Transformation of Jatropha Oil into High-Quality Biofuel over Ni-W Bimetallic Catalysts.

The production of fuel from the hydrodeoxygenation of vegetable oils has been extensively investigated on account of the decline of petroleum-based fuels and increase of ecological problems. The conversion of jatropha oil over Al-MCM-41-supported Ni, W, and Ni-W catalysts was studied at 3 MPa and 360 °C. Over the monometallic Ni and W catalysts, the biofuel yield was low, about 19.3 and 12.5 wt %, respectively, whereas the highest biofuel yield reached 63.5 wt % over the Ni-W bimetallic catalysts. X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), transmission electron microscopy (TEM), and high-resolution TEM results suggested that the proper amount of Ni and W would form a Ni17W3 active phase, the particle size of which varied with the content of Ni and W or preparation methods. The crystalline Ni17W3 phase formed when the content of both Ni and W reached 10%. With further increase of the content of W or Ni to 15%, the crystal size of Ni17W3 grew from 7 to 14 nm or to 20 nm, whereas the biofuel yield decreased with the increase of the Ni17W3 crystal size. The 10Ni-10W/Al-MCM-41 catalyst with the Ni17W3 crystal size of 7 nm showed the best performance for the transformation of jatropha oil into high-grade biofuel.

Introduction

The exhaustion of traditional petroleum-based fuels and the deterioration of ecological environment have triggered the development of biomass conversion technology to obtain alternative products to replace petroleum-based fuels.[1−7] Typical feedstocks for renewable fuels were lipids or fatty acids originated from different biomass (e.g., jatropha,[8−10] microalgae or rapeseed oil,[11,12] waste cooking oils, and fats[13,14]). Among these, vegetable oil-based feedstocks were potential resources because of their high energy content and similar structure to traditional fossil fuels.[15] Especially, the hotspot was concentrated on inedible vegetable oils. Jatropha can be cultivated mostly in wasteland and produces much more bio-oil than canola, sunflower, and soyabean.[16] Compared with traditional fuels, jatropha oil exhibited an excellent flash point, solidifying point, ignition point, and kinematic viscosity.[17] But the application of crude jatropha oil was limited due to its oxygen content which caused some deleterious properties: poor heating values, high viscosity, instability, and a tendency for polymerization.[18]

Catalytic deoxygenation (DO) (in the presence of H2) has been developed to upgrade vegetable oils using heterogeneous catalysts.[19] Currently, the DO reaction includes three distinct parallel approaches: hydrodeoxygenation (−H2O), decarbonylation (−CO), and decarboxylation (−CO2), as displayed in Scheme .[15,20−22] Mainly two types of catalysts were widely researched to catalytically convert vegetable oils into hydrocarbons: noble-metal catalysts and transition-metal sulfides catalysts. Although noble metals (mainly Pd and Pt) were conducive to DO reactions, the availability of noble metals was limited for their scarcity and high costs.[14] It has been widely reported that sulfided NiMo, CoMo, and NiW oxides were effective for DO.[23] Zhang and co-workers[13] investigated the sulfided CoMo catalysts to upgrade waste cooking oil, and found that hydrodecarbonylation/decarboxylation reactions were the major pathways during the removal of oxygen. Toba and colleagues[24] researched the hydrodeoxygenation performance of sulfided catalysts. They found that sulfided Ni–Mo and Ni–W catalysts had better hydrodeoxygenation activity than sulfided Co–Mo catalysts, and low-grade waste oils could be completely converted into hydrocarbons at 350 °C. Unfortunately, in these systems, sulfur leaching caused the deactivation of catalysts and contamination of alkane products.

Scheme 1

Pathways for the DO of Fatty Acids

Peng and co-workers[11] researched the catalytic DO of microalgae oil and disclosed that triglycerides were hydrogenated and the C–O of fatty acid esters was selectively cleaved due to the function of Ni nanoparticles. Hence, nonexpensive Ni-based catalysts could be promising alternatives for noble-metal catalysts. Furthermore, Ni–W bimetallic catalysts have been recently considered as prospective materials for the DO process. Echeandia et al.[25] conducted phenol hydrodeoxygenation using Ni–W/active carbon (AC) catalysts, and proposed that an electronic effect between nickel and tungsten occurred, and Ni–W/AC showed a good hydrodeoxygenation performance. Subsadsana et al.[26] reported the catalytic transformation of palm oil and concluded that binary NiW-ZSM-5/MCM-41 catalysts exhibited higher activity for the conversion of palm oil into hydrocarbons than the monometallic catalysts. Wang et al.[27] studied Ni(Co)–W–B catalysts for cyclopentanone hydrodeoxygenation. It was found that WO3 existed in Ni(Co)–W–B catalysts, which was a great Brønsted acid site, resulting in low selectivity of oxygenates. Rana and co-workers[28] investigated the conversion of waste vegetable oils and gasoline mixtures into biofuel on the Ni–W catalyst supported on SiO2–Al2O3. It was proved that the supported Ni–W catalyst favored decarboxylation and decarbonylation and gave considerable jet range products. Meanwhile, it was proposed that mesoporous supports played a crucial role in the transformation of triglyceride for good diffusion of feedstocks.

