Maximizing Biojet Fuel Production from ... - ACS Publications

Aug 7, 2017 - Facile Synthesis of KFI-type Zeolite and Its Application to Selective Catalytic Reduction of NOx with NH3. ACS Catalysis. Kim, Cho, and ...
1 downloads 0 Views 6MB Size
Research Article pubs.acs.org/acscatalysis

Maximizing Biojet Fuel Production from Triglyceride: Importance of the Hydrocracking Catalyst and Separate Deoxygenation/ Hydrocracking Steps Myoung Yeob Kim,† Jae-Kon Kim,‡ Mi-Eun Lee,‡ Songhyun Lee,† and Minkee Choi*,† †

Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, Republic of Korea ‡ Alternative Fuel R&D Team, Korea Petroleum Quality & Distribution Authority, Cheongju-City, Chungcheongbuk-do 28115, Republic of Korea S Supporting Information *

ABSTRACT: Various parameters in the catalytic hydroconversion of triglycerides (palm oil) were carefully investigated for maximizing the production of biojet fuel. The results showed that the deoxygenation of triglyceride via hydrotreatment should be carried out in a separate reactor prior to the hydrocracking step (i.e., two-step reaction process). Otherwise, the CO generated during deoxygenation can poison the metal components in the metal/acid bifunctional catalysts (Pt/ zeolites), which can cause significant imbalance between the metal and acid functions in hydrocracking. This leads to fast catalyst deactivation via coke formation, heavy formation of aromatics, and overcracking of hydrocarbons, resulting in the reduction of final biojet fuel yield. In the two-step process, the second hydrocracking step mainly determines the final biojet fuel yield, and thus, a rational design of the hydrocracking catalysts that can suppress overcracking is essential. The diffusion characteristics of the multibranched hydrocarbon (e.g., 2,2,4trimethylpentane) in the hydrocracking catalysts could be correlated with the yields of the jet fuel-range C8−C16 hydrocarbons and the iso/n-paraffin ratios. The result indicates that the facile diffusion of multibranched isomers out of catalysts before excessive cracking is important for the suppression of the formation of light hydrocarbons (≤C7). Consequently, Pt supported on nanocrystalline large-pore BEA zeolite showed the largest biojet fuel yield with the highest iso-paraffin content. Under the optimized conditions, 55 wt % of biojet fuel with respect to palm oil was achieved after final distillation, which satisfied all the required fuel specifications. KEYWORDS: biojet fuel, triglyceride, hydrotreating, hydrocracking, zeolite, mesoporous

1. INTRODUCTION Catalytic processes to produce transportation fuels (e.g., gasoline, diesel, and jet fuel) from biomass have been extensively investigated by the requirement to replace petroleum-based fuels with sustainable carbon sources and thus to decrease CO2 emissions.1,2 In the case of jet fuel, the International Air Transport Association (IATA) forecasted that fuel consumption in the aviation sector will increase annually by 5% until 2030.3 Owing to the combination of increasing passenger demand and reinforcement of regulations to reduce anthropogenic CO2 emissions, IATA decided to establish carbon neutral growth by 2020 and 50% reduction by 2050.3 In this regard, the use of jet fuel derived from biomass resources, often referred to as biojet fuel to differentiate it from conventional petroleum-based jet fuel, has been considered as one of the most promising solutions to satisfy the global demand. Thus far, four major processes4 have been developed to convert biomass-derived feedstock into jet fuel: (a) oil-to-jet © 2017 American Chemical Society

(deoxygenation of triglyceride and subsequent hydrocracking),5−11 (b) gas-to-jet (gasification/Fischer−Tropsch reaction followed by hydrocracking),12−14 (c) alcohol-to-jet (dehydration of alcohols and subsequent oligomerization),15−19 and (d) sugar-to-jet (various catalytic conversions of sugars).20,21 Among these technologies, the oil-to-jet and gas-to-jet technologies have been considered as the most realistic options in the short term, and the biojet fuels produced in these ways are currently allowed by ASTM specification D7566-14 for blending into commercial jet fuel at levels of up to 50%.22 In particular, the oil-to-jet process, also known as HEFA (hydroprocessed esters and fatty acids), is most widely adapted in industry, and the produced fuels have been tested by commercial airlines.23−26 This is because bioderived oils (triglycerides and fatty acids) contain paraffinic units (generally Received: April 25, 2017 Revised: July 13, 2017 Published: August 7, 2017 6256

