Support Effect on the Performance of Ni 2 P Catalysts in the Hydrodeoxygenation of Methyl Palmitate

The effect of support nature, SiO2 and γ-Al2O3, on physicochemical and catalytic properties of nickel phosphide catalysts in methyl palmitate hydrodeoxygenation (HDO) has been considered. Firstly, alumina-supported nickel phosphide catalysts prepared by temperature-programmed reduction method starting from different precursors (phosphate–Ni(NO3)2 and (NH4)2HPO4 or phosphite–Ni(OH)2 and H3PO3) were compared using elemental analysis, N2 physisorption, H2-TPR, XRD, TEM, NH3-TPD, 27Al and 31P MAS NMR techniques and catalytic experiments. The mixture of nickel phosphide phases was produced from phosphate precursor on alumina while using of phosphite precursor provides Ni2P formation with the higher activity in methyl palmitate HDO. Besides, the comparative study of the performances of Ni2P/SiO2 and Ni2P/Al2O3 catalysts demonstrates the apparent superiority of alumina-supported Ni2P in the methyl palmitate hydrodeoxygenation. Considering the tentative scheme of methyl palmitate transformation, we proposed that cooperation of Ni2P and acid sites on the surface of alumina provides the enhanced activity of alumina-supported Ni2P through the acceleration of acid-catalysed hydrolysis.


Introduction
Depletion of fossil oils resources, as well as the environmental issues of increased carbon emission, stimulates the development of new catalytic technologies for the production of transportation fuels from renewable [1][2][3].Triglyceride-based feedstocks, such as non-edible vegetable oils, animal fats and waste cooking oils, are the attractive resources that give a mixture of C 14 -C 18 alkanes in the hydrodeoxygenation (HDO) process [2,4,5].This product called green diesel has a high cetane number and stability, low density and can be mixed with the fossil-derived fuels to improve their quality [6][7][8].
Recently, noble metal-based and sulphide hydrotreating catalysts have been studied intensively in the HDO of triglycerides and model compounds, such as aliphatic esters [3][4][5].However, the practical application of noble metal catalysts, despite their high HDO activity, is restricted by the high cost and shortage of noble metals.As to hydrotreating sulphide catalysts, the S-containing agent has to be added to the feed to avoid deactivation that, in turn, leads to the appearance of undesirable S-containing products [9][10][11].Besides, the catalytic properties of supported noble metals and sulphide phase can be deteriorated by the reaction products-carbon oxides and water [12][13][14].Thus, new systems containing base metals or their carbides, nitrides and phosphides have emerged as the catalysts for HDO of aliphatic esters and vegetable oils in recent years [2][3][4]15,16].Among them, silica-supported nickel phosphide catalysts attract the particular attention of researchers due to their high stability and activity in HDO reactions [3,[16][17][18][19][20][21][22][23][24][25][26][27][28].
Effects of the precursor, initial Ni/P ratio and preparation conditions on the catalytic properties of Ni x P y /SiO 2 systems in the HDO of aliphatic esters were studied using mainly methyl oleate [18,20], methyl laurate [15,17,19,28,29] and methyl palmitate [21,[24][25][26][27].In most studies, the activity of Ni x P y /SiO 2 catalysts was shown to be higher than the activity of Ni/SiO 2 samples and the decrease of Ni/P ratio in the impregnation solution improved the catalytic activity and selectivity towards direct HDO (formation of C n hydrocarbons through H 2 O elimination) [21,24,27].It was stated that the Ni 2 P phase is the most active in hydrodeoxygenation of fatty acid methyl esters among the variety of supported nickel phosphides such as Ni 3 P, Ni 2 P, Ni 5 P 4 and Ni 12 P 5 [17,20,23,27].The optimal calcination and reduction conditions of Ni x P y /SiO 2 precursors for obtaining the most active catalysts in HDO were found [25,26,30].
In our previous study we have found out the synergetic effect of Ni 2 P/SiO 2 and γ-Al 2 O 3 physical mixture in hydrodeoxygenation of methyl palmitate, which was explained by the cooperation of the metal sites of Ni 2 P/SiO 2 and the acid sites of γ-Al 2 O 3 in consecutive metal-catalysed and acid-catalysed reactions of methyl palmitate conversion [31].Therefore, we proposed, that the employment of γ-Al 2 O 3 as the support can improve the activity of the Ni 2 P catalyst due to the ability of the former to accelerate the aliphatic esters hydrolysis [32,33].
It is well known, that support nature can influence the activity of the catalysts by affecting the active component properties, namely active phase and dispersity, at the stage of catalyst preparation [22,[34][35][36][37] or as a result of the participation of the support in catalytic conversions [33,35,[37][38][39][40].The support effect on the catalytic properties of nickel phosphide catalysts has been studied in the hydrodechlorination of chlorobenzene [41,42], hydrodesulfurization of thiophene [43] and hydrodeoxygenation of guaiacol [44].Recently, deoxygenation of methyl laurate over nickel phosphide catalysts supported on SiO 2 , CeO 2 , TiO 2 , γ-Al 2 O 3 and SAPO-11 was also investigated [22].However, in this study different nickel phosphide phases were formed depending on the support: Ni 2 P was produced on SiO 2 , CeO 2 , TiO 2 and SAPO-11, while Ni 3 P and Ni 12 P 5 were received on γ-Al 2 O 3 .The authors concluded that the catalyst activity of nickel phosphide catalysts in methyl laurate HDO depends on several factors, namely surface density of Ni sites, the electron property of metal sites, the Ni 2 P crystallite size and the synergism between the Ni sites and the acid sites.
The aim of our current study is to elucidate the real support effect, without the interference of other crucial factors, like phase composition, Ni content or Ni 2 P particle size through the comparative study of the performance of Ni 2 P/SiO 2 and Ni 2 P/Al 2 O 3 catalysts in the methyl palmitate hydrodeoxygenation.Firstly, the alumina-supported nickel phosphide catalysts differing in the precursor and reduction temperature have been investigated by elemental analysis, N 2 physisorption, H 2 -TPR, XRD, TEM, NH 3 -TPD, 27 Al and 31 P MAS NMR.It turned out that the reduction of the phosphite-type precursor at 550-650 • C makes it possible to support Ni 2 P nanoparticles on the alumina surface with the particle size distribution resembling that on the surface of Ni 2 P/SiO 2 catalysts, obtained in our previous study from phosphide precursor [26].The comparison of the catalyst's behaviour in the wide temperature interval showed the superior activity of the alumina-supported catalyst in comparison with the silica-supported one.The higher activity of Ni 2 P/Al 2 O 3 catalyst in methyl palmitate HDO along with the higher amounts of oxygen-containing intermediates in the reaction products can be explained by the acceleration of acid-catalysed hydrolysis of methyl palmitate.

