Effect of Water and Glycerol in Deoxygenation of Coconut Oil over Bimetallic NiCo/SAPO-11 Nanocatalyst under N2 Atmosphere

The catalytic deoxygenation of coconut oil was performed in a continuous-flow reactor over bimetallic NiCo/silicoaluminophosphate-11 (SAPO-11) nanocatalysts for hydrocarbon fuel production. The conversion and product distribution were investigated over NiCo/SAPO-11 with different applied co-reactants, i.e., water (H2O) or glycerol solution, performed under nitrogen (N2) atmosphere. The hydrogen-containing co-reactants were proposed here as in-situ hydrogen sources for the deoxygenation, while the reaction tests under hydrogen (H2) atmosphere were also applied as a reference set of experiments. The results showed that applying co-reactants to the reaction enhanced the oil conversion as the following order: N2 (no co-reactant) < N2 (H2O) < N2 (aqueous glycerol) < H2 (reference). The main products formed under the existence of H2O or glycerol solution were free fatty acids (FFAs) and their corresponding Cn−1 alkanes. The addition of H2O aids the triglyceride breakdown into FFAs, whereas the glycerol acts as hydrogen donor which is favourable to initiate hydrogenolysis of triglycerides, causing higher amount of FFAs than the former case. Consequently, those FFAs can be deoxygenated via decarbonylation/decarboxylation to their corresponding Cn−1 alkanes, showing the promising capability of the NiCo/SAPO-11 to produce hydrocarbon fuels even in the absence of external H2 source.


Introduction
Hydrotreating is a typical process in the production of petroleum fuels from fossil crude oils using active sulfided catalysts, e.g., supported NiMo, CoMo, and NiW, operated under hydrogen atmosphere [1]. However, in order to secure economic feasibility as well as the environmental sustainability, many studies have been carried out to improve the process by finding new kinds of alternative resources, applying an inert ambient to the process, and/or using non-sulfided catalysts [2][3][4]. Nowadays, renewable oils, animal fats, and biodiesel from low-grade oils/fats are considered as alternative sources for hydrocarbon fuel production [5][6][7]. Lately, an upgrading of hydrocarbon fuels, which mostly consists of linear alkanes, to higher fraction of branched alkanes for the use as jet fuel-like In this work, we studied the deoxygenation of TG to hydrocarbon alkanes via catalytic deoxygenation (under N 2 ambient) and in-situ HDO (by H 2 O or glycerol addition) in comparison with conventional HDO (under H 2 ambient) to yield oxygen-free hydrocarbons. The NiCo (about 10 wt% metal loading) supported on commercially available silicoaluminophosphate-11 (SAPO-11) was used as the deoxygenation catalyst. SAPO-11 is a molecular sieve material with 10-ring porous structure that has been extensively utilized due to its stability and potential activity in many catalytic processes [23,24]. The catalytic hydro/deoxygenation reactions of medium-chain (mostly C 12 ) TGs in coconut oil, without hydrogenation treatment prior to deoxygenation, were examined to describe the reaction pathway via the degree of oxygen removal and product distribution, depending on the variable feed molecules. To the best of our knowledge, it is reported for the first time that SAPO-11-supported NiCo nanoparticles could directly convert coconut oil to jet fuel-like hydrocarbons with the in-situ hydrogen donors.

Catalyst Preparation
The catalyst was prepared by co-impregnation technique by dissolving nitrate salts of Ni and Co in D.I. water. Then, SAPO-11 powder was further added to obtain 5 wt% loading of each metal. Mixing as well as drying were employed using a magnetic stirrer hot plate at 85 • C for 3 h and transferred to vacuum oven at the same temperature for overnight. The resultant powder was calcined in static air at 500 • C for 5 h. The final sample was denoted as NiCo/SAPO-11. Before catalytic testing, the catalyst powder was pelleted, sieved into 0.5-1.0 mm. in size.

