Dried fruits of Cerbera manghas were obtained from plant that grows in Keputih, Surabaya, Indonesia. It is also known as Bintaro plant in Java Island. Cerbera manghas oil was extracted from seeds through the process steps that according to previous literature, as shown in Fig. 1 (Roesyadi, 2016). At first the fruit was split and the seeds were removed from the fruit. These seeds were cleaned from the skin and obtained white oval-shaped seeds. Furthermore, the seeds were dried in the sun for 7 days. After drying, the dry seeds were chopped and extracted by hydraulic press to produce oil. Oil was stored in an airtight container to prevent oxidation, which can increase free fatty acid levels. Cerbera manghas oil from the hydraulic extraction process was obtained as much as 2.5 L from 6 kg of seeds. It can be said that 2 kg of Bintaro seeds produce about 0.8 L of oil. Figure 2 specify the different conditions of the seed and the obtained oil.

Figure 1.
Extraction process of Cerbera manghas seeds. (a) the fruit was split into two parts, (b) the seeds were removed from the fruit, (c) the seeds were dried in the sun, (d) the dry seeds were chopped, finally (e, f) the seeds were extracted by hydraulic press.
Adapted from Roesyadi (2016).

Figure 2.
Different condition of Cerbera manghas fruit and the obtained oil. (a) a dried fruit, (b) seed with skin, (c) white oval-shaped seeds seed, (d) oil.
Adapted from Roesyadi (2016).

ZSM-5 zeolite in ammonium form (CBV 8014, SiO2/Al2O3 ratio of 25, surface area of 400 m² g^–1, Na₂O of 0.05 wt.%) was provided from Zeolyst International. Meanwhile, Ni(NO₃)₂.6H₂O and Zn(NO₃)₂.6H₂O as the metal precursor were obtained from Merck with 98% purity. The calcination process on NH₄-ZSM-5 zeolite was applied to obtain HZSM-5 zeolite at temperature of 550 °C for 5 h (Sartipi et al., 2013). By incipient wetness impregnation method (Haber et al., 1995; Marlinda et al., 2016; Sartipi et al., 2013), HZSM-5 after calcined was impregnated with nickel and zinc to obtain Ni-Zn/HZSM-5 catalyst. Before impregnation, the HZSM-5 zeolite was dried overnight at a temperature of 120 °C. The amount of aqueous metal solutions was calculated according to the total pore volume of HZSM-5 as support, which was obtained from Brunauer–Emmett–Teller (BET) analysis, as shown in Table 1. After preparing aqueous metal solutions, firstly the amount of aqueous Ni(NO₃)₂.6H₂O solution of 1.81 mol L^–1 was sprayed slowly to HZSM-5 zeolite while stirring and then kept overnight in a desiccator. Finally, this catalyst was dried at 120 °C for 12 h. Furthermore, in the same way as before, the amount of aqueous Zn(NO₃)₂.6H₂O solution of 1.62 mol L^–1 was sprayed slowly to HZSM-5 zeolite while stirring. Before drying at temperature of 120 °C for 12 h, the catalyst was kept overnight in a desiccator. Catalysts containing Ni and Zn salt solution were calcined with air at temperature of 400 °C for 2 h. After calcination, active metal usually exists as a form of metal oxide (Kim et al., 2013; Romero et al., 1998). The metallic phases of Ni and Zn were obtained by reducing the metal oxides under flowing hydrogen at a temperature of 450 °C for 3 h. By atomic absorption spectrometry, the metal content was 5.43 wt.% of Ni and 1.23 wt.% of Zn of the total catalyst weight. The catalyst is denoted as Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst. Other prepared catalyst is Ni(5.42%)-Zn(1.11%)/HZSM-5.

Table 1.
The characterization of H-ZSM-5 and Ni-Zn/HZSM-5 catalyst.

Elaborated by the authors using data from Roesyadi et al. (2016). a = total surface area BET; b = the surface area mesoporous and microporous obtained from the t-plot; c = total pore volume; d = micropore volume obtained from the t – plot; e = mesoporous volume = Vmeso = Vtotal – Vmikro; f = average pore diameter; g = atomic absorption spectrometry (AAS) analysis.

