Preparation and Performance of the Lipid Hydrodeoxygenation of a Nickel-Induced Graphene/HZSM-5 Catalyst

Abstract

Graphene-encapsulated nickel nanoclusters are a feasible strategy to inhibit the nickel deactivation of nickel-based catalysts. In this work, graphene-encapsulated catalysts (Ni@C/HZSM-5) were prepared by a compression forming process, using pseudo-boehmite, Al₂O₃, and ZrO₂ as binders. The pseudo-boehmite was gradually transformed from amorphous to crystalline alumina at high temperatures, which destroyed the nucleation of Ni@C. In contrast, the crystal-stabilized zirconia was more favorable for the nucleation of Ni@C. The extensive dispersion of alumina on the surface of HZSM-5 covers the acid sites of HZSM-5. In contrast, when zirconia was used as the binder, the binder existed in the form of the direct aggregation of ~100 nm zirconia spheres; this distribution form reduced better the damage of the binder to the acid site of the catalyst. Furthermore, the particle size of Ni crystals in the graphene-encapsulated catalysts decreased significantly (mostly <11 nm), and no evident agglomeration of nickel particles appeared. It was found that the stabilization of the metal interface delayed, to an extent, the accumulation rate of carbon deposits and, thus, postponed the deactivation of the acid sites. After 8 h of continuous reaction, the conversion of the traditional catalyst Ni/Z5+Zr dropped significantly to 60%. In contrast, the conversion of Ni@C catalysts prepared with ZrO₂ remained above 90%. The regeneration test shows that air roasting could effectively remove carbon deposits and restore the catalyst activity.

1. Introduction

The global carbon reduction plan is related to the sustainable development of the environment and ecology. Biomass energy is considered an alternative to fossil fuels, with the advantages of low carbon emissions, environmental protection, and sustainable development of new energy sources [1]. The development of biomass energy will help to accelerate the process of global carbon reduction. The conversion of oil into bio-jet fuel through hydrodeoxidation, cracking, aromatization and isomerization is an important biomass conversion route [2]. Compared to conventional petroleum-based jet fuels, bio-jet fuel can reduce CO₂ emissions by more than 55% and up to 90% during the whole life cycle [3,4,5]. Therefore, it is more and more attractive to produce bio-jet from renewable oil, such as saturated or unsaturated fatty acids (SFAs) found in vegetable oils, triglycerides, animal fats, and waste cooking oils (WCOs) [6].

The conversion process of bio-jet fuel mainly consists of two steps. The first step is to convert triglyceride and unsaturated fatty acids into saturated fatty acids by hydrogenation. Then, saturated fatty acids are converted to C17–C18 linear alkanes through hydrodeoxygenation and decarboxylation reactions. The second step is the selective hydrocracking and deep isomerization of the deoxygenated linear alkanes to generate mixed liquid fuel of highly branched alkanes [7]. The common catalysts for these steps require Lewis and Brønsted acid sites and metal centers [8,9]. Conventional sulfide state catalysts NiMo/γ-Al₂O₃ and CoMo/γ-Al₂O₃ exhibited high activity, but they deactivated rapidly due to potential toxicity in an aqueous solution and carbon deposition [10]. Noble metals (such as Pt, Pd, and Ru) pose high activity and selectivity for fatty acid conversion, but the high prices hinder their large-scale applications [11]. Therefore, the researchers turned their attention to non-noble metal catalysts. Nickel-based catalyst is currently the most widely used non-noble metal catalyst in the industry. The cost of nickel is only one-thousandth that of noble metals, and it is more suitable for industrial applications as metal centers [12,13]. Nickel has high hydrogenation activity due to the electron holes in the d orbital. In addition, nickel can effectively activate the C–C bond during the deoxygenation of fatty acids (Cn) to generate Cn-1 alkanes [8,9,11]. However, traditional nickel-based catalysts are prone to deactivation during a continuous reaction. On the one hand, metal particles are easily leached or agglomerated. On the other hand, polycyclic aromatic hydrocarbons are formed during lipid hydrodeoxygenation, resulting in carbon deposits that cover the active sites of the catalyst [14,15]. HZSM-5 zeolite molecular sieve poses Lewis and Brønsted acid sites required for hydrodeoxygenation, high specific surface area, thermostability, and hydrothermal stability. It has become the most promising catalyst for the preparation of bio-jet fuel. In our previous research, the catalyst Ni@C/HZSM-5 was obtained by loading the graphene-encapsulated nickel nanoparticle structure (Ni@C) on the zeolite HZSM-5. It found that the Ni@C effectively protected the metal nanoparticles from the acid solution, enhanced the activity of the catalyst, and slowed down the deactivation rate of the metal center due to the synergistic effect of carbon materials and metal nanoparticles [2].