The Effect of Preparation Conditions on the Physicochemical Properties of Ni x P y /γ-Al 2 O 3 Catalysts
Two sets of Ni x P y /γ-Al 2 O 3 catalysts were prepared by the incipient wetness impregnation from the phosphate-or phosphite-containing precursors (NiP_A/Al 2 O 3 or NiP_I/Al 2 O 3 samples, correspondingly) with the subsequent temperature-programmed reduction at 550, 600 and 650 • C. The physicochemical properties of obtained Ni x P y /γ-Al 2 O 3 catalysts after reduction and passivation are presented in Table 1.
According to the chemical analysis (ICP-AES), the contents of Ni and P in the reduced samples of Ni x P y /γ-Al 2 O 3 catalysts were not changed remarkably in comparison with that in precursors (Table 1).The Ni/P molar ratio is varied insignificantly in the range from 0.51 to 0.55 in the reduced NiP_A/Al 2 O 3 samples (initial Ni/P ratio 0.5) and it is retained at the level of 0.34-0.35 in the reduced NiP_I/Al 2 O 3 samples (initial Ni/P ratio 0.3).The Ni/P molar ratios in reduced catalysts were slightly higher than the initial Ni/P molar ratios in oxide precursors and impregnating solutions.The increase of the Ni/P molar ratio in the catalysts after reduction is caused by the formation and elimination of volatile phosphorus-containing species (namely PH 3 or P, P 2 etc.) during high-temperature reduction.The retained P could enter into the composition of nickel phosphide particles, unreduced PO x groups [43,45], elemental phosphorus residuals [46] and AlPO 4 [43].
With the increase of the reduction temperature from 550 to 650 • C, the specific surface area (A BET ) of NiP_A/Al 2 O 3 catalysts is expanded from 127 to 158 m 2 /g along with the decrease of the average pore diameter (D pore ) from 12.7 to 10.6 nm.The specific surface area of reduced NiP_I/Al 2 O 3 catalysts is decreased approximately 2-fold in comparison with γ-Al 2 O 3 (101-120 m 2 /g vs. 205 m 2 /g of γ-Al 2 O 3 ).The average pore diameter of reduced NiP_I/Al 2 O 3 catalysts is also decreased from 13.4 nm to 8.3-9.2 nm in comparison with the γ-Al 2 O 3 support.Figure S1 shows the curves of incremental pore volume distribution depending on the average pore diameter for NiP_A/Al 2 O 3 and NiP_I/Al 2 O 3 samples.It indicates that the mesoporous structure of the support was conserved after active component loading and reduction but the total pore volume was decreased.The textural characteristics of the catalyst could have changed because of the thermal treatment [40].Also, the decrease of A BET , D pore and V pore for the catalysts in comparison with alumina support is caused with the partial pore clogging by Ni x P y particles and phosphorus-containing compounds (PO x groups, elemental phosphorus ) or probably with the support structure rearrangement because of AlPO 4 formation as will be illustrated later by NMR data.C and the hydrogen uptake does not run out up to 900 • C. According to the literature data [22,37,44], similar H 2 -TPR profiles are characteristic for the reduction of nickel phosphate precursors deposited on alumina and also for the reduction of phosphate species (PO 4 3− , P 2 O 7 4− and (PO 3 − ) n ) at a temperature above 720 • C [47,48].There are no low-temperature peaks corresponding to the reduction of nickel oxide particles [49].According to the literature data [50], the reduction of commercial aluminium phosphate and alumina-supported phosphates begins at a temperature above 700 and 800  C for the NiPO x _I/Al 2 O 3 precursor, hydrogen absorption is associated with the reduction of phosphate-containing components formed during the decomposition of phosphorous acid.Thus, it can be expected that the formation of nickel phosphide phase in the case of the phosphite-containing precursor occurs at lower reduction temperature.The H 2 -TPR curve of the reference NiPO x _I/SiO 2 precursor shows a similar dependence of the hydrogen uptake from temperature, however, the amount of absorbed hydrogen in the temperature range of 600-900 • C is higher for the silica-supported sample in contrast to the alumina-supported one.This fact was also confirmed by chemical analysis data showing the Ni/P molar ratio equal to 0.53 in NiP_I/SiO 2 (600) sample in comparison to 0.35 in NiP_I/Al 2 O 3 (600) (Table 1).Such a difference between the behaviours of SiO 2 -and Al 2 O 3 -supported catalysts is due to a stronger interaction of phosphate groups with the surface of aluminium oxide [44,[54][55][56].As a result, phosphate groups supported on silica are readily reduced and removed in the form of phosphine (or volatile phosphorus species), while the interaction of phosphate groups with the surface of alumina leads to the formation of alumina phosphate and surface-bonded mono-and poly-phosphates that are stable to at least 650 • C. Based on the H 2 -TPR data of the oxide precursors and the literature data [42], the reduction temperatures of 550-650 • C were chosen to investigate the nickel phosphide phase formation on alumina.The XRD patterns of the γ-Al 2 O 3 support, as well as alumina-supported nickel phosphide catalysts differing in the precursor natures and the reduction temperatures, are shown in Figures 2  and 3.
Catalysts 2018, 8, x FOR PEER REVIEW 4 of 23 reduction of commercial aluminium phosphate and alumina-supported phosphates begins at a temperature above 700 and 800 °C, correspondingly.From the H2-TPR profile of NiPOx_I/Al2O3 precursor it is seen that the reduction of oxide precursors starts at the temperature of 300 °C with the formation of additional amount of hydrogen, as a result of phosphorous acid and nickel phosphites decomposition [26,51,52].The H2-TPR profile contains signals in the temperature range of 400-600 °C directed downward.In this temperature region phosphorous acid and phosphite-containing compounds are disproportionate to the formation of phosphine and H3PO4 or phosphates.In agreement with the literature data phosphine could interact with Ni 2+ species producing nickel phosphides already at the temperature as low as 250 °C [53].In addition, PH3 can decompose at higher temperatures into elemental phosphorus and hydrogen.From the temperature of ~600 °C for the NiPOx_I/Al2O3 precursor, hydrogen absorption is associated with the reduction of phosphate-containing components formed during the decomposition of phosphorous acid.Thus, it can be expected that the formation of nickel phosphide phase in the case of the phosphite-containing precursor occurs at lower reduction temperature.The H2-TPR curve of the reference NiPOx_I/SiO2 precursor shows a similar dependence of the hydrogen uptake from temperature, however, the amount of absorbed hydrogen in the temperature range of 600-900 °C is higher for the silica-supported sample in contrast to the alumina-supported one.This fact was also confirmed by chemical analysis data showing the Ni/P molar ratio equal to 0.53 in NiP_I/SiO2(600) sample in comparison to 0.35 in NiP_I/Al2O3(600) (Table 1).Such a difference between the behaviours of SiO2and Al2O3-supported catalysts is due to a stronger interaction of phosphate groups with the surface of aluminium oxide [44,[54][55][56].As a result, phosphate groups supported on silica are readily reduced and removed in the form of phosphine (or volatile phosphorus species), while the interaction of phosphate groups with the surface of alumina leads to the formation of alumina phosphate and surface-bonded mono-and poly-phosphates that are stable to at least 650 °C.
Based on the H2-TPR data of the oxide precursors and the literature data [42], the reduction temperatures of 550-650 °C were chosen to investigate the nickel phosphide phase formation on alumina.The XRD patterns of the γ-Al2O3 support, as well as alumina-supported nickel phosphide catalysts differing in the precursor natures and the reduction temperatures, are shown in Figures 2  and 3.The difference intensity curves obtained by the subtraction of the XRD curve of the alumina from the catalyst ones help us to identify the Ni-containing phases.
For NiP_A/Al 2 O 3 (550) sample the predominant phase was metallic Ni 0 (PDF No. 04-0850) with average crystallite size equalled to 5.5 nm (Figure 2a, Table 1).Weak intensity peaks characteristic of Ni 3 P phase (JCPDS No. 34-501) were also observed.With the reduction temperature increase to 600 • C the predominant phase in NiP_A/Al 2 O 3 (600) sample has remained the metallic Ni 0 (D XRD = 5.0 nm).Also, the presence of Ni 3 P and Ni 12 P 5 (JCPDS No. 22-1190) phases was indicated (Figure 2b).The estimated Ni 3 P and Ni 12 P 5 crystallite sizes were 4.0-4.5 nm and 12 nm, respectively.The further reduction of temperature increase to 650 • C led to the disappearance of metallic nickel and the predominant Ni 3 P phase (D XRD = 7.6 nm) formation with an additional amount of Ni 12 P 5 phase (D XRD = 11.5 nm) (Figure 2c).Thus, it was shown that for the set of NiP_A/Al 2 O 3 catalysts obtained from phosphate precursor increase in the reduction temperature from 550 to 650 • C promotes the formation of the Ni 3 P and Ni 12 P 5 phases and the average crystallite sizes growth.Nevertheless, the desired Ni 2 P phase has not been obtained.
The XRD patterns of NiP_I/Al 2 O 3 catalysts obtained from phosphite precursor are shown in Figure 3.Only the characteristic diffraction peaks corresponding to Ni 2 P phase (JCPDS No. 03-0953) are found in the difference curve.With the growth of the reduction temperature from 550 to 650 • C the estimated Ni 2 P crystallites sizes are increased slightly from 3.8 nm to 4.4 nm.So, it is shown that using the phosphite-containing precursor makes it possible to produce highly dispersed Ni 2 P crystallites on γ-Al 2 O 3 after reduction at 550-650 • C. According to XRD analysis, the NiP_I/SiO 2 (600) catalyst also contains the characteristic peaks of the Ni 2 P phase with D XRD = 5.5 nm (Figure S2) [26].
Figure 4 shows the TEM images of NiP_A/Al 2 O 3 (650) and NiP_I/Al 2 O 3 (600) catalysts prepared from different precursors.The nickel phosphide particles with different size ranged in the region from 4.0 to 50 nm were observed in the NiP_A/Al 2 O 3 (650) sample (Figure 4a).Also, individual particles with the size up to 100 nm were found.On the contrary, NiP_I/Al 2 O 3 (550-650) samples demonstrate a uniform distribution of the nickel phosphide particles on the alumina surface (Figure 4b).The mean Ni 2 P particle diameters of the NiP_I/Al 2 O 3 (550), NiP_I/Al 2 O 3 (600) and NiP_I/Al 2 O 3 (650) samples were estimated, equalling 5.1 nm, 5.6 nm and 6.0 nm, correspondingly.Thus, the mean Ni 2 P particle diameters of NiP_I/Al 2 O 3 catalysts slowly increased with the reduction temperature growth from 550 to 650 • C. The TEM image of NiP_A/Al 2 O 3 (650) catalyst (Figure S3a) reveals crystal lattice fringes with the d-spacing value of 2.16 Å, 2.47 Å and 2.83 Å corresponding to the (231), ( 031) and (130) reflections of the Ni 3 P phase (PDF No. 34-501), respectively.Also, it was found the crystal lattice fringes with the d-spacing value of 2.12 Å and 2.55 Å corresponding to the (121) and (202) reflections of Ni 3 (PO 4 ) 2 (PDF No. 70-1796) that proved the formation of a thin layer of nickel phosphate after passivation covered the nickel phosphide particles (Figure S3a).The TEM images of NiP_I/Al 2 O 3 (600) catalyst in Figure S3b exhibit crystal lattice fringes with the d-spacing value of 2.03 Å and 2.22 Å corresponding to the (021) and (111) reflections of the Ni 2 P crystalline phase (PDF No. . The nature of the precursor of the active component affects not only the morphological features of the formed nickel phosphide particles on the alumina surface but also the number of acid sites on the surface.The NH 3 -TPD technique was employed to explore the acidic properties of samples.Before NH 3 -TPD experiments, the nickel phosphide catalyst precursors were reduced in situ in Autosorb-1 apparatus.Figure 5a,b shows the NH 3 -TPD curves of NiP_A/Al 2 O 3 and NiP_I/Al 2 O 3 catalysts, as well as the γ-Al 2 O 3 support.The quantities of acid sites estimated by integration and further deconvolution of NH 3 desorption peaks of applied materials are listed in Table 2.          C is usually attributed to the sites with the weakest acidity on the boundary of physiosorbed and chemisorbed ammonia, the second desorption peak at the temperature of 337 • C was assigned to the moderate strength acid sites [57].The total acidity of γ-Al 2 O 3 support is equal to 421 µmol/g.On the NH 3 -TPD curves of NiP_A/Al 2 O 3 and NiP_I/Al 2 O 3 catalysts reduced at 550, 600 and 650 • C, two desorption peaks of ammonia centred at the temperatures of 238-249 • C and 278-334 • C were also observed.According to the literature data, the acid sites of nickel phosphides are ascribed to weak Brønsted and Lewis acid sites specified by the presence of unreduced P-OH species and nickel sites with small positive charge (Ni δ+ ) due to the electron transfer from the metal to the P atoms, respectively [44,58,59].
With the increase of the reduction temperature from 550 to 650 • C the overall acidity of catalysts is decreased from 624 to 440 µmol/g and from 477 to 326 µmol/g for the samples obtained from the phosphate and phosphite precursor, correspondingly.The number of the weakest acid sites with T max = 238-249 • C in all catalysts is increased in comparing with the alumina support.Whereas the amount of the moderate strength acid sites with T max = 278-334 • C is retained only in NiP_A/Al 2 O 3 550 sample and decreased in the other investigated catalysts.The decrease of the amount of the moderate strength acid sites is caused by the shielding of the part of the alumina surface by the surplus of phosphorus-containing species especially in the samples prepared from the phosphite precursor.Earlier it was reported that the alumina-supported catalysts prepared from phosphate precursors also contain the increased number of Brønsted acid sites referring to O=P(OH) 2 -O-Al species formed after interaction of phosphate precursor and alumina surface OH-groups or to free O=P(OH) 3 entities interacting with the alumina surface via H-bonding [54,56].
The total acidity of silica-supported NiP_I/SiO 2 (600) catalyst is lower than the acidity of alumina-supported samples (Table 2).On the NH 3 -TPD curve of NiP_I/SiO 2 (600) catalyst, the main desorption peak of ammonia centred at the temperature of 241 • C corresponding to weak acidic sites and the minor peak corresponding to moderate strength acid sites at 299 • C were observed.
Figure 6 displays the 31 P 14 kHz MAS NMR spectrum of NiP_A/Al 2 O 3 catalyst after reduction at 650 • C, which can be divided into two main regions.The narrow region around 0 ppm corresponds to compounds comprising different PO x groups, including phosphorus oxides, phosphates, phosphites and surface PO x groups [26,46,[60][61][62].The broad part of the spectrum stretching from ~500 to ~5000 ppm contains resonances of different metallic-like compounds of phosphorus, namely, nickel phosphides [26,46,63].The most prominent signal in this part of the spectrum is located at ~1800 ppm.According to literature data, such shift corresponds to Ni 3 P phase [46].This line lies upon a wide resonance covering the area from ~1000 to ~2500 ppm, which can contain signals from several other nickel phosphides.The downfield shoulder of this resonance indicates the presence of Ni 12 P 5 phase that gives rise to two signals at ~2250 and ~1950 ppm.The upfield shoulder may correspond to NiP, Ni 5 P 4 and Ni 2 P phases, however, the latter has two inequivalent sites with chemical shifts of ~1500 and ~4000 ppm and there is no signal with 4000 ppm shift on the spectrum, so it is unlikely that the sample contains Ni 2 P particles.
Since the line of Ni 3 P is relatively narrow, we can assume that the particles of this phase are regular to some extent; moreover, this assumption is confirmed by the fact that this phase was observed earlier in the XRD pattern.On the contrary, the lines corresponding to other phases are too broad and shapeless, which can mean that their particles are small or inhomogeneous.Nonetheless, we can estimate the relative amount of phosphorus in nickel phosphides (roughly 35%, Table 2) in this sample from the ratio of spectral intensities integrated over the "phosphide" region (500-4500 ppm) and over the entire spectrum.The remaining phosphorus is present in the sample as different PO x groups including AlPO 4 .
Figure 7 shows the experimental 14 kHz MAS 27 Al NMR spectrum of NiP_A/Al 2 O 3 (650) catalyst and its decomposition into three separate lines.Two of them belong to γ-Al 2 O 3 (octahedral site of Al 2 O 3 (O) at ~10 ppm and tetrahedral site of Al 2 O 3 (T) at ~70 ppm) and the third corresponds to the AlO 4 tetrahedra of AlPO 4 [54,64].All of the lines possess an irregular shape well described by the Simple Czjzek Model of Gaussian distribution of quadrupolar parameters [65].Using this model, we deconvoluted the spectrum into the signals from γ-Al 2 O 3 and AlPO 4 and determined the relative amount of the latter equal to ~25 at% (Table 2).Since the line of Ni3P is relatively narrow, we can assume that the particles of this phase are regular to some extent; moreover, this assumption is confirmed by the fact that this phase was observed earlier in the XRD pattern.On the contrary, the lines corresponding to other phases are too broad and shapeless, which can mean that their particles are small or inhomogeneous.Nonetheless, we can estimate the relative amount of phosphorus in nickel phosphides (roughly 35%, Table 2) in this sample from the ratio of spectral intensities integrated over the "phosphide" region (500-4500 ppm) and over the entire spectrum.The remaining phosphorus is present in the sample as different POx groups including AlPO4.
Figure 7 shows the experimental 14 kHz MAS 27 Al NMR spectrum of NiP_A/Al2O3(650) catalyst and its decomposition into three separate lines.Two of them belong to γ-Al2O3 (octahedral site of Al2O3 (O) at ~10 ppm and tetrahedral site of Al2O3 (T) at ~70 ppm) and the third corresponds to the AlO4 tetrahedra of AlPO4 [54,64].All of the lines possess an irregular shape well described by the Simple Czjzek Model of Gaussian distribution of quadrupolar parameters [65].Using this model, we deconvoluted the spectrum into the signals from γ-Al2O3 and AlPO4 and determined the relative amount of the latter equal to ~25 at% (Table 2).Figure 8 displays the "phosphide" regions of 14 kHz MAS 31 P NMR spectra of NiP_I/ Al 2 O 3 (550-650) catalysts reduced from phosphite precursor at the temperatures of 550, 600 and 650 • C.These spectra contain two very broad resonances centred at ~1500 and ~4000 ppm characteristic of Ni 2 P phase, though, very disordered.Another signal with very large FWHM is located around 2500 ppm and probably connected with Ni 12 P 5 phase.Quantitative analysis of spectra of samples with different reduction temperatures gives the following result: relative content of phosphorus included in phosphides in the sample reduced at 650 • C is much higher (~60 at% vs. ~40 at%) than in the samples reduced at lower temperatures (Table 2), while the absolute intensity of lines in the "phosphide" region of these samples is close (Figure S4).It means that some PO x groups are reduced only at temperatures higher than 600 • C probably resulting in the formation of phosphine or some other volatile phosphorus compound, while no additional nickel phosphides are formed.
14 kHz MAS 27 Al NMR spectra of NiP_I/Al 2 O 3 samples reduced at 550, 600 and 650 • C display the same three lines belonging to γ-Al 2 O 3 (~10 and ~70 ppm) and AlPO 4 (~40 ppm) as the spectrum of NiP_A/Al 2 O 3 sample (Figure 9).From deconvolution of these spectra into separate lines, we calculated the relative content of Al in the form of AlPO 4 that was equal to 22-30 at% for all samples (Table 2), while the support maintained the near to perfect ratio Al 2 O 3 (O)/Al 2 O 3 (T) = 1/2.From the results of 27 Al and 31 P NMR, it is likely that while the amount of AlPO 4 slightly grows with increasing reduction temperature, the overall phosphorus load of samples decreases due to the reduction of excess supported PO x groups with the formation of phosphine or some other volatile phosphorus compound.
Thus, 31        S5).The higher activity of NiP_I/Al 2 O 3 catalysts points out that Ni 3 P/Ni 12 P 5 phases display lower activity in the HDO of methyl palmitate then the Ni 2 P one.These results agree with the data obtained earlier for silica-supported systems, demonstrating the higher activity of Ni 2 P/SiO 2 catalyst in comparison with the Ni 12 P 5 /SiO 2 system [17,20,27].NiP_I/Al 2 O 3 catalysts also revealed higher selectivity to C 18 in comparison with NiP_A/Al 2 O 3 , the same tendency was observed in the HDO of methyl laurate [17], methyl oleate [20] and methyl palmitate [27] over Ni 2 P/SiO 2 and Ni 12 P 5 /SiO 2 catalysts.According to the proposed explanation, the enhancing of the positive charge of metal Ni sites upon transition from Ni 3 P and Ni 12 P 5 to Ni 2 P favours the activation of C=O groups and promotes the direct HDO reactions [17].The activity of NiP_I/Al 2 O 3 catalysts goes through the maximum with the increase of reduction temperature from 550 to 650 • C the optimal temperature being 600 • C (Figure 10).Consequently, this sample, designated further as NiP_I/Al 2 O 3 was chosen for detailed study in the HDO of methyl palmitate.
reactions [17].The activity of NiP_I/Al2O3 catalysts goes through the maximum with the increase of reduction temperature from 550 to 650 C the optimal temperature being 600 C (Figure 10).Consequently, this sample, designated further as NiP_I/Al2O3 was chosen for detailed study in the HDO of methyl palmitate.

