Catalytic Behaviors of Supported Cu, Ni, and Co Phosphide Catalysts for Deoxygenation of Oleic Acid

Catalytic behaviors of copper phosphide supported on various oxides (SiO2, γ-Al2O3, and USY zeolite) have been evaluated for deoxygenation of oleic acid and compared with nickel and cobalt phosphides. All catalysts were prepared by the hydrogen reduction of metal phosphate precursors. CoP and Ni2P were obtained on USY zeolite, while Cu3P was formed on USY and SiO2 supports. On the contrary, the metallic Cu phase was stabilized on γ-Al2O3 support. Metal phosphide particles were highly dispersed on the USY support. Cu3P/USY exhibited much larger surface area and higher acidity compared to Cu3P/SiO2, owing to the textural and acidic properties of the USY zeolite support. All supported catalysts gave an oleic acid conversion close to 100% at 340 °C. Ni2P/USY, CoP/USY, and Cu/γ-Al2O3 favored the deoxygenation of oleic acid to alkane products such as heptadecane and octadecane. Highly selective production of octadecane (98%) through hydrodeoxygenation pathway occurred on Cu/γ-Al2O3. In contrast, the supported Cu3P catalysts favored cyclization and aromatization to form cyclic and aromatic compounds such as dodecylcyclohexane, heptylcyclopentane, and dodecylbenzene. Cu3P/SiO2 provided dodecylbenzene in higher yield (46%) than Cu3P/USY (33%).


Introduction
As oil reserves have become exhausted and the rate of fossil fuel consumption for energy continues to increase, a renewable biofuel such as the biodiesel, aliphatic ester, vegetable oil, and biomass-derived fatty acid has been receiving considerable attention due to economic, environmental, and social benefits [1][2][3]. However, renewable biofuels contain a significant amount of unsaturated bonds and oxygenated functional groups, resulting in undesirable properties, such as low energy density, high acidity, poor chemical stability, and high viscosity. Therefore, upgrading the renewable biofuel by removing the oxygen atoms and producing linear hydrocarbons, also known as bio-hydrogenated diesel (BHD) or green diesel, is required before using as a drop-in replacement for traditional petroleum fuels.

Morphology and Surface Properties
The morphologies of supported copper and metal phosphide catalysts are illustrated in Figure  2. SEM micrographs, recorded with a backscattering mode (BSE), show specimen contrast due to the variation of atomic weights. In general, the brighter areas were Cu, Cu3P, Ni2P, and CoP particles deposited on each support. Small particles of Cu3P, Ni2P, and CoP were well dispersed on USY support, while Cu3P particles on SiO2 formed agglomerates of diameters ranging from 300 nm to 1.2 μm. It seems likely that the morphology and the large surface area of USY zeolite favored the dispersion of metal phosphide particles, thus shifting the average particle size of catalysts toward smaller values. For Cu/γ-Al2O3, Cu particles tended to agglomerate on γ-Al2O3 support, and different sizes of Cu metal clusters (~100 to 800 nm) were formed.