In a packed bed reactor (PBR), the powder of the catalysts may block the reaction pipe and cause an increase in the reactor pressure [16]. Therefore, catalyst molding is crucial for the industrial application of catalysts. During the molding process of a catalyst, binders are indispensable additives so that the catalyst can meet the mechanical strength requirements for industrial applications [17]. Among them, the enhancement of mechanical strength is realized by the adhesion force of the binder and the cross-linking of terminal hydroxyl groups between adjacent binder particles [18]. Commonly used binders are Al₂O₃ [19,20], SiO₂ [21,22], pseudo-boehmite [23,24], and ZrO₂ [25]. The ion exchange capacity and acid catalytic activity of the shaped HZSM-5 was significantly enhanced when the catalyst was extruded with an Al₂O₃ binder [26]. Al₂O₃ provided more acid sites, and the shaped catalyst also had a good pore structure [21]. After adding a pseudo-boehmite binder to the catalyst, the shaped catalyst PtS/Z-A30-650 showed a higher n-C5/C6 isomerization activity and mechanical strength [24]. Furthermore, the added ZrO₂ provided an additional diffusion path for mass transfer between SAPO-34 crystals, making the shaped catalyst SAPO-34/ZrO₂ show a higher catalytic lifespan than SAPO-34 [25]. However, adding a binder also generates negative effects on the catalyst, such as the silica as a binder decreasing the catalytic activity by blocking the porosity of zeolites and reducing mass transfer efficiency [14,17]. Since the formation of the Ni@C catalyst is a crucial step towards its industrial application, which has never been reported, the influence of the binder on the nucleation process of graphene-encapsulated Ni clusters remains unclear.

In this work, we use three common binders, pseudo-boehmite, alumina, and zirconia, as raw materials to study the molding preparation of Ni@C coupled with the traditional zeolite HZSM-5. Furthermore, the stability of various-shaped catalysts for continuous hydrogenation to prepare bio-aviation kerosene was investigated in a fixed bed high temperature and high-pressure reactor. The performance was compared with that of traditional Ni/Z5+Zr. In this paper, we use zeolite HZSM-5 as support and study the ability of different binders to form carbon-encapsulated nickel structures by adding nickel, carbon source, and binder at one time. Moreover, the study investigates the utilization of oleic acid as a model compound, analyzes the hydrodeoxygenation performance and catalytic lifespan of the shaped catalysts and assesses their capability to produce bio-jet fuel.

2. Results and Discussion

2.1. Analysis Nucleation of Ni@C

According to our previous work [2], carbohydrates, such as citric acid, glucose, and sorbitol, can be used as carbon sources to generate graphene layers. In comparison, sucrose is more conducive to reducing the size of the nickel cores under the same addition amount and is more conducive to industrial application for its low cost. The nucleation process of Ni@C was carefully studied by TG-FTIR in previous reports [2]. The impregnation process ensured that the nickel nitrate and sucrose were uniformly mixed and well dispersed on the surface of the molecular sieve. During the roasting and heating process, the sucrose and the nickel nitrate were decomposed into amorphous carbon and nickel oxide, respectively. Due to the high surface energy of the nickel species, the nickel species were very easy to agglomerate. In this system, the nickel nitrate originally mixed in sucrose was gradually decomposed into nickel oxide and rapidly enriched into small clusters with the calcination temperature increase. Then, a solid dispersion was formed in amorphous carbon. In the subsequent process, the nickel nucleus increased to a certain extent, and an amorphous-carbon inert layer was formed, which blocked the continuous increase in the nickel species. The nickel oxide was subsequently reduced to Ni⁰ by the surrounding amorphous carbon, and Ni⁰ as a “catalyst” induced the rearrangement of the amorphous carbon to form multilayer graphene.

In this study, it was found that the nucleation must be completed at a structurally stable interface. Through the TG analysis of pseudo-boehmite (Figure 1), it could be seen that pseudo-boehmite first lost water at about 100 °C and then was gradually transformed from an amorphous state to crystalline alumina (Figure 1b). The Ni@C structure could not be formed completely when the catalyst was prepared by a one-step method with pseudo-boehmite as the binder (Figure 2c). During this process, the mass loss of pseudo-boehmite is about 24 wt%. This process may be the key factor affecting the nucleation of Ni@C.

To verify the above speculation, firstly, the pseudo-boehmite and HZSM-5 were fully calcined at 600 °C for 2 h to stabilize the composition, and then the Ni@C nucleation test was carried out; in this way, nickel clusters were well-coated by the graphene layers (Figure 2). It proves that the stable interface is the key to the nucleation of Ni@C. In addition, zirconia has a stable crystal structure and does not affect the nucleation of Ni@C, so zirconia is more suitable for the one-step molding of the Ni@C catalysts.

2.2. Characterization of the Shaped Catalysts

In order to investigate the effect of different binders on catalyst particles and micromorphology, the TEM technique was used to characterize the catalysts Ni@C/Z5+PB, Ni@C/Z5+Al, Ni@C/Z5+Zr, and Ni/Z5+Zr. Ni@C was composed of inner Ni⁰ clusters and 2–4 outer layers of graphene. The exposed crystal planes of the internal nickel were dominated by (1 1 1) (Figure 2), and the outer graphene layers had typical corrugated intervals of 3.4 angstroms, which confirmed its graphene structure.