Catalyst Characterization
Crystalline phase of the prepared catalyst was analysed by X-ray diffraction (XRD) measurement using an X-ray diffractometer (D8 ADVANCE, Bruker, Ltd., Karlsruhe, Germany) with Cu Kα radiation (wavelength of 1.5406 Å). The diffraction angle was scanned from 2θ of 5 • to 80 • , with steps of 0.5 • /min and counting at 0.02 s/step. The specific surface area and pore structures of the samples were calculated by the Brunauer-Emmett-Teller (BET) and the density functional theory (DFT)/Barrett-Joyner-Halenda (BJH) methods, respectively, via nitrogen adsorption-desorption technique (Quantachrome Autosorb-iQ3, Boynton Beach, FL, USA). The total pore volume was measured at the relative pressure (P/P0) of 0.98. Before measurement, the sample of approximately 0.2 g was outgassed at 120 • C for 3 h under helium flow. Total acidity and strength of acid sites of catalyst were evaluated by ammonia (NH 3 )-temperature programmed desorption (NH 3 -TPD) carried out using Quantachrome Chemisorption Analyzer ChemStar TPX Series (Boynton Beach, FL, USA). Scanning electron microscopy (SEM) analysis was performed with a S-3400N (Hitachi-Science & Technology, Berkshire, UK), operated at 5 kV. Transmission electron microscopy (TEM) images were acquired by JEOL-JEM-2100Plus instrument (Akishima, Tokyo, Japan) equipped with an energy dispersive X-ray spectrometer (EDS). The TEM machine was operated at an accelerating voltage of Nanomaterials 2020, 10, 2548 4 of 15 200 kV. Before measurement, the sample was reduced at 700 • C in a flow of H 2 for 2 h. The obtained powder was grounded then dispersed in ethanol, followed by sonication for 30 min to make the particle well-dispersed, and dropped onto a gold grid.

Catalytic Deoxygenation Testing in Continuous-Flow Fixed-Bed Reactor
The deoxygenation reaction was operated in a custom-made continuous-flow trickle-bed reactor made of the stainless steel 316, with an external diameter of 1.28 cm, internal diameter of 1 cm, and length of 70 cm. The catalyst bed was located at the middle of the fixed-bed reactor and held in position with quartz wool plugs at uniform temperature position (confirmed by inside thermocouples inserted in the catalyst bed). The reaction in the continuous system consists of a feed tank, a reactor unit, and a gas-liquid product separation unit, as shown in Figure 1. The temperature was controlled by using electrical furnace with two positions of temperature controller. A K-type thermocouple was inserted to contact the catalyst bed for temperature monitoring. The pressure of the reactor was manually controlled by using back pressure regulator (BPG) for two-phase flow. High-performance liquid chromatography (HPLC) pump and mass flow controller (MFC) were used to control the flow rate of liquid and gas feed, respectively.

Catalytic Deoxygenation Testing in Continuous-Flow Fixed-Bed Reactor
The deoxygenation reaction was operated in a custom-made continuous-flow trickle-bed reactor made of the stainless steel 316, with an external diameter of 1.28 cm, internal diameter of 1 cm, and length of 70 cm. The catalyst bed was located at the middle of the fixed-bed reactor and held in position with quartz wool plugs at uniform temperature position (confirmed by inside thermocouples inserted in the catalyst bed). The reaction in the continuous system consists of a feed tank, a reactor unit, and a gas-liquid product separation unit, as shown in Figure 1. The temperature was controlled by using electrical furnace with two positions of temperature controller. A K-type thermocouple was inserted to contact the catalyst bed for temperature monitoring. The pressure of the reactor was manually controlled by using back pressure regulator (BPG) for two-phase flow. High-performance liquid chromatography (HPLC) pump and mass flow controller (MFC) were used to control the flow rate of liquid and gas feed, respectively.