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis

affected by the porous structure of the acidic zeolite, which induces shape-selectivity.42−44 Nevertheless, the impact of zeolite structure on biojet fuel production has not been comprehensively studied to date. In the present work, we investigated rigorously the effects of different reaction processes (two-step vs single-step reaction processes), reaction conditions, and types of hydrocracking catalysts on the production of biojet fuel from triglycerides (palm oil). The investigation demonstrated that the two-step reaction process composed of separate hydrotreating and hydrocracking steps is markedly more advantageous than the single-step direct hydroconversion in terms of high biojet fuel yield, suppressed production of aromatic compounds, and catalyst lifetime. Therefore, the effects of the structural properties of hydrocracking catalysts (i.e., zeolite-supported Pt catalysts) and reaction conditions on biojet fuel yields and hydrocarbon distributions were carefully investigated in the two-step reaction process. The hydrocarbon yields in every catalytic step and after final distillation were clearly noted, and the resultant biojet fuel properties such as freezing point, flash point, viscosity, density, and net heat of combustion were also analyzed.

C12−C22) which can be facilely converted into jet fuel without the use of an energy-intensive gasification. Biojet fuel production from triglycerides generally consists of the following reactions (Scheme 1): (i) hydrogenation of Scheme 1. Reaction Pathways for the Hydroconversion of Triglycerides into Biojet Fuel (Oil-to-Jet Process)

2. EXPERIMENTAL SECTION 2.1. Fatty Acid Composition of Palm Oil. Fatty acid composition of palm oil (Sigma-Aldrich, analytical standard) was determined by analyzing the corresponding fatty acid methyl esters (FAMEs) after transesterification with methanol. For the analysis, 52.5 g of palm oil was reacted with 12.0 g of methanol (molar ratio of methanol/palm oil = 6) using 0.6 g of NaOH as a catalyst at 333 K for 2 h. The upper layer of liquid products (FAMEs) was analyzed using a gas chromatograph (GC) equipped with a flame ionization detector (FID) and an HP-5 ms capillary column (Agilent, 30 m × 0.25 mm). 2.2. Zeolite Synthesis. Conventional ZSM-5 was synthesized using tetrapropylammonium bromide (TPABr, TCI, > 98%) as a structure-directing agent (SDA). Colloidal silica (Ludox AS-30, Simga-Aldrich, 30 wt % SiO2) and Al2(SO4)3· 18H2O (Acros, 98%) were used as silicon and aluminum source, respectively. The starting molar composition was 1.25 Al2O3/9.75 Na2O/100 SiO2/20 TPABr/3.75 H2SO4/4000 H2O. The resultant gel was transferred into a Teflon-lined stainless-steel autoclave and hydrothermally treated at 443 K for 3 d under tumbling (30 rpm). After crystallization, the resultant product was collected by filtration, washed with deionized water, and dried at 373 K. The sample was calcined at 823 K (ramp: 2 K min−1) under dry air for 3 h in a plug-flow reactor. The calcined zeolite was ion-exchanged with NH4+ three times using an aqueous 0.1 M solution of NH4NO3 (57 cm3 solution per g of zeolite sample) and calcined again at 823 K for 3 h to obtain the H+-form. The resultant sample was denoted as Bulk-ZSM-5. Nanocrystalline ZSM-5 was synthesized using a gemini-type quaternary ammonium surfactant with a formula of C6H13− N+(CH3)2−C6H12−N+(CH3)2−C6H13·2Br− as the SDA.45 Colloidal silica (Ludox AS-30, Sigma-Aldrich, 30 wt % SiO2) and Al2(SO4)3·18H2O (Acros, 98%) were used as the source of silicon and aluminum, respectively. The starting molar composition was 1.25 Al2O3/9.75 Na2O/100 SiO2/10 SDA/ 3.75 H2SO4/4000 H2O. The resultant gel was transferred into a Teflon-lined stainless-steel autoclave and hydrothermally treated at 403 K for 6 d under tumbling (30 rpm). After the crystallization, the resultant product was collected by filtration,

