, hydroxyapatite supports with different Ca/P molar ratios have been synthesised. Interesting conclusions will be presented by correlating the very rich information provided by a wide battery of techniques (including BET, XRD, TEM, FTIR, NH3-TPD, CO2-TPD and TPO) and the catalytic performance of the investigated samples. To the best of our knowledge the influence of the composition, by varying the Ca/P ratio, on the catalytic properties of Ni/HAP in DRM has not been yet investigated. 2. Experimental Four HAP samples presenting different stoichiometry were synthesized by precipitation method, adding drop wise a calcium nitrate solution (MERCK) to a solution of (NH4)2HPO4 (SIGMA-ALDRICH). In order to obtain different stoichiometries the pH of the precipitation was adjusted, by adding ammonia solution, at 8.6, 9.3, 10 and 11. The mixture was, then, stirred at 80 °C for 16 h. After filtration, the recovered solid was washed with distilled water, dried overnight at 120 °C and finally calcined in air at 750 °C for 4 h. The prepared samples were named HAP-S, HAP-D1, HAP-D2 and HAP- -deficient samples -enriched sample (Ca/P > 1.67). Four Ni/HAP samples were prepared by incipient wetness impregnation of the synthesised HAP supports, described above, from aqueous solutions containing equal amount of Ni acetate tetrahydrate (SIGMA-ALDRICH, 98%) solution. The impregnated samples were dried overnight at 120 ºC and, then, calcined at 750 ºC for 4 h (with a ramp of 5 ºC min-1). The reduction of the prepared samples was carried out under flowing 5%H2/Ar (60 cm3 min-1) at 750 ºC for 2 h with a ramp of 5 ºC min-1. The N2 physisorption experiments were performed on an automatic device (Micromeritics, model TRISTAR II 3020). Prior analysis the samples were evacuated at 300 °C under nitrogen flow for 12 h. Fourier transform infrared spectroscopy (FTIR) spectra for the prepared samples were recorded with a Cary 600 Series FTIR apparatus using thin disks of samples (about 3 wt.%) diluted in KBr. The structural properties of the bare supports and Ni-modified samples were investigated by X-ray diffraction (XRD). The XRD experiments were conducted on a -MPD X- = 1.5406 Å) and a X-ray tube operated at 40 kV and 40 mA. The samples were scanned between 10° and 100° at a scan rate of 0.13º s-1. The weight fractions of crystalline phases in the as calcined supports were determined using a Rietveld program by trial and error method. This analysis concerned only the samples showing, besides the hydroxyapatite structure, the presence of CaO as additional phase . The full profile refinement of LaB6 was used for the instrumental calibration. For the estimation of the average metallic Ni particle size, by Scherrer equation, the samples were scanned between 43º and 46º with a relatively slow rate (0.0074º s-1) in order to improve the precision and the quality of the most intense peak (111) at 44.7º (± 0.1). The morphology and the Ni particle size distributions, for the samples reduced at 750 ºC, were investigated by Transmission Electron microscopy (TEM). The corresponding observations were performed using a JEOL 2010F microscope, working at 200 kV, with a structural resolution of 0.19 nm. Likewise, high-angle annular dark-field scanningtransmission (HAADF-STEM) technique was used for the observations of the spent catalysts. The Ni dispersion values were also estimated by H2 chemisorption at 35 ºC using a Micromeritics ASAP 2020 instrument. Prior to analysis, the samples were reduced insitu flowing 5%H2/Ar (60 cm3 min-1) at 750 ºC for 2 h. Thereafter, the samples were purged under high vacuum at 750 ºC for 1 h and cooled to 35 ºC (during 3h). The H2 adsorption measurements (at 35 ºC) consisted of an isotherm in the pressure range of 0200 torr. The Ni dispersion was estimated assuming a unity adsorption stoichiometry (H/Nis= 1). Temperature-programmed reduction (H2-TPR) experiments were conducted on a Micromeritics AutoChem 2920 instrument equipped with a TCD detector and coupled to a MKS Cirrus LM99 mass spectrometer. The samples were pre-treated in a flow of 5%O2/He at 750 °C for 30 min and then cooled down