The Production of Green Diesel Rich Pentadecane (C15) from Catalytic Hydrodeoxygenation of Waste Cooking Oil using Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2

Hydrodeoxygenation (HDO) is applied in fuel processing technology to convert bio-oils to green diesel with metalbased catalysts. The major challenges to this process are feedstock, catalyst preparation, and the production of oxygen-free diesel fuel. In this study, we aimed to synthesize Ni catalysts supported on silica-zirconia and aluminazirconia binary oxides and evaluated their catalytic activity for waste cooking oil (WCO) hydrodeoxygenation to green diesel. Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 were synthesized by wet-impregnation and hydrodeoxygenation of WCO was done using a modified batch reactor. The catalysts were characterized using X-ray diffraction (XRD), Xray fluorescence (XRF), and scanning electron microscopy energy dispersive X-ray spectroscopy (SEM-EDS), and N2 isotherm adsorption-desorption analysis. Gas chromatography mass spectrometry (GC-MS) analysis showed the formation of hydrocarbon framework n-C15 generated from the use of Ni/Al2O3-ZrO2 with the selectivity of 68.97% after a 2 h reaction. Prolonged reaction into 4 h, decreased the selectivity to 58.69%. Ni/SiO2-ZrO2 catalyst at 2 h showed selectivity of 55.39% to n-C15. Conversely, it was observed that the reaction for 4 h increased selectivity to 65.13%. Overall, Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 catalysts produced oxygen-free green diesel range (nC14-C18) enriched with n-C15 hydrocarbon. Reaction time influenced the selectivity to n-C15 hydrocarbon. Both catalysts showed promising hydrodeoxygenation activity via the hydrodecarboxylation pathway. Copyright © 2021 by Authors, Published by BCREC Group. This is an open access article under the CC BY-SA License (https://creativecommons.org/licenses/by-sa/4.0).