In a typical run, 4 mL of catalyst was loaded into a reactor and reduced with 50 cm 3 /min of pure H2 at 650 °C for 2 h. After the catalyst reduction step, the continuous reaction system was immediately flushed with reaction gas (N2 or H2) and set up to the desired temperature under pressurized condition. Liquid feeds-coconut oil, water, and glycerol solution-were separately introduced by two-suction line in the HPLC pump. Four different experimental practices were performed as follows: Practice #1 was designed as a reference experiment for the purpose of investigating the catalytic performance on HDO of feed oil over NiCo/SAPO-11 catalyst. The reaction was performed at 330 °C under 50 bar of H2 atmosphere, and liquid feed (i.e., coconut oil) was introduced into the reactor with liquid hourly space velocity (LHSV) of 1 h −1 .
Practice #2 aimed to investigate the catalytic performance on the deoxygenation with absent H2. The procedure was similar to that described in practice #1, except the operating gas which was N2.
Practice #3 aimed to investigate the catalytic performance on the coupling deoxygenation of feed oil with in-situ H2 production by glycerol-APR. In short, the feed was introduced into the reactor by two-suction lines in HPLC pump, one for oil feed (80% by vol.) and one for glycerol solution feed (20% by vol.). The glycerol solution of 36 wt% glycerol in water (140 g of glycerol in 250 g of water) which was 0.1 mole ratio of glycerol: water) was used at a total LHSV of 1 h −1 .
Practice #4 aimed to investigate the catalytic performance on the solvent-free HDO by adding 5% by volume of water into the reaction. In a typical run, 4 mL of catalyst was loaded into a reactor and reduced with 50 cm 3 /min of pure H 2 at 650 • C for 2 h. After the catalyst reduction step, the continuous reaction system was immediately flushed with reaction gas (N 2 or H 2 ) and set up to the desired temperature under pressurized condition. Liquid feeds-coconut oil, water, and glycerol solution-were separately introduced by two-suction line in the HPLC pump.
Four different experimental practices were performed as follows: Practice #1 was designed as a reference experiment for the purpose of investigating the catalytic performance on HDO of feed oil over NiCo/SAPO-11 catalyst. The reaction was performed at 330 • C under 50 bar of H 2 atmosphere, and liquid feed (i.e., coconut oil) was introduced into the reactor with liquid hourly space velocity (LHSV) of 1 h −1 .
Practice #2 aimed to investigate the catalytic performance on the deoxygenation with absent H 2 . The procedure was similar to that described in practice #1, except the operating gas which was N 2 .
Practice #3 aimed to investigate the catalytic performance on the coupling deoxygenation of feed oil with in-situ H 2 production by glycerol-APR. In short, the feed was introduced into the reactor by two-suction lines in HPLC pump, one for oil feed (80% by vol.) and one for glycerol solution feed (20% by vol.). The glycerol solution of 36 wt% glycerol in water (140 g of glycerol in 250 g of water) which was 0.1 mole ratio of glycerol: water) was used at a total LHSV of 1 h −1 .
Practice #4 aimed to investigate the catalytic performance on the solvent-free HDO by adding 5% by volume of water into the reaction.

Liquid and Gaseous Product Analysis
Two-phase back pressure regulator was used to reduce a mixed-phase flow streams from high pressure into ambient before they went to a liquid-gas phase separator placed downstream. The liquid products were analysed by gas chromatography-mass spectrometry (GC-MS, Shimadzu GCMS-QP2020, Kyoto, Japan) equipped with a capillary column (DB-1HT, 30 m × 0.32 mm × 0.1 µm) to inform the quantitative/qualitative data of FFAs, hydrocarbons, and other intermediate products.