to 50 ºC under He flow. Thereafter, the samples were reduced under a 5%H2/Ar flow (50 cm3 min-1) raising the temperature from 50 ºC to 750 ºC with a ramp of 10 ºC min-1. Temperature-programmed oxidation (TPO) experiments were carried out in order to determine the amounts and the nature of the coke deposited over the post-reaction samples. The latter were heated up to 900 ºC at 10 ºC min-1, in a 5%O2/He flow (60 cm3 min-1). The CO2 signal (m/z = 44) was monitored online with a quadrupole mass spectrometer (Pfeiffer, model Thermostar GSD301T1). The temperature programmed desorption of CO2 (CO2-TPD) experiments were conducted on a Micromeritics AutoChem 2920 instrument coupled to a mass spectrometer (MKS Cirrus LM99). The pre-treatment of the samples consisted of their reduction at 750 ºC (2 h) in a flow 5%H2/He and, then, their cooling to 50 °C in a flow of He. The adsorption of CO2 was carried out under a flow of 10%CO2/He (50 cm3 min1) for 30 min. Thereafter, the samples were treated with He (at 50 ºC) for 2 h and heated up to 900 °C with a ramp of 10 ºC min-1. The acid properties of the samples were investigated by temperature programmed desorption of NH3 (NH3-TPD). The samples were submitted to the same pre-treatment used for CO2-TPD. The adsorption of NH3 was performed at 50 ºC under a 10%NH3/He flow (50 cm3 min-1) for 30 min. Then, they were treated with He for 2 h and heated up to 750 °C (10 ºC min-1). The NH3 signal (m/z = 15) was followed online by a MKS Cirrus LM99 mass spectrometer. The methane dry reforming experiments were performed in a quartz fixed-bed reactor operating at atmospheric pressure. The catalysts (50 mg) gently mixed (using a spatula in an agate mortar) with silicon carbide (100 mg) were pre-reduced under a 5%H2/Ar flow (60 cm3 min-1) at 750 ºC for 2 h; then, evacuated under flowing He (60 cm3 min-1) for 1 h. The reaction mixture was composed of 50%CH4 and 50%CO2 with a total flow of 50 cm3 min-1. The experiments were carried out at 750 ºC for 24 h. The catalysts were also tested, under more severe conditions, at relatively lower temperature (600 ºC) and more prolonged time on stream (65 h). The reactants (CH4 and CO2) and the products (H2 and CO) of the reaction were analysed by using a gas chromatograph analyser (Bruker 450-GC). The CH4 and CO2 conversions were calculated as follows: (1) (2) The thermodynamic data were estimated by SimSci PRO II software by using the Gibbs Energy Minimization Method. 3. Results and discussion 3.1. Characterisation of the Ni/HAP samples 3.1.1. Textural characterisation studies The N2 adsorption/desorption studies on the bare supports and Ni/HAP samples show that their isotherms and hysteresis loops (H3) are similar (Fig. S1 and Fig. S2, respectively). According to many authors, the observed isotherms are characteristics of materials consisting of aggregated particles and/or exhibiting slit-shaped pores [20-21]. Fig. 1 displays the pore size distribution for all reduced samples. The traces for the HAP-D1 and Ni/HAP-D1 samples show an asymmetric and wide distribution peaked at 53 nm. By increasing the Ca/P molar ratio from 1.57 to 1.62 the distribution becomes narrower and the observed peak significantly shifts towards a lower value (33 nm). An almost similar distribution could be observed for the stoichiometric samples (HAP-S and Ni/HAP-S). The HAP-E support (Ca/P = 1.73) presents a relatively wider distribution, but after addition of Ni (Ni/HAP-E), the resulting distribution appears to be quite similar to that of the Ni/HAP-S and Ni/HAP-D2 samples. Table 1 resumes the experimental data extracted from the analysis of the resulting isotherms. The specific surface areas of Ca-deficient and stoichiometric supports (HAPD1, HAP-D2 and HAP-S) are found to be ranged between 25-31 m2 g-1. The Caenriched support (HAP-E), however, shows the lowest SBET value which does not Ni/HAP-D1, Ni/HAP-S and Ni/HAP-E catalysts and the particle size distribution traces for all analysed samples, respectively. The samples do not show significant differences in morphology (Fig. 5 and Fig. S5). In all cases spherical particles of metallic