Metal Oxide Nanoparticles Supported on Macro-Mesoporous Aluminosilicates for Catalytic Steam Gasiﬁcation of Heavy Oil Fractions for On-Site Upgrading

: Catalytic steam gasiﬁcation of extra-heavy oil (EHO) fractions was studied using functionalized aluminosilicates, with NiO, MoO 3 , and/or CoO nanoparticles with the aim of evaluating the synergistic effect between active phase and the support in heavy oil on-site upgrading. Catalysts were characterized by chemical composition through X-ray Fluorescence, surface area, and pore size distribution through N 2 adsorption/desorption, catalyst acidity by temperature programmed desorption (TPD), and metal dispersion by pulse H 2 chemisorption. Batch adsorption experiments and catalytic steam gasiﬁcation of adsorbed heavy fractions was carried out by thermogravimetric analysis and were performed with heavy oil model solutions of asphaltenes and resins (R–A) in toluene. Effective activation energy estimation was used to determine the catalytic effect of the catalyst in steam gasiﬁcation of Colombian EHO. Additionally, R–A decomposition under inert atmosphere was conducted for the evaluation of oil components reactions with active phases and steam atmosphere. The presence of a bimetallic active phase Inc.reases the decomposition of the heavy compounds at low temperature by an increase in the aliphatic chains decomposition and the dissociation of heteroatoms bonds. Also, coke formation after steam gasiﬁcation process is reduced by the application of the bimetallic catalyst yielding a conversion greater than 93%. the heavy fractions. Also, available acid sites are responsible for the adsorptive and catalytic capacity of the materials. Metakaolin support has a suitable structure for heavy molecules adsorption due to the presence of larger pores in comparison with clay structure. Although the clay sample presents a higher total acidity, the degree of heavy molecules diffusion through the clay structure is a function of the average size of the aggregates. In this sense, metakaolin could have more available sites for adsorbent–adsorbate interactions, and this is corroborated by K and H estimated parameters of the SLE model. R–A adsorption is enhanced by a heterogeneous adsorbent with different selectivies towards the presence of metal oxide nanoparticles on support surface. An improvement in the catalytic behavior could be done with the functionalization process, where the presence of a bimetallic active phase increases the decomposition of the heavy compounds at lower temperatures, due to an enhancement of the dissociation of heteroatoms bonds and the aliphatic chains and aromatic ring decomposition. Also, coke formation after steam gasiﬁcation process is reduced by the application of the bimetallic catalyst yielding a conversion greater than 93% in the following order KMoNi > KCoMo > CMoNi > CMoCo > metakaolin > clay. This study opens a wider landscape about the use of nanoparticle-functionalized catalysts for the production and transport of heavy in extra-heavy crude oils.


Introduction
With the depletion of conventional sources of crude oil, heavy oil (HO) and extra heavy oil (EHO) became an important alternative for the oil and gas industry [1][2][3]. Alberta in Canada, the Orinoco belt between Colombia and Venezuela, as well as China, Russia, and Mexico are regions with the most important heavy oil resources [4,5]. In Colombia, the production of HO and EHO represents about 60% of the current oil production, and the key reserves are located in the Eastern Plains and the Middle Magdalena Valley areas with 9 and 1.7 billion barrels of estimated reserves respectively [6][7][8]. Due to the severe declining of light and medium oil reserves in Colombia, it is expected that the production of HO and EHO will constitute more than 90% of the Colombian production [6].
reported. The adsorption behavior of the crude oil heavy compounds was described by the solid-liquid equilibrium (SLE) model and was correlated with the catalytic activity of the synthesized materials. This work provides potential applications of nanoparticle technology for heavy oil processing and production, which could be a viable alternate green technology.

Catalyst Characterization
The chemical composition of fresh clay and metakaolin supports obtained from X-ray Fluorescence (XFR) is shown in Table 1. Compounds as SiO 2 and Al 2 O 3 are the main oxides presents in the supports structure, with a minor content of Fe 2 O 3 and TiO 2 . Other oxides such, CaO, MgO, Na 2 O and BaO are present with contents <1 wt %. More than 80% of the content in both materials are SiO 2 and Al 2 O 3 , with a similar SiO 2 composition but with a difference of more than 10% of the Al 2 O 3 content between both supports. In addition, from Table 1 it can be observed that the SiO 2 /Al 2 O 3 ratios are 2.4 and 1.5 for Clay and Metakaolin, respectively. Surface area (S BET ) pore size distribution (PSD) of the materials were obtained through the N 2 adsorption/desorption experiments, the results are listed in Table 2. The PSD of the virgin supports showed range of sizes corresponding to mesopores in the structure of the support for values between 20 to 100 Å [77]. For metakaolin, a bimodal behavior is observed with peaks at 39 and 27 Å. However, for clay a single peak is observed at 23 Å. Table 2. Specific surface area, total acidity, mean crystallite size and metal dispersion for functionalized and non-functionalized clay and metakaolin.

