The Deoxygenation Pathways of Palmitic Acid into Hydrocarbons on Silica-Supported Ni 12 P 5 and Ni 2 P Catalysts

Pure Ni12P5/SiO2 and pure Ni2P/SiO2 catalysts were obtained by adjusting the Ni and P molar ratios, while Ni/SiO2 catalyst was prepared as a reference against which the deoxygenation pathways of palmitic acid were investigated. The catalysts were characterized by N2 adsorption, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), transmission election microscopy (TEM), infrared spectroscopy of pyridine adsorption (Py-IR), H2-adsorption and temperature-programmed desorption of hydrogen (H2-TPD). The crystallographic planes of Ni(111), Ni12P5(400), Ni2P(111) were found mainly exposed on the above three catalysts, respectively. It was found that the deoxygenation pathway of palmitic acid mainly proceeded via direct decarboxylation (DCO2) to form C15 on Ni/SiO2. In contrast, on the Ni12P5/SiO2 catalyst, there were two main competitive pathways producing C15 and C16, one of which mainly proceeded via the decarbonylation (DCO) to form C15 accompanying water formation, and the other pathway produced C16 via the dehydration of hexadecanol intermediate, and the yield of C15 was approximately twofold that of C16. Over the Ni2P/SiO2 catalyst, two main deoxygenation pathways formed C15, one of which was mainly the DCO pathway and the other was dehydration accompanying the hexadecanal intermediate and then direct decarbonylation without water formation. The turn over frequency (TOF) followed the order: Ni12P5/SiO2 > Ni/SiO2 > Ni2P/SiO2.


Introduction
Research on alternative resources for hydrocarbon fuel has received wide attention due to diminishing fossil fuel reserves and the environmental crisis during the past decades [1][2][3][4][5][6].Furthermore, hydrocarbon fuel from biomass can directly substitute for that from coal, natural gas and petroleum in the current energy system.Biomass-based oil is abundant, and it has the potential to significantly displace petroleum in the production of fuels for the transportation sector.Recently, Richard et al. reported that biofuel blending reduced particle emissions from aircraft engines at cruise condition [7].Therefore, increased attention has been focused on biomass conversion into hydrocarbon fuel [8][9][10][11][12][13][14][15].Palmitic acid from palm oil has often been chosed as model compound to investigate the conversion pathway into hydrocarbon fuel by deoxygenation.Currently, three approaches have been used to remove oxygen from fatty acids, that is, decarboxylation (DCO 2 ), decarbonylation (DCO) and hydrodeoxygenation (HDO) as shown in Scheme 1 [3,[15][16][17][18], where the degree of H 2 consumption follows the order of DCO 2 < DCO < HDO.Among the three routes, the HDO pathway was characterized by the sequential reductions of oxygenates to a saturated hydrocarbon, with oxygen being removed as water and the number of carbon atoms being retained in the whole reaction sequence.DCO 2 and DCO pathways removed oxygen by the formation of carbon oxides, which would decrease the number of carbon atoms of the reactant.DCO 2 and DCO pathways were denoted as "carbon-loss reactions" (CLR) [19].In fact, catalytic deoxygenation of fatty acid and bio-oil has been widely investigated in the past few decades [2,[20][21][22][23].Researchers have attempted heterogeneous catalysis to convert fatty acids and their derivatives to fuel-range hydrocarbons.Murzin and co-workers reported that noble metal catalysts exhibited high activity and selectivity for fatty acid conversion [24].Pd favored CO 2 formation by direct DCO 2 , while Pt preferentially promoted the formation of CO by direct DCO [20,22,25].Ru-based catalysts were also reported to catalyze the cleavage of the C-C bond, DCO and hydrogenolysis during the deoxygenation of acids [26][27][28].Furthermore, the deoxygenation of acids over Pd, Pt, Rh, Ru and Ni catalysts mainly proceeded through DCO or DCO 2 with the order of catalytic activity: Pd > Ru > Pt > Rh > Ni [29].
Due to the high price of noble metal catalysts, many researchers have investigated bio-based oil deoxygenation using low-cost catalysts such as Ni, Mo, W, Co-based catalysts from an economic viewpoint [19,[30][31][32][33][34].Peng et al. [25,35] reported that the hydrogenation of the carboxylic group of fatty acid led to the formation of aldehyde catalyzed either solely by metallic Ni or synergistically by Ni and ZrO 2 via ketene as intermediate on Ni/ZrO 2 catalyst, and followed by DCO of octadecanal to the n-heptadecane and carbon monoxide.
Ni and Ni 3 S 2 catalysts resulted in DCO 2 hydrocarbon products, while HDO hydrocarbon products formed over Mo, MoS 2 catalysts [19,34,36,37].Both CoMo and NiMo sulfide catalysts were efficient for deoxygenation by dehydration and DCO reactions [36,37].NiW catalyst showed a greater tendency towards DCO 2 and DCO compared to NiMo, CoMo and CoMo sulfided catalysts [36,38].Nevertheless, the sulfide catalysts suffered from deactivation due to sulfur loss, which made researchers begin to focus on non-sulfide catalysts [39,40].Transition metal phosphides exhibited particular deoxygenation activity, which were considered as a viable substitute to sulfide catalysts for HDO [2,41].It was reported that the catalyst activity followed the order: Ni 2 P > MoP > CoP-Co 2 P > WP > Fe 2 P-FeP, and the main deoxygenation route was the DCO pathway over metallic Ni, as well as Ni, Co and Fe phosphides, whereas HDO was the main deoxygenation pathway on MoP and WP [30][31][32].In comparison to the conventional NiMo/γ-Al 2 O 3 catalyst, Ni 2 P/SiO 2 had higher activities and showed a promising application for HDO [32,33] The aim of this work is to study the deoxygenation mechanism of palmitic acid on Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts with an emphasis on their different performance.To better understand the hydrodeoxgenation mechanism of palmitic acid, the reactivity of the main reaction intermediates, such as hexadecanol and hexadecanal, were also comparatively studied for the above catalysts.To eliminate the potential interference of a support, SiO 2 was selected as the support due to its extremely stable physical structure and inert response to the fatty acid hydrodeoxygentaion reaction.

