Enhancement of HDO Activity of MoP/SiO 2 Catalyst in Physical Mixture with Alumina or Zeolites

: Catalytic properties of physical mixture of MoP/SiO 2 catalyst with SiC, γ‐ Al 2 O 3 , SAPO ‐ 11 and zeolite β have been compared in hydrodeoxygenation of methyl palmitate (MP). MoP/SiO 2 catalyst (11.5 wt% of Mo, Mo/P = 1) was synthesized using TPR method and characterized with N 2 physisorption, elemental analysis, H 2 ‐ TPR, XRD and TEM. Trickle ‐ bed reactor was used for catalytic properties investigation at hydrogen pressure of 3 MPa, and 290 °C. The conversions of MP and overall oxygen ‐ containing compounds have been increased significantly (from 59 to about 100%) when γ‐ Al 2 O 3 or zeolite materials were used instead of inert SiC. MP can be converted to palmitic acid through acid ‐ catalyzed hydrolysis along with metal ‐ catalyzed hydrogenolysis, and as a consequence the addition of material possessing acid sites to MoP/SiO 2 catalyst could lead to acceleration of MP hydrodeoxygenation through acid ‐ catalyzed reactions. Isomerization and cracking of alkane were observed over the physical mixture of MoP/SiO 2 with zeolites, but the selectivity of MP conversion trough the HDO reaction route is remained on the high level exceeding 90%.


Introduction
In the last decade, the research efforts in the transportation fuels production from renewable sources have grown due to environmental issues and limitation of oil reserves [1][2][3][4]. One of the versatile and already commercial routes to produce diesel and aviation fuels is hydrodeoxygenation of triglyceride-containing feedstock, including animal fats, waste frying oils, non-edible vegetable oils and so on [5,6]. Oxygen removal from triglyceride molecules includes several alternative routes [7][8][9]: hydrodeoxygenation (HDO) or decarboxylation/decarbonylation (HDeCOx). The HDO pathway results in alkanes with the same numbers of carbon atoms in the chain and molecules of water, while HDeCOx reactions give COx molecules and alkanes with the reduced carbon chains [7]. The HDO reaction route is more attractive in terms of atom economy, greenhouse gases production and hydrogen recycle [10,11]. Thus, the current research issue is the design of the catalysts with high activity and high selectivity towards HDO pathway.
Nevertheless, the high selectivity of MoP-based catalysts in the conversion of aliphatic oxygenates through the HDO route [26] makes MoP/SiO2 the system of choice from the consideration of carbon atom economy and effluent gases composition, because carbon oxides make difficult purification of circulating hydrogen.
It was reported that acid sites can affect appreciably the HDO of aliphatic esters [26,29,[38][39][40][41][42][43][44][45][46]. The authors of [26,29,39,40] have proposed, that Brønsted acidic sites of silica-supported nickel phosphide catalysts participate in the reaction of carboxylic ester to carboxylic acid. Also, the rate of ethyl stearate HDO over Ru/TiO2 catalyst was increased when the catalyst was mixed with the grains of γ-Al2O3 [41]. The synergetic effect of Ni2P/SiO2 and -Al2O3 physical mixture was observed in methyl palmitate HDO resulting in the enhancement of methyl palmitate conversion over Ni2P/SiO2 catalyst after mixing it with alumina grains [42]. It was observed that Ni2P catalyst supported on -Al2O3 exhibits higher conversion of methyl palmitate HDO in comparison with silica-supported one, though both catalysts contain Ni2P nanoparticles with close mean particle size and nearly the same nickel content [43]. The authors [42,43] proposed that the acid sites of the alumina increase the reaction rate of methyl palmitate hydrolysis, that enhance the methyl palmitate conversion. The assumption was made earlier [44,45], that the Lewis acid sites of the support enhance the rate of hydrolysis of aliphatic esters over alumina-supported sulfide catalysts.
The above examples of activity enhancement in the presence of acidic material allowed us to suggest that the replacement of silica support with the alumina may increase the catalytic activity of MoP catalyst in methyl palmitate HDO without reducing selectivity. The synthesis of supported MoP/-Al2O3 catalysts is sophisticated task due to interaction between alumina surface and phosphate groups that makes the reduction difficult and prevents the formation of MoP nanoparticles on alumina support [47][48][49]. Up to now, the silica-supported MoP catalysts have been studied in the HDO reactions, due to relatively easy reduction of phosphate groups that makes possible the formation of highly dispersed MoP phase on silica support [26,50].