As shown in Figure 2c(I-1–I-7), the nickel particles of Ni@C/Z5+PB prepared with pseudo-boehmite appeared to have obvious agglomeration. The TEM images of the catalysts Ni@C/Z5+Al and Ni@C/Z5+Zr showed that a graphene-encapsulated nickel nanoparticle structure was formed. After the graphene encapsulation, the particle size of Ni crystals in the catalyst decreased from 15 nm in Ni/Z5+Zr to 7–8 nm in Ni@C/Z5+Zr. Then, the integrity of the graphene carbon layer of the catalyst was further analyzed by an acid pickling test. The results are shown in Figure 2c(II-1–II-3). After pickling, the catalyst Ni@C/Z5+Zr retained a complete graphene-encapsulated nickel particle structure. The acid pickling test was carried out by taking 0.1 g of catalyst and placing it in a configured 15 mL 40 wt% H₂SO₄ aqueous solution with magnetic agitation at 60 °C for 30 min. The treated samples are morphologically characterized by TEM, indicating that the complete graphene carbon layer was prepared entirely using ZrO₂ as an effective binder, which encapsulated nickel nanoparticles, and effectively avoided the loss of the metal center. However, a large amount of the graphene shell structure was observed in the catalyst Ni@C/Z5+PB. It means that the formed graphene carbon layer did not completely encapsulate the nickel particles, resulting in the loss of nickel loading ratio from 4.23 wt% to 1.02 wt% in the process of pickling (

Table 1. EDS of the shaped catalysts.

	Ni@C/Z5+N	Ni@C/Z5+Al	Ni@C/Z5+Zr	U-Ni@C/Z5+Zr	U-Ni@C/Z5+Zr after Air Calcination	Ni@C/Z5+Al after Pickling Test	Ni@C/Z5+Zr after Pickling Test
C	10.32	8.43	8.62	12.75	0.99	7.61	8.17
O	44	21.62	50.62	47.92	37.85	26.23	37.31
Al	2.65	32.06	2.68	2.28	3.40	30.39	3.17
Si	38.5	33.66	29.73	30.47	52.10	34.75	43.91
Ni	4.53	4.23	4.51	4.46	4.21	1.02	4.23
Zr	-	-	3.84	2.12	1.45	-	3.21

). Therefore, pseudo-boehmite as a binder was not conducive to the formation of Ni@C. Perhaps, it was because the pseudo-boehmite was not fully pyrolyzed at 550 °C (Figure 1a), and a small number of hydroxyl groups from the generated water inhibited the formation of graphene carbon layers [27]. So, the pseudo-boehmite was further calcinated at 600 °C, and the obtained product Al₂O₃ was used as a binder to prepare the catalyst Ni@C/Z5+Al. It is found that the Ni@C structure was formed successfully.

The crystal structures of Ni@C/Z5+Al, Ni@C/Z5+Zr, Ni/Z5+Zr, and Ni@C/Z5+N were analyzed by XRD patterns. The X-ray diffraction patterns of the catalysts are shown in Figure 2a. The diffraction peaks were observed at 2θ = 7.88, 14.78, 23.15, 23.88, 45.23°, etc. corresponding to HZSM-5. The position and intensity of the peaks were consistent with the standard card of the HZSM-5 (PDF#00-049-0657). It was also determined that there were obvious diffraction peaks of nickel at 44.5° and 51.8° (PDF#01-070-1849). The diffraction peaks of Ni@C/Z5+Al, Ni@C/Z5+N, and Ni@C/Z5+Zr at Ni⁰ 44.6° are wide and weak, showing that the relative grain size of a nickel is small. The width of the lattice diffraction fringe of nickel is 1.97 Å (Figure 2c(I-6)), which is consistent with the Ni (1 1 1) surface. In addition, the lattice fringe spacing of ZrO₂ is 2.79 Å (Figure 2c(II-3)), which matches well with the (1 1 1) surface of the ZrO₂ standard card (PDF#00-049-1642). The average size of the corresponding ZrO₂ crystals was between ~58.17 and 104.2 nm. The binder ZrO₂ exists in the form of direct aggregation of ~100 nm zirconia spheres, and this distribution reduces better the coverage of the catalyst by the binder, thereby reducing the damage of the acid site of the catalyst.