Detailed information of this method is applied elsewhere [25,26]. Briefly, 10 µL of sample was injected into the injection port in a split mode with the inlet temperature of 340 • C. In a GC oven, the oven temperature was kept at 40 • C for 5 min then increased to 240 • C with a rate of 15 • C/min, then ramped by 8 • C/min to 370 • C, which was maintained for another 15 min. The total analysis run time was about 50 min. In the MS zone, the temperatures of ion source and interface were held at 250 • C during the analysis. The m/z was scanned from 2 to 500. The calibration curves of alkane standards (n-C 8 to n-C 18 ) were used to quantify their weight in the liquid products, and the response factor for isomer products was assumed to be equal to its corresponding n-alkanes. For other oxygenated intermediates, the calculation was referenced by the same type of known standards.
Gas phase was analysed using an online GC equipped with two columns (molecular sieve 5A and Porapak Q), thermal conductivity detector (TCD) and flame ionization detector (FID). The GC was calibrated for all of the gaseous products obtained, including CO, CO 2 , CH 4 , C 2 H 6 , and C 3 H 8 .

Catalyst Characterization
The XRD pattern of NiCo/SAPO-11 ( Figure 2a) suggested an alloy formation between Ni and Co, matched with PDF 01-074-5694, in accordance with Huynh and co-workers [27] who prepared bimetallic Ni-Co supported on different acidic materials used in Phenol HDO. Figure 2b,c show porous characteristics of the NiCo/SAPO-11 and SAPO-11 support. The obtained isotherms for all samples could be classified as type IV with hysteresis loops at 0.4 < P/P 0 < 0.8 of H4 type [28]. These results showed that samples possess mesopore structures [29]. High adsorption of N 2 appeared in the low relative pressure range, and the hysteresis loops were detected, suggesting the existence of both micropores and mesopores. The pore size distribution plots calculated by DFT method revealed two types of peaks with narrow micropores (high intensity at pore diameter < 2 nm) and mesopores (broad peak at pore diameter ≥ 2 nm) at approximately 0.98 and 7.78 nm, respectively. Upon NiCo loading, the BET surface area and pore volume of NiCo/SAPO-11 were~134.4 m 2 /g and 0.140 cm 3 /g, respectively, which were slightly lower than those of pristine SAPO-11 (~154.7 m 2 /g and 0.139 cm 3 /g, respectively).
Nanomaterials 2020, 10, x FOR PEER REVIEW 6 of 15 loading, the BET surface area and pore volume of NiCo/SAPO-11 were ~134.4 m 2 /g and ~0.140 cm 3 /g, respectively, which were slightly lower than those of pristine SAPO-11 (~154.7 m 2 /g and ~0.139 cm 3 /g, respectively). The SEM image (Figure 3a) indicated that 10 wt% metal loading does not change the microstructure of SAPO-11 support, which is normally rod-like and has non-smooth surface. The bright field TEM image (Figure 3b) of the catalyst clearly shows Ni-Co nanoparticles, presented in black spots, in a spherical shape dispersed on the support, presented in the dark grey area. The metal particle size, estimated by measuring at least 100 particles, was about 9-18 nm. This range is consistent with San-Jose-Alonse and co-workers [30], who varied the proportion of Ni:Co. They showed that the particle size of samples prepared by impregnation was ~14.1 nm. (for Co), ~8.6 nm.  