Morphology and Surface Properties
The morphologies of supported copper and metal phosphide catalysts are illustrated in Figure 2. SEM micrographs, recorded with a backscattering mode (BSE), show specimen contrast due to the variation of atomic weights. In general, the brighter areas were Cu, Cu 3 P, Ni 2 P, and CoP particles deposited on each support. Small particles of Cu 3 P, Ni 2 P, and CoP were well dispersed on USY support, while Cu 3 P particles on SiO 2 formed agglomerates of diameters ranging from 300 nm to 1.2 µm. It seems likely that the morphology and the large surface area of USY zeolite favored the dispersion of metal phosphide particles, thus shifting the average particle size of catalysts toward smaller values. For Cu/γ-Al 2 O 3 , Cu particles tended to agglomerate on γ-Al 2 O 3 support, and different sizes of Cu metal clusters (~100 to 800 nm) were formed. N2 adsorption-desorption isotherms and Barrett, Joyner, and Halenda (BJH) pore-size distributions of supported metal and metal phosphide catalysts are presented in Figures 3 and 4, respectively. The characteristic textural properties of the catalysts are summarized in Table 1.    Table 1. N2 adsorption-desorption isotherms and Barrett, Joyner, and Halenda (BJH) pore-size distributions of supported metal and metal phosphide catalysts are presented in Figures 3 and 4, respectively. The characteristic textural properties of the catalysts are summarized in Table 1.     All catalysts and supports exhibited type IV isotherms ( Figure 3), with hysteresis loops attributing to capillary condensation in mesopores, which are typically observed for mesoporous materials. One can also notice that the isotherms of USY support, Cu3P/USY, N2P/USY, and CoP/USY catalysts, rose steeply at low relative pressure (P/P0 > 0.1), indicating the presence of micropores in the samples. USY is known to contain both micropores inherited from the zeolite structure and mesopores gained from the dealumination process which endow them with large surface area [30]. The USY zeolite support used in this study had the largest surface area when compared to SiO2 and γ-Al2O3 supports (Table 1). In general, the textural properties of the catalysts are strongly related to the properties of the corresponding supports. Each group of catalysts on the same support displayed the same type of hysteresis loop. The USY-supported catalysts exhibited type H4 hysteresis loop characteristic of narrow slit pores, while the hysteresis loops of catalysts on other supports were of type H2, which are often associated with the ink-bottle-neck-type pores. In Figure 4, the formation of Cu3P, N2P, and CoP and metallic Cu particles did not significantly alter the pore character of USY (3.2-4.0, median 3.7 nm) and γ-Al2O3 (3.7-11.7, median 7.8 nm) supports. In contrast, Cu3P/SiO2 showed a dramatic decrease of surface area and pore volume. The pore character of SiO2 (3.2-9.8, median 6.9 nm) was also suppressed after the deposition of Cu3P catalyst on SiO2 support. A plausible explanation for this observation is that the Cu3P catalyst particles could fill the pores of SiO2 support or completely block a great number of them. According to the textural properties shown in Table 1, the specific surface area followed the sequence of Ni2P/USY > Cu3P/USY > CoP/USY > Cu/γ-Al2O3 > Cu3P/SiO2. This difference in surface area could be influenced by the nature of the support material, All catalysts and supports exhibited type IV isotherms ( Figure 3), with hysteresis loops attributing to capillary condensation in mesopores, which are typically observed for mesoporous materials. One can also notice that the isotherms of USY support, Cu 3 P/USY, N 2 P/USY, and CoP/USY catalysts, rose steeply at low relative pressure (P/P 0 > 0.1), indicating the presence of micropores in the samples. USY is known to contain both micropores inherited from the zeolite structure and mesopores gained from the dealumination process which endow them with large surface area [30]. The USY zeolite support used in this study had the largest surface area when compared to SiO 2 and γ-Al 2 O 3 supports (Table 1). In general, the textural properties of the catalysts are strongly related to the properties of the corresponding supports. Each group of catalysts on the same support displayed the same type of hysteresis loop. The USY-supported catalysts exhibited type H4 hysteresis loop characteristic of narrow slit pores, while the hysteresis loops of catalysts on other supports were of type H2, which are often associated with the ink-bottle-neck-type pores. In Figure 4, the formation of Cu 3 P, N 2 P, and CoP and metallic Cu particles did not significantly alter the pore character of USY (3.2-4.0, median 3.7 nm) and γ-Al 2 O 3 (3.7-11.7, median 7.8 nm) supports. In contrast, Cu 3 P/SiO 2 showed a dramatic decrease of surface area and pore volume. The pore character of SiO 2 (3.2-9.8, median 6.9 nm) was also suppressed after the deposition of Cu 3 P catalyst on SiO 2 support. A plausible explanation for this observation is that the Cu 3 P catalyst particles could fill the pores of SiO 2 support or completely block a great number of them. According to the textural properties shown in Table 1, the specific surface area followed the sequence of Ni 2 P/USY > Cu 3 P/USY > CoP/USY > Cu/γ-Al 2 O 3 > Cu 3 P/SiO 2 . This difference in surface area could be influenced by the nature of the support material, the dispersion, and the crystallite size of catalysts. Therefore, a high dispersion of small Ni 2 P nanoparticles on USY support would be responsible for such a large surface area of Ni 2 P/USY catalyst.