unsaturated CC bonds in triglyceride, (ii) hydrogenolysis of triglyceride into fatty acids, (iii) deoxygenation of fatty acids into normal paraffins, and (iv) hydrocracking of the produced normal paraffins into shorter-chain (C8−C16) iso-rich hydrocarbons that are suitable as jet fuel. In principle, the first three reactions (i−iii) can be catalyzed by metal (e.g., Pt,27−30 Pd,27,30−33 and Ni27,34−36) or MoS2-based catalysts27,34,37−39 with hydrogen-activating function. The fourth reaction (iv) requires the use of metal/acid bifunctional catalysts,40,41 for which zeolites have been most widely used as a solid acid due to their strong acidity as well as shape-selectivity.42−44 Several patents5,7,11 have proposed a two-step catalytic processes composed of hydrotreating triglycerides into normal paraffins in the first reactor (reactions i−iii) and subsequent hydrocracking in the second reactor (reaction iv). In contrast, recent academic works8−10 have focused predominantly on a singlestep direct hydroconversion of triglycerides into biojet fuel using metal/acid bifunctional catalysts for the purpose of process intensification. In these studies, all reactions (i−iv) were carried out in a single reactor using only a metal/acid bifunctional catalyst. This approach is viable in principle since the metal component in the bifunctional hydrocracking catalyst can also catalyze the reactions i−iii. However, the effects of different reaction procedures (i.e., two-step vs single-step reactions) on product distributions and catalyst stabilities have not been investigated carefully thus far. In the research on biojet fuel production, maximizing the biojet fuel yields has been the most important concern. Unfortunately, in many of the earlier studies, the detailed product distributions, calculation methods for biojet fuel yield, and final fuel properties were not clearly described. In certain cases, surprisingly high biojet fuel yields (∼92 wt %), even higher than the theoretical maximum hydrocarbon yield, were claimed,9 which means that special caution is required for comparing the earlier catalytic results in the literature. In principle, the biojet fuel yield is mainly determined by the hydrocracking step (reaction iv) because excessive cracking can lead to significant production of short-chain gaseous hydrocarbons. Hydrocracking selectivity is, in turn, significantly 6257