The SEM image (Figure 3a) indicated that 10 wt% metal loading does not change the microstructure of SAPO-11 support, which is normally rod-like and has non-smooth surface. The bright field TEM image (Figure 3b) of the catalyst clearly shows Ni-Co nanoparticles, presented in black spots, in a spherical shape dispersed on the support, presented in the dark grey area. The metal particle size, estimated by measuring at least 100 particles, was about 9-18 nm. This range is consistent with San-Jose-Alonse and co-workers [30], who varied the proportion of Ni:Co. They showed that the particle size of samples prepared by impregnation was~14  (Figure 3b) that Ni and Co species appeared at identical areas of the bimetallic particles, possibly due to the alloy structure formation, in good agreement with previous XRD results ( Figure 2). Figure 3e,f display the high-resolution TEM (HRTEM) image of a single NiCo nanoparticle and its corresponding EDS line profile, respectively. These results confirm the co-existence of Ni and Co with comparable content of each metal, which was consistent with the analysis result from an inductively coupled plasma-mass spectrometry (ICP-MS) (results not shown). Additionally, lattice spacings of around 1.28 and 2.42 nm which were rotated by 53 • were evidenced [31]. Evaluated by the NH 3 -TPD technique, total acidity of the SAPO-11 was 0.456 mmol/g, whereas NiCo/SAPO-11 catalyst showed lower values at 0.338 mmol/g (results not shown), coincided with the lower surface area of the supported catalyst.     Table 1 shows fatty acid compositions directly evaluated by GC-MS. It is seen that C 12 , C 16 , and C 18 fatty acids are major components in coconut oil, with oxygen content~13 wt%, the ratio of saturated/unsaturated FFAs~3 times, which might induce different product distributions. The molecular weights for further calculation were 779.7 g/mol for coconut oil. 3.3. Hydrogen Production from Glycerol Aqueous-Phase Reforming (Gly-APR) over NiCo/SAPO-11

Feed Compositions
As of the complexity during the glycerol-APR process, the reactions of dehydration, dehydrogenation and hydrogenolysis could possibly take place on NiCo/SAPO-11 since the fact that this bifunctional catalyst comprises two main catalytic sites; (1) metal sites prefer to dehydrogenation and (2) acid sites prefer dehydration. Table 2 summarizes the main products such as propylene glycol (C 3 H 8 O 2 ), ethylene glycol (C 2 H 6 O 2 ), acetaldehyde (C 2 H 4 O), 1-propanol, 2-methyl-(C 4 H 10 O), 2-propanone, 1-hydroxy-(C 3 H 6 O 2 ), and ethanol (C 2 H 5 OH) obtained in the condensable phase, which can be confirmed by the fact that side reactions occurred. The most abundant liquid product was ethylene glycol (~17.8%), followed by propylene glycol (~11.0%), and 2-propanone, 1-hydroxy-(~10.5%). In addition, trace amounts of 1-propanol, 2-methyl-, acetaldehyde, and ethanol could be detected (~7.1%). These results show a good agreement with the literature in that the same intermediates were reported [32,33]. The comparatively higher selectivity to ethylene glycol indicated that the primary hydroxyl group of glycerol is eliminated by the dehydrogenation pathway. However, the formation of propylene glycol and 2-propanone, 1-hydroxy-indicated that both pathways of the dehydration/hydrogenolysis were presented on Lewis sites of the catalyst. It also points out the high C-C, C-H cleavage activity of catalyst, accompanied by an increased production of hydrogen. As NiCo/SAPO-11 catalyst was identified to be of weak and medium strength acid sites (i.e., Lewis sites), this would lead to the formation of oxygenated hydrocarbons in the liquid phase (i.e., acetaldehyde, 1-propanol, 2-methyl-, 2-propanone, 1-hydroxy-, and ethanol), suggesting that Lewis sites are involved in C-O cleavage. The result also signifies that dehydrogenation, which competes with dehydration/hydrogenolysis, was significantly favoured over bifunctional NiCo/SAPO-11 catalyst. Table 2. The list of investigated intermediates and their selectivity in liquid product during the glycerol aqueous-phase reforming over NiCo/SAPO-11 catalyst according to the GC-MS analysis.