XRD patterns and TEM images were used to analyze the structure and micromorphology of the shaped catalysts after the reaction. The crystallinity changes of these catalysts before and after the reaction were discussed. As shown in Figure 2b, it was observed that the diffraction peak of U-Ni/Z5+Zr (“U-” denotes the used catalyst) at the Ni⁰ 44.6° becomes sharper after the reaction, while those of U-Ni@C/Z5+Zr, U-Ni@C/Z5+Al, and U-Ni@C/Z5+N were still wide and weak. The strong peak intensity of U-Ni/Z5+Zr may be attributed to the agglomeration of nickel particles after the reaction, as shown in Figure 2c(III-1). From Figure 2c(III-3), the distribution and the structure of the graphene-encapsulated nickel particles can be obviously observed. No significant agglomeration of nickel was observed, and the nickel particles retained a smaller particle size. As shown in Table 1, the content of catalyst elements was measured by EDS, and it was found that the actual loading rate of nickel was 4~5%. This phenomenon further indicated that graphene-encapsulated nickel particles effectively reduced the leaching and agglomeration of the metal center during the reaction and improved the activity of the metal center. The dispersity of the loaded metal affects the contact between the substrate and the active site [28,29]. Improving the dispersity of metal particles can increase the contact between the substrate and the active site, which is one of the most favorable means to improve the catalytic efficiency greatly. As shown in Figure 2c(III-2), carbon deposition was observed on the surface and in the molecular sieve pores on the shaped catalyst after continuous hydrogenation.

2.3. Analysis of Acidity

Converting lipid to bio-jet fuel requires the synergistic catalysis of metal centers and acid sites. Acidity plays an important role in product selectivity as acidity is prone to yield more cracking products [30,31]. The acid content of the HZSM-5 carrier was investigated using NH₃-TPD. The NH₃-TPD profiles have three peaks at about 180, 280, and 410 °C, corresponding to weak acid, medium–strong acid, and strong acid [32], respectively (Figure 3a). The total acidity of the HZSM-5 molecular sieve was 1089 μmol/g. The contents of strong acid, medium–strong acid, and weak acid were 401.31 μmol/g, 143.44 μmol/g, and 544.25 μmol/g, respectively.

The Lewis and Brønsted acid sites for the HZSM-5 sample were determined by Py-FTIR at a desorption temperature of 150 °C [33]. As shown in Figure 3b, the peaks at 1545 and 1450 cm⁻¹ corresponded to pyridine interacting with the Brønsted and Lewis acids [16], respectively. According to Equations (1) and (2) [34], the content of Lewis and Brønsted acids was calculated to be 0.196 mmol/g and 0.358 mmol/g, respectively. The desorption analysis for different acid types was carried out at 350 °C, and the type of strong acid was studied. As shown in Figure 3b, when the temperature rose from 150 to 350 °C, the peak of the Lewis acid at 1450 cm⁻¹ weakened considerably; when the temperature was at 350 °C, there were strong adsorption peaks of Brønsted acids at 1540 cm⁻¹, indicating that most of the strong acids in ZSM-5 were Brønsted acids.

$$C_{\mathrm{Brønsted}}=1.88\times I_{\mathrm{Brønsted}}\times R^2/W$$

(1)

$$C_{\mathrm{Lewis}}=1.42\times I_{\mathrm{Lewis}}\times R^2/W$$

(2)

C = mmol/g; I = integrated absorbance of the Brønsted or Lewis band (cm⁻¹); R = radius of the catalyst disk (cm); W = weight of the disk (mg).

2.4. Analysis of Lipid Conversion Route and Reaction Mechanism

Based on the complexity of the interface composition of heterogeneous catalysts, the transformation of lipids in this system has multiple potential transformation routes, and each route is coupled with each other, which is the fundamental reason for the extremely complex final products [35].

In this system, HZSM-5 was a micron molecular sieve, the size of unwrapped nickel nanoclusters was ~15 nm, and the size of Ni@C clusters was about ~7 nm (as shown in Figure 2). Since HZSM-5 was a typical microporous molecular sieve with a micropore size of ~0.5 nm [36], oleic acid was mainly converted on the catalyst surface. The surface acting sites of molecular sieves are primarily in the vicinity of Al atoms. As a typical zeolite material, the structure of ZSM-5 follows Löwenstein’s rule [37]. The silicon atom cannot form an eight-electron stable structure due to the lack of four electrons in the outermost layer, so it coordinates with four oxygen atoms to form a stable structure during the hydrothermal nucleation process. When an Al atom replaces a certain Si atom, the outermost layer of the Al atom has three electrons; if an Al atom forms a four-coordinate structure with four oxygen atoms, it will lack one electron to form a stable eight-electron structure. Therefore, it possesses electron-withdrawing Lewis acid properties. In addition, from the perspective of the overall molecular sieve system, the substitution of one Al atom for one Si causes the system to produce an extra negative charge. To maintain the charge balance, a hydrogen atom is also introduced during the hydrothermal process, which is the Brønsted acid site.

During the lipid conversion process, the hydrogen molecules are dissociated and activated on the nickel surface, mainly attributed to the strong electron-withdrawing characteristic of the d orbital of the nickel atom. The two acid sites of HZSM-5 dominate the cracking, isomerization, and aromatization processes. Since the active site distribution of the heterogeneous catalysts cannot be uniform, the catalytic interface can be mainly divided into three types: isolated metal centers/Ni@C, isolated acid sites, and metal-solid acid coupling sites.

