Acid Modiﬁcation of the Unsupported NiMo Catalysts by Y-Zeolite Nanoclusters

: Unsupported NiMo catalyst has high hydrogenation activity due to its high active site distribution. However, low speciﬁc surface area and pore distribution greatly limit the e ﬃ cient utilization of the active components. The Y-zeolite nanoclusters were hydrothermally synthesized and introduced into the unsupported NiMo catalysts from a layered nickel molybdate complex oxide. The XRD, N 2 adsorption-desorption, FT-IR, Py-IR, SEM, NH 3 -TPD, and TEM were used to characterize all catalysts. The dibenzothiophene (DBT) hydrodesulfurization (HDS) reaction was performed in a continuous high pressure microreactor. The results showed that the speciﬁc surface area, pore volume, and average pore size of the unsupported NiMo catalysts were greatly increased by the Y-zeolite nanoclusters, and a more dispersed structure was produced. Furthermore, the Lewis acid and total acid content of the unsupported NiMo catalysts were greatly improved by the Y-zeolite nanoclusters. The HDS results showed that the unsupported NiMo catalysts modiﬁed by the nanoclusters had the same high desulfurization e ﬃ ciency as the unmodiﬁed catalyst, but had more proportion of direct desulfurization (DDS) products. The results o ﬀ er an alternative to reducing hydrogen consumption and save cost in the production of ultra clean diesel.


Introduction
Unsupported hydrogenation catalysts, such as NiMo, CoMo, NiW, NiMoW, CoMoW, etc., have high activity which is 3-4 times greater in hydrodesulfurization (HDS) activity than commonly used industrial catalysts [1][2][3][4][5]. Compared with the supported hydrogenation catalysts (NiMo/Al 2 O 3 , CoMo/Al 2 O 3 , and NiMoW/Al 2 O 3 , etc.) [6][7][8], the unsupported hydrogenation catalysts are made up of entirely active components and auxiliaries, and not limited by the lack of active components. The unsupported catalysts are usually synthesized by a co-precipitation method [9,10], and the catalyst precursors are mainly ammonium nickel molybdate crystals [11,12], which produce a lower specific surface area and have less pore distribution than the supported catalysts. Additionally, the unsupported catalysts have high HDS activity only after activation. The main activation methods are presulfidation, which mainly includes a gas phase (H 2 S/H 2 ) [12][13][14] and liquid phase (CS 2 /H 2 ) [15,16]. The sulfidation reagent is difficult to enter the inner of the unsupported catalysts due to its lower pore distribution, and only the active components distributed on the outer surface are sulfided, which greatly reduces the utilization rate of the active component and causes great waste. To increase the specific surface area and pore distribution of the unsupported hydrogenation catalysts, a pore-expanding agent, such as methylcellulose, was added before forming by Yin et al. [17], which produces more pore structures for the unsupported hydrogenation catalysts after calcination. In addition, hydrothermal aging was also used to treat the unsupported catalyst precursors to improve the dispersion of the active

Synthesis of Unsupported NiMo Catalysts
The unsupported NiMo catalysts were prepared according to the previously reported method [10,18,30]. The unsupported NiMo catalyst precursor was synthesized by preparing an aqueous solution of ammonium heptamolybdate, ammonia water, and nickel nitrate with a molar ratio of Ni and Mo (1.5:1). Typically, 200 mL deionized water in a flask was heated to 60 • C firstly. Then the proper amount of nickel nitrate hexahydrate (Ni(NO 3 ) 2 ·6H 2 O) and ammonium heptamolybdate solution were slowly added into the flask and stirred. When the temperature was stabilized to 93 • C, the ammonium hydroxide (28.8% NH 3 ) was added under stirring until the mixture changed from a suspension to a clear solution and stopped dropping, finally the pH was about 12 for the clear solution, and the clear aqueous solution was heated at 93 • C for 10 h under stirring to obtain the turbid liquid. The turbid liquid was aged overnight under stirring, and filtered, washed with deionized water, and dried 12 h at 100 • C. Then the unsupported NiMo catalyst precursor was synthesized. The precursor was calcined at 350 • C for 4 h. The unsupported NiMo catalyst (NiMo-1.5) was obtained.

Synthesis of Y-Zeolite Nanoclusters
Y-zeolite nanoclusters were synthesized hydrothermally by preparing a clear aqueous solution with a molar ratio of 1.0SiO 2 :0.15Al 2 O 3 :0.6TMA 2 O:40H 2 O [31]. Typically, a proper amount of tetramethylammonium hydroxide (TMAOH, 25 wt%) and aluminum isopropoxide (Al(OiPr) 3 , 98 wt%) were mixed together and stirred at room temperature for 3 h. Then the mixture of tetraethoxysilane (TEOS, 98 wt%) and deionized water was added under stirring. The final clear solution was hydrothermally crystallized at 120 • C for 10, 24, 36, 60, and 72 h at a Teflon-lined stainless steel autoclave. The liquid Y-zeolite nanoclusters with different crystallization times were obtained.