The acidic properties of catalysts and supports were studied by temperature-programmed desorption of ammonia (NH 3 -TPD). The NH 3 -TPD profiles and concentrations of acid sites are presented in Figure 5 and Table 1, respectively. The number of acid sites in supports decreased in the following order: USY > γ-Al 2 O 3 > SiO 2 , and the number of acid sites in catalysts was in the order of Ni 2 P/USY > Cu/γ-Al 2 O 3 > Cu 3 P/USY > CoP/USY > Cu 3 P/SiO 2 . The strength of acid site can be distinguished by the desorption temperature of the adsorbed NH 3 (T). The low-temperature desorption (T < 250 • C) is related to the weak acid sites, while the high-temperature desorption (T > 250 • C) is related to the strong acid sites [23]. USY support gave two desorption peaks at about 166 • C and 325 • C corresponding to the weak adsorption of ammonia molecules on Si-OH groups and the Brönsted acid sites, that is, bridged Si-OH-Al hydroxyl groups in Y-type zeolites, respectively [31]. The weak Catalysts 2019, 9, 715 6 of 12 and strong acid sites were also observed for Cu 3 P/USY, Ni 2 P/USY, and CoP/USY catalysts, however, these catalysts displayed a greater amount of weak acid sites and a smaller amount of strong acid sites when compared to the pure USY support. It is possible that the deposition of metal phosphide particles could cover both weak and strong acid sites of USY support, while the metal phosphide could itself contribute Lewis and Brönsted acidity. Metal phosphides were reported to have both Brönsted (~200 • C) and Lewis (~320 • C) acid sites, which are related to the P-OH group and the electron-deficient metal site, respectively [27,32]. Therefore, the highest acidity of Ni 2 P/USY could be attributed to the acidity properties of metal phosphide and USY support, the large surface area, and high dispersion of small Ni 2 P particles so that NH 3 molecules could be adsorbed more readily than other catalysts. Regarding the TPD profile of Al 2 O 3 support, a small broad envelope of unresolved peaks extended to 487 • C was observed, indicating the presence of strong acid sites which are related to Al Lewis acid centers (i.e., the electron acceptor sites formed by the coordinatively unsaturated aluminum ions) [33]. However, Cu/γ-Al 2 O 3 exhibited a larger desorption peak at low temperature when compared to γ-Al 2 O 3 support. The metallic Cu is known to possess the Lewis acidity, whereas the Brönsted acid sites could be generated from phosphate species, possibly an amorphous form that was not detectable by XRD, remaining as a consequence of their incomplete reduction [23]. Cu 3 P/SiO 2 contained the least amount of acid sites, which could be related to the small surface area and the weak acidity of SiO 2 support.