Introduction
The massive production of CO2 from the combustion of fossil fuel is suspected to initiate the cut down on carbon emissions in the atmosphere. The utilization of biomass [2] and fatty oil [3] renewable feedstock for biofuel production has been reported. Both feedstocks have a problem on a limitation for the mass production of bio-oil, but the low cost and abundant material become an interesting point that may decrease consumption of fossil oil.
Large scale production and the commercial use of the first generation fatty acid methyl ester (FAME) biodiesel is constrained due to high oxygen content which reduces energy density, high viscosity, and poor storage properties [4,5]. Therefore, many researchers are developing different routes for converting fatty oil into biodiesel which is fully compatible with fossil diesel fuel. The second-generation biodiesel is called 'green diesel' that is paraffin-rich diesel that is free from oxygen, sulfur, nitrogen, or aromatic hydrocarbon. The green diesel is produced from the catalytic hydrotreating of renewable feedstock containing triglycerides and fatty acids [6]. Green diesel has a high cetane value and flash value suitable as an alternative fuel for internal combustion engines (ICEs) [7]. Biodiesel production from waste material may ease waste management challenges of waste disposal. Waste cooking oil is an environmentally benign and non-edible feedstock rich in free fatty acid content. Waste cooking oil is a prospective feedstock in catalytic hydrodeoxygenation (HDO) for green diesel production [8].
A series of HDO reactions using WCO feedstock is reported in many aspects. A variety of product distribution mostly heavier straightchain hydrocarbon within the diesel range (C15-C18) has previously been generated from waste cooking oil and fatty acids via catalytic hydrodeoxygenation [9]. In a report by [10], the effect of catalyst support on HDO of waste cooking oil was investigated in a stainless steel autoclave reactor. Product distribution is reported to vary between n-C17 and n-C18 hydrocarbon depending on support type. In other reports, n-C17 and n-C18 were detected as major constituents in products obtained from HDO of waste cooking oil [11][12][13]. The product distribution of hydrocarbon between C15-C18 from biobased feedstock possesses diesel fuel properties similar to conventional diesel with high energy density and good oxidative stability [14].
Heterogeneous bimetallic sulfur-free catalysts, such as NiCu [15], NiMo [16], and NiW [17] on mesoporous and oxide supports, have been studied extensively in the hydrodeoxygenation of fatty acids of vegetable oil to biodiesel. Advantages of heterogeneous Ni-based catalysts are larger active surfaces for better cata-lytic activity, easy recovery from the reaction mixture, and reusability. Noble metal catalysts have shown outstanding activity in HDO of bio-oil for biodiesel production [12,18]. Despite the high catalytic activity, noble metals are expensive and inflate catalyst production costs. In our previous works, we reported an interesting phase phenomenon with SiO2 in Ni/ZrO2-SiO2 nanocomposite synthesized by the wet impregnation method [19]. SiO2 underwent crystallinity transformation from crystalline to amorphous due to composite formation with ZrO2. A reverse-phase formation from amorphous to crystalline SiO2 was detected after Ni impregnation. To understand the effect of binary support on Ni catalyst, we tested Ni/SiO2-ZrO2 for phenol hydrodeoxygenation in a stainless autoclave batch reactor [20]. The catalyst showed remarkable activity for ring hydrogenation with the selective formation of cyclohexanol and cyclohexanone. In another report, Ni supported by single oxides SiO2, ZrO2, and Al2O3 were synthesized and tested for phenol hydrodeoxygenation [21]. Ni/SiO2 showed activity for phenol ring-opening with product formation of 2-methyl-1-butanol and Ni/ZrO2 promoted hydrogenation leading to the formation of cyclohexanone.
In this work, we synthesized Ni catalysts supported on binary oxides of SiO2-ZrO2 and Al2O3-ZrO2 by the wet impregnation method. Catalyst physicochemical properties, such as: crystalline structure, Ni loading, surface morphology, BET surface area, pore volume, and pore diameter, were studied by XRD, XRF, SEM-EDS, and N2 isotherm adsorptiondesorption analysis respectively. Catalyst performance was tested for waste cooking oil hydrodeoxygenation to green diesel in a modified batch reactor at 350 °C, 2 MPa H2, 2 h, and 4 h. All reactions were conducted in a solvent-free environment. Analysis using GC-MS gave the formation of n-C15 as major constituents in green diesel fraction. To our knowledge, this is the first report on the production of green diesel rich with n-C15. It is suggested that the predominant reaction might be catalytic decarboxylation of the oleic acid.

Catalyst preparation
The binary oxide Al2O3-ZrO2 and SiO2-ZrO2 was previously prepared in a 1:1 (wt/wt) ratio according to the reported methods [19,20]. 10 g of Al2O3-ZrO2 was prepared by dissolving 13 g of ZrOCl2.8H2O in deionized water (DI) with the dropwise addition of NH4OH until pH 8 was identified by the formation of Zr(OH)4 as a white precipitate. The precipitate was directly mixed with 5 g of -Al2O3 which was then aged for 10 h at a constant temperature of 75 °C. After aging, the solid mixture was dried to constant weight and calcined under N2 gas for 5 h at 500 °C to obtain the Al2O3-ZrO2 composite. The SiO2-ZrO2 composite was also prepared by a similar procedure by mixing 10 g of SiO2 with the Zr(OH)4 precipitate, aged, dried, and calcinated to give SiO2-ZrO2 composite.
The Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 were prepared by the wet impregnation method, by mixing 7.4 g of Ni(NO3)2.6H2O with 10 g of Al2O3-ZrO2 in DI water. The mixture was agitated overnight and then evaporated to dryness to form a solid mixture. The solid mixture was calcined at 500 °C for 5 h under continuous N2 gas flow followed by reduction under continuous H2 gas flow at 500 °C for 5 h to form 15% Ni/Al2O3-ZrO2 (Ni/AZ) and 15% Ni/SiO2-ZrO2 (Ni/SZ), respectively.