Ni and P Elemental Analysis
The nickel and phosphorus contents for reduced catalysts determined by inductively coupled plasma (ICP) atomic emission spectroscopy(AES) are shown in Table 1.Ni 2 P formed on the SiO 2 and only 10.19% nickel loaded on the SiO 2 , which was the lowest among those three catalysts.Ni 12 P 5 appeared on the SiO 2 with 18.84% nickel loaded.It was noted that the actual nickel loading was in agreement with the controlled loading for pure Ni/SiO 2 catalyst.In this work, the nickel controlled loading was fixed at 24.20% in all catalysts based on our previous study [6].In fact, the actual nickel loading on SiO 2 was less than the controlled loading and the mass of P was excessive for Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts.From the above results, we might think that the mass of P might impact the actual mass of Ni loaded on the SiO 2 .The X-ray photoelectron spectroscopy (XPS) spectra in the Ni (2p) regions for the Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts are shown in Figure 2a and the XPS spectra in the P (2p) regions for the Ni 12 P 5 /SiO 2 , Ni 2 P/SiO 2 catalysts are shown in Figure 2b.All spectra were decomposed by considering the spin-orbit splitting of the Ni (2p3/2) and Ni (2p1/2) lines (17 eV) and the shake-up peaks at approximately 5 eV higher than the binding energy of the parent signal.For the Ni/SiO 2 catalyst, the peak at approximate 852.3 eV was assigned to Ni metal.For the Ni 12 P 5 /SiO 2 catalyst, the peaks at approximate 856.3 eV and 133.8 eV were assigned to Ni 2+ and P 5+ species in PO 4 3− , respectively [42].The peaks observed at 852.6 eV and 129.4 eV were attributed to reduced Ni δ+ (0 < δ < 2) and P species, respectively [43].As for the Ni 2 P/SiO 2 catalyst, the Ni (2p3/2) peaks at 852.8 eV was assigned to Ni δ+ in Ni 2 P (0 < δ < 2) [43].The magnitude of δ followed the sequence: Ni 2 P > N 12 P 5 .The above results were in good agreement with the shift of binding energies detected in references for Ni 2 P and Ni 12 P 5 [42,44].

Textural Properties
The isotherms of N 2 physisorption and pore distribution of the Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts and the support SiO 2 are presented in Figure 3, while the Brunauer-Emmett-Teller (BET) surface areas and other parameters are summarized in Table 2.All the samples possessed type IV sorption isotherms, as shown in Figure 3a.SiO 2 support was mesoporous with pore diameters of approximate 11.0 nm.Both the BET surface areas and the pore volumes of the catalysts decreased compared to SiO 2 support, whereas the average pore diameter of the Ni/SiO 2 and Ni 12 P 5 /SiO 2 catalysts were similar to that of the SiO 2 support.However, the Ni 2 P/SiO 2 catalyst suffered a severe loss of surface area and pore volumes.It was found that the small pores (d < 10 nm) of Ni 2 P/SiO 2 disappeared in Figure 3b.Combined with the P mass fraction from XPS in Table 3 and the high peak of P 5+ species in PO 4 3− observed from the XPS spectra in Figure 2b, we speculated that the deposition of phosphorous excess blocked the small pores, which hampered the access of N 2 molecules [45][46][47].Nonetheless, excess phosphorous was reported to be necessary for the formation of Ni 2 P and maintained Ni 2 P particles fully phosphided during reaction progress [42].The H 2 chemisorption capacities for the reduced catalysts provide the number of surface nickel atoms and an estimate of the active sites.According to the H uptakes at the room temperature for the nickel phosphide catalysts, the exposed metal sites and the active sites are estimated in Table 2.The metal dispersion followed the sequence of Ni 12 P 5 /SiO 2 > Ni/SiO 2 > Ni 2 P/SiO 2 , and the metal site concentration also followed this sequence.

Transmission Electron Microscopy (TEM)
The low-resolution images revealed the dispersion of the Ni, Ni 12 P 5 and Ni 2 P particles on the SiO 2 supports, as shown in Figure 4.For the Ni/SiO 2 and Ni 2 P/SiO 2 catalysts, the Ni and Ni 2 P crystallites exhibited poor dispersion.In contrast, the Ni 12 P 5 /SiO 2 catalyst exhibited a spherical shape with good dispersion on SiO 2 support, presumably due to the presence of Ni 12 P 5 [48].The above results suggested that Ni