Comparison of Ni2P/Al2O3 and Ni2P/SiO2 Catalysts in the HDO of Methyl Palmitate
Figure 11 shows the dependencies of the conversion of methyl palmitate and oxygen-containing compounds on the reaction temperature over NiP_I/Al2O3 (Ni2P/Al2O3) and NiP_I/SiO2 (Ni2P/SiO2) catalysts.NiP_I/SiO2 catalyst was prepared using the same procedure and contained the comparable amount of Ni and the mean particle diameter equalled to 5.5 nm (Table 1).In all experiments, the conversions of methyl palmitate and O-containing compounds are increased with the growth of temperature; wherein the NiP_I/SiO2 catalyst displays lower values of conversion.The minor difference between the dependencies of methyl palmitate and O-containing compounds conversions on temperature was observed over NiP_I/SiO2 catalyst, indicating that only small amounts of oxygen-containing intermediates are present in the reaction mixture.Indeed, the results of liquid product analysis showed only negligible amounts of oxygen-containing intermediates among the reaction products in the whole temperature range and the main products are hexadecane and pentadecane (Figure 12).Such behaviour is typical for all silica-supported NixPy systems, regardless of the precursor, Ni content or preparation conditions [24][25][26][27], pointing out that in the presence of NixPy/SiO2 catalysts the rate of methyl palmitate conversion is lower than the rates of oxygenated intermediates conversion.As a consequence, the rate of methyl palmitate transformation determines the overall rate of methyl palmitate HDO over NiP_I/SiO2 catalyst.The conclusion was confirmed in the additional experiments by the comparison of HDO conversion rate of the methyl laurate and the corresponding intermediate compounds: lauric acid, dodecanal and dodecanol (Figure S6).It was catalysts.NiP_I/SiO 2 catalyst was prepared using the same procedure and contained the comparable amount of Ni and the mean particle diameter equalled to 5.5 nm (Table 1).In all experiments, the conversions of methyl palmitate and O-containing compounds are increased with the growth of temperature; wherein the NiP_I/SiO 2 catalyst displays lower values of conversion.The minor difference between the dependencies of methyl palmitate and O-containing compounds conversions on temperature was observed over NiP_I/SiO 2 catalyst, indicating that only small amounts of oxygen-containing intermediates are present in the reaction mixture.Indeed, the results of liquid product analysis showed only negligible amounts of oxygen-containing intermediates among the reaction products in the whole temperature range and the main products are hexadecane and pentadecane (Figure 12).Such behaviour is typical for all silica-supported Ni x P y systems, regardless of the precursor, Ni content or preparation conditions [24][25][26][27], pointing out that in the presence of Ni x P y /SiO 2 catalysts the rate of methyl palmitate conversion is lower than the rates of oxygenated intermediates conversion.As a consequence, the rate of methyl palmitate transformation determines the overall rate of methyl palmitate HDO over NiP_I/SiO 2 catalyst.The conclusion was confirmed in the additional experiments by the comparison of HDO conversion rate of the methyl laurate and the corresponding intermediate compounds: lauric acid, dodecanal and dodecanol (Figure S6).It was shown, that in the presence of NiP_I/SiO 2 catalyst methyl laurate demonstrated the lowest conversion among the listed oxygen-containing substrates.The comparison of NiP_I/Al 2 O 3 and NiP_I/SiO 2 catalysts in methyl palmitate HDO shows that the use of γ-Al 2 O 3 as the support allows the significant increase of the methyl palmitate conversion.So, the conversion of methyl palmitate was raised from 22 to 67% at 290 • C and from 35 to 96% at 310 • C when NiP_I/Al 2 O 3 catalyst was used instead of NiP_I/SiO 2 (Figure 11).It should be noted, that both catalysts demonstrated stable time-on-stream activity, at least for 20 hours (Figure S7).As opposed to NiP_I/SiO 2 catalyst, a significant difference was observed between the conversion of methyl palmitate and the conversion of oxygen-containing compounds in the HDO of methyl palmitate over NiP_I/Al 2 O 3 catalyst at 250-290 • C. Besides, the appreciable amounts of oxygen-containing intermediates, like palmitic acid, hexadecanal, hexadecanols and palmityl palmitate were determined among the reaction products (Figure 13).The difference in the content of oxygenated compounds among the products of methyl palmitate HDO over NiP_I/SiO 2 and NiP_I/Al 2 O 3 catalysts is seen explicitly in Figure 14 demonstrating the content of oxygenated intermediates at the same conversion of methyl palmitate (about 57%).The observed results indicate that the rate of methyl palmitate conversion over alumina-supported catalysts is higher than the conversion rates of intermediate oxygenating compounds.To explain the observed difference in the behaviour of silica-supported and alumina-supported Ni 2 P catalysts the possible routes of methyl palmitate transformation should be analysed.The comparison of NiP_I/Al2O3 and NiP_I/SiO2 catalysts in methyl palmitate HDO shows that the use of γ-Al2O3 as the support allows the significant increase of the methyl palmitate conversion.So, the conversion of methyl palmitate was raised from 22 to 67% at 290 C and from 35 to 96% at 310 C when NiP_I/Al2O3 catalyst was used instead of NiP_I/SiO2 (Figure 11).It should be noted, that both catalysts demonstrated stable time-on-stream activity, at least for 20 hours (Figure S7).As opposed to NiP_I/SiO2 catalyst, a significant difference was observed between the conversion of methyl palmitate and the conversion of oxygen-containing compounds in the HDO of methyl palmitate over NiP_I/Al2O3 catalyst at 250-290 C.Besides, the appreciable amounts of oxygen-containing intermediates, like palmitic acid, hexadecanal, hexadecanols and palmityl palmitate were determined among the reaction products (Figure 13).The difference in the content of oxygenated compounds among the products of methyl palmitate HDO over NiP_I/SiO2 and NiP_I/Al2O3 catalysts is seen explicitly in Figure 14 demonstrating the content of oxygenated intermediates at the same conversion of methyl palmitate (about 57%).The observed results indicate that the rate of methyl palmitate conversion over alumina-supported catalysts is higher than the conversion rates of intermediate oxygenating compounds.To explain the observed difference in the behaviour of silica-supported and alumina-supported Ni2P catalysts the possible routes of methyl palmitate transformation should be analysed.