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis washed, dried, and transformed into H+-form following the same procedure as that described for Bulk-ZSM-5. The resultant sample was denoted as Nano-ZSM-5. Conventional Beta zeolite was synthesized using tetraethylammonium hydroxide (TEAOH, TCI, 35% in water) as the SDA.46 Fumed silica (CAB-O-SIL M-5, CABOT) and sodium aluminate (Al 50−56%, Na 40−45%, Sigma-Aldrich) were used as the silicon and aluminum source, respectively. The final molar composition of the synthesis gel was 2.25 Na2O/2.0 K2O/13 (TEA)2O/1.0 Al2O3/50 SiO2/765 H2O/5.0 HCl. The synthesis gel was transferred into a Teflon-lined stainless steel autoclave and hydrothermally crystallized at 443 K for 1 d under tumbling (30 rpm). The resultant product was filtered, washed with deionized water, and dried at 373 K. The sample was calcined at 823 K and transformed into the H+-form. The resultant sample was denoted as Bulk-Beta. Nanocrystalline Beta zeolite was synthesized using a cyclic diammonium-type SDA.47 Colloidal silica (Ludox AS-30, Sigma-Aldrich, 30 wt % SiO2) and Al2(SO4)3·18H2O (Acros, 98%) were used as the source of silicon and aluminum, respectively. The final molar composition was 2.5 Al2O3/30 Na2O/100 SiO2/10 SDA/10 H2SO4/6000 H2O. The synthesis gel was hydrothermally crystallized at 443 K for 1 d under tumbling (30 rpm). The resultant product was filtered, washed with deionized water, and dried at 373 K. The sample was calcined at 823 K and transformed into the H+-form. The resultant sample was denoted as Nano-Beta. 2.3. Catalyst Preparation. As a hydrotreating catalyst, nominally 1 wt % Pt was supported on γ-Al2O3 (Strem) via incipient wetness impregnation using an aqueous solution of Pt(NH3)4(NO3)2 (Aldrich). After the impregnation, the sample was dried at 373 K for 5 h and calcined in dry air at 673 K (ramp: 2 K min−1) for 3 h. The sample was subsequently reduced under H2 flow at 673 K for 3 h (ramp: 2 K min−1). As a hydrocracking catalyst, 1 wt % Pt was supported on the prepared H+-form zeolite samples (Bulk-ZSM-5, Nano-ZSM-5, Bulk-Beta, and Nano-Beta) following a similar method as that described for 1 wt % Pt/γ-Al2O3. 2.4. Catalyst Characterization. Elemental analysis was carried out by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using an iCAP 6300 (Thermo scientific). Powder X-ray diffraction (XRD) patterns were recorded using a D2-phaser (Bruker) equipped with a Cu Kα radiation source (30 kV, 10 mA) and a LYNXEYE detector. N2 adsorption−desorption isotherms were measured using a BELSorp-max volumetric analyzer (BEL Japan) at liquid N2 temperature (77 K) after degassing at 673 K. The specific surface area was determined in the P/P0 range between 0.05 and 0.20 using the Brunauer−Emmett−Teller (BET) equation. The external surface area (Sext) and micropore volume (Vmicro) were determined using t-plot analysis. The total pore volumes were estimated at P/P0 of 0.98, and the mesopore volumes (Vmeso) were calculated by subtracting the micropore volumes (Vmicro) from the total pore volumes (Vtotal). Transmission electron microscopy (TEM) images were taken with a field emission TEM (FEI) operating at 300 kV after mounting the samples on a carbon-coated copper grid (300 mesh) using ethanol dispersion. Scanning electron microscopy (SEM) images were acquired with a SU8230 (Hitachi) microscope operating at 2 kV accelerating voltage in backscattered mode without metal coating. The amounts of Brønsted and Lewis acid sites in zeolites were analyzed using Fourier transform infrared spectroscopy

(FT-IR) after pyridine adsorption. FT-IR spectra were collected on a Thermo Nicolet NEXUS instrument using an in situ IR cell with a CaF2 window. A self-supporting wafer consisting of a 15 mg sample was placed in the IR cell and connected to a vacuum system. The wafer was dehydrated at 723 K for 4 h under vacuum, and the background spectra of the samples were collected. Pyridine adsorption was carried out at 423 K for 1 h and evacuated at the same temperature for 2 h to detach weakly adsorbed pyridine molecules. IR spectra were collected with a resolution of 4.0 cm−1. For calculating the amounts of Brønsted and Lewis acid sites, IR bands at 1545 and 1455 cm−1 were used, respectively. Molar extinction coefficients of 1.67 cm μmol−1 and 2.22 cm μmol−1 were used for the Brønsted and Lewis acid sites, respectively.48 The adsorption of 2,2,4-trimethylpentane (Sigma-Aldrich, ≥ 99%) in zeolite samples was carried out using a modified thermogravimetric analysis instrument (TGA N-1000, Thermo Co.). Prior to the adsorption experiment, 5 mg of sample was degassed at 823 K under air flow (200 cm3 min−1) for 1 h and cooled down to 423 K under He flow (200 cm3 min−1). Subsequently, hydrocarbon adsorption was carried out at the same temperature by flowing He (200 cm3 min−1) saturated with 2,2,4-trimethylpentane vapor at 298 K. The diffusion kinetics were analyzed using Fick’s second law of diffusion:

⎛ ∂ 2C ⎞ ∂C = D⎜ 2 ⎟ ∂t ⎝ ∂x ⎠ During the initial stage of adsorption in a slab-like crystal, the following solution applies:49 q(t ) 2 = q(∞) π