Infrared Spectroscopy of Pyridine Adsorption (Py-IR)
The surface acidity of the phosphide catalysts were determined by infrared spectroscopy of pyridine adsorption.The infrared spectra in the region of 1700~1400 cm −1 are shown in Figure 5.
According to the literatures [30,51], the bands at about 1446, 1490, 1577 and 1607 cm −1 were attributed to the vibrations of pyridine adsorbed on Lewis acid sites, and the bands at about 1544 and 1638 cm −1 corresponded to the vibrations of pyridine adsorbed on Brönsted acid sites.In addition, the band at 1595 cm −1 was ascribed to the vibration of hydrogen-bonded pyridine [52].For the Ni/SiO 2 catalyst, the adsorption of pyridine gave rise to the bands at about 1446, 1490, 1577 and 1607 cm −1 attributed to Lewis acidity, which was in agreement with the literature [30].Meanwhile it was found that only Lewis acidity existed on the Ni 12 P 5 /SiO 2 catalyst, as shown in Figure 5 [48].In contrast, both Lewis acidity (1446, 1490, 1577 and 1607 cm −1 ) and Brönsted acidity (1544 cm −1 and 1638 cm −1 ) were observed on the Ni 2 P/SiO 2 catalyst.Furthermore, the band at 1446 cm −1 was stronger than that at 1544 cm −1 , indicating the amount of Lewis acidic sites were larger than that of Brönsted acidic sites.The catalytic activity of the SiO 2 support, the Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts were evaluated in the deoxygenation reaction of palmitic acid in heptane.The gas chromtaograph (GC) of products, and the activity results of the SiO 2 support and the above catalysts are shown in Figures 6 and 7, respectively.As indicated in Figure 6, only the chromatograph peak of palmitic acid was observed on SiO 2 support, showing that the palmitic acid could not be converted into hydrocarbons on SiO 2 support, which suggested that the SiO 2 support was inert for the palmitic acid conversion.On the Ni 2 P/SiO 2 catalyst, palmitic acid could be partly converted into hydrocarbons, whereas palmitic acid could be completely converted into hydrocarbons and other products on the Ni/SiO 2 and Ni 12 P 5 /SiO 2 catalysts.To explore the deoxygenation pathway of palmitic acid over different catalysts, the evolution of several main products over palmitic acid conversion was plotted as a function of time in Figure 7. On the Ni/SiO 2 catalyst, the highest yield to C15 (about 69.1%) was obtained, as shown in Figure 7a, while the yield of lighter alkane C12-C14 hydrocarbons increased with reaction time and finally reached 16.2%.Furthermore, the conversion of palmitic acid on Ni/SiO 2 produced about 13.3% other substances, mainly including alkanes of less than 12 carbon atoms, and no branched paraffin was produced.The yield of C16 hydrocarbons was only about 1.4%, which indicated that C16 was the minor side product.The above results displayed that Ni/SiO 2 was favorable for the cracking reaction and CLR [30].During the first 2 h of reaction, a trace amount of hexadecanol was observed, which then decreased to zero after 2 h reaction.This result was in good agreement with the literature [25].Figure 7b showed the selectivity of C15 was about 78.6% at the first hour of reaction, and then it slightly decreased to 71.1% with prolonged reaction time.The selectivity of C12-C14 slightly increased during the reaction and finally reached 27.3%.It is worth noting that the selectivity of C16 and hexadecanol were very low.The gas phase products were collected and analyzed by GC to provide additional information on the reaction.The major gas products were CH 4 , C 2 H 6 , C 3 H 8 , C 4 H 10 and CO 2 , whereas CO was not observed.The quantity of CH 4 was the largest among the gas products.The CH 4 formation might be caused by methanation of CO 2 occurring (CO 2 + 3H 2 CH 4 + H 2 O) on the Ni/SiO 2 catalyst [15].Thus, we speculated that DCO 2 was the main palmitic acid conversion pathway, since CO 2 could still be observed in the gas products.For the Ni 12 P 5 /SiO 2 catalyst, the catalytic performance is given in Figure 7c,d.The palmitic acid conversion reached 100% after 6 h, and the main products were C15 and C16 hydrocarbons with about 59.8% yield for C15, and about 33.7% yield for C16.That is to say, the C15 yield was about twofold that of the C16 yield after the reaction finished.Besides, 0.5% C12-C14 hydrocarbons and about 5.8% other products were also detected, while only a small amount of hexadecanol was observed, and no branched paraffin was produced.The selectivity of C15 was maintained at about 60% during the reaction, while the C16 selectivity increased from 15.1% to 37.2%.It was noticed that the selectivity of C12-C14 and hexadecanol were very low, as shown in Figure 7d.The major component of gas products was CO accompanied by a small amount of CH 4 , C 2 H 6 and C 3 H 8 .However, CO 2 was not observed in the gas products.The above product distribution suggested that the main conversion ways of palmitic acid were the HDO and CLR reactions.However, the competition reactions for the C16 and C15 formation existed; that is, there would be two deoxygenation routes.Based on the above results, we could speculate that the main deoxygenation routes of palmitic acid were the DCO and HDO pathways with the intermediate hexadecanol.In contrast, the conversion of palmitic acid on Ni 2 P/SiO 2 was low (about 55.2%), and the main products were hydrocarbons with 32.2% C15, 9.2% C16 and 1.8% C12-C14 as well as a small amount of branched alkanes, as shown in Figure 7e.During the palmitic acid conversion process, a small amount of hexadecanal was observed, whereas hexadecanol was not found.The selectivity of C15 increased from 53.2% to 62.7% and the selectivity of C16 increased from 8.6% to 25.8%, whereas the selectivity of C12-C14 first increased and then decreased to 4.6%; however, the selectivity of hexadecanal decreased monotonically with time, as shown in Figure 7f.The major gas product was CO accompanied by a small amount of CH 4 and C 2 H 6 , while CO 2 could not be observed.The above product distribution disclosed that the catalytic performance of Ni 2 P/SiO 2 was poor compared to Ni/SiO 2 and Ni 12 P 5 /SiO 2 .The main deoxygenation pathway of palmitic acid might be CLR reactions.The yield of C16 (9.2%) over Ni 2 P/SiO 2 was less than that of C16 (33.7%) over Ni 12 P 5 /SiO 2 and much more than that of C16 (1.4%) over Ni/SiO 2 , which suggested that the HDO route was following the sequence of Ni 12 P 5 /SiO 2 > Ni 2 P/SiO 2 > Ni/SiO 2 .The Brönsted acid sites on the Ni 2 P/SiO 2 catalyst might be responsible for the formation of branched alkanes under such reaction conditions and Lewis acid sites were beneficial to the hydrocracking reaction [48].
The distinct product distributions implied the different catalytic mechanisms of the deoxygenation reactions.The different Ni species would induce different deoxygenation ways.Combined with the TOF results in Table 2, it was found that the TOF followed the sequence of Ni 12 P 5 > Ni > Ni 2 P. The high TOF on Ni 12 P 5 /SiO 2 depended on the high-index facets of Ni 12 P 5 and the higher metal dispersion.The Lewis acid sites that existed on the three catalysts contributed to the formation of the CLR reaction products, while the Brönsted acid sites that existed on Ni 2 P/SiO 2 were beneficial to the formation of the isomerization products.To further explore the deoxygenation mechanism of palmitic acid, control experiments using the feasible intermediate products, that is, hexadecanol and hexadecanal, as the reactants were carried out.The initial reaction rate based on palmitic acid conversion was obtained from the above activity data and compared with those of hexadecanal and hexadecanol conversion.
For the Ni/SiO 2 catalyst, the initial hexadecanal consumption rate (6.2 mmol•g −1 •h −1 ) was higher than that of both palmitic acid (4.0 mmol•g −1 •h −1 ) and hexadecanol (4.5 mmol•g −1 •h −1 ), as shown in Table 4.The distribution of hydrocarbons using sole hexadecanal, or hexadecanol, as well as the 1:1 ratio mixture of hexadecanal and hexadecanol as reactants, is shown in Figure 8.The hexadecanol conversion over the Ni/SiO 2 catalyst is shown in Figure 8a; the hexadecanol conversion in the presence of H 2 yielded mainly C15 accompanied by trace amounts of C14 and C16; the gas products were CH 4 accompanying with a trace amount of C 2 H 6 ; however, CO and CO 2 were not observed.Peng et al. reported that C-C bond cleavage for alcohols with terminal hydroxyl groups occured via DCO of aldehydes (generated by dehydrogenation of alcohols) over Pt/Al 2 O 3 catalysts [53,54], so we considered that the hexadecanol was mainly converted into C15 hydrocarbon via the route of DCO over the Ni/SiO 2 catalyst, and CH 4 might be formed via the methanation of CO.As can be seen from Figure 8b, the hexadecanal conversion also yielded mainly C15 accompanied by trace amounts of C14 and C16, and the C15 yield obviously increased in the first 1.5 h; hexadecanol was also observed, and its yield decreased quickly from 20% to 0.5% within 2 h, and then became zero with the reaction time extending.Obviously, the sum amount of C15 and the intermediate hexadecanol yield was approximately equal to the conversion of hexadecanal, which suggested that the conversion of hexadecanal might proceed via the intermediate hexadecanol, which then formed the C15 by DCO on the Ni/SiO 2 catalyst.The trace amount of C16 was from the dehydration of the intermediate hexadecanol, since there was also a trace amount of C16 formation in the separated study using hexadecanol as reactant.The only gas product was CH 4 ; while CO and CO 2 were not observed, therefore the CH 4 formation might be caused by the methanation of CO.The results demonstrated that the product distribution from hexadecanal was similar to that from hexadecanol.As shown in Figure 8c, when the hexadecanal and hexadecanol mixture in a 1:1 ratio were chosen as the reactants, it was found that the hexadecanal conversion was higher than the hexadecanol in the first 2 h and, finally, their conversion reached 100%; and the initial consumption of the hexadecanal (6.2 mmol•g −1 •h −1 ) was higher than that of the hexadecanol (4.5 mmol•g −1 •h −1 ), which suggested that the Ni/SiO 2 catalyst was beneficial for the hexadecanal conversion compared with the hexadecanol; the main product C15 was about 90% after the reaction finished, which also suggested that the hexadecanal and hexadecanol were mainly converted into C15 accompanied by CO released via DCO; the gas CO was converted into CH 4 by the methanation reaction, consistent with the results of the separate study of sole hexadecanal or hexadecanol.
Nevertheless, we found that C15 was the main product accompanied by trace amounts of C12-C14 and C16 all the time of the reaction process when palmitic acid was used as reactant, as shown in Figure 7.It was interesting that the palmitic acid conversion tendency was the same as the yield of C15.Furthermore, the ratio of the main products, C15/C16, from the palmitic acid, hexadecanol, hexadecanal and hexadecanal/hexadecanol (1:1 weight ratio) conversion are shown in Figure 9.It can be observed that the amount of C15 was much higher than that of C16, varying from 45-to 70-fold, while the C15/C16 ratio from palmitic acid conversion was the lowest among those reactions.If the palmitic acid conversion into alkanes via hexadecanal and/or hexadecanol were the main deoxygenation pathway on Ni/SiO 2 , we would get a higher C15/C16 ratio of hydrocarbons.However, the C15/C16 ratio was the lowest among the reactions using different starting materials.Therefore, we reasonability speculated that there was another reaction route for the formation of C15.Combined with the gas products from hexadecanal and hexadecanol conversion, CH 4 was main gas product and might be from CO by methanation [3], whereas the main gas products from palmitic acid conversion were CH 4 , C 2 H 6 and CO 2 .So, we can conclude that the formation of C15 might be from the C-C cleavage of palmitic acid accompanying the CO 2 release, which was the main deoxygenation route for the palmitic acid conversion into hydrocarbon over the Ni/SiO 2 catalyst.Furthermore, the trace amount of intermediate hexadecanol was detected in the first 2 h of palmitic acid conversion on Ni/SiO 2 , which finally converted into C15 by DCO.Because the hexadecanol was almost completely converted into C15 according to the above results, there was a minor deoxygenation route via the intermediate hexadecanol, most of which converted into C15 via DCO, while the remaining hexadecanol produced the trace amount of C16 via HDO.Consequently, we can conclude that the main deoxygenation pathways of palmitic acid on the Ni/SiO 2 catalyst were the C-C cleavage via DCO 2 with minor intermediate hexadecanol formation, and then the hexadecanol further converted into alkane via DCO and HDO.On the Ni 12 P 5 /SiO 2 catalyst, the initial rate of palmitic acid consumption (2.6 mmol•g −1 •h −1 ) was less than that of the hexadecanal (4.1 mmol•g −1 •h −1 ) and the hexadecanol (3.7 mmol•g −1 •h −1 ).The hexadecanol conversion over Ni 12 P 5 /SiO 2 catalyst is shown in Figure 10a.The conversion of hexadecanol yielded C15 and C16, while the C16 yield was much higher than that of the C15 yield after the reaction finished, and the gas products were also mainly CO accompanied by a small amount of CH 4 and C 2 H 6 .The C16 was formed by the hexadecanol conversion via HDO, whereas the C15 was formed by the hexadecanol conversion via DCO, which suggested that the hexadecanol conversion over the Ni 12 P 5 /SiO 2 catalyst was mainly through the HDO route.The conversion of hexadecanal yielded C15 and C16 accompanying the hexadecanol formation, and the C15 yield was slightly higher than that of C16 within the 2.5 h reaction, as shown in Figure 10b, whereas the amount of hexadecanol increased first and then decreased; but the C16 yield was finally slightly higher than that of C15 after the reaction finished, and the gas products were mainly CO accompaned by a small amount of CH 4 and C 2 H 6 , which suggested that the hexadecanal conversion over the Ni 12 P 5 /SiO 2 catalyst was via both HDO and direct DCO.At the beginning of the reaction, the direct DCO was the main deoxygenation route because the amount of C15 was higher than that of C16.Prolonging the reaction time, the amount of C16 increased fast since the intermediate hexadecanol converted into C16 via HDO, which was consistent with the above results of the hexadecanol conversion.So, we can conclude that the hexadecanal conversion was through two routes over the Ni 12 P 5 /SiO 2 catalyst, that is, the HDO and direct DCO of hexadecanal, which proceeded competitively.When the hexadecanal and hexadecanol mixture of 1:1 ratio were chosen as the reactants, as shown in Figure 10c, it was found that the hexadecanal conversion was higher than the hexadecanol, which also suggested that the Ni 12 P 5 /SiO 2 catalyst was beneficial for the hexadecanal conversion compared with the hexadecanol, which agreed with the fact that the comsuption rate of hexadecanal (4.1 mmol•g −1 •h −1 ) was slightly higher than that of the hexadecanol (3.7 mmol•g −1 •h −1 ).The C16 yield was higher than that of C15 after the reaction finished, which was caused by the fact that the main product was both C16 in the separate study of sole hexadecanol or hexadecanal.The gas products from hexadecanal and hexadecanol conversion were mainly CH 4 and CO, while the main gas products from palmitic acid conversion were mainly CH 4 , C 2 H 6 and CO.Furthermore, the separate study of the intermediate hexadecanal showed that the amount of C15 was slightly less than that of C16, and the separate study of hexadecanol discovered that the amount of C15 was half of that of C16.In fact, the C15/C16 ratio was approximately twofold during the palmitic acid conversion process, which implied that the main route of C15 formation might be from the C-C cleavage via DCO of palmitic acid.Besides, the C15/C16 ratio of hydrocarbons from the palmitic acid conversion was the highest among all the reactions on Ni 12 P 5 /SiO 2 , as shown in Figure 9.If the palmitic acid conversion into alkanes via hexadecanal and/or hexadecanol were the main deoxygenation pathway on Ni 12 P 5 /SiO 2 , we would get the lower C15/C16 ratio of hydrocarbons.Accordingly, there was still a C-C cleavage of palmitic acid accompanying the CO and H 2 O release.On the other hand, hexadecanol was observed, while hexadecanal could not be detected during the palmitic acid conversion on Ni 12 P 5 /SiO 2 , which could be accounted for by the consumption rate of hexadecanal being faster than that of hexadecanol.Therefore, for the Ni 12 P 5 /SiO 2 catalyst, we can confirm that there were two main deoxygenation pathways of palmitic acid: one pathway was the C-C cleavage via direct DCO of palmitic acid to form C15 alkane; and the other pathway was HDO of palmitic acid directly generating hexadecanol; the subsequent dehydration of hexadecanol yielded C16 hydrocarbon via HDO, and those two routes proceeded competitively over the Ni 12 P 5 /SiO 2 catalyst.
For the Ni 2 P/SiO 2 catalyst, the initial consumption rate of hexadecanal and hexadecanol were both 1.9 mmol•g −1 •h −1 .The hexadecanol conversion on Ni 2 P/SiO 2 resulted in C16 alkane accounting for approximately 80% of the products, whereas the C15 alkane was observed as a minor product as shown in Figure 11a.The gas products were also mainly CH 4 and C 2 H 6 , which suggested that the conversion of hexadecanol was mainly through the route of HDO on the Ni 2 P/SiO 2 catalyst.The hexadecanal conversion was lower, as seen in Figure 11b.The C15 yield was higher than that of C16, and hexadecanol was also obviously observed.The gas products were mainly CH 4 and C 2 H 6 , while the CH 4 formation might be caused by methanation of CO.From the above results, we can conclude that the hexadecanal conversion was mainly through the direct DCO pathway, while the HDO route via the intermediate hexadecanol also existed, which seemed to be the subordinate deoxygenation route.When the hexadecanal and hexadecanol mixture of 1:1 ratio were chosen as the reactants, the hexadecanol conversion was higher than that of the hexadecanal conversion, as shown in Figure 11c.The main product was C16.The intermediate hexadecanol could not be completely converted after 3 hours' reaction in the separate study of hexadecanal, and the product was mainly C16 in the separate study solely of hexadecanol.The gas products were detected from the separate study of hexadecanal and it was found that CH 4 was the main gas product; which might be from CO transformation by the methanation reaction [3].The C15/C16 ratio of hydrocarbons from the palmitic acid conversion was the highest among the reactions using different starting materials on Ni 2 P/SiO 2, as shown in Figure 9.If the palmitic acid conversion into alkanes via hexadecanal and/or hexadecanol were the main deoxygenation pathways on Ni 2 P/SiO 2 , we would get a low C15/C16 ratio of hydrocarbons.This tendency was similar to that of the Ni 12 P 5 /SiO 2 catalyst.Accordingly, there was still a C-C cleavage of palmitic acid accompanying the CO and H 2 O release.Combined with the above results, we inferred that the C-C cleavage of palmitic acid could also occur on Ni 2 P/SiO 2 .On the other hand, the intermediate hexadecanal was observed in the palmitic acid conversion experiment.Thus, combined with the gas and liquid products, we can conclude that there were two main deoxygenation pathways of palmitic acid on the Ni 2 P/SiO 2 catalyst: one pathway was the C-C cleavage of palmitic acid via DCO to form C15 alkane; and the other pathway was HDO of palmitic acid to generate hexadecanal, which subsequently directly released CO to form C15 alkane.Based on a little amount of C16 alkane formed, we speculated that a side reaction was the palmitic acid conversion into C16 alkane by the HDO pathway.