Synthesis of Unsupported NiMo Catalysts Modified by Y-Zeolite Nanoclusters
Twenty grams of liquid Y-zeolite nanoclusters with different crystallization times were dripped into the unsupported NiMo catalyst precursor turbid liquid after the aqueous solution of nickel nitrate hexahydrate, ammonium heptamolybdate, and ammonia water was reacted for 9 h. After an hour, the heating was stopped, and the mixture was aged overnight under stirring to obtain the complexes of unsupported NiMo catalyst precursors and Y-zeolite nanoclusters. The complexes were filtered, washed with deionized water, and dried 12 h at 100 • C to get the modified unsupported NiMo catalyst precursor. And the precursor was calcined at 350 • C for 4 h to obtain the modified unsupported NiMo catalysts. The catalysts after modification were labeled as Y-NiMo-10h, Y-NiMo-24h, Y-NiMo-36h, Y-NiMo-60h, and Y-NiMo-72h, respectively.

Characterization of Catalysts
X-ray diffraction (XRD) analysis was performed with an X'Pert Pro MPD instrument (Panalytical, Almelo, Netherlands) using a Cu Kα operated at 45 kV and 40 mA. The samples were ground to powder and pressed into a smooth sheet before measurement. The specific surface area and pore structure of the unsupported NiMo catalyst were determined by N 2 adsorption-desorption using a Tristar3000 multifunctional adsorption instrument (Micromeritics, Norcross, GA, USA) with the N 2 adsorption temperature of −196 • C, and the pore structure properties (specific surface area, pore volume, and average pore size) were calculated by BET and BJH methods. The FT-IR analysis was carried out via a Nexus infrared spectrometer (Nicolet, Madison, WI, USA) with the background of KBr. The Py-IR analysis was performed by Bruker Vertex-70. Before measurement, the samples were calcined at 300 • C for 3 h, and pyridine was adsorbed for 24 h at room temperature and vacuum desorbed at 150 • C for 5 h. The NH 3 temperature-programmed desorption (NH 3 -TPD) and H 2 temperature-programmed reduction (H 2 -TPR) of all catalysts were performed on a Micromeritics AutoChem HP 2950 Chemisorption Analyzer (Micromeritics, Norcross, GA, USA) with a thermal conductivity detector (TCD). The morphology of the all samples was investigated by a cold field emission scanning electron microscope (SEM) (S-4800, Hitachi Company, Tokyo, Japan). TEM measurements were performed on a high resolution electron transmission microscope (JEM-2100, Hitachi Company, Tokyo, Japan) operated at 200 KV. The slab length and stacking layers of MoS 2 were statistically analyzed by about 15 micrographs, including 200 particles for each catalyst. The average slab length and stacking layers of MoS 2 were calculated according to Equations (1) and (2) [32,33], and The dispersion of Mo atoms on the edge surface of the MoS 2 crystals (f Mo ) was calculated by Equation (3) [34,35]: where L i is the slab length, and n i of Equation (1) is the number of MoS 2 slabs whose slab length is L i . For Equation (2), N i represents the stacking layers, n i represents the number of MoS 2 particles whose stacking layer number is N i . For Equation (3), x i is the number of atoms along one edge of a MoS 2 slab determined from L =3.2(2x i −1)A, and n represents the total number of MoS 2 slabs shown in the TEM micrographs.

Catalytic Activity Evaluation
The HDS performance of the unsupported NiMo catalysts before and after modification was evaluated using a 1.0 wt% DBT/n-heptane solution. The HDS reaction for DBT was conducted in a fixed bed high pressure microreactor loaded with 1 mL catalysts (20-40 mesh) diluted with 2 mL quartz (20-40 mesh). The catalysts were presulfided in a mixture of 3.0 wt% CS 2 /n-heptane solution at 330 • C for 8 h (3.0 MPa, H 2 -to-oil ratio of 300, liquid hourly space velocity (LHSV) of 6.0 h −1 ). Reaction tests were performed after sulfidation, and the reaction conditions were 300 • C, 3.0 MPa, H 2 -to-oil ratio of 300, liquid hourly space velocity (LHSV) of 10 h −1 . The liquid samples were collected and analyzed by a Varian 3800 gas chromatography. Additionally, the desulfurization pathways, including direct desulfurization (DDS) and hydrogenation (HYD), were defined by Equations (4) and (5), and the selectivity of the desulfurization pathway can be reflected by DDS/HYD: HYD = (THDBT + HHDBT + CHB + BCH) × 100% THDBT + HHDBT + BP + CHB + BCH (5) Figure 1 shows the XRD spectra of Y-zeolite nanoclusters with different crystallization time. As shown in Figure 1, when the crystallization time is less than 72 h, all samples are in an amorphous state. When the crystallization time is prolonged to 144 h, the characteristic diffraction peaks of the Y-zeolite were detected at 2θ = 6.2 • , 10.3 • , 21.8 • , 26.5 • , and 31.3 • , respectively [31,36], which indicates that the Y-zeolite can be successfully synthesized. Therefore, no Y-zeolite crystals were formed in the samples with crystallization time less than 72 h, and these samples could be considered as the Y-zeolite nanoclusters.