Sample
S BET ± 0.1 m 2 /g Total Acidity ± 0.02 µmol/g Mean Particle Size of Active Phase ± 0.  41 13. 5 15 The TPD profile obtained with the adsorption/desorption of NH 3 molecules in the samples gives information related to the type of acidity present in the sample based on the desorption peaks [78]. For aluminosilicates, the Al-OH and Si-OH functional groups are responsible for the NH 3 adsorption [79,80]. TPD-NH 3 profiles of ammonia desorption for clay, metakaolin, and their functionalization are shown in Figure 1 and results of total acidity are listed in Table 2. From Figure 1 two predominant peaks at 590 K and 845 K for fresh clay can be observed due to the presence of Lewis and Brønsted sites [79,81]. The low-temperature peak observed in a TPD profile is related to the Lewis sites. Furthermore, the high-temperature peak is associated with the Brønsted sites, because the energy required to promote the NH 3 desorption is higher [80]. In the case of metakaolin, the thermal process in the material preparation could contribute to the collapse of the layers in its structure, hindering the diffusion of the NH 3 in the sample. Hence, it can be said that clay structure presents a high contribution of the acidity for the acid sites inside the material through an easier NH 3 diffusion. It can be observed a reduction of the total acidity for the clay-supported catalysts after the metal oxide impregnation, and this could be due to the pores blockage by the metal oxide formation, decreasing the NH 3 diffusion and adsorption through the clay structure. Also, when comparing Ni-containing materials, higher acidity values are observed and is in agreement with Peña et al. [82] who evaluated the supported metal oxide (Mn, Cu, Cr, Co, Fe, V, and Ni) in TiO 2 for catalytic reduction of NO with NH 3 and observed that TiO 2 -Ni catalyst offers a higher acidity in comparison with the Co supported catalyst.
Catalysts 2017, 7, 319 5 of 24 the supported metal oxide (Mn, Cu, Cr, Co, Fe, V, and Ni) in TiO2 for catalytic reduction of NO with NH3 and observed that TiO2-Ni catalyst offers a higher acidity in comparison with the Co supported catalyst. Pulse H2 chemisorption measured the covered metal surface after functionalization process to identify the formation of metal oxide nanoparticles troughs the H2 consumption. Results of the metal oxide mean particle size for the studied samples is shown in Table 2. For this metal combinations, the alloy formation is expected between the different evaluated active phases [83][84][85], and the results of mean particle size and the dispersion degree are based on the cluster size and dispersion degree of the alloy formed. The obtained results confirm the formation of metal oxide nanoparticles over the support surface. It can be observed from Table 2 that mean particle size for compounds synthesized over the clay surface have an average value of 4.5 nm, while for the KMoNi and KMoCo samples the mean particles sizes are higher with values of 8.9 and 13.5 nm, respectively. Hence, it can be said that the formation of the oxide nanoparticles over the material surface may be influenced by the porous structure of the material. Dispersion of the active material over the support surface was about 10% for clay and between 15 and 20% for metakaolin.

Adsorption Isotherms of Resins and Asphaltenes
Due to the high molecular weight of the heavy fractions compounds, the diffusion of these molecules through the pores in a porous material is low [43]. For this reason, textural properties such the volume and pore diameter are more important in the heavy fraction adsorption than the specific surface area of the material [52]. Furthermore, adsorption isotherms of R-A are shown in Figure 2a,b for (a) clay-supported and (b) metakaolin-supported catalysts together with the SLE model fit. In accordance with the International Union of Pure and Applied Chemistry (IUPAC), the obtained adsorption isotherms follow a Type Ib behavior, which means a strong interaction between the available active zones in the adsorbent and the R-A mixture [77,86,87]. However, the asphaltenesresins interactions and a lower amount of functional groups, as well as low polarity in resins structure, could affect the interactions between the adsorbent and adsorbate [69,88,89]. The low asphaltenes concentration and the colloidal state of the adsorbate affects the asphaltene-asphaltene interactions, indicating that the adsorption process could be governed by the interaction of the adsorbent with the asphaltene-surrounding resins [25,89]. A heterogeneous adsorbent with multiple selectivities for heavy fraction would be more efficient attracting the heavy compounds to the catalyst surface. In this sense, the MoCo catalysts show the highest adsorptive capacity of the heavy fraction for both clay and metakaolin supports. The selectivity towards the active phase generates an Pulse H 2 chemisorption measured the covered metal surface after functionalization process to identify the formation of metal oxide nanoparticles troughs the H 2 consumption. Results of the metal oxide mean particle size for the studied samples is shown in Table 2. For this metal combinations, the alloy formation is expected between the different evaluated active phases [83][84][85], and the results of mean particle size and the dispersion degree are based on the cluster size and dispersion degree of the alloy formed. The obtained results confirm the formation of metal oxide nanoparticles over the support surface. It can be observed from Table 2 that mean particle size for compounds synthesized over the clay surface have an average value of 4.5 nm, while for the KMoNi and KMoCo samples the mean particles sizes are higher with values of 8.9 and 13.5 nm, respectively. Hence, it can be said that the formation of the oxide nanoparticles over the material surface may be influenced by the porous structure of the material. Dispersion of the active material over the support surface was about 10% for clay and between 15% and 20% for metakaolin.