following order: USY > γ-Al2O3 > SiO2, and the number of acid sites in catalysts was in the order of Ni2P/USY > Cu/γ-Al2O3 > Cu3P/USY > CoP/USY > Cu3P/SiO2. The strength of acid site can be distinguished by the desorption temperature of the adsorbed NH3 (T). The low-temperature desorption (T < 250 °C) is related to the weak acid sites, while the high-temperature desorption (T > 250 °C) is related to the strong acid sites [23]. USY support gave two desorption peaks at about 166 °C and 325 °C corresponding to the weak adsorption of ammonia molecules on Si-OH groups and the Brönsted acid sites, that is, bridged Si-OH-Al hydroxyl groups in Y-type zeolites, respectively [31]. The weak and strong acid sites were also observed for Cu3P/USY, Ni2P/USY, and CoP/USY catalysts, however, these catalysts displayed a greater amount of weak acid sites and a smaller amount of strong acid sites when compared to the pure USY support. It is possible that the deposition of metal phosphide particles could cover both weak and strong acid sites of USY support, while the metal phosphide could itself contribute Lewis and Brönsted acidity. Metal phosphides were reported to have both Brönsted (~200 °C) and Lewis (~320 °C) acid sites, which are related to the P-OH group and the electron-deficient metal site, respectively [27,32]. Therefore, the highest acidity of Ni2P/USY could be attributed to the acidity properties of metal phosphide and USY support, the large surface area, and high dispersion of small Ni2P particles so that NH3 molecules could be adsorbed more readily than other catalysts. Regarding the TPD profile of Al2O3 support, a small broad envelope of unresolved peaks extended to 487 °C was observed, indicating the presence of strong acid sites which are related to Al Lewis acid centers (i.e., the electron acceptor sites formed by the coordinatively unsaturated aluminum ions) [33]. However, Cu/γ-Al2O3 exhibited a larger desorption peak at low temperature when compared to γ-Al2O3 support. The metallic Cu is known to possess the Lewis acidity, whereas the Brönsted acid sites could be generated from phosphate species, possibly an amorphous form that was not detectable by XRD, remaining as a consequence of their incomplete reduction [23]. Cu3P/SiO2 contained the least amount of acid sites, which could be related to the small surface area and the weak acidity of SiO2 support.

Deoxygenation of Oleic Acid
The oleic acid conversion and product yields of CoP/USY, Ni 2 P/USY, Cu 3 P/USY, Cu 3 P/SiO 2 , Cu/γ-Al 2 O 3 were plotted as a function of reaction temperature, as illustrated in Figure 6, taking into account that a complete conversion was achieved over all catalysts after reaching 340 • C for 4 h. It is observed that CoP/USY (Figure 6a) and Ni 2 P/USY (Figure 6b) catalysts exhibited similar catalytic behavior. At 4 h, these catalysts favored the production of heptadecane and octadecane as main components with the presence of heptylcyclopentanone and heptylcyclopentane as intermediates. These two compounds were formed at low temperature and further converted to heptadecane and octadecane at high temperature and longer reaction time. A close examination of the reaction over Ni 2 P/USY at 4 h reveals that % selective yield of octadecane and heptadecane decreased because these products were cracked to small alkanes. In comparison, CoP/USY showed good performance for both HDO and DCO/DCO 2, pathways according to their almost equal % selective yield, while Ni 2 P/USY tended to favor the DCO/DCO 2 pathway as heptadecane was more abundant than octadecane. Note that the catalytic activity in decarbonylation (DCO) and decarboxylation (DCO 2 ) reactions could not be directly correlated with the amount of CO and CO 2 detected in the gas phase because methanation and water gas shift reaction were involved in the main gas-phase reaction. The calculation was therefore based on the liquid products.