On the isolated metal center, the hydrogen molecule is induced to dissociate by nickel, whose electrons are transferred to the nickel atom. Hence, the electrostatic potential of a hydrogen atom is positive. The middle carbon–carbon double bond of oleic acid and the vicinity of the carbonyl oxygen is strongly negatively charged due to the concentration of electrons. Due to electrostatic interactions, the carbon–carbon double bond or the carbonyl oxygen will be first hydrogenated into stearic acid or oleyl alcohol, then further hydrogenated into octadecanol completely converted into noctadecane. At the same time, the carboxyl site is very active and easily converted into heptadecane by decarbonylation or decarboxylation reaction, which can be confirmed by the products of carbon dioxide and carbon monoxide in the reaction.

Under the action of isolated acid sites, oleic acid is very easy to undergo a decarboxylation or decarbonylation reaction and transform into a series of heptadecane isomeric components. When adsorbed on the metal-solid acid coupling interface, the long-chain alkanes or alkenes obtained in the early stage will be converted into isoalkanes by hydroisomerization. However, if adsorbed on an isolated metal center, all the long-chain alkanes and alkenes will hydrogenolysis to short-chain alkanes occurs with methane production. If they were adsorbed on an isolated acid site, long-chain alkanes and alkenes would be converted into short-chain olefins or alkanes by β-cracking and further aromatized to aromatics [38].

It can be seen that, no matter what route oleic acid undergoes to convert into long-chain hydrocarbons, as long as the isolated acid site is strong enough, the previous components will be converted into aromatic hydrocarbons. The continued aromatization of aromatic hydrocarbons will produce polycyclic aromatic hydrocarbons with a large molecular weight, which will eventually form carbon deposits and cause catalyst deactivation [15].

2.5. Catalytic Performance of the Catalysts

The setting of the test conditions in this part was based on the previous research basis [2]. This work aims to discuss the influence of adding different binders on the product components, and the component generation process is based on the above analysis. The continuous hydrodeoxidation performance of the shaped catalysts Ni/Z5+Zr, Ni@C/Z5+Zr, Ni@C/Z5+Al, and Ni@C/Z5+N was studied. Oleic acid was used as feedstock, and the reaction was carried out in a continuous fixed-bed reactor under the conditions of 370 °C, LHSV (ratio of oleic acid volume to catalyst volume) of 2.92 h⁻¹, H₂/Feed (ratio of H₂ volume intake to oleic acid volume) of 1214 mL/mL and 4 MPa. Figure 4 shows the conversion and yield curve of the C8–C18 hydrocarbons over time. It can be seen that the initial activity of the three graphene-encapsulated catalysts is more or less the same as that of the traditional catalyst Ni/Z5+Zr, and all of the catalysts exhibit good stability and a high catalytic activity during the first 6 h of continuous reaction. The conversions of oleic acid to jet fuel remained above 90% over 6 h of reaction time. However, the yields of jet fuel were relatively low (<40%). This may be due to the rich, strong Brønsted acid content in the catalyst, which led to excessive cracking of the product and reduced the yield of C8–C18 hydrocarbons [19,30].

However, after 6 h of continuous reaction, the four catalysts showed different activity and stability. Ni@C/Z5+N and Ni/Z5+Zr showed a rapid decrease in the conversion rate, which decreased to less than 60% after 8 h, while the catalyst was prepared with binder Al₂O₃ or ZrO₂ still showed a conversion rate of more than 90%. Especially for the catalyst Ni@C/Z5+Zr, the continuous reaction time was extended to 12 h, and the conversion rate remained above 60%. The reason for the sharp decline in the catalytical efficiency of Ni@C/Z5+N may be that the strength of the binder-free catalyst Ni@C/Z5+N was weak; the catalyst was prone to cracks in the reaction environment with a high temperature and high pressure for a long time, and rapidly lost the partial metal and active zeolite centers, resulting in a reduced catalytic activity [39]. The decrease in the activity of the traditional catalyst Ni/Z5+Zr was related to nickel agglomeration, as shown by the TEM results (Figure 2c(III-1)). On the contrary, the shaped catalysts Ni@C/Z5+Zr and Ni@C/Z5+Al had a better stability and longer lifespan because of the good dispersion of nickel particles and the protection of the graphene carbon layer, as well as the improvement of the catalyst strength caused by the addition of the binder, which avoided the cracking of the catalyst and the loss of the metal center and active zeolite center. Nevertheless, after the reaction time exceeded 9 h, Ni@C/Z5+Al and Ni@C/Z5+Zr also showed a reduced conversion rate (Figure 4a), mainly caused by carbon deposition [15]. During the continuous reaction from 3 h to 11 h, the catalyst Ni@C/Z5+Al showed a 39.6% decrease from 97.0% to 58.6%, and the Ni@C/Z5+Zr catalyst showed a 29.1% decrease in conversion from 98.4% to 69.8%. It indicates that the Ni@C/Z5+Zr catalyst had a better stability than the Ni@C/Z5+Al catalyst.