Reaction Scheme
The reaction networks of methyl palmitate transformation over silica-supported nickel phosphide catalysts were widely discussed in the literature [17,19,[22][23][24][25]27].Scheme 1 represents the reaction network for methyl palmitate HDO, including the intermediate and final product of methyl palmitate conversion.Palmitic acid and hexadecanal are usually considered as the primary products of methyl palmitate hydroconversion.Palmitic acid can be produced through the hydrogenolysis of the C-O bond in the methoxy group over metal sites or via hydrolysis over acid sites, giving methane or methanol, correspondingly [17,25,27,31].Hydrogenolysis of the ester C-O bond over metal sites leads to hexadecanal and methanol.Further conversion of palmitic acid was proposed to give pentadecane through the decarboxylation reaction or hexadecanal as a result of hydrogenation.Decarboxylation of palmitic acid can proceed to some extent over NiP_I/SiO2, NiP_I/Al2O3 and NiP_A/Al2O3 catalysts because the small amounts of CO2 were observed among the gas products but the primary route of pentadecane formation is decarbonylation of hexadecanal.The sum of CO and CO2 correlates with pentadecane content (Figure S8).In addition, hexadecanal gives hexadecanol-1 via hydrogenation that in turn is transformed to hexadecane through the subsequent reactions of dehydration to hexadecene and hydrogenation to hexadecane.