nickel are observed. However, as shown in Fig. S6, the distribution of the Ni particle sizes seems to depend on the used support. The Ca-deficient samples (Ni/HAP-D1 and Ni/HAP-D2) present almost similar shape consisting of a relatively narrow distribution peaked around 11-15 nm. Over the stoichiometric sample (Ni/HAP-S) the sizes of the detected Ni population shift towards higher values presenting a peak at 17-19 nm. By contrast, the distribution over the Ca-enriched sample (Ni/HAP-E) is significantly much broader (ranged between 15 and 55 nm) and appears to be heterogeneous compared to the other samples. Table 2 lists the average size and the dispersion of Ni particles. The Ni/HAPD2 exhibits the smallest size (17 nm) followed by Ni/HAP-D1 (18.5 nm). Moreover, increasing the Ca/P ratio to reach values close to 1.67 (Ni/HAP-S) and 1.73 (Ni/HAP-E) would imply a significant increase in the size of Ni particles (22 and 34.3 nm, respectively). It is worth outlining that, though they differ slightly from each other, the general trend of these observed values is fully consistent with the XRD results, discussed above. 3.1.5. H2 chemisorption study Table 2 also reports the Ni dispersion and the Ni particle size for all reduced catalysts as determined by H2 chemisorption. The objective of these experiments is to determine quantitatively the capacity of the surface Ni active phases to chemisorb H2 molecule. As can be deduced from Table 2, the general trend of Ni particle size values matches very well with those estimated by TEM and XRD techniques. However, the dispersion data indicate that the values determined by H2 chemisorption are in all cases lower than those given by XRD and TEM. This suggests that a fraction of the deposited Ni particles is not accessible to the gas phase. For this reason, in the following, the reactivity of the sites associated to Ni species will be correlated to the metallic surface area estimated by H2 chemisorption. 3.1.6. Acid-base properties of the Ni/HAP catalysts 3.1.6.1. Temperature programmed desorption of NH3 (NH3-TPD) The thermal stability of pre-adsorbed NH3 on the acid sites of the reduced HAP and Ni/HAP catalysts was studied by means of NH3-TPD techniques (Fig. 6). Table 3 and Table S2 summarise the total amounts and the distribution of the acid sites, respectively, calculated by integration of their respective desorption peaks. Data corresponding to the bare supports are also included in order to be used as reference. As shown in Fig. 6, the traces of all analysed supports consist of two desorption domains, suggesting the presence of at least two adsorption sites with different natures. The first peak centred around 300-320 ºC is ascribed to medium-strength acid sites on the surface of the solids whereas the second feature (T > 600 ºC) could be assigned to strong acid sites. The latter was also reported by Lonyi et al. [34] in their study on the adsorption of ammonia over zeolite materials. They attributed its occurrence to a desorption of NH3 from strong Lewis acid sites. Interestingly, the Ca-deficient samples (HAP-D1 and HAP-D2) show similar surface density of acid sites (0.53-0.54 mol m-2) (Table 3) and exhibit almost identical distribution, being composed of 66-68% of medium-strength acid sites (0.35-0.37 mol m-2) and 32-34% of strong acid sites (0.17-0.18 mol m-2) (Table S2). Furthermore, the effect of increasing the Ca/P molar ratio from 1.62 (HAP-D2) to 1.67 (HAP-S) would mainly consist of an increase in the medium-strength acid sites fraction (from 66% to 80%, respectively) at the expense of the strong acid sites (from 34% to 20%, respectively). As expected, a significant decrease in the overall density of adsorption sites is observed for the Ca-enriched sample (HAP-E) which does not exceed 0.46 mol m-2. Silvester et al. [18] attributed the acidic properties of hydroxyapatites to surface HPO42and OHvacancies mainly occurring on Ca-deficient samples, denoted as Ca10z(HPO4)z(PO4)6-z(OH)2-z (0 < z 1) [23]. Moreover, they linked the observed decrease in the number of acid sites with the Ca/P increase to the formation of CaO species