The Kinetic Parameters
The rate constants for the consumption of palmitic acid, hexadecanal and hexadecanol on the Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts are summarized in Table 5.According to previous reports [55,56], the deoxygenation of palmitic acid, hexadecanal and hexadecanol were considered a first order reaction.The obtained rate constants are not very sensitive to reaction materials on the Ni/SiO 2 and Ni 12 P 5 /SiO 2 catalysts.The fact that k (palmitic acid) > k (hexadecanal) ≈ k (hexadecanol) suggested that the palmitic acid conversion was faster than hexadecanol and hexadecanal on Ni/SiO 2 and Ni 12 P 5 /SiO 2 .Hexadecanol was detected during the palmitic acid conversion progress, while hexadecanal was not observed on the Ni/SiO 2 and Ni 12 P 5 /SiO 2 catalysts.In addition, the rate constant of hexadecanal conversion (k) to products was equal to that of hexadecanol conversion (k), which was perhaps why only hexadecanol was obversed as the intermediate product on Ni 12 P 5 /SiO 2 during the palmitic acid conversion.On the contrary, the fact that k (palmitic acid) ≈ k (hexadecanal) < k (hexadecanol) on the Ni 2 P/SiO 2 catalyst suggested that the conversion of palmitic acid and hexadecanal were lower than that of hexadecanol.Furthermore, hexadecanal was detected during the palmitic acid conversion.So, the hydrogenation of palmitic acid to hexadecanal was prevalent, relative to other parallel reactions of the C-C cleavage via DCO and HDO.The extremely low rate constants of palmitic acid and hexadecanal on Ni 2 P/SiO 2 demonstrate the poor catalytic performance of palmitic acid and hexadecanal conversion.According to the kinetic experiments, the apparent activation energy of and pre-exponential factor for the palmitic acid conversion were obtained and are listed in Table 5.The activation energy followed the sequence: Ea (Ni 2 P/SiO 2 ) < Ea (Ni/SiO 2 ) < Ea (Ni 12 P 5 /SiO 2 ), it was apparent that the pre-exponential factor followed the same sequence.However, the activity experiments results from TOF disclosed that Ni 12 P 5 /SiO 2 was exhibited excellently, while Ni 2 P/SiO 2 displayed a poor performance.Furthermore, the catalyst Ni 12 P 5 index facet was (400) according to the HRTEM results.We speculated that the high-index facets might influence for the catalysis activity for the conversion of palmitic acid.Combined with the above results, we concluded that the excellent catalytic performance of Ni 12 P 5 /SiO 2 was ascribed to high-index facets, high metal dispersion and high pre-exponential, whereas the poor catalytic performance of Ni 2 P/SiO 2 was attributed to poor metal dispersion and the big crystallite size.