The goal of the current study is the investigation of the catalytic properties of mechanical mixtures of MoP/SiO2 catalyst and materials with different acidity (SiC, γ-Al2O3, SAPO-11 and zeolite β) in methyl palmitate hydrodeoxygenation as the representative model component of triglyceride-based feedstock.

Catalyst Characterization
Physicochemical properties of MoP/SiO2, prepared using phosphate and phosphite precursors by TPR are shown in Table 1. MoP_A was prepared using ammonium paramolybdate and ammonium phosphate, MoP_I was prepared using ammonium paramolybdate and phosphorous acid. MoP is a bulk sample, prepared for comparison and XRD analysis. Initial Mo/P molar ratio in impregnating solution was 1. This ratio is usually used to prepare MoP catalysts [26,51] and the samples with this ratio were shown to be more active in hydrotreating reactions [52]. After reduction Mo/P ratio is slightly decreased to 0.91 for MoP_A and MoP and to 0.83 for MoP_I (Table 1). This can be due to Mo losses at impregnation step. Mo/P ratio determined from EDX data differ slightly from the chemical analysis results, but still remains close to Mo/P = 1. Initial SiO2 support has SBET = 300 m 2 /g and Dpore = 10.6 nm. MoP_A and MoP_I have decreased value of SBET and Dpore in comparison with the silica support. Both active component and unreduced phosphate groups can block pores and decrease SBET. Bulk MoP has quite high area due to presence of large quantities of citric acid in the precursor, which are transformed to carbon after reduction.
The precursors and POx/SiO2 samples were investigated by H2-TPR ( Figure 1). There are two peaks for the bulk MoP precursor: at 490 and at 720 °С. The first corresponds to Mo +6 → Mo +4 reduction, the second corresponds to Mo +4 → Mo δ+ and Р +5 → Р δ-phosphide formation. Reduction of phosphates in POx/SiO2 sample occurs at higher temperatures (> 750 °C, Figure 1). H2-TPR of MoP_A and MoP_I precursors have the maximums at the temperatures similar to the reduction temperature of bulk MoP. H2-TPR shows that precursor nature does not influence the temperatures of reduction of silica-supported MoP precursors. It was shown that in the case of Ni-based phosphide catalysts during H2-TPR of phosphite precursors the additional amount of hydrogen is produced, which is detected in H2-TPR curves [40,43]. For the Mo-phosphide precursors there is only slight difference between H2-TPR curves of MoP_A and MoP_I precursors ( Figure 1). This can be attributed to oxidation of HPO3 2-by paramolybdate after impregnation:  XRD analysis of unreduced bulk MoP precursor did not show presence of any crystal phases ( Figure 2), only increased intensity at low 2θ angles, which corresponds to amorphous carbon residues. XRD pattern of reduced bulk MoP sample include well defied maximums, corresponding to MoP phase (PDF № 24-0771). This pattern also has some amorphous phase corresponding to an increased base line in a 2θ range of 35-40°. DXRD of MoP particles is 12 nm (Table 1). In supported MoP_A and MoP_I samples, there were broad signals of SiO2 at ~20°; low-intensive broadening is also observed in the range of 35-45° that can be due to highly dispersed MoP (<3 nm) nanoparticles. Intensity, a.u.   Table 1) and narrow particle size distributions. MoP sample has particles of 10-50 nm; therefore, there is no particle size distribution on Figure 3 for this sample. The diluters used for catalytic experiments in this work were SiC, γ-Al2O3, SAPO-11 and zeolite β ( Table 2). XRD analysis proved corresponding crystal structures for these materials ( Figure S1 in Supporting Information). Unlike SiC, an inert material with extremely low SBET (1 m 2 /g), γ-Al2O3, SAPO-11 and zeolite β have developed textural structures ( Table 2). The acidity of these materials, in accordance with NH3-TPD data, is decreased in the following order: zeolite β > SAPO-11 > γ-Al2O3 (Table 2).

2 ()
Zeolite β is characterized by Brønsted acid sites and three-dimensional system of channels, with sizes of 0.56  0.56 and 0.66  0.67 nm [53]. SAPO-11 has one-dimensional system of channels with size of 0.40  0.65 nm. SAPO-11 is a mesoporous structure, which has Brønsted acid sites of middle strength [54].