Although, some reports concerning the catalytic transformation of vegetable oils on Ni–W bimetallic catalysts without sulfuration have been researched, the crystal size effects of nickel and tungsten active components on the catalytic performance were still not clear. Besides, the active species with different particle sizes had different catalytic performances during the reactions of heterogeneous catalysis.[29] Thus, in the present work, a series of Ni–W/Al-MCM-41 bimetallic catalysts with various contents of nickel and tungsten were synthesized. Al-MCM-41 was used as support for its large surface area and uniform mesopore.[30,31] The DO of jatropha oil over these catalysts was conducted. Catalyst characterizations were performed by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), N2-physisorption, temperature-programmed reduction (TPR), transmission electron microscopy (TEM) [including high resolution TEM (HRTEM)], and so forth. Ni17W3 alloy was found to be the active phase for the conversion of jatropha oil to biofuel, and the crystal size effect on the biofuel yield is discussed in detail. This work provides new thoughts about the preparation of a potential kind of catalyst with versatile functions to convert a vegetable oil into high-performance biofuel.

Results and Discussion

Inductively Coupled Plasma Atomic Emission Spectroscopy and XRD of All Samples

The actual contents of Ni and W in all catalysts were detected by the inductively coupled plasma atomic emission spectroscopy (ICP–AES) test (seen in Tables S1 and S2), and it was found that the actual contents of both Ni and W were somewhat lower than the controlled ones, whereas those of Ni were closer to the controlled ones compared to those of W.

All precursors and catalysts were tested by the XRD technique. The XRD patterns are displayed in Figures and 2. All precursors had a peak at around 23° corresponding to the amorphous silica in the support.[32] As shown in Figure A, when only 10% Ni was loaded on the Al-MCM-41, the peaks of 37.2°, 43.3°, 62.9°, and 75.3°, corresponding to NiO (PDF 65-6920) were observed. When the W content increased from 2.5 to 15%, the intensity of NiO diffraction peaks decreased, and when W loading was less than 10%, only NiO crystallite peaks were observed, whereas when W loading reached or surpassed 10%, a new diffraction peak at 2θ of 30.8° attributed to NiWO4 (PDF 15-0755) and peaks at 2θ = 23.7°, 28.5°, and 33.6° attributed to WO3 (PDF 20-1324) were observed, respectively. The intensity of NiWO4 and WO3 peaks did not enhance with further increasing W loading. When loading 10% W solely, WO3 crystallite peaks appeared at 2θ = 23.7°, 28.5°, 33.6°, and 41.5° (PDF 20-1324), as displayed in Figure B. When fixing the W loading at 10%, with the Ni loading increasing from 2.5 to 15%, the NiO peaks appeared and the intensity increased. When the Ni loading reached or surpassed 10%, the peak of 2θ = 30.8° attributed to NiWO4 (PDF 15-0755) was observed.

Figure 1

(A) XRD patterns of precursors of 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15) and (B) XRD patterns of precursors of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15). Intensity is given in arbitrary units (a.u.).

Figure 2

(A) XRD patterns of 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15) and (B) XRD patterns of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15). Intensity is given in arbitrary units (a.u.).