The various techniques used to characterize catalysts were X-ray diffraction (XRD), atomic absorption spectrometry (AAS) and N. adsorption-desorption. The metal contents of Ni dan Zn were confirmed with AAS. In principle, the distribution of metal on ZSM-5 is influenced by the preparation method used (Chen et al., 2015). X-ray diffraction was applied to identify phase analysis and the Ni-Zn/HZSM-5 catalyst crystallinity. X-ray diffraction patterns of the solid catalyst (Phillip X-Pert diffractometer with Cu radiation Ka) were collected from 2q of 5–90° to identify the peak of crystalline zeolite ZSM-5. The existence of the crystalline phase of NiO, ZnO, Ni and Zn particles was confirmed from the diffraction patterns recorded on 2q = 30–80°.

The catalyst surface area was measured by BET calculation on Quantachrome NovaWin Version 10.0 according to the adsorption data with the relative pressure (P/Po) in range from 0.095 to 0.297 through the recording process of isotherm BET on five points. Before the measurement of nitrogen adsorption and desorption isotherms at a temperature of 77 K, the sample was outgassed for 16 h at 300 °C. The pore size distributions were calculated using adsorption model of Barrett–Joyner–Halenda (BJH). Micropores volumes were obtained from the t-plot analysis. Total pore volumes were obtained according to the amount of nitrogen adsorbed at relative pressure about 0.99034. Mesoporous volumes were obtained from the reduction of the total pore volume by the volume of micropore.

Hydrocracking reactions were carried out in the Parr USA pressure batch reactor equipped with a mechanical stirrer. Cerbera manghas oil of 150 mL and catalysts of 1 g were transferred in the reactor. To remove air, nitrogen was flowed into the reactor at least 30 min for purging. Depending on temperature condition, reaction pressure was changed between 15 and 25 bar after flowing H. for at least 1 h, as reported in previous study (Marlinda et al., 2016). The reaction proceeded at various temperatures, i.e., 350, 375 and 400 °C for 2 h. After a reaction time was reached, the reactor temperature was allowed to reach room temperature. Furthermore, biofuel was analyzed by GC-MS, i.e., Agilent HP 6890 GC equipped with a capillary column, model Agilent 19091S-433 (HP–5MS, phenylmethylsiloxane of 5%). The column dimensions used have an internal diameter of 0.25 mm, film thickness of 0.25 mm and a length of 30 m. All hydrocarbon components are identified using the Wiley275 and NIST02 mass spectral library of data. Based on GC-MS analysis, it is possible to estimate the relative percentages of hydrocarbon composition and distribution of products to their carbon numbers. According to Barrón et al. (2011), hydrocarbon compounds (including n-paraffin, isoparaffin, cycloparaffin, aromatic, olefin) were grouped in gasoline-like hydrocarbon (C5-C9), kerosene-like hydrocarbon (C10-C13), and gasoil-like hydrocarbon (C14-C22).

The clear yellow color of Cerbera manghas oil without purification was analyzed by GC-MS. Table 2 indicated that the most abundant compound in Cerbera manghas oil is oleic acid at 77.76 area%, which agrees with that obtained by Marlinda et al. (2016). Oleic acid is also the largest constituent of Calophyllum inophyllum oil, Jatropha curcas oil (Chuah et al., 2016) and Pongamia pinnata oil (Dwivedi and Sharma, 2015), as much as 39.8 ± 0.4, 51.2 ± 0.6 and 65.3 wt.%, respectively.

Table 2.
Compounds of Cerbera manghas oil tested by GC-MS.

Elaborated by the authors using data from Prajitno et al. (2015).

Some different techniques were used to obtain the textural properties and metal content of the Ni-Zn/HZSM-5 catalyst. Figure 3 shows the type of adsorption isotherms classified by International of Pure Applied Chemistry (IUPAC) for HZSM-5 and Ni-Zn/HZSM-5 catalyst. It was clear that a combined pattern of type I and type IV was exhibited in Fig. 3b. The type of microporous solids owned by HZSM-5 is type I isotherm, as shown in Fig. 3a. It shows an initial curve that rises sharply at very low relative pressure, as reported in previous studies (Hao et al., 2012; Sartipi et al., 2013; Wang et al., 2013). The mesopores and the pore size distribution calculated from the sorption isotherm can be called hysteresis. The presence of hysteresis exhibited with type IV isotherm was seen on high relative pressure of 0.65 towards 0.85, at which N. desorption does not pass through the adsorption original path (Hao et al., 2012; Vitale et al., 2013; Wang et al., 2013). Nickel and zinc impregnated on support can create mesoporous sites in catalyst, so that bulky molecules can diffuse easily to the pores of catalyst, as reported in previous studies (Hao et al., 2012; Vitale et al., 2013; Wang et al., 2013).