The effects of the four shaped catalysts on the selectivity distribution of aromatics, C8–C16 alkanes and C17–C18 alkanes in the conversion products were further studied. The results are shown in Figure 5. It was found that hydrodeoxidation catalyzed by the four catalysts generated distinguish products over time. The aromatic content gradually decreased, and the content of the C17–C18 component gradually increased, showing the activity of the acid center in these catalysts decreased and the cracking activity weakened [40]. Compared with Ni@C/Z5+Zr, the aromatic selectivity of the Ni@C/Z5+N catalysts decreased faster. During the continuous reaction period of 6–8 h, the aromatic selectivity of Ni@C/Z5+N rapidly reduced by 41.2%, from 72.9% to 42.9%, as shown in Figure 5c. The aromatic selectivity of Ni@C/Z5+Zr decreased by 10.1%, from 93.7% to 83.7%, as shown in Figure 5b. In addition, during the continuous reaction period of 8–12 h, the C17–C18 alkane selectivity of Ni@C/Z5+Zr increased from 9.6% to 38.9%, while that of Ni@C/Z5+N had little change.

As shown in Figure 6, the amount of strong Brønsted acid of Ni@C/Z5+Zr is 0.053 mmol/g, while Ni@C/Z5+Al is 0.041 mmol/g. Furthermore, it can be seen from Figure 5 that Ni@C/Z5+Zr also maintains a better catalytic activity in aromatic hydrocarbon formation and exhibits more active acid centers than Ni@C/Z5+Al. When the conversion rate was greater than 90%, the aromatic selectivity of Ni@C/Z5+Zr was 93.7%, while that of Ni@C/Z5+Al was only 51.6% (Figure 5a). The catalyst Ni@C/Z5+Zr showed high aromatics selectivity, which was mainly attributable to the better maintenance of the exposure of the molecular sieve surface by the support in the zirconia system. As we discussed above (Section 2.4), the generation of aromatics was mainly attributed to the isolated acid sites of the molecular sieve. Without a binder, the surfaces of the molecular sieve were directly overlapped and formed. This interface overlap would inevitably cause a large loss of surface acid sites. When the agent was used, alumina was widely dispersed on the surface of HZSM-5(Figure 7), so that the surface acid sites were largely covered. When zirconia is used as the binder, the binder exists in the form of direct aggregation of ~100 nm zirconia spheres, and this distribution form better reduces the damage of the binder to the acid site of the catalyst.

The aromatic selectivity of the traditional catalyst Ni/Z5+Zr decreased faster than that of the graphene-encapsulated Ni@C/Z5+Zr catalyst. As shown in Figure 5b,d, the aromatic selectivity of the conventional catalyst Ni/Z5+Zr and the Ni@C/Z5+Zr catalyst decreased by 53.4% and 20.0%, respectively, during the continuous reaction time of 2–8 h. The rapid decrease in aromatic selectivity of the catalysts was attributed to the rapid weakening of the activity of the acid center [41] covered by carbon deposition [42]. There is a competitive relationship between the metal center and the acid site corresponding to the adsorption of the substrate, and the two active sites have different conversion routes for the substrate components. When the acid center dominates the reaction, the products are mainly aromatics. When there is a competitive relationship between the metal center and the acid site center, the products are mainly alkanes [35]. The stability of the metal center effectively inhibits the activity of the acid site. This is because the Ni@C catalyst has a slower rate of aromatics selectivity decline compared with traditional catalysts, indicating that its acid site deactivation is slower. This is precisely because Ni@C effectively inhibits the acid-site-dominated aromatization process, which delays the accumulation of carbon deposition. However, as the heterogeneous catalysis we analyzed above cannot avoid the existence of isolated acid sites, the aromatization process cannot be prevented, so the catalyst deactivation caused by carbon deposition in this system cannot be avoided. In contrast to traditional catalyst Ni/Z5+Zr, the stabilization of the metal interface of Ni@C/Z5+Zr delayed the accumulation rate of carbon deposits [2]. In other words, the traditional catalyst Ni/Z5+Zr is easier to deactivate than the Ni@C/Z5+Zr catalyst.

As shown in Figure 8, the acid site of the catalyst was analyzed by pyridine infrared before and after the reaction, and Brønsted and Lewis acids were reduced from 0.077 mmol/g and 0.133 mmol/g to 0.004 mmol/g and 0.029 mmol/g, respectively, after the enrichment reaction due to carbon deposition. Based on this, in the catalyst regeneration process, the carbon deposits on the catalyst surface are removed by oxygen-enriched roasting.