For reduced monometallic 10Ni catalyst, the diffraction peaks at 2θ = 44.4°, 51.7°, and 76.2° were attributed to the metal nickel (PDF 04-0850), as shown in Figure A. Fixing the Ni loading at 10%, when the mass loading of W increased from 2.5 to 15%, the intensity of the metal Ni diffraction peaks decreased. When the W loading reached 10%, the peaks for crystalline WO2 appeared at 25.7°, 36.8°, and 53.1° (PDF 32-1393) and the peak for W3O appeared at 39.7° (PDF 41-1230). It could be observed that the diffraction peaks at about 44°, 51°, and 75° had broad peaks and relatively poor symmetry. The PDF card showed that Ni17W3 diffraction peaks should appear at 2θ = 43.9°, 51.2°, and 74.9° (PDF 65-4828). Considering the fact that the diffraction peak of NiWO4 observed on the precursors vanished, NiWO4 could be transformed to Ni17W3, according to the literature.[33,34] It was speculated that Ni17W3 phase might form, the diffraction peaks of which overlapped with those of Ni. That is, the diffraction peaks at about 44°, 51°, and 75° were the overlapped peaks of Ni17W3 and Ni phases. To further prove the overlapping phenomenon, the following three catalysts were prepared: (1) 10Ni–10W–Ni catalyst (impregnated Ni species first), where only the diffraction peak at about 2θ = 44.4° for Ni phase was observed; (2) 10Ni–10W-700-Redu. catalyst (increased the reduction temperature to 700 °C), where only the diffraction peak at about 2θ = 43.9° for Ni17W3 phase was observed; and (3) 10Ni–10W-700-Calc. catalyst (increased the calcination temperature to 700 °C), where the overlap of diffraction peaks at about 2θ = 44° for Ni17W3 and Ni was observed. Comparing the XRD patterns of these catalysts and enlarging the profile of the diffraction peak near 2θ = 44° (seen in Figure S1), the individual diffraction peak at 2θ = 43.9° of Ni17W3 and the individual diffraction peak at 2θ = 44.4° of Ni were relatively narrow and symmetrical, whereas the overlapped diffraction peak of Ni17W3 and Ni at about 2θ = 44° became broad and asymmetric. Besides, when the reduction temperature increased to 700 °C, there existed the individual diffraction peaks of W at 2θ = 40.4°, 58.4°, and 73.4° (PDF 89-3728) and Ni17W3 without Ni phase. When the calcination temperature increased to 700 °C, the overlapping phenomenon of Ni17W3 and Ni phases on 10Ni–10W-700-Calc. catalyst was more obvious than that of the 10Ni–10W catalyst. HRTEM measurement was carried out to further confirm the existence of Ni17W3 on these catalysts. Lattice fringes on the catalysts were observed, as shown in Figure S2. The value of lattice distance (d) was about 0.21 nm based on the HRTEM results, and there existed both Ni and W on these particles based on energy-dispersive X-ray (EDX) results (seen in Figure S3). The standard value of the Ni17W3 lattice distance was 0.207 nm provided by the PDF card of MDI Jade 6.0 software, thus, it further proved the formation of Ni17W3 alloy phase, and Ni17W3 alloy phase mainly exposed the crystal face of (111). Thus, the peak-fitting of these peaks was conducted based on the PDF 65-4828 card provided by MDI Jade 6.0 software; then, according to Scherrer’s formula, their particle sizes were separately calculated. Furthermore, all of these peaks strengthened with the increase of W loading. The WO3 diffraction peaks at 23.5°, 33.1°, and 40.4° (PDF 32-1395) and those for WO2 at 25.7°, 36.8°, and 53.1° (PDF 32-1393) were observed on pure 10W catalyst in Figure B. Fixing the loading of W at 10%, when the loading of Ni was 2.5 and 5%, the crystalline W3O peaks appeared at 35.4°, 39.7°, and 43.7° (PDF 41-1230) and the diffraction peaks of WO3 disappeared. When the loading of Ni reached or surpassed 10%, the diffraction peaks of Ni17W3 appeared and the intensity enhanced with increasing Ni loading. At the same time, the intensity of Ni peaks enhanced with increasing the content of nickel as well. The crystal sizes of different species were calculated by the Scherrer equation and are listed in Table . For 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15), the Ni crystal size on 10Ni catalyst was 15 nm and its crystal size slightly decreased with adding W species. When the W content reached 10%, the crystal size of Ni17W3 was about 7 nm, and grew up to 14 nm with continuously increasing the content of W to 15%. Besides, for 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15), the crystal Ni17W3 phases were also observed when the addition of Ni species was equal to 10%. Although with consistently increasing the content of Ni from 10 to 15%, the crystal size of the Ni17W3 phase increased from 7 to 20 nm. For Ni–W bimetallic catalysts, when the W content was equal to or more than 10%, W3O and WO2 species could be observed. The crystalline size of W3O varied little with varying W content and that of WO2 varied from 17 to 29 nm. Although the deconvolution might not be very accurate, the variation trend of the particle sizes would provide some valuable information about the size of particles.

Table 1

The Crystal Sizes of Different Phases for All Catalystsa

	crystal sizes (nm)
catalysts	Ni	Ni17W3	W3O	WO2	WO3
10Ni	15	---	---	---	---
10Ni–2.5W	13	---	---	---	---
10Ni–5W	14	---	---	---	---
10Ni–10Wa	11	7	12	17	---
10Ni–12.5W	12	12	12	22	---
10Ni–15W	11	14	11	19	---
10W	---	---	---	20	10
10W–2.5Ni	---	---	14	29	---
10W–5Ni	---	---	13	21	---
10Ni–10Wb	11	7	12	17	---
10W–12.5Ni	11	13	16	23	---
10W–15Ni	9	20	14	21	---

a and b were the same catalysts; ---: not observed.