Figure 3.
Nitrogen adsorption and desorption and BJH pore size distribution of catalyst. (a) HZSM-5; (b) Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst.

The pore size distributions of the two types of the catalyst were centered at 3.8 nm. However, the slightly narrower pore size distribution and lower peak height were seen on Ni–Zn impregnated on HZSM-5. Reduction of peak height on the catalyst showed that the presence of Ni and Zn had a considerable influence on the pore structure of HZSM-5. Both of the catalysts have a mesoporous structure as the result of the interparticle voids. This pore structure will greatly affect the increase of reactivity of the catalysts, as reported by Wang et al. (2013).

Table 1 shows that micropores volume decreased from 0.156 to 0.101 cm. g^–1, while pore diameter increased from 2.709 to 3.109 nm after Ni–Zn metals were impregnated on HZSM-5. Nevertheless, Chen et al. (2016) stated that the internal nickel particles can inhibit the nitrogen molecules into the pore by blocking the micropores of HZSM-5. However, the increase of pore diameter and the presence of mesopores size distribution indicated that fatty acids had a significant probability to diffuse into the pores successfully and react on the active sites. According to Bockisch (1998), assuming a typical triglyceride molecule to have a spherical form, its diameter would be 1.5 nm. Therefore, if the fatty acids cannot diffuse and react with the active sites located in micropores channel, then these fatty acids will diffuse and react with the active sites located in mesopores channel. While the actual metals content obtained were shown in Table 1.

X-ray diffraction data of catalyst system shows that the full diffractograms were recorded in the range 2q° of 5–90 to see the diffraction peak of NiO, ZnO, Ni and Zn in the catalyst samples. The characteristic of diffraction peaks of commercial HZSM-5, as shown in Fig. 4a, located at 2q of 7.9, 8.8, 23.09, 23.31, 23.69, 23.9 and 24.4°, which indicated a structure type mordenite framework inverted (MFI), as reported by a previous study (Marlinda et al., 2016). As shown in Fig. 4b and c, diffractograms of catalyst used for the hydrocracking on Cerbera manghas oil were almost similar to those of commercial HZSM-5.

Figure 4.
X-ray diffraction pattern of catalysts. (a) Commercial HZSM-5 after calcined (Marlinda et al., 2016), (b) Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst, (c) Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst.

After impregnation with Ni–Zn, the HZSM-5 framework structure was almost unchanged, but only the intensity of each diffraction peak decreased (Chen et al., 2016; Niu et al., 2014; Wang et al., 2013). Figure 4b shows that a small intensity diffraction peak of ZnO crystallites was observed at 2q of 34.55, 36.32, 47.56, 56.66 and 69.08°. The Zn and Ni particles were also observed at 2q of 70.94 and 52.06° with low intensity, respectively. While the presence of NiO crystallite was not found. Figure 4c shows that a small peak diffraction intensity of ZnO crystallites were observed at 2q of 34.53, 36.19, 47.58 and 69.10°. NiO crystallite and Ni particle were observed at 2q of 79.23 and 51.96° with low intensity, respectively. While the presence of Zn metal was not detected. When incipient wetness impregnation method was applied by means of physical mixing, diffraction peak for ZnO was obtained at 2q of 36.21°, as observed by Niu et al. (2014). In principle, the reduction of ZnO and NiO crystallites with hydrogen will lead to the formation of proton ions (H⁺), as the Bronsted acid. As reported by Chen et al. (2016), after reduction of NiO with hydrogen, proton ion (H⁺) was formed (Ni²⁺ + H₂®Ni + 2H+) so that the catalysts which contained Bronsted acid sites (weak strength sites) have contributed to direct isomerization reaction. While Lewis acid sites (strong acid sites) may direct the ability of cracking.

Biofuel produced from hydrocracking of Cerbera manghas oil at a temperature of 400 °C with a Ni-Zn(2%)/HZSM-5 catalyst and Ni-Zn(4%)/HZSM-5 (Prajitno et al., 2015) still contained many carboxylic acids and compounds containing more oxygen, i.e., 9-octadecenal in the range of 5.35–6.11% and ethenone in the range of 0.20–0.37%. Table 3 shows that olefins are found in very small quantities while the n-paraffins and cycloparaffins can only be obtained from catalyst with metal loading of 4%. It indicated that the metals impregnated on HZSM-5 have not been able to hydrogenate fatty acid. The GC-MS spectra of biofuel is shown in Fig. 5.