After 7 h of continuous hydrogenation, the lipid conversion rate remained at ~95%, but showed an obvious downward trend, and the conversion rate decreased sharply from 8 h. This may be because carbon accumulation is similar to the seed effect during crystal nucleation, and a small amount of seed template accelerates the crystallization process. Therefore, the catalyst was selected to be regenerated when the accumulation of carbon deposits was not too great. The results show that, after regeneration, the catalyst activity increased by 8.8% from 91.0% to 99.8% at the 8th hour compared with that without regeneration (Figure 8). Although the conversion rate decreased after four hours of reaction, it remained above ~90%. Correspondingly, the lipid conversion rate of the regenerated catalyst was reduced to ~70%. This showed that the catalyst regeneration method could prolong the catalyst life, which is crucial for reducing the production cost in industrial applications. However, after regeneration, the refresh catalyst could not restore the initial activity of the fresh catalyst, mainly because air roasting could not completely remove all carbon deposits. These carbon deposits acted as a “template” for subsequent carbon deposits, which accelerated the accumulation of carbon deposits.

3. Materials and Methods

3.1. Materials

The HZSM-5(SiO₂/Al₂O₃ = 32) used in this work was purchased from Shengli Chemical Technology Co., Ltd., (Changsha, China). Others were purchased from different commercial suppliers, as shown in

Table 2. Chemical reagents used in this work.

Entry	Name	Formula	Grade	Company
1	Nickel nitrate hexahydrate	Ni(NO3)2·6H2O	≥98%	Guangzhou Chemical Reagent Factory, (Guangzhou, China)
2	Oleic acid	CH3(CH2)7CH=CH(CH2)7COOH	AR	Guangzhou Chemical Reagent Factory, (Guangzhou, China)
3	Ethanol	C2H5OH	99.5%	Shanghai Macklin Biochemical Co., Ltd., (Shanghai, China)
4	Sucrose	C12H22O11	AR	Damao Chemical Regent Factory, (Tianjing, China)
5	Pseudo-boehmite	AlOOH·H2O	AR	Shanghai Macklin Biochemical Co., Ltd., (Shanghai, China)
6	Zirconium dioxide	ZrO2	AR	Shanghai Macklin Biochemical Co., Ltd., (Shanghai, China)

in detail.

3.2. Catalyst Preparation and Regeneration

Ni@C/HZSM-5 was prepared by the dipping-tablet-calcination method; the material ratio of the catalyst was prepared according to the molar ratio of sucrose to nickel of 0.375:1, and the calcination temperature was 550 °C, which was studied in detail in our previous work [2]. Earlier studies in the autoclave showed that higher than 3 wt% nickel loading could achieve better product conversion; so, in this work, the nickel loading was set to 5 wt%.

In this work, two strategies, the one-step method and a two-step method, were used to study the catalyst formation. A typical one-step preparation process is as follows: A total of 5 g of HZSM-5, 1.4865 g of Ni(NO₃)₂·6H₂O, and 0.5 g of the binder were added to a 250 mL flask, and dissolved in 120 mL ethanol to form solution A; then, 0.6561 g of sucrose was dissolved in 5 mL of deionized water to form solution B, (the molar ratio of sucrose to Ni(NO₃)₂·6H₂O was 0.375). Solutions A and B were mixed and then magnetically stirred openly at 60 °C until the ethanol evaporated completely. The obtained solid component was transferred to a tableting die and tableted at 5 MPa for 2 min. After that, the catalyst cake was broken, granulated, and sieved through a 14–40 mesh screen. It was transferred to a tube furnace and calcined for 2 h at a heating rate of 10 °C/min to 550 °C in a nitrogen atmosphere.

The typical two-step process is as follows: First, 5 g of HZSM-5 and 0.5 g of pseudo-boehmite were mixed in ethanol, and evaporated under magnetic stirring at 60 °C, and then transferred to a muffle furnace for calcination at 600 °C for 2 h to obtain HZSM-5+Al₂O₃ powder. The corresponding amounts of nickel nitrate, sucrose, and ethanol solutions were stirred, followed by a one-step method for tablet compression, calcination, and other procedures.

The graphene-encapsulated catalyst with pseudo-boehmite, Al₂O₃, and ZrO₂ as binders were named Ni@C/Z5+PB, Ni@C/Z5+Al, and Ni@C/Z5+Zr, respectively; the catalyst without binder was named Ni@C/Z5+N; the traditional catalyst with ZrO₂ as a binder was named Ni/Z5+Zr.

The regeneration of the spent catalyst: A total of 1 g of spent catalyst was calcined in air at 500 °C for 2 h. After calcination at 500 °C, the carbon content was largely eliminated, and the carbon content was decreased from 12.75 wt% to 0.99 wt% (Table 2). The previous experimental study showed that, after calcination in air at 500 °C, the nickel would agglomerate, thereby reducing the activity of the metal center. Therefore, it is considered that the calcined catalyst is reimpregnated with a corresponding stoichiometric amount of nickel nitrate and sucrose solution. After calcination, the graphene-encapsulated nickel particle structure is reformed. The sample calcined was then transferred to the solution containing the 0.2 g of Ni(NO₃)₂·6H₂O and 0.2537 g of sucrose, and then magnetically stirred openly at 60 °C until the ethanol evaporated completely. Tableting and calcination steps are consistent with the one-step method.