Adsorption Isotherms of Resins and Asphaltenes
Due to the high molecular weight of the heavy fractions compounds, the diffusion of these molecules through the pores in a porous material is low [43]. For this reason, textural properties such the volume and pore diameter are more important in the heavy fraction adsorption than the specific surface area of the material [52]. Furthermore, adsorption isotherms of R-A are shown in Figure 2a,b for (a) clay-supported and (b) metakaolin-supported catalysts together with the SLE model fit. In accordance with the International Union of Pure and Applied Chemistry (IUPAC), the obtained adsorption isotherms follow a Type Ib behavior, which means a strong interaction between the available active zones in the adsorbent and the R-A mixture [77,86,87]. However, the asphaltenes-resins interactions and a lower amount of functional groups, as well as low polarity in resins structure, could affect the interactions between the adsorbent and adsorbate [69,88,89]. The low asphaltenes concentration and the colloidal state of the adsorbate affects the asphaltene-asphaltene interactions, indicating that the adsorption process could be governed by the interaction of the adsorbent with the asphaltene-surrounding resins [25,89]. A heterogeneous adsorbent with multiple selectivities for heavy fraction would be more efficient attracting the heavy compounds to the catalyst surface. In this sense, the MoCo catalysts show the highest adsorptive capacity of the heavy fraction for both clay and metakaolin supports. The selectivity towards the active phase generates an improvement in the self-association (i.e., K parameter of the SLE model) of the heavy oil fraction over the catalyst surface and follows the trend KMoCo > KNiMo > CMoCo > KNiMo > Metakaolin > Clay. A suitable catalyst for the heavy fraction adsorption by longer pores and higher surface acidity results in a better adsorptive behavior. In addition, a higher acidity in the active sites generates an improvement in the affinity between the adsorbate and the adsorbent, showing a reduction in the H parameter. For a better comprehension of the adsorption process, the experimental data was described by the SLE model [90]. The values of the obtained model parameters and their associated R 2 and Root Mean Square Error (RSME%) values are shown in Table 3. improvement in the self-association (i.e., parameter of the SLE model) of the heavy oil fraction over the catalyst surface and follows the trend KMoCo > KNiMo > CMoCo > KNiMo > Metakaolin > Clay. A suitable catalyst for the heavy fraction adsorption by longer pores and higher surface acidity results in a better adsorptive behavior. In addition, a higher acidity in the active sites generates an improvement in the affinity between the adsorbate and the adsorbent, showing a reduction in the parameter. For a better comprehension of the adsorption process, the experimental data was described by the SLE model [90]. The values of the obtained model parameters and their associated R 2 and Root Mean Square Error (RSME%) values are shown in Table 3. For KNiMo material, a higher total acidity could lead a strong adsorption process, due to the relation between the adsorption affinity (i.e., lower) and the total acidity reported in Table 2. Franco et al. [68] and Nassar et al. [69] showed that relationship exists between the surface acidity and the adsorption capacity of asphaltenes. Attraction forces between the O-and N-containing functional groups of the heavy fraction with active sites available for the adsorption process is expected to be an important adsorption mechanism [91,92]. The high clay acidity is associated with a better diffusion of NH3 molecules through the pores of the particle due to their low particle diameter [93]. However, the available acid sites for heavy fraction adsorption is less, due to the aggregates For KNiMo material, a higher total acidity could lead a strong adsorption process, due to the relation between the adsorption affinity (i.e., H lower) and the total acidity reported in Table 2. Franco et al. [68] and Nassar et al. [69] showed that relationship exists between the surface acidity and the adsorption capacity of asphaltenes. Attraction forces between the O-and N-containing functional groups of the heavy fraction with active sites available for the adsorption process is expected to be an important adsorption mechanism [91,92]. The high clay acidity is associated with a better diffusion of NH 3 molecules through the pores of the particle due to their low particle diameter [93]. However, the available acid sites for heavy fraction adsorption is less, due to the aggregates formed in the R-A systems that decrease the diffusion capacity through the porous materials. It has been reported that the addition of resins to asphaltene solutions changes the self-aggregation of the asphaltenes [94]. The stabilization effect of resins has been attributed to form smaller structures preventing the flocculation of the asphaltenes particles [95,96]. Spiecker et al. [97] found that with an R-A ratio of about 10:1 in toluene solutions, R-A aggregates correlated a length between 11-14 Å. This indicates that the active acid sites where the adsorption could happen are less than the total acid sites of the particle. In this sense, a surface impregnation with the metals leads to active sites that are more available for R-A adsorption.