D L2

t

where q(t)/q(∞) is normalized uptake, D is diffusivity, L is characteristic diffusion path length, and t is time. H2 chemisorption on supported Pt catalysts was measured at 323 K using an ASAP2020 instrument (Micromeritics). Prior to these measurements, samples were vacuum-degassed at 573 K for 6 h, reduced at 673 K for 2 h under H2, and finally evacuated for 2 h at the same temperature. Amount of chemisorbed H2 was calculated by extrapolation of the linear portion (7−30 kPa) of the isotherm to zero pressure. 2.5. Catalytic Measurements. All catalytic reactions were carried out in a stainless steel down-flow fixed-bed reactor. The liquid hydrocarbon products collected in a liquid trap were analyzed with GC equipped with FID and an HP-5 ms capillary column (Agilent, 30 m × 0.25 mm). n-Dodecane (TCI, > 99%) was used as an external standard for the quantification of liquid hydrocarbon products. The gaseous products were analyzed using an online GC equipped with FID and a GS-GasPro capillary column (Agilent, 30 m × 0.32 mm). In the two-step reaction, palm oil was first hydrotreated to produce n-paraffins with 1 wt % Pt/γ-Al2O3 as the catalyst. 2.4 g of sieved 1 wt % Pt/γ-Al2O3 catalyst (150−300 μm) was loaded in the reactor after physical mixing with 5 g of acid-washed quartz particles. After pretreatment at 673 K under H2 for 2 h, the reaction was carried out at 633 K and 2.0 MPa total pressure. The H2 flow rate was 116 cm3 min−1, and the liquid injection rate of palm oil was 2.4 g h−1 (WHSV = 1.0 h−1). The liquid hydrocarbons collected in the liquid condenser were separated from condensed water by decanting and used as the reactant for the second hydrocracking step. In the hydrocracking reaction, zeolite-supported Pt catalysts (1 wt % Pt on 6258

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis Table 1. Physicochemical Properties of Catalysts sample

Si/Ala

Pta (wt %)

SBETb (m2 g−1)

Sextc (m2 g−1)

Vmicroc (cm3 g−1)

Vmesod (cm3 g−1)

Vtotale (cm3 g−1)

nBrønstedf (μmol g−1)

nLewisf (μmol g−1)

H/Ptg

nPt/nBAh

Pt/γ-Al2O3 Pt/Bulk-ZSM-5 Pt/Nano-ZSM-5 Pt/Bulk-Beta Pt/Nano-Beta

0 27 28 12 13

0.98 0.92 1.00 0.95 1.08

200 398 496 540 789

200 43 167 35 316

0 0.16 0.15 0.25 0.22

0.43 0.06 0.36 0.04 1.14

0.43 0.22 0.51 0.29 1.36

0 266 227 393 289

179 21 56 122 112

0.80 0.91 1.08 0.88 0.93

0.16 0.24 0.11 0.18

a

Si/Al ratios and Pt contents were analyzed by ICP-AES. bBET surface areas were evaluated in the P/P0 range of 0.05−0.20. cExternal surface areas and micropore volumes were determined using t-plot analysis. dMesopore volumes were calculated by Vtotal − Vmicro. eTotal pore volumes were evaluated at P/P0 = 0.98. fThe number of acid sites was measured by FT-IR spectroscopy following pyridine adsorption. gH/Pt was determined by H2 chemisorption at 323 K. hThe moles of accessible Pt sites measured by H2 chemisorption divided by the moles of Brønsted acid sites determined by FT-IR spectroscopy following pyridine adsorption.

structures, i.e., ZSM-5 (MFI structure) and zeolite Beta (BEA structure), were prepared. The medium-pore ZSM-5 contains three-dimensional interconnected 10-membered-ring (MR) microporous channels, whereas the large-pore zeolite Beta contains three-dimensionally connected 12-MR channels. Further, to study the catalytic effects of zeolite crystallite size, both zeolite structures (ZSM-5 and zeolite Beta) were synthesized in two different crystallite sizes (i.e., bulk and nanocrystallites). As shown in Figure 1a, all zeolite samples