observed that CoP/USY (Figure 6a) and Ni2P/USY (Figure 6b) catalysts exhibited similar catalytic behavior. At 4 h, these catalysts favored the production of heptadecane and octadecane as main components with the presence of heptylcyclopentanone and heptylcyclopentane as intermediates. These two compounds were formed at low temperature and further converted to heptadecane and octadecane at high temperature and longer reaction time. A close examination of the reaction over Ni2P/USY at 4 h reveals that % selective yield of octadecane and heptadecane decreased because these products were cracked to small alkanes. In comparison, CoP/USY showed good performance for both HDO and DCO/DCO2, pathways according to their almost equal % selective yield, while Ni2P/USY tended to favor the DCO/DCO2 pathway as heptadecane was more abundant than octadecane. Note that the catalytic activity in decarbonylation (DCO) and decarboxylation (DCO2) reactions could not be directly correlated with the amount of CO and CO2 detected in the gas phase because methanation and water gas shift reaction were involved in the main gas-phase reaction. The calculation was therefore based on the liquid products. Figure 6. Influence of temperature on deoxygenation of oleic acid over supported metal and metal phosphide catalysts; selectivity (a -e) and the conversion (f) Figure 6. Influence of temperature on deoxygenation of oleic acid over supported metal and metal phosphide catalysts; selectivity (a-e) and the conversion (f). Cu 3 P/USY catalyst showed different catalytic behavior. The main product for Cu 3 P/USY catalyst was dodecylbenzene, and the minor products were identified as dodecylcyclohexane and heptylcyclopentane. Intermediates were cyclic compounds, such as 2-dodecylcyclohexanol and decyl-2-cyclopenta-1-one, being produced at low temperature with oxygen atom and converted to dodecylbenzene at 340 • C. This indicates that several reactions, including HDO, DCO, DCO 2 , aromatization, and hydrogen transfer, could occur in this reaction system as reported by Tian et al. [34]. The structure of CoP, Ni 2 P, and Cu 3 P possesses different Lewis acid sites (metal sites) and Brönsted sites (P-OH sites) that may affect their intrinsic activities. Peroni and coworkers reported that the intrinsic activities of different transition metals have significant impact on the surface reaction of catalysts [35]. However, the support may also have an influence on the activity of catalysts during the reaction. The influence of support materials on the catalytic performance was investigated over supported copper and copper phosphide catalysts (i.e., Cu 3 P/USY, Cu 3 P/SiO 2 , and Cu/γ-Al 2 O 3 ). It was found that Cu 3 P/SiO 2 produced similar products as Cu 3 P/USY. Interestingly, Cu/γ-Al 2 O 3 showed the best selectivity for octadecane production through the HDO pathway (Figure 7) with the highest selective yield (98%), even better than other catalysts. The ester compounds are intermediates which could be observed at lower temperature.
In general reactions, the metal active site favored the alkane products that were obtained from the hydrogenation of oleic acid to octadecanoic acid, and then the water molecule was removed from the octadecanoic acid by dehydration reaction. The Cu/γ-Al 2 O 3 could prove that the hydrogen was easier to be adsorbed on metal sites and then added to the oleic acid compound. For Brönsted sites, the heptadecenoic acid was produced from oleic acid by decarboxylation, and then H 2 gas was added into the compound with hydrogenation to provide the heptadecane product. In contrast, for Cu 3 P, the oleic acid was transformed to cyclic compounds that have a hexyl ring group or pentyl ring group by cyclization. The water was removed from the compounds by dehydration to form the main product, dodecylbenzene, and then H 2 gas was added into the compounds to produce dodecylcyclohexane and heptylcyclopentane.
heptylcyclopentane. Intermediates were cyclic compounds, such as 2-dodecylcyclohexanol and decyl-2-cyclopenta-1-one, being produced at low temperature with oxygen atom and converted to dodecylbenzene at 340 °C. This indicates that several reactions, including HDO, DCO, DCO2, aromatization, and hydrogen transfer, could occur in this reaction system as reported by Tian et al. [34]. The structure of CoP, Ni2P, and Cu3P possesses different Lewis acid sites (metal sites) and Brönsted sites (P-OH sites) that may affect their intrinsic activities. Peroni and coworkers reported that the intrinsic activities of different transition metals have significant impact on the surface reaction of catalysts [35]. However, the support may also have an influence on the activity of catalysts during the reaction. The influence of support materials on the catalytic performance was investigated over supported copper and copper phosphide catalysts (i.e., Cu3P/USY, Cu3P/SiO2, and Cu/γ-Al2O3). It was found that Cu3P/SiO2 produced similar products as Cu3P/USY. Interestingly, Cu/γ-Al2O3 showed the best selectivity for octadecane production through the HDO pathway (Figure 7) with the highest selective yield (98%), even better than other catalysts. The ester compounds are intermediates which could be observed at lower temperature.