Summary of Reaction Pathway
Combination of the reaction of palmitic acid, hexadecanal and hexadecanol allowed us to formulate the overall reaction pathways for the conversion of palmitic acid on the Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 .
The process of the main deoxygenation pathways of palmitic acid on the Ni/SiO 2 catalyst was the C-C cleavage to form the main product of C15 and CO 2 in Scheme 2a [19].Another side reaction was palmitic acid converted into hexadecanol via HDO.After the formation of hexadecanol, two parallel reactions occurred, that is, the majority of hexadecanol was converted into C15 via DCO and a small amount of hexadecanol was converted into C16 by HDO catalyzed by metallic Ni sites.
The Ni 12 P 5 /SiO 2 catalyst has DCO and HDO activity at 543 K under H 2 .Therefore, we can confirm that there were two main competition deoxygenation pathways of palmitic acid on the Ni 12 P 5 /SiO 2 catalyst.One pathway was the C-C cleavage via DCO of palmitic acid to form C15 alkanes (Step 1, Scheme 2b), and the other pathway was HDO of palmitic acid to generate hexadecanol as the intermediate, and then the hexadecanol converted into C16 via hydrodeoxygeantion.Furthermore, C16 selectivity was higher compared with other reactions.Apparently, hexadecanol via HDO was favored for C16 formation over Ni 12 P 5 /SiO 2 .
On Ni 2 P/SiO 2 , the palmitic acid conversion had two main reaction pathways.One pathway was the direct C-C cleavage via DCO of the palmitic acid to form C15 alkane (Step 1, Scheme 2c), and the other pathway was that palmitic acid was firstly converted into hexadecanal, and then hexadecanal released CO to form C15 (Step 2, Scheme 2c).In addition, the hexadecanal of HDO was inhibited and C16 selectivity was markedly low.