on the hydroxyapatite surface. Nevertheless, when compared to the others supports, the HAP-E sample exhibits relatively the highest density of strong sites (0.22 mol m-2) which represent 48% of the acid sites (Table S2). The impregnation of nickel onto the studied supports influences significantly the amounts and the distribution of the acid sites (Fig. 6, Table 3 and Table S2). For instance, after deposition of nickel onto HAP-D1 support the peak due to mediumstrength acid sites tends to split into two peaks, having maxima at 275 ºC and 380 ºC, which implies the occurrence of new weak adsorption sites. Similar observation could be noticed in the case of the other Ni modified samples. However, the position of the two resulting peaks seems to depend on the Ca/P molar ratio. The low temperature peak, for example, shifts from 275 ºC for the Ni/HAP-D1 to 260 ºC for Ni/HAP-D2, to 220 ºC for Ni/HAP-S and to 192 ºC for Ni/HAP-E. On the other hand, as can be deduced from Table 3, among all investigated samples the Ni/HAP-D1 bears the largest amounts and the highest surface density of acid sites (17.6 molNH3 g-1 and 0.73 molNH3 m-2, respectively). Moreover, depending on the nature of the support, the effect of Ni addition on the amounts and the distribution of acid sites seem to proceed via different pathways (Table S2). This would mainly consist of an increase in the surface density of acid sites for Ni/HAP-D1 (from 0.54 to 0.73 molNH3 m-2) and Ni/HAP-E (from 0.46 to 0.69 molNH3 m-2). By contrast, the values corresponding to Ni/HAP-D2 (0.53 molNH3 m-2) and Ni/HAP-S (0.62 molNH3 m-2) reveal that the impregnation of Ni does not affect the surface acid sites density measured on their supports (HAP-D2 and Ni/HAP-S, respectively). It is worth mentioning that, irrespective of the Ni-modified sample an increase in the density of strong acid sites could be observed (compared with their bare supports) which becomes almost constant (0.24-0.26 molNH3 m-2). This implies that these strong sites are associated with the effective predominance of surface Ni species. In other words, we claim that these strong acid sites might act as anchoring sites for stabilising Ni metals, as reported in previous work [4]. In this since, a linear relationship between the metallic surface area, determined by H2 chemisorption, and the number of strong acid sites could be observed (Fig. S7). 3.1.6.2. Temperature programmed desorption of CO2 (CO2-TPD) The surface basicity of the reduced HAP and Ni/HAP samples has been characterised by means of CO2-TPD techniques. Fig. 7 displays the corresponding desorption profiles after adsorption of CO2 at 50 ºC. Four desorption regions associated with different strengths can be observed on the CO2-TPD diagrams (Fig. 7). The samples show desorption peaks, located at low temperatures (< 300 ºC), and ascribed to weak basic sites. However, the peaks associated with medium-strength basic sites, occurring in the second temperature range (300-550 ºC), can only be observed for stoichiometric (HAPS and Ni/HAP-S) and Ca-enriched (HAP-E and Ni/HAP-E) samples. Thus, the occurrence of such sites could be due to the formation of CaO species on the hydroxyapatite surface [18]. This is in good agreement with our FTIR and XRD data, discussed before, where a contribution of CaO phase inherent to the excess Ca has been detected in the reduced HAP-E and HAP-S samples. At high temperatures (T > 550 ºC), irrespective of the examined sample, the CO2 desorption process gives two well resolved peaks. The first one (550-800 ºC) is probably due to strong surface basic. According to previous reports these sites are principally consisting of surface OHspecies which strongly interact with CO2 molecules [18,30]. The second feature (T > 800 ºC) could be ascribed to the decomposition of structural carbonate rather than to surface species. In order to evaluate this hypothesis a blank TPD experiment was carried out on the HAP-E bare support (Fig. S8). In this case the sample was submitted to the same pre-treatment applied for the CO2-TPD experiment (reduction at 750 ºC for 2 h), but it was not exposed to CO2 