Liquid Product Retention Time (min) Selectivity (%)
Acetaldehyde Nanomaterials 2020, 10, x FOR PEER REVIEW 9 of 15                  Overall, the reaction pathways could be proposed that the reaction involves the C-O bond scission, followed by hydrogenation which consumes H 2 . Therefore, a part of H 2 could be consumed and resulted in limited hydrogen production. The formation of ethylene glycol generated from dehydrogenation would support this assumption. Table 3 shows the main components of the gaseous products: CH 4 > CO > CO 2 > H 2 , respectively. As mentioned above, H 2 yield was hardly maintained because H 2 was consumed by side reaction (in liquid phase). Notably, the formation of CH 4 also confirmed that H 2 was used in CO/CO 2 methanation (CO + 3H 2 ↔ CH 4 + H 2 O, ∆H 298 = −206 kJ/mol and CO 2 + 4H 2 ↔ CH 4 + 2H 2 O, ∆H 298 = −164.94 kJ/mol) [33]. Generally, Ni species prefer to generate more CO. However, the amount of CO formed was higher than that of CO 2 , suggesting that bimetallic NiCo/SAPO-11 catalyst was favourable to the decomposition of methanol. Co-based catalysts have been reported as less effective materials for the reforming of methanol [34,35]. However, Xue and co-worker [36] promoted Co into NiCu/Al catalyst for methanol reforming. The increased H 2 amount showed that Co could promote the reaction by enhancement of water gas shift reaction. Papadopoulou and co-worker [37] investigated the adsorption of methanol on Co-based catalysts. TPD results indicated that the number of active sites provided the catalyst to be able to adsorb or decompose methanol molecules. In Table 3, methanol was not detected among the condensable phase products suggesting the possible reactions of methanol reforming (CH 3 OH + H 2 O ↔ CO 2 + 3H 2 ) or methanol decomposition (CH 3 OH ↔ CO + 2H 2 ) over NiCo/SAPO-11 catalyst. This behaviour coincided with the synergistic effects of Ni-CO alloy in a NiCo/SAPO-11 catalyst in a positive way for H 2 production. Figure 4 displays the results of coconut oil HDO according to the change of co-reactant feed at 330 • C, LHSV 1 h −1 , and 50 bar of reaction gas over 4 mL of NiCo/SAPO-11 catalyst. Under H 2 ambient, the catalyst exhibited high activity with 100% conversion, while more than 85% of coconut oil conversion could yield the liquid product. Under an inert N 2 atmosphere, the conversion was 30% and liquid product accounted for 15%. With the help of co-reactants, H 2 O and aqueous glycerol, the conversion increased to 70% and 90%, whereas liquid product yield also increased to 47% and 87%, respectively. Compared to the conventional H 2 ambient, the results verified that the presence of H 2 O and glycerol helped to promote both coconut oil conversion and product yield. In addition, the experimental practices applying H 2 below 50 bar (i.e., 30 bar) with and without glycerol solution were also performed. Even though 100% conversion was also achieved, hydrocarbon liquid yields were dropped to 65-67%, compared with using either 50 bar H 2 without glycerol (Practice #1) or 50 bar N 2 with glycerol solution (Practice #3). This could be because the existence of H 2 O under H 2 ambient caused side reaction. Thus, the glycerol did not clearly selective to glycerol reforming.    Figure 5 shows the liquid product distribution over NiCo/SAPO-11 catalyst. With the help of H 2 , the catalyst could provide excellent selectivity to alkanes at 97%. Several intermediates (i.e., ketones, aldehydes, alcohols, cycloalkanes esters, FFAs, and alkenes) were detected in the liquid phase, while the C 3 H 8 , H 2 , CO, and CO 2 were mainly formed in the gas phase. However, the product distribution was altered (see Figure 6). Under N 2 atmosphere, main products are alkanes (C 11 -C 18 ) and a small amount of FFAs, alkenes, and some alcohols. This finding is consistent with Morgan and co-workers [13], who established the thermal deoxygenation of TGs without H 2 over 1 wt% Pt/C, 5 wt% Pd/C, and 20 wt% Ni/C catalysts. However, the existence of fatty acid composition was an important intermediate for alkane formation. Long chains of alkenes and alcohols were produced via hydrocracking. It can be implied that H 2 generated from cracking affected the hydrocarbon production [38]. In the presence of H 2 O, CO and CO 2 were mainly formed, while propane was not found in the bulk of gas product (see Figure 6). It can be considered that FFAs are successfully produced by the TG hydrolysis reaction route rather than deoxygenation which requires H 2 molecules. However, the produced FFAs will either undergo DCO 2 or DCO without H 2 gas.