Mass Loss Analysis
The catalytic steam gasification tests of the heavy fraction in the presence and absence of the catalysts were performed in an N 2 atmosphere saturated with H 2 O (g) . Figure 3 shows the plots of rate of mass loss for catalytic steam gasification in the presence and absence of (a) clay and (b) metakaolin-supported materials. Two characteristic zones can be observed for the rate of mass loss for the R-A mixture, and are associated with the aliphatic chains and aromatic rings decomposition in both heavy fractions [25]. From the asphaltenes and resins structure, it is supposed that R-A blend presents a higher aliphatic chains proportion than the aromatic rings, because of a low asphaltenes content. For lower temperatures (before 700 K), alkyl chains break up and dissociation of S-C and N-C bonds of the surrounded aliphatic chains in the micelle structure is expected [25,98,99]. R-A decomposition begins at approximately 470 K in the absence of the materials and is reduced about 100 K in the presence of the catalysts associated with an enhancement of the functionalized supports in the aliphatic chains decomposition for both active phases [7]. Hence, metakaolin surface contribution generates a synergetic effect in the R-A steam gasification in comparison with the results observed for functionalized clay. The main decomposition peak for R-A mixture is observed at 773 K, which is attributed to aromatics rings decomposition [13]. However, in the presence of the synthesized metakaolin-supported catalysts, this temperature can be reduced significantly, probably related to the chemical content (Al 2 O 3 and SiO 2 ) and the pore size distribution that could determine the interaction degree of the adsorbed species and the adsorbate. Metakaolin bimetallic catalysts reduce the R-A main peak decomposition more than 100 K, with a clear contribution of both active phases. KMoNi profile present 3 main decompositions peaks instead of the KMoCo series, which presents 2 decompositions peaks. The main decomposition peak for R-A in presence of KMoNi is observed at 695 K and 707 K for KMoCo. The magnitude of the mass changes as a function of temperature at 500 K is higher for KMoNi than KMoCo catalyst, suggesting a higher rupture of short and long alkyl groups for KMoNi sample [7]. It could be inferred that peak observed at 780 K would be due to the presence of MoO 3 nanoparticles in the decomposition test of the MoNi and MoCo series, nevertheless, a higher magnitude in KMoNi decomposition profile propose a better synergistic effect between Mo-Ni than Mo-Co in the polycyclic aromatics hydrocarbons cracking [72]. For this reason, the behavior differences between the magnitudes of the decomposition profile of the adsorbed R-A species in MoNi-and MoCo-containing materials could be associated such a higher catalytic activity of the MoNi couple. NiO nanoparticles present in the MoNi series enhanced the catalytic behavior of a Mo bimetallic catalyst for heavy fractions decomposition as a result of a high acidity in comparison with MoCo series.  The R-A decomposition in the presence and absence of KMoNi sample under an inert atmosphere is shown in Figure 4. The KMoNi catalyst under inert atmosphere enhances the R-A cracking at lower temperatures by the presence of active sites which catalyzes the chains rupture [100]. Virgin R-A decomposition begins at 545 K and 485 K for pyrolysis and steam gasification process respectively, which suggests that the steam atmosphere improve the short and long chain rupture by the saturation and partial oxidation of the molecules [101]. For steam gasification, water catalytic dissociation products are used for organic free radicals saturation and oxidation/reforming reactions [59]. At lower temperatures, a higher profile magnitude for steam gasification of adsorbed R-A onto KMoNi, which suggest the formation of free radical saturations and reforming reactions with the short and longs chains of the surrounding aromatics rings in R-A species. Differences between the KMoNi profiles in steam and inert atmosphere in the high-temperature zone suggest that NiO nanoparticles in presence of steam minimize the condensation of free radicals by the addition of hydrogen in the radicals structures [102]. Deriv.Weight (%/K) Deriv.Weight (%/K) The R-A decomposition in the presence and absence of KMoNi sample under an inert atmosphere is shown in Figure 4. The KMoNi catalyst under inert atmosphere enhances the R-A cracking at lower temperatures by the presence of active sites which catalyzes the chains rupture [100]. Virgin R-A decomposition begins at 545 K and 485 K for pyrolysis and steam gasification process respectively, which suggests that the steam atmosphere improve the short and long chain rupture by the saturation and partial oxidation of the molecules [101]. For steam gasification, water catalytic dissociation products are used for organic free radicals saturation and oxidation/reforming reactions [59]. At lower temperatures, a higher profile magnitude for steam gasification of adsorbed R-A onto KMoNi, which suggest the formation of free radical saturations and reforming reactions with the short and longs chains of the surrounding aromatics rings in R-A species. Differences between the KMoNi profiles in steam and inert atmosphere in the high-temperature zone suggest that NiO nanoparticles in presence of steam minimize the condensation of free radicals by the addition of hydrogen in the radicals structures [102]. After the steam gasification process, the coke formation was evaluated by heating the samples in an oxidative atmosphere at a flow of 100 mL/min and a heating rate of 20 K/min. Conversion degree of the R-A mixture in the absence and presence of the catalysts is shown in Figure 5 for (a) clay and (b) metakaolin-supported materials. Heavy oil fractions conversion was enhanced by the presence of metal oxide nanoparticles as active phase of the materials, remaining the following coke trend for R-A steam gasification: R-A (18.6%) > Clay (20.4%) > Metakaolin (13.8%) > CMoCo (5.6%) > CMoNi (5.3%) > KMoCo (3.4%) > KMoNi (2.1%). The results suggest that the evaluated bimetallic catalyst promote a low coke yield, enhance the conversion of the heavy compounds at low temperatures and reduce the feasibility of deactivation by the coke presence on the catalyst surface. After the steam gasification process, the coke formation was evaluated by heating the samples in an oxidative atmosphere at a flow of 100 mL/min and a heating rate of 20 K/min. Conversion degree of the R-A mixture in the absence and presence of the catalysts is shown in Figure 5 for (a) clay and (b) metakaolin-supported materials. Heavy oil fractions conversion was enhanced by the presence of metal oxide nanoparticles as active phase of the materials, remaining the following coke trend for R-A steam gasification: R-A (18.6%) > Clay (20.4%) > Metakaolin (13.8%) > CMoCo (5.6%) > CMoNi (5.3%) > KMoCo (3.4%) > KMoNi (2.1%). The results suggest that the evaluated bimetallic catalyst promote a low coke yield, enhance the conversion of the heavy compounds at low temperatures and reduce the feasibility of deactivation by the coke presence on the catalyst surface. After the steam gasification process, the coke formation was evaluated by heating the samples in an oxidative atmosphere at a flow of 100 mL/min and a heating rate of 20 K/min. Conversion degree of the R-A mixture in the absence and presence of the catalysts is shown in Figure 5 for (a) clay and (b) metakaolin-supported materials. Heavy oil fractions conversion was enhanced by the presence of metal oxide nanoparticles as active phase of the materials, remaining the following coke trend for R-A steam gasification: R-A (18.6%) > Clay (20.4%) > Metakaolin (13.8%) > CMoCo (5.6%) > CMoNi (5.3%) > KMoCo (3.4%) > KMoNi (2.1%). The results suggest that the evaluated bimetallic catalyst promote a low coke yield, enhance the conversion of the heavy compounds at low temperatures and reduce the feasibility of deactivation by the coke presence on the catalyst surface.   Figure 6 shows the coke yield for virgin and adsorbed R-A on KMoNi surface under steam and inert atmospheres. Steam injection as a source of hydrogen allows the coke yield reduction for virgin and adsorbed R-A by the saturation and reforming of free radicals obtained fom thermal cracking [59]. In this sense, for inert atmosphere the coke yield of virgin and adsorbed R-A is 25.8% and 6.19%. Nevertheless, coke yield under steam atmosphere for virgin and adsorbed R-A is 20.4% and 2.01%.  Figure 6 shows the coke yield for virgin and adsorbed R-A on KMoNi surface under steam and inert atmospheres. Steam injection as a source of hydrogen allows the coke yield reduction for virgin and adsorbed R-A by the saturation and reforming of free radicals obtained fom thermal cracking [59]. In this sense, for inert atmosphere the coke yield of virgin and adsorbed R-A is 25.8% and 6.19%. Nevertheless, coke yield under steam atmosphere for virgin and adsorbed R-A is 20.4% and 2.01%.   Figure 6 shows the coke yield for virgin and adsorbed R-A on KMoNi surface under steam and inert atmospheres. Steam injection as a source of hydrogen allows the coke yield reduction for virgin and adsorbed R-A by the saturation and reforming of free radicals obtained fom thermal cracking [59]. In this sense, for inert atmosphere the coke yield of virgin and adsorbed R-A is 25.8% and 6.19%. Nevertheless, coke yield under steam atmosphere for virgin and adsorbed R-A is 20.4% and 2.01%.