MP HDO Results
Catalytic properties of physical mixtures of MoP/SiO2 with SiC, γ-Al2O3, SAPO-11 or zeolite β were investigated in MP HDO at T = 290 °C, PH2 = 3 MPa, H2/feed = 600 cm 3 /cm 3 , LHSV = 60 or 10 h -1 , using 0.4 or 1.4 g of catalyst and volume ratio of Vcat:Vdiluter = 1:10.5 or 1:1.5. Stable conversion as a function of time on stream was obtained for all tested systems ( Figure S2 in Supporting Information). The results of comparative study of these systems in the methyl palmitate HDO are presented on Figure 4, which displays the values of MP conversion (XMP) and oxygen conversion (XO). MoP/SiO2 sample diluted with inert SiC grains displays the lowest XMP (59%) and XO (55%) values, whereas XMP and XO data for systems containing γ-Al2O3, zeolite β and SAPO-11 are higher than 90%. It is well known that HDO of fatty acid esters includes consecutive and parallel reactions: hydrogenolysis of C-O (HDO) and C-C (HDeCOx) bonds, hydrolysis of ester, hydrogenation of double bounds, dehydration of alcohols and esterification of acids, which occur over metal and acid active sites of catalysts, supports and diluters [26,42,55]. In particular, MP can be transformed at metal centers through hydrogenolysis to acid or aldehyde, or via acid-catalyzed hydrolysis to acid [56]. Thus, the higher conversions of MP over МоP/SiO2 catalyst in the presence of Al2O3, SAPO-11 or zeolite β can be explained by the higher rate of methyl palmitate hydrolysis, which is favored by the acid sites of these materials.  The main HDO product over MoP/SiO2-SiC and MoP/SiO2-γ-Al2O3 is hexadecane at 100% conversion of MP; the selectivity of MP conversion through the HDO pathway exceeds 96% ( Figure  5). Thus, the use of γ-Al2O3 instead of SiC in the mixture with MoP/SiO2 increases the MP conversion, but has minor impact on the product distribution. The dilution of MoP/SiO2 with the same amount of SAPO-11 or zeolite β instead of SiC decreases the selectivity of C16 alkane formation in the course of MP HDO and increases the quantity of iso-C16, iso-C15 and C6-C14 hydrocarbons among the liquid products. Especially increased formation of hydrocarbons with 14 carbon atoms in the hydrocarbon chains is observed over MoP/SiO2-zeolite β catalytic system. Apparently, Brønsted acid sites in МоP/SiO2-SAPO-11 and МоP/SiO2-zeolite β result in partial cracking and structural isomerization of linear C15 and C16 hydrocarbons, which are formed in the HDO reaction of the ester. In these cases, the secondary reactions of the alkanes that are formed in MP HDO make it unattainable to calculate selectivities of HDO and HDeCOx pathways using the results of liquid phase analysis. To estimate the effect of SAPO-11 and zeolite β on the selectivity of MP conversion the amounts of carbon oxides were compared for MoP/SiO2-SiC, МоP/SiO2-SAPO-11 and МоP/SiO2-zeolite β ( Figure 6). The results obtained let us see that the use of МоP/SiO2-SAPO-11 and МоP/SiO2-zeolite β induces only negligible increase in the amounts of carbon oxides in the gas phase in comparison with MoP/SiO2-SiC system. Appreciable amounts of light hydrocarbons with the carbon number 5 were detected in the gas-phase products of MP HDO over МоP/SiO2-zeolite β that is the evidence of alkane overcracking in the presence of zeolite β.
The above results show that the use of the MoP/SiO2 catalyst in the mixtures with acidic materials, like γ-Al2O3, SAPO-11 and zeolite  improves the catalytic activity of MoP catalyst in the hydrodeoxygenation of MP without reducing selectivity through HDO pathway. MoP/SiO2-γ-Al2O3 system gives hexadecane with the selectivity up to 96%, while МоP/SiO2-SAPO-11 produces iso-alkane and some amount of C6-C14 hydrocarbons. Therefore, MoP catalysts on SAPO-11 are perspective in one-pot HDO/hydroisomerization of aliphatic esters.
Preparation method of MoP/SiO2 samples Incipient wetness impregnation method was used to prepare precursors of MoP/SiO2 samples. Mo/P molar ratio in impregnation water solution was taken equal 1 (Mo content~11.5 wt%). The label MoP_A was chosen for the MoP/SiO2 sample, which was prepared from phosphate solution, the label MoP_I-for the MoP/SiO2 sample, which was prepared from phosphite solution.