XPS Results

XPS measurements were conducted to analyze the chemical states of Ni and W. For 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15) as shown in Figure A, the binding energy of Ni 2p3/2 near 852.3 and 855.5 eV was attributed to Ni0 and Ni2+, separately.[35] With increasing W loading from 2.5 to 15%, the Ni 2p3/2 binding energy moved from 852.3 to 852.8 eV. The increase of Ni 2p3/2 binding energy was caused by the electron transformation from Ni to W in Ni17W3 alloy as observed in XRD. The electronegativity of tungsten was stronger than nickel, causing charge transfer from Ni to W.[34,36−38] The peak located at 861.4 eV was attributed to a satellite peak of Ni 2p3/2 due to multielectron excitation.[25] Accordingly, the binding energy of W 4f doublets at 35.6 and 37.7 eV decreased to 34.8 and 37.0 eV when the W loading increased from 2.5 to 15% in Figure B. The W 4f doublets with binding energy near 35.6 and 37.7 eV were corresponded to W6+ 4f7/2 and W6+ 4f5/2 species in WO3 or NiWO4.[39]

Figure 3

(A) XPS spectra of Ni 2p and (B) XPS spectra of W 4f for 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15).

As described in Figure A, the binding energy of Ni 2p3/2 of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15) also increased from 852.3 to 853.7 eV, which resembled the 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15). For the results of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15) in Figure B, there also existed similar decrease of the W 4f binging energy to that in Figure B. The above results confirmed the formation of Ni17W3 alloy on these catalysts.

Figure 4

(A) XPS spectra of Ni 2p and (B) XPS spectra of W 4f for 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15).

Brunauer–Emmett–Teller Results

As depicted in Figure A,B, the isotherms with a hysteresis loop were identified as a type IV characteristic of mesoporous materials. It was disclosed that the support and all catalysts belonged to mesoporous materials.[40] According to the pore size distributions from Figure A,B, it was found that the pore size of support was mostly in a range of 2 and 30 nm, which also indicated the presence of mesopores in the material. Ni–W bimetallic catalysts distributed mainly in 2–20 nm compared with the support. Besides, fixing 10% Ni or 10% W, the pores of 2–30 nm obviously decreased with increasing the other metal content from 2.5 to 15%. Thus, the results proved that some Ni and W species entered into these pores, whereas others supported on the surface of the carrier without blocking the 2–30 nm mesopores.

Figure 5

(A) N2 adsorption and desorption isotherms of 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15) and (B) N2 adsorption and desorption isotherms of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15).

Figure 6

(A) Pore size distribution of 10Ni–yW catalysts (y = 0, 2.5, 5, 10, 12.5, 15) and (B) pore size distribution of 10W–xNi catalysts (x = 0, 2.5, 5, 10, 12.5, 15).

Tables and 3 demonstrated the physicochemical properties of all catalysts. As observed from Tables and 3, the surface area and pore volume of Al-MCM-41 were 808 m2·g–1 and 1.16 cm3·g–1, respectively. The surface areas of 10Ni and 10W catalysts were 522 and 708 m2·g–1, respectively. When the W loading increased from 2.5 to 15%, the surface areas of binary Ni–W catalysts were slightly larger than the 10Ni catalyst because of the interaction of active components or the smaller active Ni–W particles.[41] With the Ni content increasing from 2.5 to 15%, the surface area of binary Ni–W catalysts decreased compared with the 10W catalyst. As a whole, fixing 10% Ni or 10% W, with increasing the other metal content from 2.5 to 15%, the surface areas and pore volumes of Ni–W bimetallic catalysts declined comparing to the carrier as expected, disclosing that the active components were indeed loaded on the surface and channels of the support.

Table 2

Physicochemical Properties of 10Ni–yW Catalysts (y = 0, 2.5, 5, 10, 12.5, 15)

sample	BET surface area (m2/g)	pore volume (cm3/g)	pore size (nm)
Al-MCM-41	808	1.16	5.4
10Ni	522	0.35	3.9
10Ni–2.5W	626	0.63	4.4
10Ni–5W	586	0.5	5.3
10Ni–10W	559	0.52	4.5
10Ni–12.5W	553	0.56	4.6
10Ni–15W	540	0.48	4.4

Table 3

Physicochemical Properties of 10W–xNi Catalysts (x = 0, 2.5, 5, 10, 12.5, 15)

sample	BET surface area (m2/g)	pore volume (cm3/g)	pore size (nm)
Al-MCM-41	808	1.16	5.4
10W	708	0.73	4.6
10W–2.5Ni	572	0.46	4.1
10W–5Ni	586	0.53	4.4
10Ni–10W	559	0.52	4.5
10W–12.5Ni	574	0.54	4.5
10W–15Ni	538	0.46	4.3