Catalyst characterization
The crystal structures of the catalysts were analyzed using an X-Ray Diffractometer (XRD) PAN Analytical E'xpert Pro operated at 40 kV, 35 mA with Cu-K  radiation of λ = 1.54060 and 2θ = 10°-80°. SEM measurements at a scan rate of 1°/min were conducted to determine the catalyst surface morphology. The SEM data were collected from PHENOM. Specific surface area is determined from BET isotherms using nitrogen gas at 77 K measured by Belsorp Mini II (Microtrac BEL) at 77 K. X-Ray Fluorescence (XRF) Pan Analytical Type Minipal 4 was used to determine actual Ni loading on Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2, respectively. Potassium Bromide (KBr) pellets with 400-4000 cm −1 spectral range in a Shimadzu 8400S were used to record the Fourier-Transform Infrared Spectra (FT-IR) of the catalysts.

Waste cooking oil pre-treatment method
Waste cooking oil pre-treatment was conducted according to a report by Li et al. [22]. A 500 g of the waste cooking oil collected from a halal food court center in Brawijaya University was mixed with 700 mL of saturated salt solution and heated at 80 °C for 3 h under continuous magnetic stirring. After heating, the solution was placed in a separatory funnel for 12 h and the upper oil layer was collected and mixed with 500 mL of DI water containing 4 g of activated carbon (AC). The mixture was heated at 80 °C for 3 h with stirring. Afterward, the activated carbon was removed by vacuum filtration and the oil was placed in a separatory funnel for layering. The upper oil layer was collected and dried with anhydrous Na2SO4. The Na2SO4 was removed by filtration and the obtained oil was then stored in an oven at 105 °C for 12 h.
A 10 g of the pre-treated waste cooking oil was trans-esterified with methanol and NaOH catalyst as reported [23]. The free fatty acid (FFA) content of the oil was determined by Gas Chromatography-Mass Spectrometer injected on a Shimadzu (GCMS QP2010s) with an HP-5MS column and FID detector.

Waste cooking oil hydrodeoxygenation reaction
The hydrodeoxygenation reaction (HDO) was according to published methods by our group [24]. Before the reaction, each catalyst Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 (0.2 g) was activated by heating at 350 °C for 1 h under 2 MPa H2 gas in the reactor before each run. HDO reaction was started by inserting 25 mL Figure 1. XRD spectra of (a) Ni/Al2O3-ZrO2 and (b) Ni/SiO2-ZrO2. of the pre-treated waste cooking oil in the reactor containing the activated catalyst. Each reaction was tested at 350 °C, 2 MPa pure H2 gas pressure for 2 h and 4 h for each run. The HDO liquid organic products obtained were analyzed using GC-MS injected on a Shimadzu QP2010s. GC-MS analysis was conducted at an oven temperature of 40 °C, gradually increased to 250 °C, and held for 2 mins with a splitless injection mode. The gas pressure was 49.5 kPa at a flow rate of 30 mL/min. A Shimadzu (8400S) Fourier-Transformed Infrared Spectrophotometer instrument was used to analyze the liquid organic products before GC-MS analysis. Figure 1 illustrates the XRD patterns of Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 catalysts. From the patterns, the metallic nickel phase is detected at 2θ = 44° (111) and 76° (220) according to JCPDS data 96-210-2279. The absence of NiO at 2θ = 37°, 43°, and 64° was confirmed [25]. The additional heating at 500 °C for 5 h under H2 flow completely reduced ionic nickel into metallic nickel. Analysis of the support showed that zirconia is detected in two phases, monoclinic (m-ZrO2) at 2θ = 30° and 50°, and tetragonal (t-ZrO2) at 2θ = 35°, 51°, and 60°, similar to Ulfa et al. [19]. Distinctive missing patterns are observed for Al2O3 and SiO2. This is attributed to the formation of amorphous oxides of Al2O3-ZrO2 and SiO2-ZrO2 as suggested by Xie et al. [26] and Zitouni et al. [27]. Table 1 shows the physical properties of Ni/Al2O3-ZrO and Ni/SiO2-ZrO2 catalysts. X-ray fluorescence (XRF) analysis revealed 15.7 wt% Ni loading on Al2O3-ZrO support for Ni/Al2O3-ZrO and 16.45 wt% Ni loading on SiO2-ZrO2 support for Ni/SiO2-ZrO2. The slight wt% Ni increment which differs from the theoretical calculations is attributed to Al2O3-ZrO2 and SiO2-ZrO2 support weight loss due to calcination at 500 °C for 5 h before Ni impregnation similarly in a report by Fertal et al. [28]. The crystallite size of nickel was determined from the XRD data using Scherrer's equation (Table 1).