Phosphate method (MoP_A). To synthesis samples using this technique 2.357 g of (NH4)2HPO4 (Alfa Aesar, Kandel, Germany) and 3.151 g of (NH4)6Mo7O244H2O (Reakhim, Moscow, Russia) were dissolved under stirring in distilled water. Subsequently the support (8.5 g SiO2) was impregnated by this solution. After drying on air overnight (at room temperature) the precursors were dried for 4 h at 110 °C. Calcination of the dried samples was carried out at 500 °C for 4 h, and reduction of the calcined samples (TPR) was performed in H2 flow (100 mL/(min•g)). TPR has the following program: heating to 370 °C (with the rate of 3 °C/min), heating to 650 °C (with the rate of 1 °C/min), keeping the samples at 650 °C (for 1 h). Passivation of the reduced samples (to prevent oxidation on air atmosphere) was carried out at room temperature in 1 vol% O2/He (at 40 mL/(min•g) for 2 h).
Phosphite method (MoP_I). First, silica support (8.5 g) was impregnated with aqueous solution containing 3.151 g of (NH4)6Mo7O244H2O and 1.464 g of H3PO3. Then, after drying overnight and subsequently at 80 °C (for 24 h), the samples were reduced (TPR) in H2 flow (100 mL/(min•g)). TPR was carried as follows: heating to 650 °C (with the rate of 1 °C/min), keeping at 650 (for 1 h). The same passivation technique was carried out for MoP_I samples as in case of MoP_A samples.
POx/SiO2 sample. SiO2 (5 g) was impregnated by the solution of 2.357 g of (NH4)2HPO4 in distilled water. Drying was performed at room temperature overnight and at 110 °C (for 4 h). After calcination at 500 °C (for 4 h), the precursor was reduced and passivated according the same TPR and passivation programs as for MoP_A and MoP_I precursors.
Bulk MoP reference sample preparation method. The bulk molybdenum phosphide sample (bulk MoP) was prepared as follows: to the water solution of (NH4)2HPO4 was added water solution of (NH4)6Mo7O24. The Mo/P initial molar ratio was equal to 1. After evaporation of the water, the obtained sample was dried at 87 °C (for 48 h) and then at 124 °C (for 24 h). Calcination of the sample was performed at 500 °C (for 5 h). Afterwards, TPR was carried out (at H2 flow of 150 mL/(min•g)): heating to 650 °C (with the rate of 1 °C/min), keeping at 650 °C (for 1 h). The obtained bulk sample was passivated in flow of 1 vol.% O2/He (80 mL/(min•g), for 3 h).

Catalysts Characterization
Nitrogen physisorption was used to determine textural characteristics of the MoP, MoP/SiO2 and diluters. It was carried out at 77 K on ASAP 2400 instrument (USA).
Inductively coupled plasma atomic emission spectroscopy (ICP-AES) was used to perform elemental analysis of the MoP and MoP/SiO2 samples. ICP-AES was carried out on Optima 4300 DV (Perkin Elmer, Villebon-sur-Yvette, France).
Bruker D8 Advance X-ray diffractometer (Bruker, Bremen, Germany) was used to carry out XRD analysis. CuKα radiation (with λ = 1.5418 Å) was used, and 2θ scanning range was from 10 to 70. To determine crystal phases JCPDS database was used [57]. Scherrer equation was used to calculate average crystallite size (DXRD).
Transmission electron microscope JEM-2010 (JEOL, Tokyo, Japan) was used to obtain high resolution transmission electron microscopy (TEM) images. The accelerating voltage and resolution were 200 kV and 0.14 nm correspondingly. EDX spectrometer was used to analyze elemental composition.
H2 temperature-programmed reduction (H2-TPR) was performed to determine reduction temperatures of the prepared samples. The sample mass was about 0.10 g, it was heated in 10 vol.% H2/Ar (60 mL/min, with rate of 10 °C/min) from 80 °C to 900 °C. Thermal conductivity detector (TCD) was used to determine H2 consumption.
A trickle-bed reactor was used to evaluate catalytic activity of the prepared MoP catalysts in methyl palmitate HDO ( Figure S3 in Supporting Information). The reaction conditions: T = 290 °C, PH2 = 3.0 MPa, volume ratio H2/feed = 600, LHSV = 60 or 10 h -1 . The liquid feed consisted of: 10 wt% of methyl palmitate, 0.5 wt% of n-octane (internal standard), the rest is n-dodecane. The catalysts particles were diluted with SiC, Al2O3, SAPO-11 or zeolite β (0.2-0.3 mm) in a ratio 1:10.5 or 1:1.5, placed in the catalytic reactor between. Above and under the catalytic layer there were SiC layers. Before methyl palmitate HDO experiments reduction of the catalyst precursors was performed in situ by TPR technique described earlier in catalysts preparation section. Collecting of liquid product effluent was carried out hourly until the reaction reaches steady-state. For all experiments time on stream = 8 h ( Figure S2 in Supporting Information).