Catalytic Performance of NiCo/SAPO-11 over In-Situ HDO of Coconut Oil
As also shown in Figure 6, the existence of methane (CH 4 ) showed the consumption of H 2 . As a result, in-situ H 2 produced via the water gas shift reaction (CO + H 2 O ↔ CO 2 + H 2 , ∆H 0 298 = −41 kJ/mol) [37] circulated in the system. Interestingly, two major side products were detected: high molecular weight esters or fatty esters (such as dodecanoic acid, dodecyl ester) and acetate compounds (such as octacosyl acetate). The formation of fatty esters might be due to the dehydration reaction of FFAs and alkyl alcohols on acid sites. It is believed that large molecular weight hydrocarbons lead to the deactivation of the catalyst by blocking the catalyst pores and depositing as coke. In addition, with an acidic functionality, acetate can be obtained from dehydrogenation of alcohol couples with aldehyde under an inert atmosphere to yield an acetate [38].  As also shown in Figure 6, the existence of methane (CH4) showed the consumption of H2. As a result, in-situ H2 produced via the water gas shift reaction (CO + H2O ↔ CO2 + H2, ∆ = −41 kJ/mol) [37] circulated in the system. Interestingly, two major side products were detected: high molecular weight esters or fatty esters (such as dodecanoic acid, dodecyl ester) and acetate compounds (such as octacosyl acetate). The formation of fatty esters might be due to the dehydration reaction of FFAs and alkyl alcohols on acid sites. It is believed that large molecular weight hydrocarbons lead to the deactivation of the catalyst by blocking the catalyst pores and depositing as coke. In addition, with an acidic functionality, acetate can be obtained from dehydrogenation of alcohol couples with aldehyde under an inert atmosphere to yield an acetate [38].
In H2O, the amount of FFAs, originating from hydrolysis of TGs, was much higher. This is in accordance with the literature [26,39,40], in which the disappearing of C=O of esters (in the TGs molecules) and the obtaining of C=O of carboxylic acids (in the FFAs molecules) in the reaction by using H2O were reported. This result confirmed that hydrolysis was the dominant pathway. The presence of H2O accelerated and promoted the conversion of TG molecules to FFAs. There was no consumption of hydrogen for the hydrolysis of the TGs. However, the H2 could be further utilized to promote deoxygenation of FFAs (DCO or DCO2). Therefore, the formation of the desired hydrocarbons was very low. It is known that when there was a lack of H2, which inhibits the catalytic performance, the reaction underwent cracking, for example: C10H22 → C3H6 + C7H16. Initially, hydrogen is present in the system because of some hydrocarbon molecules rearranging itself into a cycloalkane or an aromatic hydrocarbon with the loss of H2-for example: C6H14 → C6H12 + H2-and thus after prolonged reaction times, when this reservoir was consumed, an enhanced formation of aromatic Cn-1 hydrocarbons was observed. Those of FFAs will either undergo DCO2 or DCO. This interesting phenomenon showed that alkenes were formed using either N2 or H2O. In the latter case, there was enough hydrogen and the formation hydrocarbons was almost constant. In H 2 O, the amount of FFAs, originating from hydrolysis of TGs, was much higher. This is in accordance with the literature [26,39,40], in which the disappearing of C=O of esters (in the TGs molecules) and the obtaining of C=O of carboxylic acids (in the FFAs molecules) in the reaction by using H 2 O were reported. This result confirmed that hydrolysis was the dominant pathway. The presence of H 2 O accelerated and promoted the conversion of TG molecules to FFAs. There was no consumption of hydrogen for the hydrolysis of the TGs. However, the H 2 could be further utilized to promote deoxygenation of FFAs (DCO or DCO 2 ). Therefore, the formation of the desired hydrocarbons was very low. It is known that when there was a lack of H 2 , which inhibits the catalytic performance, the reaction underwent cracking, for example: C 10 H 22 → C 3 H 6 + C 7 H 16 . Initially, hydrogen is present in the system because of some hydrocarbon molecules rearranging itself into a cycloalkane or an aromatic hydrocarbon with the loss of H 2 -for example: C 6 H 14 → C 6 H 12 + H 2 -and thus after prolonged reaction times, when this reservoir was consumed, an enhanced formation of aromatic C n-1 hydrocarbons was observed. Those of FFAs will either undergo DCO 2 or DCO. This interesting phenomenon showed that alkenes were formed using either N 2 or H 2 O. In the latter case, there was enough hydrogen and the formation hydrocarbons was almost constant.