TPR Results

TPR graphs of all precursors were obtained to decide the reduction temperature of nickel and tungsten species. As displayed in Figure A, for 10Ni precursor, the α1 reduction peak centered at about 368 °C was caused by the reduction of bulk NiO.[32] The α2 reduction peak appeared at higher temperature (about 483 °C), and the shoulder was attributed to highly dispersed NiO interacting strongly with the support.[35,42] On the 10W precursor in Figure B, the β1 peak at about 500 °C resulted from WO3 interacting weakly with the carrier, whereas the intense β2 and β3 peaks at around 623 °C and at 735 °C were caused by WO3 and WO2 reduction, respectively.[25,39] Fixing the loading of Ni at 10% in Figure A, with loading W species, the α1 reduction peak shifted toward higher temperature (370–390 °C), implying that W species enhanced the interaction between NiO with the support. Additionally, with increasing the loading of tungsten, the intensity of the α1 reduction peak weakened, and the α2 reduction peak became broader and shifted toward higher temperature. Combined with XRD, the results demonstrated that the broad peaks in the range of 528–590 °C were caused by coreduction of NiO and WO3 or NiWO4.[33,43−46] When the W loading reached 10% or above, the fitted peak at 535–550 °C was attributed to the reduction peak of NiWO4,[20,39,40] which led to the formation of the Ni17W3 alloy.

Figure 7

(A) TPR curves of 10Ni–yW precursors (y = 0, 2.5, 5, 10, 12.5, 15) and (B) TPR curves of 10W–xNi precursors (x = 0, 2.5, 5, 10, 12.5, 15).

When the W loading was fixed at 10% as depicted in Figure B, the NiO reduction peak (366–443 °C) appeared and its intensity increased with increasing nickel contents. When the nickel loading reached 10% or above, the β1 peak disappeared and the β2 peak became broader and shifted toward a lower-temperature range (513–618 °C). Similarly, the fitted peak at 513–550 °C resulted from the NiWO4 reduction peak. Additionally, the reduction peak in the range of 698–772 °C resulted from the reduction of WO2. In comparison with monometallic precursors, the reduction temperature of Ni increased and that of W species decreased on Ni–W bimetallic precursors. It was suggested that there existed interaction between Ni and W on Ni–W bimetallic precursors.

TEM Results

Representative TEM images of the catalysts showed metal particles with different shapes, and the particle size distributions were also determined by TEM pictures. As displayed in Figure , the nearly spherical particles of metal Ni species slightly formed agglomerates on the 10Ni catalyst. With the addition of metal W, the particles gradually dispersed well and there existed both spherical and elliptical particles on the Ni–W bimetallic catalysts. Besides, the particle sizes on 10Ni–2.5W and 10Ni–5W catalysts mainly distributed between 14–18 and 13–16 nm, respectively. When the loading of W increased to 10 and 15%, the particle size distributions over 10Ni–10W, 10Ni–12.5W, and 10Ni–15W catalysts changed from 9–12 to 17–20 nm (Table ).

Figure 8

TEM images of the catalysts. (a) 10Ni, (b) 10Ni–2.5W, (c) 10Ni–5W, (d) 10Ni–10W, (e) 10Ni–12.5W, and (f) 10Ni–15W.

Table 4

The Particle Size Distributions of All Catalystsa

catalysts	particle sizes (nm)	catalysts	particle sizes (nm)
10Ni	12–15	10W	---
10Ni–2.5W	14–18	10W–2.5Ni	16–20
10Ni–5W	13–16	10W–5Ni	16–19
10Ni–10W	9–12	---	---
10Ni–12.5W	16–19	10W–12.5Ni	15–28
10Ni–15W	17–20	10W–15Ni	18–27

---: not measured.

Similarly, some groups of dark-thread-like floccules could be observed on the 10W catalyst in Figure . With the increase of Ni content, the spherically shaped metallic particles could be clearly seen and had a better dispersion degree than the 10W catalyst. As a whole, the particles of Ni–W bimetallic catalysts had a better dispersion than the monometallic ones, and both spherical and elliptical particles were observed. Besides, 10W–2.5Ni and 10W–5Ni catalysts had broad distributions between 16–20 and 16–19 nm, respectively. But when the content of nickel increased, the particle size distributions changed from 9–12 to 18–27 nm over 10Ni–10W, 10W–12.5Ni, and 10W–15Ni catalysts, seen in Table . Hence, it was indicated that adjusting Ni and W loadings to an optimum value could obtain small particles with better dispersion. The particle size distributions by TEM were a little bit bigger than those based on XRD results.

Figure 9

TEM images of the catalysts. (a) 10W, (b) 10W–2.5Ni, (c) 10W–5Ni, (d) 10Ni–10W, (e) 10W–12.5Ni, and (f) 10W–15Ni.