adsorption. As expected, the corresponding TPD diagram shows that no desorption peak could be observed at temperatures lower than 800 ºC. However, it evidences the desorption of significant amounts of CO2, retained probably in the hydroxyapatite structure, at high temperatures (> 800 ºC). As stated previously, these carbonate species are located in the hydroxyapatite framework by substitution of PO43ions. Table 3 reports the amounts and the distribution of CO2 desorption sites, classified according to their strength/nature, determined by integration of the peaks centred in the four temperature regions (50-300 ºC, 300-550 ºC, 550-800 ºC and > 800 ºC). For comparison, data corresponding to the bare supports are also included. As expected, the overall density of desorption sites increases with the Ca/P molar ratio increase (2.3 molCO2 m-2, for Ni/HAP-D1, 2.9 molCO2 m-2, for Ni/HAP-D2, 4.6 molCO2 m-2, for Ni/HAP-S, and 11.2 molCO2 m-2, for Ni/HAP-E). Likewise, in line with previous findings [18,19], the density of the medium-strength basic sites, strong basic sites and structural carbonate increase systematically as the Ca/P molar ratio is increased. This tendency is explained by a structural evolution where the hydroxyapatite surface is progressively enriched by CaO, OHand structural carbonate, respectively. In order to separate the concept of this distribution, in the following, the basic sites occurring on the HAP phase surface will be called HAP basic sites whereas those related to the CaO phase will be named CaO basic sites. 3.2. Catalytic activity in DRM The in-situ reduced Ni/HAP catalysts were assayed in the DRM reaction at 750 ºC for 24 h. Fig. 8 (a, b and c) shows CH4 and CO2 conversions and the resulting H2/CO molar ratio, respectively. Table 4 reports the catalytic activity data at 24 h and the coke deposition amounts (wt.%), estimated by TPO. In addition, in order to be used as reference Ni catalysts data, extracted from the literature, are also reported. As shown in Fig. 8a the Ni/HAP-D2 sample exhibits the highest initial conversion of methane (80%) followed by NiHAP-D1 (78%). These values are significantly lowered, with the increase in the Ca/P ratio, which become ranged between 72 and 73% over the Ni/HAP-E and Ni/HAP-S, respectively. A similar trend can be observed when the comparison is made on the basis of their initial CO2 conversion values. Irrespective of the tested catalyst the latter gives values (82-89%) higher than those of CH4 conversion (Fig. 8b). Moreover, the resulting initial H2/CO values registered over all catalysts are lower than 1 (0.77-0.80) (Fig. 8c). This could be explained by the occurrence of the reverse water-gas shift reaction, where a fraction of produced H2 reacts with CO2 producing carbon monoxide and water [35-46]. On the basis of their performances the assayed catalysts follow this general trend: Ni/HAP-D2 > Ni/HAP-D1 > Ni/HAP-S > Ni/HAP-E. Then, it is clear that the activity is improved over the catalysts presenting a deficiency in calcium. Furthermore, except for NiHAP-E, the catalysts demonstrate a very good stability, in the studied time on stream. For instance, the Ni/HAP-D2 sample exhibits only 2.5% activity loss after 24 h reaction (from 80 to 78%). However, the Caenriched sample (Ni/HAP-E) suffers from deactivation to a larger extent (5.5%); over which methane conversion decreases from 72 to 68%. On the other hand, the TPO results for the post-reaction catalysts show a good resistance against coke deposition which, in all cases, does not reach 1% (Table 4). This is actually expected considering the relatively high temperature of the reaction (750 ºC) [31,36,37]. In parallel, HAADFSTEM analysis of the used samples show that there is no significant changes in the Ni particle size when compared with the freshly reduced samples. Moreover, images of Ni/HAP-D2 (Fig. S9), Ni/HAP-S (Fig. S10) and Ni/HAP-E (Fig. S11) spent catalysts show that the deposited carbon is mainly filament-shaped which explain the high resistance of the catalysts to deactivation. Nevertheless, as can be observed in Fig. 9 (corresponding to Ni/HAP-E spent