Gas chromatography-mass spectrometry spectra of Cerbera manghas oil and biofuel produced in the various reaction temperatures and metal loading of 5% are shown in Fig. 6 and 7, respectively. When Fig. 6 and 7 are compared, it can be seen that the compounds with retention time of 10–18 min in GC-MS spectra of Cerbera manghas oil disappear after reaction, as reported by a previous study (Marlinda et al., 2016). It indicated that the triglycerides containing long chain molecules were converted into short chain molecules by hydroprocessing (including cracking, deoxygenation, and isomerization reaction) with HZSM-5 catalyst, as reported by a previous study (Zheng et al., 2015). Meanwhile, the number of different compounds with retention time of 0–9 min increased, as shown in Fig. 7. The presence of Ni and Zn metals have contributed to the hydrogenation process of unsaturated fatty acids into n-paraffins.

Figure 5.
Gas chromatography-mass spectrometry spectra of (a) Cerbera manghas oil (Prajitno et al., 2015), and liquid products produced at 400 °C with (b) Ni-Zn(2%)/HZSM-5 catalyst, (c) Ni-Zn(4%)/HZSM-5 catalyst, under pressure 20 ± 5 bar in the batch reactor.

Table 3.
Hydrocarbon composition of biofuel produced by hydrocracking at temperature of 400 oC under pressure 20 bar in batch reactor, tested GC-MS.

Figure 6.
Gas chromatography-mass spectrometry spectra of Cerbera manghas oil.
Adapted from Marlinda et al. (2016).

Figure 7.
Gas chromatography-mass spectrometry spectra of biofuel produced at (a) 350 °C, (b) 375 °C, (c) 400 °C with Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst, and GC-MS spectra of biofuel produced at (d) 350 °C, (e) 375 °C, (f) 400 °C using Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst, under pressure 20 ± 5 bar in the batch reactor. Saturated chain hydrocarbons (n-paraffins) have the number of carbon atom from C12 to C24.

3.3.1 Effect of temperature on hydrocarbon composition

Figure 8a shows that biofuel contains a different hydrocarbon composition for each temperature of hydrocracking reactions with Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst. It can be seen that, at temperature of 350 °C, biofuel contained n-paraffins of 42.03 area%, but carboxylic acids of merely 15.48 area%, and compounds containing oxygen of 8.69 area%, i.e., octacosanol, were still found. Meanwhile, biofuel containing olefins of 20.31 area% and cycloparaffins of 12.48 area% was produced more at temperature of 375 °C. On the other hand, polycyclic aromatic hydrocarbons (PAHs) and aromatic are obtained at temperature of 400 °C, i.e., 21.32 and 35.47 area%, respectively. The existence of PAHs compounds (including 1-methyl fluorene, phenanthrene, and anthracene) showed that the heating of vegetable oil was not complete. These compounds were not expected to present in biofuel. As reported by Vichaphund et al. (2015), deactivation occurred on HZSM-5 surface because of coke derived from PAHs compounds. Coke on the catalyst surface can be inhibited in the presence of hydrogen during hydrocracking reaction. However, overall, the compounds containing oxygen were still found in biofuel produced at all temperature variables. In principle, this compound should no longer exist in biofuel because the presence of oxygen atoms affects the viscosity and heating value, as reported by a previous study (Zheng et al., 2015). Furthermore, isoparaffins were not found in biofuel for each reaction temperatures.

Figure 8.
Effect of reaction temperature on (a) the hydrocarbon composition, (b) gasoline/kerosene/gasoil yields of liquid hydrocarbon fuel produced using Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst; (c) gasoline/kerosene/gasoil yields of liquid hydrocarbon fuel produced using Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst, under 20 bar in the batch reactor (“c” is elaborated by a previous study (Roesyadi et al., 2016)).