Reaction Scheme
The reaction networks of methyl palmitate transformation over silica-supported nickel phosphide catalysts were widely discussed in the literature [17,19,[22][23][24][25]27].Scheme 1 represents the reaction network for methyl palmitate HDO, including the intermediate and final product of methyl palmitate conversion.Palmitic acid and hexadecanal are usually considered as the primary products of methyl palmitate hydroconversion.Palmitic acid can be produced through the hydrogenolysis of the C-O bond in the methoxy group over metal sites or via hydrolysis over acid sites, giving methane or methanol, correspondingly [17,25,27,31].Hydrogenolysis of the ester C-O bond over metal sites leads to hexadecanal and methanol.Further conversion of palmitic acid was proposed to give pentadecane through the decarboxylation reaction or hexadecanal as a result of hydrogenation.Decarboxylation of palmitic acid can proceed to some extent over NiP_I/SiO 2 , NiP_I/Al 2 O 3 and NiP_A/Al 2 O 3 catalysts because the small amounts of CO 2 were observed among the gas products but the primary route of pentadecane formation is decarbonylation of hexadecanal.The sum of CO and CO 2 correlates with pentadecane content (Figure S8).In addition, hexadecanal gives hexadecanol-1 via hydrogenation that in turn is transformed to hexadecane through the subsequent reactions of dehydration to hexadecene and hydrogenation to hexadecane.
The scheme of methyl palmitate conversion includes some parallel and sequential transformations that can take place with the participation of acidic (ether hydrolysis, alcohol dehydration) and metal sites (hydrogenation and hydrogenolysis reactions).It is known that nickel phosphides are characterized by the presence of metallic and acidic centres, caused by an ensemble (geometric) and/or ligand (electronic) effect of phosphorus [66,67].Taking in mind the inertness of SiO 2 , the mechanism of ester conversion over NiP_I/SiO 2 catalyst is determined by the electronic properties of the metal sites and the acidity of the nickel phosphide phase.It was shown, that nickel phosphide has both Lewis and Brønsted acid sites, that are usually ascribed to the positively charged metal cations (Ni δ+ ) and surface P-OH groups, respectively [17,58].The correlation was also found between the acidity of Ni 2 P/SiO 2 catalysts, differing in precursor and preparation conditions and the content of phosphate groups determined by means of 31 P NMR [26,27].It was proposed, that P-OH groups of silica-supported Ni 2 P participate in the hydrolysis of aliphatic ester (methyl laurate) as well as assist the conversion of methyl laurate adsorbed on the metal sites via decarbonylation or hydrogenation reactions [17].
metal cations (Ni δ+ ) and surface P-OH groups, respectively [17,58].The correlation was also found between the acidity of Ni2P/SiO2 catalysts, differing in precursor and preparation conditions and the content of phosphate groups determined by means of 31 P NMR [26,27].It was proposed, that P-OH groups of silica-supported Ni2P participate in the hydrolysis of aliphatic ester (methyl laurate) as well as assist the conversion of methyl laurate adsorbed on the metal sites via decarbonylation or hydrogenation reactions [17].Scheme 1. Probable reaction network for HDO of methyl palmitate.Red lines-the decarboxylation and decarbonylation routes, green lines-the hydrodeoxygenation route.
The higher activity of NiP_I/Al2O3 catalyst in methyl palmitate HDO along with the higher production of palmitic acid in the reaction products can be explained by the participation of acid sites on the alumina support in the conversion of methyl palmitate through hydrolysis route.Indeed, the alumina-supported Ni2P catalyst is characterized by the higher amount of acid sites in comparison with NiP_I/SiO2 sample according to NH3-TPD data (Table 2).It is well established, that alumina can provide the acid-catalysed reactions, such as hydrolysis, dehydration and esterification [32,35,40,68].In our case the exact nature of acid sites on the surface of NiP_I/Al2O3 catalyst is undisclosed.NiP_I/Al2O3 catalyst produces the increasing amounts of the product of acid-catalysed esterification, namely palmityl palmitate (Figure 14), thereby indirectly confirming our assumption about the acceleration of acid-catalysed reactions over alumina-supported Ni2P in comparison with silica-supported one.Despite on the higher acidity of NiP_A/Al2O3 catalyst in comparison with NiP_I/Al2O3 sample, NiP_A/Al2O3 catalyst displays lower activity in methyl palmitate hydrodeoxygenation.This observation can be explained by the assumption that lower activity of Ni3P/Ni12P5 in metal-catalysed reactions led to the accumulation of oxygen-containing intermediates that retard the reversible hydrolysis reaction.Therefore, the balance between active metal sites of Ni2P and acid sites of support is necessary for the design of the efficient catalyst for the fatty acid esters hydrodeoxygenation.The obtained results prove our assumption that the activity of the Scheme 1. Probable reaction network for HDO of methyl palmitate.Red lines-the decarboxylation and decarbonylation routes, green lines-the hydrodeoxygenation route.
The higher activity of NiP_I/Al 2 O 3 catalyst in methyl palmitate HDO along with the higher production of palmitic acid in the reaction products can be explained by the participation of acid sites on the alumina support in the conversion of methyl palmitate through hydrolysis route.Indeed, the alumina-supported Ni 2 P catalyst is characterized by the higher amount of acid sites in comparison with NiP_I/SiO 2 sample according to NH 3 -TPD data (Table 2).It is well established, that alumina can provide the acid-catalysed reactions, such as hydrolysis, dehydration and esterification [32,35,40,68].In our case the exact nature of acid sites on the surface of NiP_I/Al 2 O 3 catalyst is undisclosed.NiP_I/Al 2 O 3 catalyst produces the increasing amounts of the product of acid-catalysed esterification, namely palmityl palmitate (Figure 14), thereby indirectly confirming our assumption about the acceleration of acid-catalysed reactions over alumina-supported Ni 2 P in comparison with silica-supported one.Despite on the higher acidity of NiP_A/Al 2 O 3 catalyst in comparison with NiP_I/Al 2 O 3 sample, NiP_A/Al 2 O 3 catalyst displays lower activity in methyl palmitate hydrodeoxygenation.This observation can be explained by the assumption that lower activity of Ni 3 P/Ni 12 P 5 in metal-catalysed reactions led to the accumulation of oxygen-containing intermediates that retard the reversible hydrolysis reaction.Therefore, the balance between active metal sites of Ni 2 P and acid sites of support is necessary for the design of the efficient catalyst for the fatty acid esters hydrodeoxygenation.The obtained results prove our assumption that the activity of the phosphide catalysts in the HDO of aliphatic esters can be improved by the employment of the support with the acidic properties instead of silica [31].Detailed kinetic experiments may subsequently throw light on the mechanism of methyl palmitate hydrodeoxygenation and help to evaluate the contribution of metal-and acid-catalysed reactions that take place over multifunctional catalysts.