XMP-methyl palmitate conversion was calculated as follows: 1 100% c o MP and cMP (mol/L)-methyl palmitate concentration in the initial feed and in the liquid product correspondingly.
XO-oxygen-containing compounds conversion was calculated as follows: c o O and cO (mol/L)-oxygen concentration in the initial feed and in the liquid product correspondingly.
Si-selectivity was calculated as follows: ci (mol/L)-concentration of the ith compound in the liquid product.

Analysis of the Reaction Products
7000 GC/MS Triple QQQ GC System 7890A (Agilent Technologies, Santa Clara, CA, USA) was used to identify HDO products. The GC was equipped with VF-5MS column (length 30 m, inner diameter 0.25 mm, film thickness 0.25 μm). Quantitative analysis was performed using GC with flame ionization detector (FID, Agilent 6890N, USA) and an HP-1MS column (length 30 m, inner diameter 0.32 mm, film thickness 1.0 μm). Carbon balance for all experiments was not lower than 97%.
Elemental analyzer Vario EL Cube (Elementar Analysensysteme GmbH, Langenselbold, Germany) was used to determine oxygen content in the liquid reaction mixture.
The concentration of gas-phase products was determined using GC Khromos-1000 with FID (Khromos, Dzerzhinsk, Russia). CO and CO2 were analyzed after methanation over Pd catalyst.

Conclusions
In this work catalytic properties of MoP/SiO2 catalyst in MP HDO were studied in mechanical mixture with materials having different acidity: SiC, γ-Al2O3, SAPO-11 and zeolite .
A method was chosen for the preparation of SiO2-supported MoP catalyst. Two samples were prepared using (NH4)6Mo7O24 and different P precursors: (NH4)2HPO4 and H3PO3 (MoP_A and MoP_I samples, correspondingly). It was shown that H2-TPR curves for MoP_A and MoP_I samples are coinciding regardless of P precursor that can be explained by the oxidation of H3PO3 in the reaction with ammonium paramolybdate during impregnation and drying of support. The silica-supported MoP catalysts samples were synthesized by reduction (TPR) of the precursors at 650 C, and according to TEM data, the particle sizes of MoP nanoparticles were 1.4 and 2.0 nm for MoP_A and MoP_I samples, respectively. Both catalysts (MoP_A and MoP_I) display similar activity in the MP hydrodeoxygenation; MoP_A sample (designated as MoP/SiO2) was chosen for the experiments with different diluters.
The comparative study of MoP/SiO2-SiC, MoP/SiO2-γ-Al2O3, МоP/SiO2-SAPO-11 and МоP/SiO2-zeolite β systems has demonstrated the enhancement of MoP/SiO2 catalyst activity in MP HDO in mechanical mixtures with alumina, SAPO-11 and zeolite β. The higher conversions of MP over МоP/SiO2 catalyst in the presence of Al2O3, SAPO-11 and zeolite β can be explained by the higher rate of methyl palmitate hydrolysis, which is favored by the acid sites of these materials. The use of γ-Al2O3, SAPO-11 and zeolite β instead of SiC in the mixture with MoP/SiO2 increases the MP conversion, but has minor impact on the HDO selectivity. It was shown that selectivities of methyl palmitate conversion through the HDO pathways are nearly the same for the different systems, because the amounts of CO and CO2 in the gas-phase products differ only slightly. The main HDO product observed in the liquid products of MP HDO over MoP/SiO2-SiC and MoP/SiO2-γ-Al2O3 is hexadecane; the selectivity of HDO pathway exceeds 96% at 100% of MP conversion. The use of MoP/SiO2 in the mixture with the SAPO-11 or zeolite β instead of SiC changes the product distribution due to partial cracking and structural isomerization of linear C15 and C16 hydrocarbons, which are formed in HDO reaction of ester. Thus, МоP/SiO2-SAPO-11 system produces iso-alkane and some amount of C6-C14 hydrocarbons, while increased formation of hydrocarbons with  14 carbon atoms in the hydrocarbon chains is observed over MoP/SiO2-zeolite β catalytic system.
The obtained results show, that the use of the MoP/SiO2 catalyst in the mixtures with γ-Al2O3, SAPO-11 and zeolite  enhances the catalytic activity of MoP catalyst in the hydrodeoxygenation of MP without reducing selectivity through HDO pathway. The product distribution in the liquid phase depends on the dilutor nature. Linear alkanes are produced in the presence of -Al2O3. SAPO-11 can be considered as the perspective material for the one-pot HDO/hydroisomerization of aliphatic esters.