In general reactions, the metal active site favored the alkane products that were obtained from the hydrogenation of oleic acid to octadecanoic acid, and then the water molecule was removed from the octadecanoic acid by dehydration reaction. The Cu/γ-Al2O3 could prove that the hydrogen was easier to be adsorbed on metal sites and then added to the oleic acid compound. For Brönsted sites, the heptadecenoic acid was produced from oleic acid by decarboxylation, and then H2 gas was added into the compound with hydrogenation to provide the heptadecane product. In contrast, for Cu3P, the oleic acid was transformed to cyclic compounds that have a hexyl ring group or pentyl ring group by cyclization. The water was removed from the compounds by dehydration to form the main product, dodecylbenzene, and then H2 gas was added into the compounds to produce dodecylcyclohexane and heptylcyclopentane.

Synthesis of Supported Metal Phosphide Catalysts
Copper phosphide catalyst supported on γ-Al 2 O 3 , SiO 2 , and ultrastable zeolite Y (USY) were synthesized by hydrogen reduction of phosphate precursors, according to the previously reported procedure [26,36]. In the first step, the supported copper phosphate precursors were prepared by incipient wetness impregnation method with metal loading of 10 wt%. Copper nitrate and (NH 4 ) 2 HPO 4 were dissolved in water with the initial Cu/P molar ratio of 2 and maintained under magnetic stirring. A few drops of nitric acid were added to dissolve some precipitates. Then, SiO 2 , γ-Al 2 O 3 , or USY supports were added into the solution under continuous stirring for 30 min, followed by ultrasonication for 3 h. The obtained phosphate precursor mixture was dried at 80 • C for 12 h and calcined in air at 450 • C for 3 h. In the second step, the metal phosphate precursors were reduced to phosphide catalysts in hydrogen atmosphere with a ramp rate of 5 • C/min from room temperature to 650 • C and kept at isothermal conditions for 5 h.
Other supported metal phosphides (i.e., Ni 2 P/USY, CoP/USY) were prepared from the corresponding metal nitrates, with the same metal loading and metal/P molar ratio, using the described procedure. As the reduction of cobalt phosphate has been reported to occur at higher temperature (~700-720 • C) than that of nickel phosphate (~600-650 • C) [37,38], the hydrogen reduction temperatures of Ni 2 P/USY and CoP/USY were 650 • C and 750 • C, respectively.

Characterization of Supported Metal Phosphide Catalysts
X-ray diffraction (XRD) patterns were measured on a powder diffractometer (D8 ADVANCE, Bruker, Karlsruhe, Germany) using Cu Kα radiation with Ni filter, operated at 40 kV and 40 mA, in the 2θ range of 10-80 • . The average crystallite size of catalysts, D, was calculated using the Debye-Scherrer formula and FullProf software [39] as described below.
where λ is the X-ray wavelength (nm), β is the integral breadth of diffraction peak, and K is a constant related to crystallite shape. LaB 6 was used as a standard to determine the instrumental resolution of the X-ray diffractometer. The scanning electron microscopy (SEM) analysis was performed using a HITACHI SU5000 FE-SEM microscope operating at 10 kV in back-scattering electron (BSE) mode. Nitrogen adsorption-desorption isotherms were recorded at −196 • C using a Nova 2000e analyzer (Quantachrome Instruments) after each sample was degassed at 300 • C for 3 h. The surface area and the pore-size distribution of catalysts were determined by using the Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) methods, respectively. T-plot analysis was additionally applied to evaluate the surface area and volume of micropores. Temperature-programmed desorption (TPD) of NH 3 analysis was conducted using an automated ChemBET Pulsar TPR/TPD chemisorption analyzer (Quantachrome instruments). Catalyst (100 mg) was loaded and pretreated in He at 120 • C for 1 h. Afterward, NH 3 -TPD was performed in flowing H 2 /Ar gas mixture (H 2 /Ar = 1.5; total flow 30 cm 3 min −1 ) and heated to 800 • C at a heating rate of 5 • C/min.