Catalytic Activity Evaluation

Tungsten-based catalysts were considered as promising hydrocracking catalysts with the DO performance.[20,47,48] It was found that the jatropha oil could be completely converted into hydrocarbons, and no linoleic acid (C18:2), palmitic acid (C16:0), stearic acid (C18:0), and oleic acid (C18:1) were seen in the products, as shown in Figure S4. Shown in Table , the low bio-oil yield of 19.3 wt % was obtained on 10Ni catalyst, and the main liquid products were composed of C6–C18 straight-chain alkanes, the detailed composition of which is given in Table S3. When fixing the Ni loading at 10%, with increasing W content from 2.5 to 10%, the yield of bio-oil increased from 11.7 to 63.5 wt %. Yet, with the loading of W continuously going up to 15%, the bio-oil yield decreased to 28.7 wt %. The major liquid products were also composed of C6–C18 straight-chain paraffins, a few i-paraffins, aromatics, and naphthenes on Ni–W bimetallic catalysts, as shown in Table S3. Combined with XRD results, it was observed that there was no Ni17W3 phase on the 10Ni, 10Ni–2.5W, and 10Ni–5W catalysts, and the bio-oil yields were low. When the Ni17W3 phase was observed when the W loading reached 10%, the bio-oil yield was the highest and the contents of i-paraffins and aromatics were the lowest. With the W content going up from 10 to 15%, the Ni17W3 crystal size grew up from 7 to 14 nm, the bio-oil yield decreased, and the contents of i-paraffins and aromatics increased.

Table 5

The Bio-Oil Yields of the Jatropha Oil Conversion on All Catalysts

catalysts	bio-oil yield (wt %)	catalysts	bio-oil yield (wt %)
10Ni	19.3	10W	12.5
10Ni–2.5W	11.7	10W–2.5Ni	11.9
10Ni–5W	22.6	10W–5Ni	16.1
10Ni–10W	63.5
10Ni–12.5W	31.6	10W–12.5Ni	42.6
10Ni–15W	28.7	10W–15Ni	31.9

Nickel catalysts were well-known for their hydrogenation performance and had a good selectivity for oxygen removal under mild hydrodeoxygenation conditions.[49−53] The jatropha oil was completely transformed into alkanes with no residual linoleic acid (C18:2), palmitic acid (C16:0), stearic acid (C18:0), and oleic acid (C18:1), as seen in Figure S5. For 10W catalyst, the C6–C18 straight-chain alkanes were the main liquid products as seen in Table S4. Besides, fixing the W loading at 10%, when Ni loading was 2.5%, seen in Table , the bio-oil yield exhibited a slight decrease compared with the 10W catalyst. When the Ni loading changed from 2.5 to 10%, the yield of bio-oil increased from 11.9 to 63.5 wt %. However, when Ni content increased from 10 to 15%, the yield of bio-oil declined to 31.9 wt %. Except a minor amount of i-paraffins, aromatics, and naphthenes on Ni–W bimetallic catalysts, the major liquid products were composed of C6–C18 alkanes, as shown in Table S4. Hexadecanoic acid (C16) and octadecanoic acid (C18) might be transformed to C15–C18 alkanes by hydrodeoxygenation, decarbonylation, or decarboxylation reaction, then the formed long alkanes would be further cracked into short alkanes (C6–C14), ascribing to the existence of acidic sites in the catalysts (seen in Figures S6 and S7 and Table S5).[54] i-Paraffins and aromatics were obtained by isomerization and aromatization reactions. Combined with XRD results, when there existed no Ni17W3 alloy, the bio-oil yields were not high. When the content of Ni was equal to 10%, the bio-oil yield enhanced and the amounts of i-paraffins and aromatics declined with the Ni17W3 alloy formed. When the nickel loading went up continuously from 10 to 15%, the Ni17W3 alloy crystal size became bigger from 7 to 20 nm, while the bio-oil yield decreased from the highest (63.5 wt %) to 31.9 wt %, and i-paraffins and aromatics increased. It implied that when there existed only Ni, W3O, WO2 phases, the bio-oil yields were not high. With the formation of the Ni17W3 alloy phase, the bio-oil yield increased greatly. Besides, the crystal sizes of Ni, W3O, and WO2 phases varied little and that of the Ni17W3 alloy phase increased with increasing Ni or W from 10 to 15%, while the bio-oil yield decreased. Therefore, Ni, W3O, and WO2 phases might not be the main active phases and the crystal sizes slightly affected the bio-oil yield. The Ni17W3 alloy phase was the main active phase, and the increase of its crystal size caused the decrease of bio-oil yield. Ni17W3 might prevent the formation of i-paraffins and aromatics, and the inhibition of isomerization and aromatization reactions decreased with the increase of Ni17W3 particle size.