Differential Scanning Calorimetry (DSC) Analysis
Differential scanning calorimetry (DSC) experiments under air atmosphere were carried out for the adsorbed R-A on the bimetallic catalyst surface and represented in Figure 7, with the aim to identify the reactions zones in the R-A decomposition. The catalytic behavior of supported clay catalyst is denoted by an exothermic reaction for temperatures below to 700 K, indicating the enhancement of the aliphatic chains decomposition. Two characteristics peaks on the heat flow are observed during the decomposition of adsorbed species in the presence of the catalysts with metakaolin as support, with similar results as the profile mass loss, representative to a high aliphatic and aromatic rings decomposition.
behavior of supported clay catalyst is denoted by an exothermic reaction for temperatures below to 700 K, indicating the enhancement of the aliphatic chains decomposition. Two characteristics peaks on the heat flow are observed during the decomposition of adsorbed species in the presence of the catalysts with metakaolin as support, with similar results as the profile mass loss, representative to a high aliphatic and aromatic rings decomposition.

Analysis of the Evolution of the Gaseous Product during the Steam Gasification Process
The selectivity behavior during the catalytic process was studied by tracing the produced gasses using FTIR analysis [72,100,103]. Steam gasification of R-A fractions as carbonaceous source contains reactions such as water-gas shift reaction, partial oxidation, steam reforming reactions, boundary equilibrium reaction and methanation [72]. Different gasses could be involved in R-A decomposition, namely CO, CH4, O2, CO2 and H2. Nevertheless, H2 and O2 were eliminated for the present study, because of water signal interference [7]. Similarly, CO2 production was low in comparison with that of CH4. The CH4 production during the R-A gasification in the presence and absence of nanomaterials are shown in Figure 8. Because the CH4 band intensity for metakaolin bimetallic catalysts was always higher than the intensity reported in the others systems, the results were normalized based on the CH4 produced by the presence of KMoNi. The CH4 production by the steam gasification of the adsorbed heavy fraction compounds in the virgin supports does not show an increase in light

Analysis of the Evolution of the Gaseous Product during the Steam Gasification Process
The selectivity behavior during the catalytic process was studied by tracing the produced gasses using FTIR analysis [72,100,103]. Steam gasification of R-A fractions as carbonaceous source contains reactions such as water-gas shift reaction, partial oxidation, steam reforming reactions, boundary equilibrium reaction and methanation [72]. Different gasses could be involved in R-A decomposition, namely CO, CH 4 , O 2 , CO 2 and H 2 . Nevertheless, H 2 and O 2 were eliminated for the present study, because of water signal interference [7]. Similarly, CO 2 production was low in comparison with that of CH 4 . The CH 4 production during the R-A gasification in the presence and absence of nanomaterials are shown in Figure 8. Because the CH 4 band intensity for metakaolin bimetallic catalysts was always higher than the intensity reported in the others systems, the results were normalized based on the CH 4 produced by the presence of KMoNi. The CH 4 production by the steam gasification of the adsorbed heavy fraction compounds in the virgin supports does not show an increase in light hydrocarbons conversion, suggesting a low selectivity of the evaluated aluminosilicates in the methanation reaction [104]. The CH 4 production is lower for clay support in comparison with the metakaolin material, probably due to a suitable structure that endorses the interaction between the active sites and the heavy fractions. However, supports functionalization process enhanced the CH 4 production by the presence of different active sites, confirming the catalytic activity of the functionalized catalysts. Metal oxide nanoparticles guarantee a high CH 4 production at lower temperatures [100], also promote the R-A conversion into light products with the following trend KMoCo < KMoNi < CMoCo < CMoNi < K < R-A < C. Partial oxidation, water-gas and steam reforming are associated with CO production, however, the evolution profile does not show a representative CO production in the evaluated catalyst, suggesting the promotion of the methanation reaction. hydrocarbons conversion, suggesting a low selectivity of the evaluated aluminosilicates in the methanation reaction [104]. The CH4 production is lower for clay support in comparison with the metakaolin material, probably due to a suitable structure that endorses the interaction between the active sites and the heavy fractions. However, supports functionalization process enhanced the CH4 production by the presence of different active sites, confirming the catalytic activity of the functionalized catalysts. Metal oxide nanoparticles guarantee a high CH4 production at lower temperatures [100], also promote the R-A conversion into light products with the following trend KMoCo < KMoNi < CMoCo < CMoNi < K< R-A< C. Partial oxidation, water-gas and steam reforming are associated with CO production, however, the evolution profile does not show a representative CO production in the evaluated catalyst, suggesting the promotion of the methanation reaction.  Figure 9 show the CH4 and LHC production profiles under inert and steam atmosphere and panel (c) in Figure 9 displays the CO2 profiles under both atmospheres. Under inert atmosphere R-A is converted mainly to CO2, light hydrocarbons, and CH4 with a less extent. Also, oxygen sources present on catalyst structure and R-A functional groups are probably used in CO2 production during the thermal cracking by the reaction of free carbon radicals and the oxygen sources [13]. From Figure  9a,b it is observed that steam injection has hydrogen source contribute to a less condensation of light hydrocarbons by the cracking of LCH and saturation of free radicals [59], which means a higher yield in methane production under steam atmosphere. Carbon oxides profiles from Figure 9 shows a higher CO2 production under inert atmosphere instead of the CO2 production under steam atmosphere, suggesting that steam atmosphere improve the methanation of carbon oxides produced in reactions such partial oxidation, steam reforming and water-gas shift reactions [59,72].  Figure 9 show the CH 4 and LHC production profiles under inert and steam atmosphere and panel (c) in Figure 9 displays the CO 2 profiles under both atmospheres. Under inert atmosphere R-A is converted mainly to CO 2 , light hydrocarbons, and CH 4 with a less extent. Also, oxygen sources present on catalyst structure and R-A functional groups are probably used in CO 2 production during the thermal cracking by the reaction of free carbon radicals and the oxygen sources [13]. From Figure 9a,b it is observed that steam injection has hydrogen source contribute to a less condensation of light hydrocarbons by the cracking of LCH and saturation of free radicals [59], which means a higher yield in methane production under steam atmosphere. Carbon oxides profiles from Figure 9 shows a higher CO 2 production under inert atmosphere instead of the CO 2 production under steam atmosphere, suggesting that steam atmosphere improve the methanation of carbon oxides produced in reactions such partial oxidation, steam reforming and water-gas shift reactions [59,72].