The cracking of olefins into lighter hydrocarbon compounds was shown by the decrease of olefin at temperatures from 350 to 400 °C, as reported by Kim et al. (2013). On the other hand, aromatic hydrocarbon compounds, i.e., 1-methyl-2-(2-propenyl) benzene of 4.5 area%, 1-Methylnaphthalene of 4.79 area%, 1,6-dimethylnaphthalene of 3.39 area%, 1-ethyl-naphthalene of 2.64 area% were found at a temperature of 400 °C. In fact, aromatics increased with the increasing reaction temperature. This suggests that aromatization reaction on olefin occurred with the presence of HZSM-5 zeolite, as reported by Pinto et al. (2014). The cycloparaffins also indicated the functionality of acidic site on HZSM-5. As stated by a previous study (Chen et al., 2016), the acidity of HZSM-5 zeolite decreased with the increase of active metal loading and Si/Al ratio. In addition, the H2 to oil ratio would help the process of hydrogenation on olefins to n-paraffins. The determination of hydrocarbon composition and the degree of isomerization is useful to provide information to evaluate the properties of biofuel, such as cetane number (Kim et al., 2013), freezing point, flash point and viscosity (Liu et al., 2015).

Table 4 shows the components containing abundant compounds in biofuel based on group of hydrocarbons at a temperature of 350 °C. The main compound of cycloparaffin was cyclohexane with branched chain resulting in hydrogenation reaction of unsaturated hydrocarbon compounds, as found by Zheng et al. (2015). The n-hexadecanoic acid was formed greatly. It was indicated that the saturation of the double bond occurred quite well. The presence of oxygen-containing compounds showed that it would be better for hydrodeoxygenation reaction to occur at temperatures above 350 °C. Oxygen-containing compounds decreased from 8.69 to 4.33% at 350 and 400 °C, respectively. The increase of temperature led to the increasing cracking activity.

Table 4.
Several abundant compounds in each group for biofuel produced at 350 oC, under 20 bar in the batch reactor* with Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst.

*Adopted from table format in literature Zheng et al. (2015).

As shown in Tab. 4, the presence of pentadecane (C15) and heptadecane (C17) indicated that HDC was the more dominant reaction pathway than HDO. Although two of these reactions allowed the formation of n-paraffin, but the H. supplied for both reaction routes was different so that n-paraffin produced was different based on the number of carbons (Pinto et al., 2014; Sotelo-Boyás et al., 2012; Zheng et al., 2015). Hydrogenation process on Cerbera manghas oil containing tripalmitin proceeded to obtain three palmitic acids and propane. By following the route of decarboxylation and/or decarbonylation without going through the saturation of the first double bond by hydrogenation reaction, pentadecane (C₁₅H₃₂) was immediately produced. In fact, the highest triglycerides in Cerbera manghas oil were triolein. Triolein was hydrogenated to obtain three oleic acids and propane. It still contained a double bond, so the hydrogenation process on oleic acid into stearic acid occurred. By following the route of decarboxylation and/or decarbonylation, stearic acid was converted into heptadecane (C_17.36). Furthermore, the aromatic compounds were produced through dehydrogenation process on olefin, i.e., the cracking and aromatization reactions as reported by Zheng et al. (2015).

With Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst, the highest yield for n-paraffin was found at a temperature of 375 °C. On the other hand, isoparaffin reaches the highest value of 4.58 area% at 400 °C (Roesyadi et al., 2016). A single type of methyl isomer was obtained as isomerization product at 400 °C, as shown in Tab. 5. This is similar to the result obtained by Chen et al. (2016), who reported that the short-branched isoparaffins were formed because methyl alkanes can pass easily through the HZSM-5 pore. Ethyl isomers cannot be found because the molecular size of ethyl alkanes is larger than that of methyl alkanes. Isoparaffin increased significantly with the increasing reaction temperature because the literature (Chen et al., 2016) stated that isomerization reaction proceeded in endothermic condition.

Table 5.
Several abundant compounds in each group for biofuel produced at 400 oC, under 20 bar in the batch reactor*, using Ni(5.42%)-Zn(1.11%)/HZSM-5 catalyst.

*Adopted from table format in literature Zheng et al. (2015).

With the increasing reaction temperature, aromatics increased significantly so that n-paraffin (including C15 and C17) and cycloparaffins also decreased for Ni(5.43%)-Zn(1.23%)/HZSM-5 catalyst (Fig. 8a). It indicated that aromatization and cyclization reaction were favored at temperatures over 350 °C (Rocha Filho et al., 1993). According to Šimáček et al. (2009; 2011), aromatics also increased because of the increasing cyclization reactions rate. Tamiyakul et al. (2016) also reported that the presence of Zn²⁺ and ZnO detected in HZSM-5 was responsible for n-paraffins dehydrogenation and oxygenates decarboxylation, respectively. As a result, more aromatic yield was produced.

marlindalenny@unja.ac.id