sample) some analysed zones evidence the presence of other forms of carbonaceous species which encapsulate Ni particles. According to many reports, the deposition of filamentous carbon penalises at much lesser extent the activity of the Ni catalysts in DRM compared to those encapsulating the active species [13,38]. Fig. S12a and Fig. S12b displays the XRD patterns and the TPO diagrams, respectively, recorded over the spent catalysts. In contrast to their freshly reduced counterpart, the diffractograms of the tested Ni/HAP-S and NiHAP-E samples evidence the absence of the CaO phase -CPP phase becomes significantly more intense (Fig. S12a) when compared to that exhibited by the reduced samples (Fig. 4). There is a second difference worth of outlining. The used Ca-enriched sample (Ni/HAP-E) presents additional intense peak due to the formation of a graphitic carbon phase (Fig. S12a). This result is consistent with the relatively highest conversion loss (5.5%) observed for this sample and the encapsulation of Ni particles by carbonaceous species detected by HAADF. Nevertheless, the observed deactivation mainly occurs during the first hour of TOS after what the catalyst appear to be more stable (Fig. 8). This behaviour is typical for carbon diffusion in the Ni lattice which proceeds until its saturation [39,46]. As can be deduced from Table 4 methane and CO2 conversions achieved by the Ni/HAP-D2 (78% and 86%, respectively) after 24 h reaction are close to those corresponding to the thermodynamic equilibrium (83.7% and 90.3%, respectively). These results clearly outperform those shown by Ni/alumina and Ni/CZ catalysts, reported in previous works [5,35]. Likewise, it is worth outlining the superiority of our investigated home-made and free-promoter materials compared to a Ni-Co catalyst supported on a commercial hydroxyapatite support [13]. Even though, the latter was tested under less severe conditions (Table 4). Fig. 10a and Fig. 10b display the dependence of the catalytic activity, after 24 h reaction, on the metallic surface area (measured by H2 chemisorption) and the number of strong acid sites, respectively. The reported data reveal a linear increase in methane and CO2 conversion, respectively, and H2/CO ratio with increasing the specific Ni surface area and the number of strong acid sites, respectively. As discussed before, the distribution of Ni on the catalyst surface is linked to the effective abundance of the strong acid sites. This feature is markedly gathered on the Ca-deficient samples. It seems that anchoring active metallic phase on these sites induces a strong metal-support interaction and, then, a positive effect on the observed activity and stability of the catalysts. In a previous study [4] on the behaviour of F-modified Ni/Al2O3 catalysts in DRM, it was found that their activity and stability were associated to the presence of strong Lewis acid sites which stabilised the Ni particles from sintering through an enhanced metal-support interaction. It is also worth analysing the effect of the basic sites on the activity of the investigated catalysts. Fig. 10c shows a relationship between the activity and the number of surface HAP basic sites. It seems that the abundance of the latter appears to be beneficial for the Ni/HAP activity in DRM. In this sense, it is well-known that surface basicity leads to increased adsorption of CO2 producing intermediate species which in turn react with deposited carbon to form CO [39-41]. However, no clear effect, neither positive nor negative, could be associated to the CaO species, detected on the freshly reduced stoichiometric (Ni/HAP-S) and the Ca-enriched (Ni/HAP-E) samples. As revealed by CO2-TPD these species induce specially the formation of surface medium-strength basic centres. We claim that their presence, at least under our experimental conditions, does not affect the catalytic properties of the tested catalysts. Our conclusion is in line with previous studies which claimed that, despite providing an advantageous basic character, modification of Ni catalysts by CaO does not significantly influence their activity in DRM [39,42]. In order to examine the effect of Ca/P on the capacity of the catalysts to endure more severe conditions, inducing an increase in the coke formation [31,36,37], additional experiments have been carried out at relatively low temperature (600 ºC) and more prolonged time on stream (65 h). For this study the catalysts exhibiting the lowest and the highest Ca/P ratio (1.57 and 1.73, respectively) have been assayed (Fig. 11). As expected, the two catalysts exhibit lower initial CH4 and CO2 conversions compared to their respective values shown at 750 ºC. The initial methane conversion becomes about 59% over Ni/HAP-D1 whereas 40% is shown over Ni/HAP-E. Likewise, the Ni/HAPD1 maintains its superiority in terms of its initial CO2 conversion (47% vs. 33% over Ni/HAP-E). Nevertheless, the two catalysts seem to suffer from a deactivation process in similar way. Over Ni/HAP-D1 catalyst the methane conversion decreases quickly during the first seven hours from 59% to 48%. Thereafter, the deactivation rate appears to be lower where the catalyst gives a CH4 conversion close to 42% at 65 h. Table S3 resumes the obtained results on the two tested catalysts. As stated before, the extent of deactivation occurring on the two samples is almost the same. After 65 h reaction their activity decreases by 28.2±0.5 %. The TPO analyses performed on the two spent catalysts show that their resistance against coke formation are very similar (coke formation around 7-8%) (Table S3) and the shapes of their TPO traces (Fig. S13) consist mainly of a CO2 formation peak centred at 610±10 ºC. Moreover, XRD analyses evidence the formation of graphitic carbon on the two tested samples. This suggests that carbonaceous species exhibiting the same nature are deposited on the two catalysts. These results confirm that the activity of the tested catalysts might be associated to the abundance of the same active sites which suffer from a similar deactivation process. 4. Conclusions The present study sheds light on the crucial role that plays the composition of the hydroxyapatite materials when used as support for metallic Ni nanoparticles in their catalytic properties in the DRM reaction. It seems that a preparation starting from a substoichiometric composition (Ca/P < 1.67) results in suitable properties which give the highest catalytic performance compared with stoichiometric (Ca/P = 1.67) and overstoichiometric (Ca/P = 1.73) compositions, respectively. We assume that the texture, the surface chemistry and the Ni species distribution shown by the investigated materials are mainly derived from the structural properties of the support used. Indeed, the XRD data showed that after their calcinations at 750 ºC the [44 carbonaceous deposits via the dry reforming reaction using transition metal catalysts, Catal. Today 253 (2015) 155-162. [45] M.S. Aw, M. Zorko, I.G. A.Pintar, Progress in the Synthesis of Catalyst Supports: Synergistic Effects of Nanocomposites for Attaining Long-Term Stable Activity in CH4-CO2 Dry Reforming, Ind. Eng. Chem. Res. 54 (2015) 37753787. [46] D.L. Trimm, The Formation and Removal of Coke from Nickel Catalyst, Catal. Rev. 16 (1977) 155-189. Acknowledgements The financial support for this work provided by Ministerio de Economía y Competitividad (CTQ2015-73219-JIN (AEI/FEDER/UE) and MAT2017-87579-R), Gobierno Vasco (GIC IT-657-13) and the Junta de Andalucia (FQM-110 group) are gratefully acknowledged. Likewise, the technical support provided by SGIker (UPV/EHU) is gratefully acknowledged. CAPTIONS FOR TABLES AND FIGURES Table 1 Elemental analyses and textural properties data for the prepared catalysts Table 2 H2-TPR, XRD, TEM and H2 chemisorption data for the Ni/HAP catalysts Table 3 Acid and base properties of the HAP and Ni/HAP samples as determined by NH3-TPD and CO2-TPD, respectively Table 4 DRM catalytic performance (at 750 ºC) of the Ni/HAP catalysts . Comparison with Ni catalysts reported in the literature. Figure 1 Pore size distribution for the reduced HAP bare supports and Ni/HAP catalysts Figure 2 FTIR spectra for the calcined (a) HAP supports and (b) Ni/HAP catalysts Figure 3 