Catalyst Characterizations
In Table 1, the BET surface area and porous structure of SiO2-ZrO2 support, Ni/Al2O3-ZrO2, and Ni/SiO2-ZrO2 catalysts are exhibited. The specific surface area of SiO2-ZrO2 support significantly decreased from 77.591 (m 2 /g) before Ni impregnation to 17.985 (m 2 /g), therefore, indicating that Ni addition has resulted in changes in the physical properties of SiO2-ZrO2. However, the pore volume and pore diameter of SiO2-ZrO2 remained almost the same before and after Ni addition. Compared with Ni/Al2O3-ZrO2 (65.938 m 2 /g), the SBET results

Catalyst
SBET (m 2 /g) a Pore volume (cm 3 /g) b   revealed a smaller specific surface area for Ni/SiO2-ZrO2. In addition, Ni/SiO2-ZrO2 showed a larger pore volume while the pore diameter remained smaller.
Isotherm adsorption-desorption of the catalysts is depicted in Figure 2. Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 showed a hysteresis loop on P/Po = 0.6-0.8, typical of type IV mesopore catalyst. The hysteresis type H2 suggested that there is a blocking on desorption process suggesting the complex pore system with bottle-ink like pore shape with wide neck size [29]. The pore distribution of Ni/Al2O3-ZrO2 is slightly larger compared to Ni/SiO2-ZrO2. This is consistent with a previous study by Ulfa et al. [30].
From FT-IR analysis (Figure 3) [31]. The SEM images illustrated in Figure 4 show the morphology of Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2, as a close structure parked with tiny particles of irregular shapes which are attributed to Ni deposition on the surface of the support [30]. A       , 17 (1), 2022, 141 Copyright © 2021, ISSN 1978-2993 close comparison of the images showed that the surface morphology of Ni/Al2O3-ZrO2 consists of tinier particles than Ni/SiO2-ZrO2. The SEM-EDS analysis revealed 16.88 wt% nickel content on the surface of Ni/Al2O3-ZrO2 (Figure 4a) and 26.15 wt% on Ni/SiO2-ZrO2 (Figure 4b).

Bulletin of Chemical Reaction Engineering & Catalysis
The differences in metal distribution on the surface might be explained by different penetration modes of Ni on the SiO2-ZrO2 surface which is neglected even in higher metal loading [32].