Bulk-ZSM-5, Nano-ZSM-5, Bulk-Beta, and Nano-Beta) were used as catalysts. Typically, 1.2 g of a sieved catalyst was loaded into the reactor after physical mixing with 5 g of acid-washed quartz particles. After pretreatment at 673 K under H2 for 2 h, the reaction was carried out at 488−523 K and 2.0 MPa total pressure. The H2 flow rate was 128 cm3 min−1, and the liquid injection rate of hydrocarbons (produced from the first hydrotreating reactor) was 2.4 g h−1 (WHSV = 2.0 h−1). For single-step direct hydroconversion, 1.2 g of the sieved 1 wt % Pt on Nano-Beta catalyst was loaded into the reactor after physical mixing with 5 g of acid-washed quartz particles. The loaded catalysts were pretreated at 673 K under H2 for 2 h. The reaction was carried out at 568 K and 2.0 MPa total pressure. The H2 flow rate was 116 cm3 min−1, and liquid injection rate of palm oil was 2.4 g h−1 (WHSV = 2.0 h−1). 2.6. Separation of Biojet Fuel by Distillation and Fuel Characterization. The liquid hydrocarbons obtained after the two-step reaction process were fractionated by distillation into light (boiling point 0.05). 3.2. Fatty Acid Composition of Palm Oil. The fatty acid composition of palm oil was determined by analyzing the composition of fatty acid methyl esters (FAMEs) after transesterification with methanol (Table 2). The results showed

contrast, Bulk-ZSM-5 and Bulk-Beta samples were found to be composed of much larger crystallites (>100 nm). In order to study the effects of the zeolite microporous structures and crystallite sizes on hydrocarbon diffusion, 2,2,4trimethylpentane adsorption kinetics were measured at 423 K. 2,2,4-Trimethylpentane was chosen as a model adsorbate because it represents well the bulky tribranched hydrocarbons produced during hydrocracking.40,50 As shown in Figure 3, the

Figure 3. Adsorption rates of 2,2,4-trimethylpentane in zeolite samples at 423 K.

Table 2. Fatty Acid Composition of Palm Oil

D/L2 values increased in the order of Bulk-ZSM-5 (0.0011 s−1) < Nano-ZSM-5 (0.0077 s−1) ≪ Bulk-Beta (0.0317 s−1) < Nano-Beta (0.0479 s−1). These results suggest that diffusion of the tribranched model hydrocarbon is drastically slower in ZSM-5 (MFI) containing 10-MR microporous channels than in zeolite Beta (BEA) with 12-MR channels. Within zeolites belonging to the same structural group, the nanocrystalline zeolites exhibited substantially faster molecular diffusion than bulk zeolites as a result of shorter diffusion path lengths along the microporous channels of these zeolites. The metal/acid bifunctional hydrocracking catalysts were prepared by supporting 1 wt % Pt on the synthesized zeolites. It has been well established that sufficient hydrogenation/ dehydrogenation functions (i.e., accessible Pt sites) relative to the amount of Brønsted acid sites are required for ideal bifunctional catalysis.41,51−53 Metal/acid bifunctional hydrocracking mechanism consists of five steps: (1) dehydrogenation of alkane on metal site to produce olefin; (2) transport of olefins from metal sites to Brønsted acid sites; (3) isomer-

fatty acid

composition (wt %)

myristic acid (C14:0)a palmitic acid (C16:0)a stearic acid (C18:0)a oleic acid (C18:1)a linoleic acid (C18:2)a

1.0 43.0 5.0 42.0 9.0

a

In the notation Cx:y, x is the carbon length in a fatty acid, and y is the number of unsaturated CC bonds.

that palm oil was composed mainly of C16 (43 wt %) and C18 (56 wt %) fatty acids. Palmitic acid (C16:0) with completely saturated hydrocarbon chain was the only C16 acid species detected. On the other hand, oleic acid (C18:1) containing a single unsaturated CC bond was detected as the major C18 acid species. 3.3. Two-Step Hydroconversion of Palm Oil into Biojet Fuel. Palm oil was converted catalytically via a twostep process consisting of separate hydrotreating and hydro6260

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis Table 3. Product Composition Obtained after Hydrotreating of Palm Oil Using a Pt/γ-Al2O3 Catalysta hydrotreating product composition (wt %)

a

CO

CO2

C1

C2

C3

C4−C14

C15

C16

C17

C18

H2O

3.0

3.6

5.5

2.9

3.3

3.2

32.8

3.2

36.3

3.1

3.1

Reaction conditions: 1 d, 2.0 MPa, 633 K, and WHSV = 1.0 h−1.