Effective Activation Energies
Effective activation energy (Eα) was calculated using isoconversional OFW method by thermogravimetric measurements at three different heating rates between 10 and 30 K/min to validate the catalytic activity of the evaluated catalysts. Figure 10 shows the estimated effective activation energies as a function of the conversion degree. R-A effective activation energies decreased as the conversion increase, while the Eα value for adsorbed R-A has the opposite trend. In virgin supports, for a conversion degree lower than 40%, the required energy is lowered in comparison with R-A compounds. The interaction between small aggregates adsorbed onto support surface is the first step in steam gasification of adsorbed species, requiring low amounts of energy. However, the decomposition of bigger aggregates and new compounds from addition reactions generates an increase in the activation energy required, suggesting the weakness of the active sites on the support surface in the gasification of heavy fractions compounds. In this sense, metakaolin supports exhibit a decrease in activation energy in all the conversion range in comparison with clay support, probably due to more available active sites from a higher pore size distribution and greater dispersion of the active phase over the support surface. Functionalized catalysts show reductions greater than 20% in the effective activation energy estimation for all the samples. Differences between the activation energy of MoNi-and MoCo-containing materials are associated with a higher affinity (lower H) between the R-A compounds and the MoNi-containing materials observed during the adsorption process, which lead a lower activation energy than MoCo-containing materials. Effective activation energy (Eα) was calculated using isoconversional OFW method by thermogravimetric measurements at three different heating rates between 10 and 30 K/min to validate the catalytic activity of the evaluated catalysts. Figure 10 shows the estimated effective activation energies as a function of the conversion degree. R-A effective activation energies decreased as the conversion increase, while the Eα value for adsorbed R-A has the opposite trend. In virgin supports, for a conversion degree lower than 40%, the required energy is lowered in comparison with R-A compounds. The interaction between small aggregates adsorbed onto support surface is the first step in steam gasification of adsorbed species, requiring low amounts of energy. However, the decomposition of bigger aggregates and new compounds from addition reactions generates an increase in the activation energy required, suggesting the weakness of the active sites on the support surface in the gasification of heavy fractions compounds. In this sense, metakaolin supports exhibit a decrease in activation energy in all the conversion range in comparison with clay support, probably due to more available active sites from a higher pore size distribution and greater dispersion of the active phase over the support surface. Functionalized catalysts show reductions greater than 20% in the effective activation energy estimation for all the samples. Differences between the activation energy of MoNi-and MoCo-containing materials are associated with a higher affinity (lower H) between the R-A compounds and the MoNi-containing materials observed during the adsorption process, which lead a lower activation energy than MoCo-containing materials.  Figure 11 displays the estimated effective activation energy for virgin and adsorbed R-A onto KMoNi for steam and inert atmospheres. The activation energy required for virgin R-A decomposition at α = 60% is reduced 18% by the steam injection. Furthermore, the activation energy for adsorbed R-A fraction onto KMoNi surface at steam atmosphere is 10% less than the inert atmosphere. Steam injection as hydrogen source promote hydrocarbon steam gasification reactions, which reduce the coke formation, the estimated activation energy by the OFW method and enhance the production of valuable gases as CH4.   Figure 11 displays the estimated effective activation energy for virgin and adsorbed R-A onto KMoNi for steam and inert atmospheres. The activation energy required for virgin R-A decomposition at α = 60% is reduced 18% by the steam injection. Furthermore, the activation energy for adsorbed R-A fraction onto KMoNi surface at steam atmosphere is 10% less than the inert atmosphere. Steam injection as hydrogen source promote hydrocarbon steam gasification reactions, which reduce the coke formation, the estimated activation energy by the OFW method and enhance the production of valuable gases as CH 4 . The comparison between adsorptive and catalytic behavior is presented in Figure 12a-c. Figure  12a suggest a relationship between parameter for SLE model with the activation energy required for the decomposition of the heavy fractions, whereas a suitable support structure contributes to the reduction of the energy required after the functionalization process. Hence, it can be said that there is a synergistic effect between the active phases and the supports in the catalytic steam gasification of the adsorbed compounds. The presence of active phases with different selectivities improves the R-A aggregation at the primary active site, but also contribute in a heavy fraction cracking by the improvement in the catalyst affinity as observed in Figure 12b. For this reason, evaluated bimetallic catalyst exhibits and improvement in the aggregate adsorption with available active sites for reaction and subsequently heavy fractions decomposition into lighter products. The comparison between adsorptive and catalytic behavior is presented in Figure 12a-c. Figure 12a suggest a relationship between H parameter for SLE model with the activation energy required for the decomposition of the heavy fractions, whereas a suitable support structure contributes to the reduction of the energy required after the functionalization process. Hence, it can be said that there is a synergistic effect between the active phases and the supports in the catalytic steam gasification of the adsorbed compounds. The presence of active phases with different selectivities improves the R-A aggregation at the primary active site, but also contribute in a heavy fraction cracking by the improvement in the catalyst affinity as observed in Figure 12b. For this reason, evaluated bimetallic catalyst exhibits and improvement in the aggregate adsorption with available active sites for reaction and subsequently heavy fractions decomposition into lighter products. The comparison between adsorptive and catalytic behavior is presented in Figure 12a-c. Figure  12a suggest a relationship between parameter for SLE model with the activation energy required for the decomposition of the heavy fractions, whereas a suitable support structure contributes to the reduction of the energy required after the functionalization process. Hence, it can be said that there is a synergistic effect between the active phases and the supports in the catalytic steam gasification of the adsorbed compounds. The presence of active phases with different selectivities improves the R-A aggregation at the primary active site, but also contribute in a heavy fraction cracking by the improvement in the catalyst affinity as observed in Figure 12b. For this reason, evaluated bimetallic catalyst exhibits and improvement in the aggregate adsorption with available active sites for reaction and subsequently heavy fractions decomposition into lighter products.

Materials and Chemical
Kaolin and clay microparticles with 10 μm and 2 μm in mean particle size, respectively, were provided by Suministros de Colombia (Sumicol S.A.S.) and were used as supports of the active phase. The kaolin was first calcinated at 1073 K yielding a Metakaolin support. Salt precursors, Ni(NO3)2·(6H2O) (Merck KGaA, Darmstadt, Germany), (NH4)6Mo7O24 (Merck KGaA, Darmstadt Germany) and CoCl2 (J.T Baker, USA), as well as distilled water, were used for functionalization of the supports [105,106]. Toluene (99.5%, Merk KGaA, Darmstadt, Germany) was used to prepare heavy oil model solutions. An extra heavy crude oil, produced from a reservoir in the center of Colombia was the source of the heavy oil compounds used in this work. This crude oil has 6.4° API (with a specific gravity of 0.9745), the viscosity of 6 × 10 6 at 298 K and contents of saturates, aromatics, resins, and asphaltenes of 19.2, 16.2, 52.0 and 12.6 wt %, respectively. Also, almost the 51 wt % of the crude has normal boiling point greater than 573 K. For resins and asphaltenes isolation, the extraheavy crude oil was mixed with n-heptane (99%, Sigma-Aldrich, St. Louis, MO, USA) in a ratio of 40 mL of n-heptane per 1 g of crude oil. The solution was sonicated for 2 h and then stirred for 20 h at 200 rpm. Later, the mixture was centrifuged for 45 min at 4500 rpm and asphaltenes were recovered by filtration [107]. The deasphalted oil (DAO) obtained from asphaltenes extraction process was mixed with chromatographic silica (Sigma-Aldrich, St. Louis, MO, USA) in a 1:1 mass ratio and further stirred at 300 rpm for 24 h for separating the resins. The silica is filtered from the solution and then added to a Chloroform (Sigma-Aldrich, St. Louis, MO, USA) solution in a ratio of 10 mL per gram of crude for the purpose to generate the resins desorption [108,109]. The solution was stirred for 24 h at 200 rpm and then centrifuged for 45 min at 4500 rpm. Finally, the supernatant was evaporated and the resins were obtained.