Hydrodeoxygenation of Waste Cooking Oil
Before hydrodeoxygenation, WCO was hydrolyzed into free fatty acid using NaOH. The composition of free fatty acid content is tabulated in Table 2. The analysis revealed that 11-Octadecenoic acid is the main fatty acid content in our WCO samples ( Figure 5). Our analysis agreed with Toba et al. [33], in which the main fatty acid content in a household collected WCO was Octadecenoic acid. Similarly, the Octadecenoic acid was reported as a major fatty acid content in WCO from an unidentified source by Zhang et al. [13].
Ni/Al2O3-ZrO2 (Ni/AZ) and Ni/SiO2-ZrO2 (Ni/SZ) catalysts were tested for waste cooking oil (WCO) hydrodeoxygenation to green-diesel. The reaction is carried out independently for each variation of the reaction. The product conversion and distribution were determined by GC-MS analysis ( Figure 6) using an internal standard.
All reaction conditions are summarized in Table 3. In all reactions, the HDO products conversion is 100% except for the reaction using Ni/AZ at 2 h (entry 1). In this reaction by prolonging the time until 4 h increased the conversion up to 100% (entry 2). For the reaction using Ni/SZ catalyst, either at 2 h or 4 h gave 100% conversion (entry 3, entry 4). It is shown that the HDO reaction using Ni/SZ is faster compared to Ni/AZ. Our results are in line with Lycourghiotis et al. [34] and Ding et al. [10], on higher conversion rates of waste cooking oil fatty acid with sulfur-free Ni-based catalysts. HDO activity and selectivity can be significant-ly influenced by catalyst support material. The acidity of support material is an important property to consider. Generally, Al2O3 exhibits higher acidic strength than SiO2 therefore prone to coke formation [35]. In a report by Wu et al. [36], the effect of Al2O3 and SiO2 support on Ni2P catalytic HDO of guaiacol was compared. Al2O3 showed higher activity for demethylation (DME) and SiO2 promoted deoxygenation (DDO). SiO2 also showed weaker acidic strength with the least coke formation. From our results, it is shown that Ni/SZ facilitated higher deoxygenation leading to better WCO conversion and Ni/AZ promoted demethylation by producing more n-C15.
Next, the total green diesel yield is varied from 90.57%-97.12% (Table 3) with a major product yield of n-C15 followed by n-C17. The trend of the product is similar for the use of Ni/AZ and Ni/SZ. It is revealed that both catalysts probably promoted hydrogenation followed by decarboxylation and subsequent carbon-chain cracking to give a series of hydrocarbon products consisting of n-C14 to n-C17 as indicated by Zhang et al. [13]. Ni content on AZ and SZ supports (see XRF analysis in Table 1) contributed to higher yields of n-C15 for Ni/AZ and Ni/SZ as observed because of high Ni content enhances carbon-chain cracking due to Ni active species high electron density favorable for C−C bond cleavage [35]. However, product n-C18 gives the lowest distribution among all reactions considering that the complex reaction is preferable during the one-step reaction.
In Figure 7, the effect of reaction time on the product distribution is displayed. Ni/AZ gave a higher yield of n-C15 at 2 h, compared to Ni/SZ under the same reaction time. In this case, the difference in catalyst support and Ni loading and/or distribution should affect the yield of n-C15. In reports by Hellinger et al. [37] and Popov et al. [38], Al2O3 is suggested to show a higher specific surface area than SiO2. In our case, the higher BET specific surface area of Ni/AZ is suspected to influence the catalytic activity of Ni/AZ due to the adequate space for active Ni + species deposition on the  surface of the AZ support (Table 1, SBET data). It is also observed from the SEM image analysis that Ni/AZ catalyst exhibits a smaller particle size compared to Ni/SZ (Figure 4a, 4b). On the other hand, the SEM-EDS data suggest that the Ni distribution on the surface of Ni/SZ is higher compared to Ni/AZ even though the total Ni loading is almost similar in both catalysts according to the XRF analysis (Table 1, Ni loading). The high distribution of Ni content observed on the surface of SZ in Ni/SZ ( Figure  4b) could be attributed to the better fatty acid conversion for Ni/SZ as suggested by Sánchez-Cárdenas et al. [39].  Interestingly, a reserve phenomenon is noticed when the HDO reaction time is prolonged to 4 h. It showed decreasing amount of n-C15 yield for Ni/AZ and an increase of n-C15 by using Ni/SZ. It is predicted that Ni/AZ deactivated in longer reaction time due to a high reaction temperature of 350 °C which agrees with Monnier et al. [40], on Al2O3 support deactivation under similar experimental conditions used in this study. In Figure 8, the predicted reaction pathway is highlighted.

Conclusions
In this study, deposition of Ni into Al2O3-ZrO2 and SiO2-ZrO2 composites by wet impregnation gave mesopore catalysts Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 based on BET adsorptiondesorption analysis. The obtained Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 were tested for waste cooking oil hydrodeoxygenation to green diesel. Analysis of gas chromatography -mass spectrometry revealed that both catalysts gave green diesel fraction n-C14 to n-C17 with rich content of n-C15 fraction. Notwithstanding, Ni/SiO2-ZrO2 showed better activity for waste cooking oil conversion even after prolonging the reaction from 2 h to 4 h. While Ni/Al2O3-ZrO2 promoted higher selectivity of n-C15 hydrocarbon. Therefore, we conclude that Ni/Al2O3-ZrO2 and Ni/SiO2-ZrO2 should be potential catalysts for waste cooking oil conversion to green diesel.