To further verify the role of Ni17W3, the 10Ni–10W–Ni catalyst (first impregnating Ni, calcination, then impregnating W), 10Ni–10W–W catalyst (first impregnating W, calcination, then impregnating Ni), and the solid mixture of 10Ni and 10W as catalyst (which was mechanically mixed by 3 g total mass of 10Ni and 10W catalysts at the ratio of 1:1 to make the total contents of Ni and W the same as 10Ni–10W) were prepared and the activity experiments under the same conditions were conducted. There were no Ni17W3 diffraction peaks on the above catalysts as seen in Figure S8, consequently, the bio-oil yields were very low compared with 10Ni–10W as shown in Table S6. Therefore, the best activity of 10Ni–10W might really come from the presence of Ni17W3 alloy in this system.

The preparation methods of the catalyst, such as calcination temperature (increased to 700 °C), reduction temperature (increased to 700 °C), the loading amount of Ni and W (increased the loading of Ni and W, they were both 20%, respectively), the impregnation procedure of the components (impregnated Ni or W species first), and the rate of reduction and reduction time (at a fast or slow rate of reduction temperature) were varied, and the catalytic performances of the samples were tested under the same conditions. The results were as follows: (1) when increasing the loading amount of both Ni and W to 20%, the particle size of Ni17W3 increased and the bio-oil yield decreased. (2) When increasing the reduction temperature to 700 °C, the active phases of the catalyst changed. The diffraction peaks of pure Ni17W3 were observed at 2θ = 43.9°, 51.2°, and 74.9° (PDF 65-4828), and the diffraction peaks of W were observed at 2θ = 40.4°, 58.4°, and 73.4° (PDF 89-3728), and the particle size of Ni17W3 grew up to 10 nm, and consequently, the bio-oil yield was low. (3) When increasing the calcination temperature to 700 °C, the overlapped peaks of Ni17W3 and Ni were more obvious than before, and the particle size of Ni17W3 grew up to 13 nm, while the bio-oil yield was 16.9 wt %. (4) When changing the impregnation order of Ni and W, no matter which order was followed for first impregnating Ni or W, the main diffraction peak at about 2θ = 44° was only Ni phase. There were no Ni17W3 phases observed and the bio-oil yields were also low. (5) When increasing or decreasing the rate of reduction temperature, the particle size of Ni17W3 could increase to 11 or 12 nm and the bio-oil yields were not high. Based on the above XRD results as seen in Figure S9, the amount of Ni17W3 phase actually increased, accompanied by the particle size of Ni17W3 increasing by varying the preparation methods of the catalysts, but the bio-oil yields all decreased, as shown in Table S7. These results also supported the previous discussion, that is, the Ni17W3 alloy was the main active phase and the small particle size of Ni17W3 made a big contribution to the high bio-oil yield.

The thermogravimetry (TG)–MS experiment for the best catalyst 10Ni–10W was carried out after the first run. The weight loss rate of the used catalyst was 33%, and there were two weight loss peaks at the temperature of about 350 and 500 °C (seen in Figure S10). The TG results suggested that carbon deposition was serious. The best catalyst 10Ni–10W was picked out to study the reuse. The catalyst was regenerated by calcination and reduction, and then, the catalytic activity was tested under the same conditions. In the second run, the bio-oil yield was 46.0 wt % and the obtained bio-oil was clear, and in the third run, the bio-oil yield was 39.0 wt % and the obtained bio-oil was still clear. The components of the bio-oil were both C6–C18 straight-chain alkanes on the regenerated catalysts (seen in Figure S11). After three runs, the active phases of the best catalyst had no obvious change, as seen in Figure S12, whereas the particle size grew a little bigger than the fresh catalyst and the bio-oil yield decreased, as shown in Table S8. It was further proved that the small particle size of Ni17W3 favored the increase of bio-oil yield. How to keep the stability of this catalytic system needs further study.

Conclusions

In this article, Ni–W bimetallic catalysts realizing complete transformation of jatropha oil was discussed. Compared with monometallic catalysts, the 10Ni–10W/Al-MCM-41 bimetallic catalyst exhibited excellent catalytic performance with 63.5 wt % bio-oil yield. The high catalytic activity of the 10Ni–10W catalyst was due to the existence of small particles of Ni17W3 alloy (about 7 nm). Excessive Ni or W loading would increase Ni17W3 alloy particle size and decrease the bio-oil yield. The Ni–W bimetallic catalyst is a promising hydrodeoxygenation catalyst for biofuel production, and small crystal size of Ni17W3 for high bio-oil yield and desired products could be obtained by adjusting the loading of Ni and W.

Experiments

Synthesis of Support

The MCM-41 support was synthesized by the hydrothermal method.[55] Appropriate amounts of hexadecyl trimethyl ammonium bromide and sodium hydroxide were added to 200 mL of deionized water, and the mixed solution was stirred until these components were fully dissolved at room temperature. Ethyl orthosilicate (TEOS) was added carefully, then the pH value of the solution was adjusted to 5.5 by addition of dilute nitric acid, and stirring for 3 h. The synthesis gel was formed at room temperature and then filtered, water-washed until neutral, dried for 3 h at 110 °C, and subsequently calcined at 540 °C for 8 h.