Catalyst Preparation
The supports were first dried at 393 K for 6 h to eliminate any humidity, and impregnated using an aqueous solution based on the desired precursor salts. The salt precursors concentration in the aqueous solution used for the impregnation was achieved at a metal oxide content from 2.5 wt % for each metal on the catalyst surface. For this, Mo-Ni and Mo-Co combinations were employed for clay and metakaolin functionalization. The solid was calcined at 450 °C for 1 h, yielding an oxide catalyst [110]. The supported catalysts obtained in this study are denoted by the initial letter of the support

Materials and Chemical
Kaolin and clay microparticles with 10 µm and 2 µm in mean particle size, respectively, were provided by Suministros de Colombia (Sumicol S.A.S.) and were used as supports of the active phase. The kaolin was first calcinated at 1073 K yielding a Metakaolin support. Salt precursors, Ni(NO 3 ) 2 ·(6H 2 O) (Merck KGaA, Darmstadt, Germany), (NH 4 ) 6 Mo 7 O 24 (Merck KGaA, Darmstadt, Germany) and CoCl 2 (J.T Baker, USA), as well as distilled water, were used for functionalization of the supports [105,106]. Toluene (99.5%, Merk KGaA, Darmstadt, Germany) was used to prepare heavy oil model solutions. An extra heavy crude oil, produced from a reservoir in the center of Colombia was the source of the heavy oil compounds used in this work. This crude oil has 6.4 • API (with a specific gravity of 0.9745), the viscosity of 6 × 10 6 at 298 K and contents of saturates, aromatics, resins, and asphaltenes of 19.2, 16.2, 52.0 and 12.6 wt %, respectively. Also, almost the 51 wt % of the crude has normal boiling point greater than 573 K. For resins and asphaltenes isolation, the extra-heavy crude oil was mixed with n-heptane (99%, Sigma-Aldrich, St. Louis, MO, USA) in a ratio of 40 mL of n-heptane per 1 g of crude oil. The solution was sonicated for 2 h and then stirred for 20 h at 200 rpm. Later, the mixture was centrifuged for 45 min at 4500 rpm and asphaltenes were recovered by filtration [107]. The deasphalted oil (DAO) obtained from asphaltenes extraction process was mixed with chromatographic silica (Sigma-Aldrich, St. Louis, MO, USA) in a 1:1 mass ratio and further stirred at 300 rpm for 24 h for separating the resins. The silica is filtered from the solution and then added to a Chloroform (Sigma-Aldrich, St. Louis, MO, USA) solution in a ratio of 10 mL per gram of crude for the purpose to generate the resins desorption [108,109]. The solution was stirred for 24 h at 200 rpm and then centrifuged for 45 min at 4500 rpm. Finally, the supernatant was evaporated and the resins were obtained.

Catalyst Preparation
The supports were first dried at 393 K for 6 h to eliminate any humidity, and impregnated using an aqueous solution based on the desired precursor salts. The salt precursors concentration in the aqueous solution used for the impregnation was achieved at a metal oxide content from 2.5 wt % for each metal on the catalyst surface. For this, Mo-Ni and Mo-Co combinations were employed for clay and metakaolin functionalization. The solid was calcined at 450 • C for 1 h, yielding an oxide catalyst [110]. The supported catalysts obtained in this study are denoted by the initial letter of the support and the symbols of the cations of the resulting metal oxides after calcination. For instance, the material CNiMo is a catalyst impregnated with 2.5 wt % of Ni and 2.5 wt % of Mo supported in clay and the material KMoCo is a catalyst impregnated with 2.5 wt % of Mo and 2.5 wt % of Co supported in metakaolin.

Catalyst Characterization
The virgin supports were analyzed using X-ray Fluorescence (XFR) analysis with a Rigaku spectrometer, equipped with a graphite monochromator using CuKα radiation (40 kV) and angular domain scanned 6 • ≤ 2θ ≤ 70 • . N 2 adsorption and desorption experiments at 77 K were conducted in an Autosorb-1 from Quantachrome after outgassing samples overnight at 413 K. The surface area of the materials was calculated using the Brunauer-Emmett-Teller (BET) method [111], and the pore size distribution curves were obtained from desorption isotherms by the Barret-Joyner-Halenda (BJH) method [112]. Temperature programmed desorption with NH 3 (TPD-NH 3 ) experiments were performed to measure the total acidity of the prepared catalyst with the aim to correlate the surface acidity with the catalytic behavior in the decomposition tests. Approximately 100 mg of each prepared catalyst was dried at 473 K for 1 h under He flow in a U-shaped quartz tube. Then, a fixed amount of 10 vol % NH 3 in He at 80 mL/min was introduced to be adsorbed at 373 K for 1 h. The carrying gas was changed again to pure He at 80 mL/min at 373 K for 1 h and finally heated up to 1143 K at 10 K/min [72]. A calibrated Thermal conductivity detector (TCD) with a constant flow of He at 80 mL/min senses the changes in the flow of the gas through the instrument, due to the desorption of the NH 3 in the heating ramp [80,113]. The instrument used to TPD-NH 3 experiments was a Chembet 3000 (Quantachrome Instruments, Boynton Beach, FL, USA). The determination of metal dispersion and average metal particles sizes in supported catalyst were performed by pulse chemisorption using H 2 titration with the Chembet 3000 (Quantachrome Instruments, Boynton Beach, FL, USA). About 100 mg of the samples were placed in a U-shaped quartz tube dried at 473 K for 1 h. Then, the catalysts were reduced at 973 K for 1 h in 10 vol % H 2 in Ar at 80 mL/min and purged with flowing Ar for 1 h until the samples reach atmosphere temperature (298 K). Hydrogen pulses continued until no additional uptake of H 2 was observed. Due to the interactions between the metals on the support surface, the formation of alloys generates different adsorption behaviors between each of the components and their mixtures. For this reason, average metal particles sizes and metal dispersion are related to the bimetallic cluster formed in the support surface [83][84][85].