Catalyst Synthesis
The Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts were synthesized by incipient wetness impregnation followed by temperature programmed reduction (TPR).The controlled Ni content calculated using Equation (1) for all catalysts was 24.20 wt %, and the content of P was adjusted by the molar ratios of Ni/P.The controlled Ni/P molar ratios of 0.5/1 and 1.5/1 led to the formation of Ni 2 P and Ni 12 P 5 catalysts, respectively.Before impregnation, the silica support was sieved to 40-60 mesh and dried at 873 K for 4 h in ambient air.The dried silica support was first impregnated with a mixture aqueous solution of Ni(NO 3 ) 2 •6H 2 O and NH 4 H 2 PO 4 or only Ni(NO 3 ) 2 •6H 2 O, followed by ultrasonic treatment for 4 h, and the impregnation lasted for 24 h.Then, the catalyst precursor was dried in an oil bath at 353 K, and followed by drying at 383 K overnight and calcination at 873 K for 4 h in ambient air.Thereafter, the precursor sample was reduced from room temperature to 873 K at a rate of 2 K•min −1 and maintained at 873K for 2 h under flowing H 2 (99.999%, 30 mL•min −1 ), and then cooled to room temperature under flowing N 2 to obtain the active nickel phosphide catalysts and Ni catalyst.