Catalyst Preparation
Two sets of alumina-supported nickel phosphide catalysts with the nickel content about 7.5 wt% were prepared by the incipient wetness impregnation method from the phosphate-or phosphite-containing precursors with the subsequent temperature-programmed reduction.
Catalyst preparation from the phosphate-containing precursor (A).Support (γ-Al 2 O 3 ) was incipiently impregnated with an aqueous solution of (NH 4 ) 2 HPO 4 and Ni(NO 3 ) 2 , followed by drying in air at room temperature overnight and at 110 • C for 3 h and then calcination at 500 • C in air for 3 h.The initial Ni/P molar ratio in precursor was 0.5.Then the oxide precursor was reduced with H 2 flow (100 mL/(min•g)) using the temperature program: heating to 370 • C at a ramp rate of 3 • C/min and then to 550, 600 or 650 • C at a ramp rate of 1 • C/min and keeping at the reduction temperature for 1 h.After reduction, the samples were passivated at room temperature for 2 h in 1 vol% O 2 /He flow (40 mL/(min•g)) before taking out into the air for investigation by physicochemical methods.For the reactivity test, the catalysts were reduced in the catalytic reactor before experiments.The catalysts prepared from phosphate-containing precursor are denoted on the base of the precursor type (A) and the TPR temperature, for example, NiP_A/Al 2 O 3 (550) (shown in Table 1).
Catalyst preparation from the phosphite-containing precursor (I).The support (γ-Al 2 O 3 , SiO 2 ) was incipiently impregnated with an aqueous solution of Ni(OH) 2 and H 3 PO 3 with consequent drying in air at room temperature overnight and at 80 • C for 24 h.The initial Ni/P molar ratio in precursor was 0.3.Dried precursors were reduced directly without calcination by TPR in H 2 flow (100 mL/(min•g)) using the following temperature program: heating to 550, 600 or 650 • C at a ramp rate of 1 • C/min and keeping at the reduction temperature for 1 h.The catalysts prepared from phosphite precursor are denoted on the base of the precursor type (I) and the TPR temperature, for example, NiP_I/Al 2 O 3 (550) (Table 1).For comparison, the reference silica-supported Ni 2 P/SiO 2 catalyst (NiP_I/SiO 2 (600)) was prepared by the same procedure.

Catalyst Characterization
The elemental analysis of the samples was conducted using inductively coupled plasma atomic emission spectroscopy (ICP-AES) on Optima 4300 DV (Perkin Elmer, France).
The textural properties of the catalysts were characterized using N 2 physisorption at 77 K with an ASAP 2400 instrument (Micromeritics, Norcross, GA, USA).The specific surface area (A BET ) was determined using a multipoint Brunauer-Emmett-Teller (BET) model.The average pore diameter (D pore ) and the pore size distribution were determined by the Barret-Joyner-Halenda (BJH) method using the desorption branch of the isotherm.The total pore volume (V pore ) was estimated at a relative pressure of 0.99.
The reduction peculiarities of the catalyst precursors were determined by the H 2 temperatureprogrammed reduction method (H 2 -TPR).The calcined phosphate type precursor (or dried sample in the case of phosphite type precursor) in the amount of 0.10 g was loaded into a quartz U-tube reactor (5 mm in inner diameter) and pre-treated in argon atmosphere at 200 • C for 1 h.The experiments were carried out in 10 vol.%H 2 /Ar flow (flow rate = 60 mL/min) at a heating rate of 10 • C/min in a temperature range from 80 to 900 • C. The H 2 consumption was evaluated using a thermal conductivity detector (TCD).A cold trap with a temperature of −60 • C mounted before the TCD was used to remove water from the exhaust gas.X-ray diffraction (XRD) patterns were acquired on an X-ray diffractometer Bruker D8 Advance (Bruker, Germany) using CuK α radiation (wavelength λ = 1.5418Å) in 2θ scanning range from 10 to 70 • .The qualitative phase analysis was performed using the JCPDS database [69].The quantitative phase analysis and refining of the unit cell parameters were carried out by Rietveld analysis of a diffraction pattern using X'Pert High Score Plus software.The average crystallite size (D XRD ) was estimated using the Scherrer equation.
Transmission electron microscopy (TEM) images were obtained on a JEM-2010 instrument (JEOL, Tokyo, Japan) with an accelerating voltage of 200 kV and a resolution of 0.14 nm.To obtain statistical information, the structural parameters of ca.250 particles were measured.The Sauter mean diameter was used to determine the mean particles diameter (D s ) by the Equation: where n i is the number of particles, d i is the diameter of the i th particle.NMR experiments were carried out using a Bruker Avance-400 spectrometer (Bruker, Germany) at the resonance frequencies of 104.31 and 161.923MHz for 27 Al and 31 P nuclei respectively.Magic Angle Spinning (MAS) spectra were measured using a Bruker MAS NMR probe with 4-mm (outer diameter) ZrO 2 rotors at the spinning frequency of 14 kHz.The detailed procedure of 31 P MAS NMR measurements can be found in reference [27]. 27Al MAS NMR spectra were measured at the same spinning frequency with a simple one-pulse sequence.We used a short non-selective π/8 pulse of 0.5 µs with a 500 ms second repetition delay; 512 transients were recorded for each sample.The spectra were referenced to 0.1 mM Al[H 2 O] 6 3+ solution.Deconvolution of 27 Al NMR spectra was carried out in Dmfit program [70] using Simple Czjzek model that accounts for Gaussian quadrupole parameters distribution in disordered solids [65].To avoid oxidation of reduced catalysts by air, all samples were sealed in glass ampoules without oxygen access.Prior to NMR experiments, the catalysts were transferred from ampoule to NMR rotor in a glove box under argon atmosphere.This technique helps to minimize possible oxidation during NMR spectra acquisition.The acidic properties of the catalysts were characterized by temperature-programmed desorption of ammonia (NH 3 -TPD) using an Autosorb-1 apparatus (Quantachrome Instruments, Boynton Beach, FL, USA).Prior to ammonia adsorption, the precursor sample (0.25 g) was reduced at the given reduction temperature (550-650 • C) for 1 hour in hydrogen flow (25 mL/min) and then cooled to 120 • C. Subsequently, the sample was saturated with NH 3 for 30 min.The physically adsorbed ammonia was desorbed from the sample with He flow (25 mL/min) at 120 • C for 30 min.Desorption of ammonia was started by increasing the temperature from 120 • C up to 600 • C at a heating rate of 10 • C/min.The desorbed NH 3 was detected by a TCD.