Batch Adsorption Experiments
According to the SARA heavy fraction content (asphaltenes and resins) in the extra heavy oil, the asphaltenes and resins have a mass ratio corresponding to 20:80 respectively in the oil matrix. Hence, batch adsorption experiments were performed at 298 K for different concentrations of a 20:80 mixture of resins and asphaltenes (R-A). The R-A mixtures were added in toluene at concentrations from 100 to 5000 mg/L and for a ratio of the solution volume to the dry mass of catalyst of 10 g/L [114]. The solutions were stirred for 24 h to ensure the maximum adsorption of the heavy compounds. Finally, the catalysts with adsorbed heavy compounds were separated from the solutions and dried to remove any traces of solvent. The amount of the heavy fraction in the adsorbed phase was measured by thermogravimetric analyses (TGA) with a Q50 analyzer (TA Instruments, Inc., New Castle, DE, USA). Thermogravimetric analyses, in this case, were performed under an air flow of 100 mL/min with a heating rate of 5 K/min from 373 to 1173 K. The differences in the mass loss between the virgin catalyst and catalyst with adsorbed species determine the total amount adsorbed [89].

Catalytic Steam Gasification of Asphaltenes and Resins
For the catalytic steam gasification of adsorbed heavy compounds over the catalyst, samples were submitted to TGA using a Q50 analyzer (TA Instruments, Inc., New Castle, DE, USA) coupled with an IRAffinity-1 FTIR device (Shimadzu, Kyoto, Japan) that is equipped with a gas cell to analyze produced gasses in the decomposition process. The sample mass in the analyzer was kept low to avoid diffusions limitations (~5 mg) [115]. Also, the adsorbed amount of heavy species was kept at 1.5 mg/m 2 . For gasification experiments, N 2 flow was fixed at 100 mL/min, and H 2 O (g) was introduced to the TGA analyzer at a flow rate of 6.30 mL/min using a gas saturator filled with distilled water in a thermostatic bath [7]. The FTIR spectrophotometer was operated in transmission mode at a resolution of 2 cm −1 with 10 scans per minute in the range of 400-4000 cm −1 . Different gasses could be detected during the cracking of heavy oil compounds, predominately CO 2 , H 2 , CO, CH 4 , and O 2 in the steam gasification process [101]. The characteristic intensity of the adsorption bands at 2149, 2349, 3016 and 2750 cm −1 were analyzed respectively for CO, CO 2 , CH 4 , and other light hydrocarbons, respectively [13]. All experiments were performed in duplicate to confirm reproducibility. For the other hand, Differential Scanning Calorimeters experiments were carried out in a DSC analyzer (Q20 TA Instruments, Inc., New Castle, DE, USA) in order to determine reactions zones associated with the R-A decomposition in presence and absence of the prepared catalysts. The airflow rate was a constant 100 cm 3 /min throughout the experiment at a rate of 10 K/min. The sample mass was the same used for catalytic steam gasification of adsorbed species to avoid diffusions limitations (~5 mg) [115].
R-A pyrolysis under inert atmosphere conditions in the absence and presence of the catalyst sample with better behavior in the decomposition peak were carried out in order to evaluate the liquid-film thickness effect in coke formation and the synergistic effect between the steam atmosphere and the active phases of catalysts. The sample mass in the analyzer was the same used in the steam gasification process, the adsorbed amount of heavy species kept at 1.5 mg/m 2 and the N 2 flow was fixed at 100 mL/min.

Solid-Liquid Equilibrium (SLE) Model
The adsorption isotherms of heavy crude oil fractions onto macroparticulate materials are described by the Solid-Liquid Equilibrium (SLE) Model based on the association theory of Talu and Meunier [90,116]. The expression of the model is given by: where H (mg/g) is Henry´s law constant, which is an indicator of the adsorption affinity between the heavy fractions and the adsorbent (related with the adsorption strength interactions). In this sense, a lower H value means a higher affinity. The parameter K (g/g) is an indicator of the degree of heavy fractions association once the primary sites are occupied; finally N ads (mg/m 2 ) is the maximum adsorption capacity. The other parameters are defined by: where SA is the specific surface area of the catalyst and ξ is defined as follows:

Estimation of the Effective Activation Energy for R-A Decomposition
Based on the isoconversional principle, the method of Ozawa-Flynn-Wall (OFW) [117,118] was used for the effective activation energy estimation with the aim of evaluating the catalytic activity of the functionalized materials in the R-A decomposition. The isoconversional methods assume that the reaction rate at a constant conversion is only a function of temperature as the following equation [119]: where E α (kJ/mol) is the effective activation energy for a constant conversion degree, K α (s −1 ) is the pre-exponential factor, T (K) is the reaction temperature, R (J/mol·K) is the ideal constant. The reaction conversion at a given temperature is expressed as: With m o , m f , m T as the initial sample mass, the final sample mass and the sample mass at a given temperature. By the reaction rate integration and defining the heating rate as β = dT dt can be obtained by the following equation: The integral of the right-hand side of this equation could be evaluated by the Doyle approximation [120], yielding: The effective activation energy is estimated as the slope of the best-fit line of log(β) against 1 T plot.

Conclusions
This study evaluates the supports and active phase nature by the employment of clay and metakaolin materials as support of MoCo and MoNi active phases in the catalytic steam cracking of a resins-asphaltenes blend (R-A). Structural properties such the pore size distribution and the total acidity of the materials impact in the adsorptive and subsequent steam catalytic cracking of the heavy species. Pore size distribution defines the aggregates that could be adsorbed and the sites that would interact with the heavy fractions. Also, available acid sites are responsible for the adsorptive and catalytic capacity of the materials. Metakaolin support has a suitable structure for heavy molecules adsorption due to the presence of larger pores in comparison with clay structure. Although the clay sample presents a higher total acidity, the degree of heavy molecules diffusion through the clay structure is a function of the average size of the aggregates. In this sense, metakaolin could have more available sites for adsorbent-adsorbate interactions, and this is corroborated by K and H estimated parameters of the SLE model. R-A adsorption is enhanced by a heterogeneous adsorbent with different selectivies towards the presence of metal oxide nanoparticles on support surface. An improvement in the catalytic behavior could be done with the functionalization process, where the presence of a bimetallic active phase increases the decomposition of the heavy compounds at lower temperatures, due to an enhancement of the dissociation of heteroatoms bonds and the aliphatic chains and aromatic ring decomposition. Also, coke formation after steam gasification process is reduced by the application of the bimetallic catalyst yielding a conversion greater than 93% in the following order KMoNi > KCoMo > CMoNi > CMoCo > metakaolin > clay. This study opens a wider landscape about the use of nanoparticle-functionalized catalysts for the production and transport of heavy in extra-heavy crude oils.