H2-TPR diagrams for the HAP and Ni/HAP samples. Figure 4 XRD patterns for the (a) HAP supports and (b) Ni/HAP catalysts: (c) samples calcined at 750 ºC and (r) samples reduced at 750 ºC. Figure 5 TEM micrograph for the reduced Ni/HAP-D2 catalyst. Figure 6 NH3-TPD profiles for the reduced HAP bare supports and Ni/HAP catalysts Figure 7 CO2-TPD profiles for the reduced HAP bare supports and Ni/HAP catalysts. Figure 8 DRM activity over Ni/HAP catalysts at 750 ºC: (a) CH4 conversion, (b) CO2 conversion and (c) H2/CO ratio. Figure 9 HAADF-STEM images of the Ni/HAP-E spent catalyst (DRM at 750 ºC) evidencing the encapsulation of Ni particles . Images (d), (e) and (f) display the colour maps for Ni (red) and C (green). Figure 10 Dependence of DRM activity on (a) metallic surface area, (b) number of strong acid sites and (c) number of basic sites excluding those due to the CaO phase. Figure 11 DRM activity over Ni/HAP-D1 and Ni/HAP-E catalysts at 600 ºC: (a) CH4 conversion, (b) CO2 conversion and (c) H2/CO ratio. Sample pH(a) (±0.1) ICP (b) BET ( c ) Ca/P (±0.01) Ni, wt.% (±0.03) SBET, m 2 g - 1 Vp, cm 3 g - 1 dp, nm HAP-D1 8.6 1.57 0 25 0.28 49 HAP-D2 9.3 1.62 0 31 0.29 32 HAP-S 10 1.67 0 25 0.22 32 HAP-E 11 1.73 0 11 0.10 29 Ni/HAP-D1 - 1.57 3.41 24 0.28 44 Ni/HAP-D2 - 1.62 3.41 26 0.25 36 Ni/HAP-S - 1.67 3.41 17 0.16 32 Ni/HAP-E - 1.73 3.41 9 0.08 32 (a) pH of precipitation during the supports synthesis. (b) Elemental analysis. (c) Data corresponding to the samples reduced at 750 ºC Table 1 Sample H2-TPR XRD TEM H2 chemisorption H2/Ni , % , % , % HAP, nm Niº, nm Disp, % Niº, nm Disp, % Niº, nm Disp, % Ni/HAP-D1 1.06 63.4 28.2 8.3 41 (40) 18 6.9 18.5 6.7 22.5 5.5 Ni/HAP-D2 1.04 49.4 41.7 8.9 34 (29) 15.2 8.2 17 7.3 18.3 6.8 Ni/HAP-S 1.05 70.4 29.6 0 44 (48) 33.1 3.8 22 5.7 35.5 3.5 Ni/HAP-E 1.03 61.7 38.3 0 51 (49) 36 3.5 34.3 3.6 42.8 2.9 Data in brackets correspond to the bare supports. Table 2 NH3-TPD CO2-TPD Sample Total amount of acid sites, molNH3 g-1 Surface density of acid sites, molNH3 m-2 Total amount of desorbed CO2, molCO2 g-1 Surface density of CO2 desorption sites, molCO2 m-2 Distribution of CO 2 desorption sites, mol CO2 m-2 Weak basic sites(a) Medium - strength basic sites(b) Strong basic sites(c) Structural carbonate(d) HAP - D1 13.4 0.54 44.5 1.8 0.59 0 0.57 0.62 HAP - D2 16.5 0.53 92 3.0 0.78 0 1.14 1.04 HAP - S 15 0.60 102.4 4.1 0.67 0.28 1.71 1.43 HAP - E 5.1 0.46 97 8.8 1.15 1.42 2.94 3.31 Ni/HAP - D1 17.6 0.73 54.6 2.3 1.20 0 0.47 0.58 Ni/HAP - D2 13.7 0.53 75 2.9 1.15 0 0.86 0.86 Ni/HAP - S 10.6 0.62 78.6 4.6 0.92 0.68 1.41 1.61 Ni/HAP - E 6.2 0.69 100.8 11.2 1.12 2.87 2.53 4.69 (a) Determined by integration of the CO2-TPD peak centred (b) Determined by integration of the CO2- (c) Determined by integration of the CO2- (d) Determined by integration of the CO2-TPD peak centred at T > 800 ºC. Table 3 Catalyst XCH4(a) XCO2(a) H2/CO Reaction mixture CH 4 /CO 2 /N 2 GHSV, cm3 h-1 g-1 TOS, h Coke, % Reference Ni/HAP-D1 75 83 0.78 50/50/0 60,000 24 0.6 This work Ni/HAP-D2 78 86 0.79 50/50/0 60,000 24 0.8 Ni/HAP-S 70 80 0.75 50/50/0 60,000 24 0.7 Ni/HAP-E 67 78 0.72 50/50/0 60,000 24 0.9 Ni/P-Al2O3 63 61 0.86 40/40/20 60,000 20 8 [5] Ni/CYSZ 66 78 0.70 50/50/0 60,000 24 7 [34] Ni-Co/HAP 73 79 n.d. 20/20/60 15,882 24 n.d. [13] Ni-Mg-Al-La 65 75 n.d. 50/50/0 24,000 10 n.d. [42] NiCo/CeZrO2 78 84 0.84 50/50/0 12,000 80 0.3 [43] Ni-Co/SiC-CeZr 74 80 0.75 50/50/0 12,000 23 0.5 [44] (a) The margin of error is around 1%. Table 4 Figure 1 0 20 40 60 80 100 120 140 160 0.00 0.05 0.10 0.15 0.20 0.25 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 20 40 60 80 100 120 140 160 0.0 0.2 0.4 0.6 0.8 1.0 0 20 40 60 80 100 120 140 160 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Pore width, nm Ni/HAP-E HAP-E Ni/HAP-D1 HAP-D1 Ni/HAP-D2 HAP-D2 Pore width, nm Ni/HAP-S HAP-S Figure 2 Figure 3 Figure 4 Figure 11 0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 90 0 1 2 3 4 5 6 7 20 40 60 80 (a) Ni/HAP-D1 Ni/HAP-E TOS, h 0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 90 0 1 2 3 4 5 6 7 20 40 60 80 (b) TOS, h 0 10 20 30 40 50 60 0.0 0.2 0.4 0.6 0.8 (c) TOS, h