3.3. Characterization

Many techniques were used to characterize the morphology and performance of the prepared catalysts. The specific methods and conditions of these experiments are shown below.

The crystal phases of the samples were characterized by X-ray diffraction (XRD) with Cu Kα radiation at 40 kV and 40 mA. The patterns were in the range of 5–80°.

Transmission electron microscopy images (TEM) were recorded using a JEOL JEM 2100F instrument (JEOL, Ltd., Tokyo, Japan).

Temperature programmed desorption measurements for NH₃ (NH₃-TPD) were performed on an autochem II 2920-ms instrument (Micromeritics instrument corporation, Norcross, GA, USA). The catalyst (0.1 g) was first pretreated at 300 °C for 0.5 h under a flow of He, cooled to 100 °C and blown with ammonia for 1 h to saturation. Finally, the desorption of chemically absorbed NH₃ was conducted by a temperature program raising the temperature at a rate of 10 K/min for 600 °C.

Brønsted and Lewis acids sites were determined using a Nicolet 6700-Q50 (TA Instruments, New Castle, DE, USA), Fourier transform infrared spectroscope of pyridine (Py-FTIR). First, the samples were pretreated under a vacuum (10⁻⁴ Pa) at 350 °C for 2 h. Subsequently, the samples were treated with saturated pyridine vapor for 15 min at room temperature. Finally, the samples were heated to the preset temperature (150 °C/350 °C) and held for 0.5 h, then cooled to room temperature, and the spectra were collected.

Thermogravimetric analysis (TG) was conducted on an SDT650TG instrument (Waters Corporation, Milford, MA, USA) by raising the temperature at a rate of 10 K/min from 50 °C to 900 °C.

Energy dispersive X-ray spectroscopy (EDS) was conducted on an S-4800 instrument (Hitachi Limited, Tokyo, Japan).

3.4. Catalyst Evaluation

The catalyst evaluation test was carried out in a fixed-bed reactor; the catalyst was packed in a reaction tube with an inner diameter of 3 mm, the packing length was about 250 mm, and the bulk volume of the catalyst was approximately 1.69 mL.

The liquid products were collected every 1 h for each catalyst under different conditions. Liquid products were analyzed using an Agilent 7890A-5975C GC-MS instrument (Thermo Fisher Scientific, Waltham, MA, USA) equipped with a flame ionization detector and capillary column (DB-5, 30 m × 0.25 mm × 0.25 m). The conversion, selectivity, and yield of the product were calculated as follows:

Conversion of oleic acid to jet fuel/%:

$$conversion/\%=\frac{mole of fuel components}{moles of all the liquild components}\times 100\%$$

(3)

Jet fuel components include alkanes and arenes of C8–C18.

The selectivity of y/%

$$y/\%=\frac{mole of fuel components}{moles of all the liquild components}\times 100\%$$

(4)

where y = aromatics, alkanes of C8–C16, and alkanes of C17–C18.

The yield of jet fuel/%:

$$yield/\%=\frac{mass of \left(alkanes+aromatics\right)}{mass of oleic acid}\times 100\%$$

(5)

4. Conclusions

It was found that ZrO₂ and Al₂O₃ were suitable binders for forming a complete graphene carbon layer. At the same time, pseudo-boehmite was not conducive to the formation of Ni@C, due to the calcination of pseudo-boehmite affecting the nucleation of Ni@C. The easy leaching and accumulation of the metal center led to shortening the lifespan of the traditional-shaped catalyst Ni/Z5+Zr. The method of graphene-encapsulated nickel particles mitigated this phenomenon effectively. After graphene encapsulation, the particle size of the Ni crystals in the catalyst was (mostly < 11 nm), and no obvious accumulation of nickel particles appeared. The graphene-encapsulated catalysts exhibited a better stability and longer lifespan than the traditional catalysts. When the conversion rate was reduced to 60%, the continuous reaction time of the traditional catalyst, Ni/Z5+Zr, was 8 h, while that of Ni@C/Z5+Zr was prolonged to 12 h. Ni@C/Z5+Zr and Ni@C/Z5+Al offer more stable activity and increased catalytic lifespan than the binder-free Ni@C/Z5+N. After 8 h of continuous reaction time, the aromatic selectivity of Ni@C/Z5+Zr was 93.7%, while a 51.6% aromatic selectivity was observed with the Ni@C/Z5+Al. After 7 h of continuous hydrogenation, regeneration was carried out and the conversion increased by 8.8%. The conversion of the refresh catalyst could still be maintained above ~90% after four hours of continuous reaction, indicating that the catalyst activity could be effectively recovered after regeneration. However, the activity of the refresh catalyst was not fully restored after regeneration. It is reasonable to believe that Ni@C/Z5+Zr has a good application potential in industrial hydrodeoxygenation processes, based on the advantages of Ni@C/Z5+Zr in activity, selectivity, stability, and cost, as well as achieving an efficient regeneration to prolong the catalyst lifespan.