According to the above procedures, by adding aluminum nitrate (Al(NO3)3·9H2O) to the mixed solution after the addition of TEOS, Al-modified MCM-41 could be obtained.

Synthesis of the Catalysts

All samples were synthesized by the conventional impregnation method and TPR. In brief, a proper amount of a mixed solution of ammonia metatungstate and nickel nitrate hexahydrate (Ni(NO3)2·6H2O) was used for impregnation. The catalyst precursors were obtained by drying at 110 °C, and the impregnated supports were calcined at 600 °C. First, the precursors were reduced under a flow of H2, heated to 400 °C at a fast rate, then the temperature was raised to 600 °C at 2 °C·min–1 and maintained for 2 h. Subsequently, the catalysts were cooled down in a N2 flow overnight. The obtained catalysts were designated as xNi, yW, or xNi–yW, in which “x” and “y” represented the controlled content of nickel and tungsten, separately. Control samples were also prepared by varying the preparation conditions.

Catalyst Characterizations

Atomic absorption spectroscopy was obtained by a VG PQ ExCell to test the actual contents of Ni and W. Before the test, all catalysts were dissolved by a mixed solution of hydrofluoric acid (48%), nitric acid, ammonia (25–28%), hydrogen peroxide (27%), and citric acid.

XRD was conducted on a Shimadzu XRD-6100 diffractometer by using Cu Kα radiation. The XRD pattern was gained in the diffraction angle range of 5°–80°.

XPS spectra were taken by an AXIS Ultra DLD (Kratos) spectrometer using monochromatic Al Kα radiation at 80 eV pass energy. The contaminant carbon (C 1s = 284.6 eV) as an internal standard calibrated the binding energy of the samples.

N2 adsorption and desorption isotherms at 77 K were measured using a Micromeritics TriStar II 3020 analyzer. The surface area was calculated by the Brunauer–Emmett–Teller (BET) equation, whereas the pore volume and average pore diameter were decided by the Barrett–Joyner–Halenda method using the desorption branches of the isotherms.

TPR in H2 was tested on a Micromeritics AutoChem II 2920 instrument. The TPR program was carried out under a mixed flow of H2/Ar at 5 °C·min–1 heating rate to 900 °C, and then kept for 1 h.

An FEI Tecnai G2 20 TWIN instrument (operation voltage 200 kV) with EDX spectrometer obtained TEM images and HRTEM images.

Temperature-programmed desorption (TPD) in NH3 was measured on a Micromeritics AutoChem II 2920 instrument. The TPD program was conducted under a mixed flow of 10% NH3/He at 150 °C, and then kept for 4 h.

Catalytic Performance Evaluation

Using a fixed-bed reactor, the DO of jatropha oil was conducted.[56] Typically, 3 g of catalyst sample (40–60 mesh) was loaded and mixed uniformly by small porcelain balls in the reactor to improve heat transfer. The experiments were conducted at 360 °C and overall pressure of 3 MPa in a mixed flow of H2 and N2. Jatropha oil was continuously added to the reactor with a liquid micro pump for 5 h. The liquid yield (yield) could be calculated based on the following equationwhere Wproduct and W were the mass of total oil products and added jatropha oil during the reaction, respectively. In addition, the repeated activity experiments were carried out, and the standard deviations are listed in Table S9.

Raw Material and Product Analysis

Gas chromatography–mass spectrometry (GC–MS, Agilent 5973) equipped with a capillary column (HP-INNOWax, 30 m × 0.25 mm × 0.25 μm) detected the liquid products and jatropha oil. The temperatures of injector and detector were both 553 K. The temperature program for the GC oven was set as follows: 343 K (4 min), 5 K min–1, 373 K (6 min), 10 K min–1, 433 K, 10 K min–1, 503 K (6 min). The jatropha oil contained linoleic acid (C18:2), palmitic acid (C16:0), stearic acid (C18:0), and oleic acid (C18:1) revealed by GC–MS analysis, as shown in Figure S13, and the peak area contents were about 14.97, 9.50, 5.13, and 19.82% respectively. According to Hua et al.,[57] the composition of jatropha oil from Panzhihua in Southeast China was mainly 40.31 wt % oleic acid (C18:1), 32.69 wt % linoleic acid (C18:2), 17.25 wt % palmitic acid (C16:0), and 7.42 wt % stearic acid (C18:0). Then, the liquid products were quantitatively analyzed by a gas chromatograph (PerkinElmer Clarus 680) equipped with an Elite-Petro column (100 m × 0.25 mm × 0.5 μm). The injector and detector temperatures were both set at 553 K. The temperature program for the GC oven was set as follows: 308 K (10 min), 5 K min–1, 323 K (55 min), 1.3 K min–1, 473 K.