Deoxygenation of Oxygenated Hydrocarbon Compounds
Deoxygenation reaction of the oxygenated model compound, oleic acid, was carried out in a Parr reactor. A 1 g portion of supported metal phosphide catalyst and 60 mL of 5 wt% solution of oleic acid in dodecane were loaded into the reactor. Prior to the reaction, the reactor was purged with N 2 , then heated to 240 • C and pressurized with H 2 to 50 bar. The catalytic activities were measured at 260 • C, 280 • C, 300 • C, 320 • C, 340 • C, and at two-hour intervals for six hours at 340 • C. All obtained products were analyzed by a gas chromatograph-mass spectrometer (GC-MS) with a DB-1HT capillary column. Note that the gas-phase products in this study, such as C 4 H 10 , C 3 H 8 , C 2 H 6 , CH 4 , CO, and CO 2 , were not analyzed because methanation and water gas shift reaction were involved in the main gas-phase reaction. The calculation was therefore based on the liquid products. All standard calibration curves of C 13 -C 18 were used for quantitative analysis. Moreover, 8 mg of C 15 was added into the sample as an internal standard for Cu 3 P/SiO 2 and Cu 3 P/USY catalysts. The conversion of oleic acid was calculated according to the following equations: conversion(%) = mole of oleic acid in feed − mole of oleic acid in product mole of oleic acid in feed × 100 The selective yield (Y) of products was calculated based on carbon mass balance [3]: Y i (mol%) = n i × a i n Oleic acid × a Oleic acid × 100 where n i and a i represent the mole and carbon atom number of product i. n Oleic acid and a Oleic acid represent the mole of oleic acid and carbon atom number of oleic acid, respectively. The quantitative analyses were mostly conducted on HDO and DCO/DCO 2 products. The unspecified liquid products could refer to the cyclic or aromatic compounds and polymerized products.

Conclusions
Supported copper, nickel, and cobalt phosphide catalysts were evaluated for deoxygenation of oleic acid. The oleic acid was chosen as a model compound with the aim to gain insight into the deoxygenation reaction of fatty acids, a key step in the conversion of renewable biofuel to fuels and value-added chemicals. We have demonstrated that different catalysts exhibited different catalytic behaviors. The nature of the support materials has a profound effect on the structural, surface, and catalytic properties of Cu 3 P. All supported metal phosphides were prepared by the hydrogen reduction of impregnated metal phosphate precursors. CoP and Ni 2 P were formed on USY zeolite. Cu 3 P was formed on USY and SiO 2 supports, while the metallic Cu phase was stabilized on γ-Al 2 O 3 support. Metal phosphide particles were highly dispersed over the surface of the USY support. Cu 3 P/USY exhibited much larger surface area and higher concentration of acid sites compared to Cu 3 P/SiO 2 , owing to the textural and acidic properties of the USY zeolite support. All supported catalysts gave an oleic acid conversion close to 100% at 340 • C. The main hydrocarbon products of Ni 2 P/USY and CoP/USY were heptadecane and octadecane derived from DCO/DCO 2 and HDO pathways, respectively. Cu/γ-Al 2 O 3 facilitated the HDO reaction and inhibited cracking reactions, leading to the highest selective production of octadecane (98%). In contrast, the supported Cu 3 P catalysts favored cyclization and aromatization to form cyclic and aromatic compounds such as dodecylcyclohexane, heptylcyclopentane, and dodecylbenzene. Cu 3 P/SiO 2 gave higher selective yield of dodecylbenzene (46%) than the Cu 3 P/USY (33%). Therefore, we may conclude that the supported Cu 3 P catalysts have potential applications in the production of cyclic and aromatic compounds, while Cu/γ-Al 2 O 3 can be considered as a promising catalyst for the hydrodeoxygenation of renewable biofuel to alkane products. We believe that this study can provide new ideas and directions for the development of renewable biofuel energy in the future.