Catalyst Characterization
The BET specific surface area was determined by nitrogen adsorption-desorption at 77 K using a Micromeritics Tristar II 3020 instrument.Before measurement, the samples were activated in vacuum at 393 K and 573 K for 2 h respectively.The pore volumes and average pore diameters were determined by the Barrett-Joyner-Halenda (BJH) method from the desorption branches of the isotherms.
XRD patterns were recorded using a SHIMADZU XRD-6100 diffractometer equipped with Cu-Kα monochromatized radiation generated at 40 kV and 30 mA.Powder diffraction patterns were recorded over the 2Θ range of 30-80 • at a scan rate of 1 XPS experiments were performed on an AXIS Ultra DLD (KRATOS) spectrometer using a pass energy of 50 eV (0.1 eV per step) with a monochromatic Al-Kα X-ray source.Binding energy (BE) of various elements have been referenced to the C1s at 284.6 eV.A Shirley background was subtracted from all spectra and peak fitting was performed using an 80/20 Lorentz-Gauss function.
The actual Ni and P loadings were determined by atomic absorption spectroscopy using a VG PQExCell.Prior to measurement, the samples were dissolved in a mixture of hydrofluoric acid (48%) and nitric acid.
TEM images of the samples were obtained on an FEI Tecnai G2 20 TWIN instrument at an acceleration voltage of 200 kV.
Hydrogen chemisorption and H 2 -TPD measurements were performed using the Micromeritics Autochem II 2920 instrument.Approximately 0.1 g of catalyst was flashed firstly in a 10 mL•min −1 flow of Ar at room temperature for 2 h.The catalysts were activated at 673 K in a flow of Ar and hold for 1 h at 50 mL min −1 , then cooled down to 273 K.The H 2 chemisorption was measured by injecting pulses of 10% vol % H 2 -Ar flow (50 mL•min −1 ) at 3 min intervals until adsorption saturation at 273 K. Next, the samples were heated from 273 K to 325 K and held for 1 h in a flow of Ar at 50 mL•min −1 , the temperature was then increased to 873 K at the rate of 10 K•min −1 to obtain the desorption curve, and the hydrogen adsorption volume capacity, denoted as V(H 2 ), was obtained by manual integration of the desorption curve.The mole active sites were calculated according to Equation (2): n (the mole of Ni active sites) = 2n(H 2 ) = PV(H 2 )/RT (2) A Ni atom was assumed to adsorb one H atom, therefore, the mole of Ni active sites should be equal to 2n(H 2 ).We have revised in the manuscript.
The acid sites of catalysts were determined by infrared (IR) spectroscopy with pyridine as probe molecule.The IR spectrum was measured with a BRUKER VERTEX 70 spectrometer with a resolution of 4 cm −1 (number of scans, 16).The samples were activated at 673 K for 2 h in vacuum, and a background spectrum was recorded when the temperature was decreased to 303 K.The activated samples were exposed to pyridine vapor at 303 K for 0.5 h in vacuum.After pyridine adsorption finished, the spectrum was recorded.

Measurement of Catalytic Reaction
We selected saturated plamitic acid as the model compound.The Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts were tested.The typical experiment was carried out as follows: palmitic acid reactant (1.0 g), heptane solvent (100 mL), and catalyst (5 g•L −1 ) were charged into a batch autoclave (Parr Instrument , 300 mL, Moline, Illinois, USA,).Before starting the reaction, N 2 was charged into the batch autoclave several times to replace air, and then pure H 2 was charged to 1.2 MPa at room temperature.The reaction was carried out at 543 K with stirring at 600 rpm for 6 h.During the reaction, the liquid products were sampled at intervals of one hour and analyzed by GC when the reaction temperature reached 543 K.The separated experiments with hexadecanal, hexadecanal as reactants were carried out respectively for 3 h under the same reaction conditions except that the liquid products were analyzed at the intervals of half an hour.