Figure 4. (a) Hydrocarbon distribution in the liquid product obtained from the first hydrotreating reaction over Pt/γ-Al2O3 catalyst (Table 3). (b) Distributions of hydrocarbons in liquid and gas products obtained after hydrocracking (reaction conditions: 2.0 MPa, WHSV = 2.0 h−1) over Pt/ Nano-Beta at (b) 493 K, (c) 498 K, (d) 503 K, (e) 508 K, and (f) 513 K. Orange bars indicate iso-paraffins, and blue bars indicate n-paraffins.

cracking reaction steps. In the first hydrotreating step, monofunctional Pt/γ-Al2O3 was used as the catalyst for the hydrogenation of unsaturated CC bonds, triglyceride hydrogenolysis into fatty acids, and their subsequent deoxygenation (reactions i−iii in Scheme 1). The liquid hydrocarbons produced in this reaction step were collected using a liquid trap, separated from gas products and liquid water, and subsequently employed as a reactant for the second hydrocracking reaction. In the hydrocracking reaction, various zeolite-supported Pt catalysts with metal/acid bifunction were utilized as the catalyst. As shown in Table 3, hydrotreating over Pt/γ-Al2O 3 converted palm oil fully into hydrocarbons under the employed experimental conditions (633 K, WHSV = 1.0 h−1). The reaction produced mainly saturated C15 and C17 n-paraffins with minor formation of C16 and C18 n-paraffins. Only the saturated paraffins were detected, indicating that the unsaturated CC bonds in fatty acids (e.g., oleic and linoleic acid) were hydrogenated completely. Considering that palm oil contains predominantly C16 and C18 fatty acids (Table 2), the major formations of C15 and C17 n-paraffins containing one less carbon atom indicate that deoxygenation of fatty acids proceeded mainly via decarbonylation/decarboxylation (removal of oxygen by CO/CO2) rather than hydrodeoxygenation (removal of oxygen by H2O) (Scheme 1). This observation is consistent with the results reported in earlier studies, which showed that noble metal catalysts (Pt27−30 and Pd27,30−33) prefer decarbonylation/decarboxylation over hydrodeoxygenation. As a consequence, substantial amounts of CO and CO2 and their methanation product (C1) were detected in the gas products (Table 3). Small amounts of C2 and C3 hydrocarbons were also produced, which appeared to be generated from the

conversion of glycerol. The formation of cracking products (C4−C12) of long-chain hydrocarbons was very minor as a result of the limited Brønsted acidity of Pt/γ-Al2O3 (Table 1). The hydrocarbons collected in the liquid trap amounted to 78.0 wt % of the palm oil injected initially as a reactant. This value is very close to the theoretical liquid hydrocarbon yield of 80.5 wt %, which was calculated based on the fatty acid composition of palm oil (Table 2) and the assumption that glycerol was decomposed as light gaseous species. The hydrotreating reaction was found to be stable up to 4 d time-on-steam, and no significant changes in conversion and selectivity were observed (Figure S1), which indicates that catalyst deactivation was negligible. The liquid hydrocarbons collected in the first hydrotreating step (Figure 4a) were used as the reactant for the second hydrocracking reaction, in which zeolite-supported Pt catalysts were employed as the metal/acid bifunctional catalyst. The hydrocarbon distributions determined as a function of reaction temperature are shown for Pt/Nano-Beta catalyst in Figure 4. As will be discussed later, this catalyst was the best-performing hydrocracking catalyst tested in this study, producing the highest yield of jet fuel. To study the relationship between conversion levels and product selectivity, the hydrocracking reaction was carried out at a range of reaction temperatures (493−513 K) under fixed space velocity (WHSV = 2.0 h−1). It should be noted that the difference in activation energies for hydroisomerization and hydrocracking is known to be insignificant, and therefore, the product selectivities depend mainly on the level of conversion rather than reaction temperature.54−56 6261

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis

Figure 5. (a) Hydrocarbon distribution in the liquid product obtained from the first hydrotreating reaction over the Pt/γ-Al2O3 catalyst (Table 3). (b) Distributions of hydrocarbons in liquid and gas products obtained after hydrocracking (reaction conditions: 2.0 MPa, WHSV = 2.0 h−1) over Pt/ Bulk-Beta at (b) 488 K, (c) 493 K, (d) 498 K, (e) 503 K, and (f) 508 K. Orange bars indicate iso-paraffins, and blue bars indicate n-paraffins.