The initial in-situ H 2 can be generated from glycerol reforming (C 3 H 8 O 3 + 3H 2 O ↔ 3CO 2 + 7H 2 , ∆H 0 298 = +128 kJ/mol) [41] and glycerol decomposition (C 3 H 8 O 3 ↔ 3CO + 4H 2 , ∆H 0 298 = +250 kJ/mol) [42,43]. The formation of H 2 , CO, and CO 2 ( Figure 6) confirmed the utilization of glycerol co-feeding to generate in-situ H 2 . These H 2 initially aid the triglyceride breakdown to form propane and FFAs. Unfortunately, the abundant FFAs still remained in the liquid product. It exhibited the insufficiency of in-situ H 2 so that alkanes could not be considerably produced. However, this in-situ H 2 helps to decrease the level of high molecular weight hydrocarbon compounds. Table 4 shows the proportion of DCOx/HDO and CO 2 /CO. The ratio of DCOx/HDO declined when applying co-reactant as follows: N 2 > aqueous glycerol > H 2 O. The results indicated that generated H 2 had a significant impact on reaction pathways. At the initial of reaction with glycerol co-reactant, an abundance of generated H 2 can break C-O bonds in TG molecules through the HDO reaction to produce propane and FFAs. However, considering the final gas product, the reaction prefers to promote via DCO 2 which does not require H 2 . It can be said that the rate of breaking of the C-O bonds in FFA should be slower than the C-O bonds in TGs and C-C bonds in FFA itself. On the other hand, the reaction seems to contribute to DCO when using H 2 O co-reactant. These results indicated that the insufficient H 2 at the beginning suppressed HDO. Once FFA was generated, H 2 from water gas shift reaction promoted DCO pathway, consistent with other reports [10,14,[42][43][44][45][46]. In all conditions, the selectivity to jet fuel range (C 8 -C 14 ) was higher than that to diesel fuel range (C 15 -C 18 ).

Conclusions
The catalytic deoxygenation of coconut oil was performed on NiCo/SAPO-11 nanocatalyst in a continuous-flow reactor for hydrocarbon production. Reaction condition in assistance with in-situ H 2 produced from glycerol (50 wt% aqueous solution), without any external H 2 source (under N 2 ambient) and addition of H 2 O (5% by volume) was investigated on the deoxygenation reaction and product distribution. The results showed that under N 2 atmosphere applying co-reactants H 2 O and glycerol significantly increased the oil conversion. The main products were FFAs and their corresponding C n−1 alkanes, whereas a small fraction of the intermediates was found. Considering FFA products, the addition of H 2 O aids the triglyceride breakdown into FFAs, and the addition of glycerol solution as the hydrogen donor is favourable for the initial hydrogenolysis into FFAs. Then, those FFAs can be either decarbonylated or decarboxylated to their corresponding C n−1 alkanes, in the fraction of jet and diesel fuels. The NiCo/SAPO-11 showed great potential in the production of advanced biofuels in the absence of an external H 2 source.

Conflicts of Interest:
The authors declare no conflict of interests.