Product Analysis
The liquid products were quantitatively analyzed using a gas chromatograph (GC, PANNA A91) equipped with a HP-5 capillary column (30 m × 0.25 mm × 0.25 µm) and a flame ionization detector (FID).N 2 was used as the carrier gas.The internal standard (i.e., 2-Ethylbutyric acid) was used for quantification.Both injection and detection temperatures were 593 K.The temperature program was set as: 343 K (3 min), 5 K•min −1 , 433 K, 10 K•min −1 , 553 K (10 min).The gas products were analyzed using a gas chromatograph (GC, PANNA A91) equipped with a EC-1 (30 m × 0.53 mm × 0.5 µm) capillary column and FID, PQ column and 5A column with thermal conductivity detector (TCD).The yield (Yield) was calculated by Equation (3).Yield = (W product /M product )/(W reactant /M reactant ) × 100%. ( where W product and W reactant are the weights of the product and reactant, and M product and M reactant are the molecular weight of the product and reactant, respectively.

Conclusions
Ni 12 P 5 /SiO 2 showed higher catalytic performance compared with Ni/SiO 2 and Ni 2 P/SiO 2 due to its higher metal dispersion, high-index crystal and high pre-exponential.The Ni/SiO 2 and Ni 12 P 5 /SiO 2 catalysts were more favourable for the hexadecanal conversion than for hexadecanol.In contrast, Ni 2 P/SiO 2 was benefited for hexadecanol conversion compared with hexadecanal.The nature of Ni-species affected the activity and reaction pathway.Ni/SiO 2 mainly catalyzed DCO 2 of palmitic acid to product diesel-range hydrocarbon (C15) accompanied with the side reaction of the C-C cleavage to produce diesel-range hydrocarbon C12-C14.Ni 12 P 5 /SiO 2 exhibits DCO and HDO activity from palmitic acid, and there are competition deoxygenation pathways.One is a direct DCO pathway to yield C15, and the other is the plamitic acid conversion via hexadecanol as the intermediate product that is then converted into C16.In contrast, the palmitic acid deoxygenation on Ni 2 P/SiO 2 occurs easily via direct DCO to form C15 and HDO to the product hexadecanal, and then the hexadecanal is converted into C15 by releasing CO.
. However, nickle phosphides have different crystal forms, such as Ni 12 P 5 , Ni 2 P and so on, which might exhibit different catalytic performance in the conversion of bio-based oil.Although several research works have used Ni 2 P catalyst for the conversion of palmitic oil to alkane, the detailed mechanisms have not yet been clearly established; in particular, the possibly different performance between Ni 12 P 5 and Ni 2 P catalysts needs more attention.

Figure 1
Figure 1 presents the X-ray diffraction (XRD) patterns of the Ni/SiO 2 , Ni 2 P/SiO 2 and Ni 12 P 5 /SiO 2 catalysts.For the Ni/SiO 2 catalyst, the three obvious diffraction peaks at 44.4 • , 51.7 • and 76.4 • (PDF#04-0850) were attributed to the metal Ni.For the Ni 12 P 5 /SiO 2 catalyst, the XRD patterns showed only the diffraction peaks at 38.3 • , 41.6 • , 44.3 • , 46.8 • and 48.8 • assigned to Ni 12 P 5 (PDF#22-1190).For the Ni 2 P/SiO 2 catalyst, the sharp diffraction peaks at 40.7 • , 44.5 • , 47.2 • and 54.8 • were assigned to Ni 2 P (PDF#03-0953).The crystallite sizes of Ni, Ni 2 P, Ni 12 P 5 were calculated using the Scherrer Equation, and the results are shown in Table 2. Accordingly, the average crystallite sizes of Ni, Ni 2 P and Ni 12 P 5 were 15.6, 36.7 and 17.3 nm, respectively.The crystallite size of Ni 2 P was the biggest among these catalysts, which suggested that crystallite Ni 2 P easily aggregated into bigger particles compared to Ni and Ni 12 P 5 .It should be noted that no diffraction peaks assigned to Ni 2 P were observed on the Ni 12 P 5 /SiO 2 catalyst, meanwhile no diffraction peaks assigned to Ni 12 P 5 were observed on the Ni 2 P/SiO 2 catalyst, indicating the pure phase Ni 2 P and Ni 12 P 5 active component catalysts were obtained.
2 P and Ni were in favor of aggregation.The high-resolution images of Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 revealed that Ni, Ni 12 P 5 and Ni 2 P adopted a globular morphology on the SiO 2 supports.Therefore, a clear difference of crystal indices could be observed among Ni, Ni 12 P 5 and Ni 2 P in the high-resolution transmission electron microscopy (HRTEM) images.The high-resolution images showed lattice fringes having spacings of 2.03, 2.16 and 2.21 Å, which are assigned to the crystallographic planes of Ni (111), Ni 12 P 5 (400) and Ni 2 P (111), respectively.The observed d-spacings are consistent with the previous reports on the Ni/Al 2 O 3 [49] and Ni 2 P/SiO 2 catalysts [50].

Figure 6 .
Figure 6.Gas chromatographs (GC) of the products over different samples.

Scheme 2 .
Scheme 2. (a) The suggested main reaction pathways of palmitic acid over the Ni/SiO 2 catalyst; (b) the suggested main reaction pathways of palmitic acid over the Ni 12 P 5 /SiO 2 catalyst; (c) the suggested main reaction pathways of palmitic acid over the Ni 2 P/SiO 2 catalyst.

Table 2 .
Physicochemical and turn over frenquency (TOF) data of the samples.

Table 3 .
The Ni and P mass fraction of Ni/SiO 2 , Ni 12 P 5 /SiO 2 and Ni 2 P/SiO 2 catalysts from XPS.