Figure 6. (a) Hydrocarbon distribution in the liquid product obtained from the first hydrotreating reaction over the Pt/γ-Al2O3 catalyst (Table 3). (b) Distributions of hydrocarbons in liquid and gas products obtained after hydrocracking (reaction conditions: 2.0 MPa, WHSV = 2.0 h−1) over Pt/ Nano-ZSM-5 at (b) 503 K, (c) 508 K, (d) 513 K, (e) 518 K, and (f) 523 K. Orange bars indicate iso-paraffins, and blue bars indicate n-paraffins.

Figure 4e and f). The preferential isomerization/cracking of longer-chain hydrocarbons can be explained by their preferential adsorption in the zeolite catalyst owing to stronger van der Waals interactions. According to previously reported kinetic studies,57 apparent reaction rates in the hydrocracking of mixed hydrocarbons are mainly controlled by adsorption, increasing with the chain length of hydrocarbons. The individual intrinsic reaction rates of different hydrocarbons were found to be quite similar. As the reaction temperature increased (Figure 4), lighter hydrocarbons were produced by the preferential cracking of long-chain hydrocarbons. Up to the reaction temperature of 508 K (Figure 4e), the yield of jet fuel-range hydrocarbons (C8−C16) continued to increase gradually.

In the low conversion regime (Figure 4b−c), small amounts of cracking products ( rtype‑B1 ≈ rtype‑B2 > rtype‑C ≫ rtype‑D.40 Among the kinetically important reaction pathways (A, B1, B2, and C), the cracking of tribranched isomers (A) produces two branched hydrocarbon fragments, the cracking of dibranched isomers (B1 and B2) produces a linear and branched fragments, and the cracking of monobranched isomers (C) produces only linear fragments. This means that, the more the formation of highly branched hydrocarbons before β-scission, the higher is the iso/n-paraffin ratio in the hydrocracking products. The molecular dimension of hydrocarbons generally increase with branching: n-paraffin (∼4.5 Å) < monobranched isomer (5.8− 6.0 Å) < dibranched isomer (6.1−6.3 Å) < tribranched isomer (>6.5 Å).61 Considering the medium-size microporous

channels of ZSM-5 (5.1 × 5.5 Å along the [100] direction; 5.3 × 5.6 Å along the [010] direction), the formation of multibranched species before β-scission should be highly inhibited, which can result in low iso/n-paraffin ratios of hydrocracking products. In contrast, within the large microporous channels of zeolite Beta (6.6 × 6.7 Å along the [100] direction and 5.6 × 5.6 Å along the [001] direction), even the tribranched hydrocarbons can be readily formed without spatial constraint. As shown in Figures 4−7, the maximum achievable C8−C16 yields increased in the order of Pt/Bulk-ZSM-5 (49.8 wt %, Figure 7b) < Pt/Nano-ZSM-5 (57.0 wt %, Figure 6d) < Pt/ Bulk-Beta (64.0 wt %, Figure 5e) < Pt/Nano-Beta (71.5 wt %, Figure 4e). Under the conditions allowing the maximum C8− C16 yields, the iso/n-paraffin ratio of products also increased in the same order: Pt/Bulk-ZSM-5 (0.08, Figure 7b) < Pt/NanoZSM-5 (0.63, Figure 6d) < Pt/Bulk-Beta (3.66, Figure 5e) < Pt/Nano-Beta (7.54, Figure 4e). It is notable that high iso6263

DOI: 10.1021/acscatal.7b01326 ACS Catal. 2017, 7, 6256−6267

Research Article

ACS Catalysis paraffin content is important for obtaining low freezing point (