Catalytic Activity Measurements
The reaction of methyl palmitate HDO was performed in a high-pressure fixed-bed reactor (inner diameter 9 mm and total length 265 mm) at the temperature range of 250-330 • C, hydrogen pressure 3.0 MPa, H 2 /feed volume ratio 600 Nm 3 /m 3 and methyl palmitate LHSV in the range of 2.4-12 h −1 ((cm 3 of MP)/(cm 3 of catalysts) per hour).The composition of the reaction mixture amounted to methyl palmitate -10 wt% (oxygen content -1.183 wt%), n-octane (used as internal standard for quantification of liquid product) -0.5 wt% and the rest -n-dodecane.For catalytic activity measurements, 0.5 cm 3 of the catalyst precursor was diluted with silicon carbide (0.2-0.3 mm) in a ratio of 1:9 and placed in the reactor between two layers of SiC.Then the sample of the precursor was reduced as described in Section 3.2.The liquid reaction products were collected every hour till steady-state condition and time on stream was not less than 8 h for each stage.
The conversion of methyl palmitate (X MP ), conversion of oxygen-containing compounds (X O ) and the selectivity to C 16 hydrocarbons (S C16 ) were calculated according to Equations: where n o MP and n MP are the initial and the current methyl palmitate content (mol/L), n o O and n O are the initial and current oxygen content (mol/L), n C16 is the concentration of C 16 hydrocarbons (mol/L) in the reaction mixture at X MP = 100%.The relative error in determining of the methyl palmitate (from GC data) and oxygen-containing compounds (from elemental analysis by Vario EL Cube) conversions was no more than 1%.

Product Analysis
The reaction products were identified using a GC-MS technique (Agilent Technologies 6890N with MSD 5973, Santa Clara, CA, USA) with a VF-5MS quartz capillary column (30 m × 0.25 mm × 0.25 µm).The liquid samples were analysed with a gas chromatography system (Agilent 6890N, Santa Clara, CA, USA) equipped with HP-1MS column (30 m × 0.32 mm × 1.0 µm) and flame ionization detector (FID).Gaseous phase products were analysed online with a gas chromatograph (Chromos 1000, Dzerzhinsk, Russia) equipped with a column packed with 80/100 mesh HayeSep®(Sigma-Aldrich, Saint Louis, MO, USA) and FID.The concentrations of CO and CO 2 were analysed in the form of methane after methanation over reduced Pd catalyst at 340 • C. The carbon balance across the reactor for all experiments was not less than 95%.
The total oxygen content in the reaction mixtures was evaluated using CHNSO elemental analyser Vario EL Cube (Elementar Analysensysteme GmbH, Hanau, Germany).The method is based on high-temperature pyrolysis of a liquid sample of the reaction mixture in a reducing atmosphere at 1170 • C and the resulting oxygen-containing radicals interact quantitatively with the carbon filler to form carbon monoxide (Boudouard equilibrium).An IR detector was used to quantify the CO.

Conclusions
Effect of precursors on the physicochemical and catalytic properties of alumina-supported nickel phosphide catalysts differing in precursors was studied.Ni(NO 3 ) 2 and (NH 4 ) 2 HPO 4 (phosphate precursor, NiP_A/Al 2 O 3 catalyst) or Ni(OH) 2 and H 3 PO 3 (phosphite precursor, NiP_I/Al 2 O 3 catalyst) pairs of nickel and phosphorus compounds were used for catalyst preparation by temperature-programmed reduction.It was found that NiP_I precursor is reduced at lower temperatures apparently due to nickel phosphite and H 3 PO 3 disproportionation with the formation of PH 3 interacting readily with nickel species and producing Ni 2 P nanoparticles in the temperature range of 550-650 • C. On the other hand, starting from phosphate precursor, we did not succeed in the production of Ni 2 P and the reduction at the temperature as high as 650 • C gives the mixture of Ni 3 P and Ni 12 P 5 phases.Using 31 P and 27 Al MAS NMR, the formation of aluminium phosphates in all catalysts was revealed.Interaction of phosphate groups with alumina surface hinders the formation of the desired Ni 2 P phase and leads to an irreversible change in textural characteristics and surface properties of γ-Al 2 O 3 .Indeed, the increase in the number of weak acid sites, which are usually associated with Brønsted acidity, has been observed by means of NH 3 -TPD in the supported catalysts in comparison with the alumina support.NiP_I/Al 2 O 3 catalyst displays significantly higher activity in methyl palmitate HDO compared to the NiP_A/Al 2 O 3 sample within the temperature region of 250-330 • C that is in agreement with the literature data reported the superiority of Ni 2 P phase against Ni 3 P/Ni 12 P 5 systems in HDO of aliphatic esters over silica-supported nickel phosphide catalysts.
The performances of alumina-and silica-supported Ni 2 P catalysts were compared in methyl palmitate HDO using NiP_I/Al 2 O 3 and NiP_I/SiO 2 catalysts with nearly the same nickel content and mean particle size.The employment of alumina instead of silica for Ni 2 P supporting allows a significant increase of the methyl palmitate conversion.Considering the tentative scheme of methyl palmitate transformation and reaction products distribution over NiP_I/Al 2 O 3 and NiP_I/SiO 2 catalysts, we conclude that the acceleration of methyl palmitate hydrolysis over acid sites of the alumina-supported catalyst is the reason of increased activity of NiP_I/Al 2 O 3 catalyst.
In general, the proper balance between metal and acid sites provides the superior performance of the NiP_I/Al 2 O 3 catalyst in the complicated scheme of methyl palmitate transformation, which includes the framework of acid-and metal-catalysed reactions.

Figure 8
Figure 8 displays the "phosphide" regions of 14 kHz MAS 31 P NMR spectra of NiP_I/Al2O3(550-650) catalysts reduced from phosphite precursor at the temperatures of 550, 600 and 650 °С.These spectra contain two very broad resonances centred at ~1500 and ~4000 ppm characteristic of Ni2P phase, though, very disordered.Another signal with very large FWHM is located around 2500 ppm and probably connected with Ni12P5 phase.Quantitative analysis of spectra of samples with different reduction temperatures gives the following result: relative content

Figure 7 .
Figure 7. Blue-experimental 14 kHz MAS 27 Al NMR spectrum of NiP_A/Al 2 O 3 (650) sample, black-theoretic calculated lines corresponding to different Al 2 O 3 and AlPO 4 sites, red-sum of the calculated lines.

Figure 8 .
Figure 8. 14 kHz MAS 31 P NMR spectra of NiP_I/Al 2 O 3 samples reduced at 550, 600 and 650 • C. The region with PO x groups is not shown.

2. 2 .
Catalytic Properties of Ni x P y /γ-Al 2 O 3 Catalysts in Methyl Palmitate HDO 2.2.1.The Effect of Preparation Conditions on the Catalytic Properties of Ni x P y /γ-Al 2 O 3 Catalysts in Methyl Palmitate HDO To evaluate the impact of the preparation conditions on the catalytic properties of the Ni x P y /Al 2 O 3 systems in the HDO of methyl palmitate several alumina-supported nickel phosphide catalysts differing in the precursor nature and reduction temperature were studied.It was shown in the previous section that the reduction of the samples prepared from phosphite precursor at 550, 600 and 650 • C gives rise to Ni 2 P phase supported on alumina (NiP_I/Al 2 O 3 catalysts).The Ni 3 P phase with the admixture of Ni 12 P 5 and some unreduced species have been observed in NiP_A/Al 2 O 3 catalyst prepared from phosphate precursor, even after reduction at 650 • C. Therefore, NiP_A/Al 2 O 3 catalyst (reduced at 650 • C) and series of NiP_I/Al 2 O 3 samples, prepared by reduction at 550, 600 and 650 • C, were chosen for the comparative study in the HDO of methyl palmitate.The reaction was carried out at the temperature of 290 • C, hydrogen pressure -3.0 MPa, H/C ratio -600 Nm 3 /m 3 , methyl palmitate LHSV -2.4-12 h −1 for 8 h, during this time the concentrations of the reagent and the products have reached a constant value.The catalytic properties of Ni x P y /Al 2 O 3 catalysts are shown in Figure 10.The presented data indicate that NiP_A/Al 2 O 3 catalyst is significantly inferior in activity compared to the NiP_I/Al 2 O 3 samples and this tendency is conserved within the temperature region of 250-330 • C (Figure

Figure 10 .
Figure 10.Conversion (a) and selectivity (b) of NiP_A/Al 2 O 3 and NiP_I/Al 2 O 3 catalysts in HDO of methyl palmitate (P H2 = 3.0 MPa, T = 290 • C, H 2 /feed = 600 Nm 3 /m 3 , methyl palmitate LHSV = 9 h −1 ).2.2.2.Comparison of Ni 2 P/Al 2 O 3 and Ni 2 P/SiO 2 Catalysts in the HDO of Methyl Palmitate Figure11shows the dependencies of the conversion of methyl palmitate and oxygen-containing compounds on the reaction temperature over NiP_I/Al 2 O 3 (Ni 2 P/Al 2 O 3 ) and NiP_I/SiO 2 (Ni 2 P/SiO 2 ) catalysts.NiP_I/SiO 2 catalyst was prepared using the same procedure and contained the comparable amount of Ni and the mean particle diameter equalled to 5.5 nm (Table1).In all experiments, the conversions of methyl palmitate and O-containing compounds are increased with the growth of temperature; wherein the NiP_I/SiO 2 catalyst displays lower values of conversion.The minor difference between the dependencies of methyl palmitate and O-containing compounds conversions on temperature was observed over NiP_I/SiO 2 catalyst, indicating that only small amounts of oxygen-containing intermediates are present in the reaction mixture.Indeed, the results of liquid product analysis showed only negligible amounts of oxygen-containing intermediates among the reaction products in the whole temperature range and the main products are hexadecane and pentadecane (Figure12).Such behaviour is typical for all silica-supported Ni x P y systems, regardless of the precursor, Ni content or preparation conditions[24][25][26][27], pointing out that in the presence of Ni x P y /SiO 2 catalysts the rate of methyl palmitate conversion is lower than the rates of oxygenated intermediates conversion.As a consequence, the rate of methyl palmitate transformation determines the overall rate of methyl palmitate HDO over NiP_I/SiO 2 catalyst.The conclusion was confirmed in the additional experiments by the comparison of HDO conversion rate of the methyl laurate and the corresponding intermediate compounds: lauric acid, dodecanal and dodecanol (FigureS6).It was shown, that in the presence of NiP_I/SiO 2 catalyst methyl laurate demonstrated the lowest conversion among the listed oxygen-containing substrates.

Catalysts 2018, 8 ,
x FOR PEERREVIEW  13 of 23    shown, that in the presence of NiP_I/SiO2 catalyst methyl laurate demonstrated the lowest conversion among the listed oxygen-containing substrates.

Catalysts 2018, 8 ,
x FOR PEERREVIEW  13 of 23    shown, that in the presence of NiP_I/SiO2 catalyst methyl laurate demonstrated the lowest conversion among the listed oxygen-containing substrates.
Figure S4: Full-scale mass-normalized 14 kHz MAS 31 P spectra of NiP_I/Al 2 O 3 reduced at 550, 600 and 650 • C. A significant decrease in intensity of the line corresponding to PO x groups can be observed for the sample reduced at 650 • C. Figure S5: Temperature effect on the conversion of methyl palmitate and oxygen-containing compounds over NiP_A/Al 2 O 3 650 and NiP_I/Al 2 O 3 600 catalysts.
Figure S7: MP conversion and total oxygen-containing compounds conversion as a function of time on stream for Ni 2 P/Al 2 O 3 and Ni 2 P/SiO 2 catalysts.
Figure S8: Dependence of the CO and CO 2 sum from pentadecane content in methyl palmitate hydrodeoxygenation over NiP_A/Al 2 O 3 650 catalysts.Author Contributions: G.A.B. conceived and designed the experiments and supervised the work; I.V.D. and I.V.S. performed the catalyst synthesis; I.V.D., I.V.S. and P.V.A. performed the catalytic activity tests; E.Y.G. performed the catalyst characterization by TEM; V.P.P. performed the catalyst characterization by XRD; E.G.K. performed the catalyst characterization by NH 3 -TPD; I.V.Y. and O.B.L. performed the catalyst characterization by MAS NMR; I.V.D., P.V.A. and G.A.B. analysed the experimental data and wrote the paper.Funding: This work was conducted within the framework of the budget project No. AAAA-A17-117041710075-0 for Boreskov Institute of Catalysis.

Table 1 .
Physicochemical properties of NiP_A/Al 2 O 3 , NiP_I/Al 2 O 3 and NiP_I/SiO 2 catalysts.TPR profiles of NiPO x _A/Al 2 O 3 precursor and NiPO x _I/Al 2 O 3 precursor as well as the reference NiPO x _I/SiO 2 precursor are shown in Figure 1.H 2 -TPR profiles of the NiPO x _A/Al 2 O 3 and NiPO x _I/Al 2 O 3 precursors are entirely different.It was indicated that the reduction of NiPO x _A/Al 2 O 3 precursor starts at the temperature of 525 • C. It was observed two maximums of hydrogen consumption at 725 • C and 840 [53]51,52]spondingly.From the H 2 -TPR profile of NiPO x _I/Al 2 O 3 precursor it is seen that the reduction of oxide precursors starts at the temperature of 300 • C with the formation of additional amount of hydrogen, as a result of phosphorous acid and nickel phosphites decomposition[26,51,52].The H 2 -TPR profile contains signals in the temperature range of 400-600 • C directed downward.In this temperature region phosphorous acid and phosphite-containing compounds are disproportionate to the formation of phosphine and H 3 PO 4 or phosphates.In agreement with the literature data phosphine could interact with Ni 2+ species producing nickel phosphides already at the temperature as low as 250 • C[53].In addition, PH 3 can decompose at higher temperatures into elemental phosphorus and hydrogen.From the temperature of ~600

reduction , o C NH 3 -TPD MAS NMR T max , o C Quantity, µmol/g Overall Quantity, µmol/g Ni x P y , at.% 31 P NMR AlPO 4 , at.% 27 Al NMR
TPD curve of γ-Al 2 O 3 has two desorption peaks of ammonia centred at 237 • C and 335 • C. The first desorption peak around 237