Characterization of Y-Zeolite Nanoclusters
The skeletal structure of Y-zeolites at different crystallization times were detected by FT-IR spectra, and the results were shown in Figure 2.
All samples show the absorption peaks at 440-470 cm −1 , which can be attributed to the characteristic absorption peak of Si-O bond in the tetrahedral TO 4 structure (T = Si, Al) materials [37,38]. Additionally, when the crystallization time is 10 and 24 h, the characteristic absorption band at about 570 cm −1 was detected, which is attributed to the characteristic absorption peak of the Y-zeolite five-membered ring structure [38,39]. However, this peak shifts to the higher wavenumber from 570 to 590 cm −1 with the prolongation of the crystallization time, which is the result of the aggregation of the five-membered ring structure. In addition, when the crystallization time is 0 h, the obvious absorption peak at 614 cm −1 is the characteristic absorption peak of the Si-O-Si structure of silicate [24,26,40], and the peak shifts to the higher wavenumber from 614 to 700 cm −1 with the prolongation of the crystallization time, the reason is that the silicates are converted with the prolongation of the crystallization time. Furthermore, the obvious absorption bands between 1300 and 1700 cm −1 were detected, which could be attributed to the tetramethylammonium hydroxide (TMAOH), indicating that the template was encapsulated in the Y-zeolite nanoclusters.
Crystals 2019, 9, x FOR PEER REVIEW 5 of 17 obvious absorption peak at 614 cm −1 is the characteristic absorption peak of the Si-O-Si structure of silicate [24,26,40], and the peak shifts to the higher wavenumber from 614 to 700 cm −1 with the prolongation of the crystallization time, the reason is that the silicates are converted with the prolongation of the crystallization time. Furthermore, the obvious absorption bands between 1300 and 1700 cm −1 were detected, which could be attributed to the tetramethylammonium hydroxide (TMAOH), indicating that the template was encapsulated in the Y-zeolite nanoclusters.
Transmittance / a.u. Figure 2. FT-IR spectra of Y-zeolite nanoclusters after different crystallization times. obvious absorption peak at 614 cm −1 is the characteristic absorption peak of the Si-O-Si structure of silicate [24,26,40], and the peak shifts to the higher wavenumber from 614 to 700 cm −1 with the prolongation of the crystallization time, the reason is that the silicates are converted with the prolongation of the crystallization time. Furthermore, the obvious absorption bands between 1300 and 1700 cm −1 were detected, which could be attributed to the tetramethylammonium hydroxide (TMAOH), indicating that the template was encapsulated in the Y-zeolite nanoclusters.
Transmittance / a.u. Figure 2. FT-IR spectra of Y-zeolite nanoclusters after different crystallization times. Figure 2. FT-IR spectra of Y-zeolite nanoclusters after different crystallization times.

Characterization of the Modified Unsupported NiMo Catalysts
The pore structure distribution of the unsupported NiMo catalyst modified by the Y-zeolite nanocluster is shown in Table 1. Table 1 shows that the specific surface area and pore volume of the unsupported NiMo catalyst modified by the Y-zeolite nanocluster are greatly improved, the reason is that the TMAOH encapsulated in the Y-zeolite nanocluster structural unit can be thermally decomposed into volatile gases, and the volatilization of these gases can play a role in pore formation. In addition, as shown in Figure 3, the introduction of Y-zeolite nanoclusters improved the pore size, and broadened the pore size range of unsupported Ni-Mo catalysts. However, the average pore size of the Y-NiMo-72h is only 3.65 nm less than that of the NiMo-1.5 (3.99 nm), the main reason is that the Y-zeolite nanocluster can gradually grow into larger particles with the prolongation of the crystallization time. A larger Y-zeolite nanocluster particle could destroy the macropores of the catalyst to produce more relatively small pores, so that the overall pore diameter is reduced. Additionally, the specific surface area of the Y-zeolite nanoclusters is 577 m 2 ·g −1 which is much larger than the modified catalyst. The reason is that the structure of the nanoclusters will be changed greatly than that of the pure nanoclusters after the zeolite nanoclusters are introduced to unsupported NiMo catalysts. On the other hand, when the structure of the Y-zeolite nanoclusters was measured by the N 2 adsorption instrument, high-temperature treatment was required. In this process, the properties of the zeolite nanoclusters could be changed greatly, therefore, there is a great difference in the specific surface area between the Y-zeolite nanoclusters and the modified catalysts. nanocluster is shown in Table 1. Table 1 shows that the specific surface area and pore volume of the unsupported NiMo catalyst modified by the Y-zeolite nanocluster are greatly improved, the reason is that the TMAOH encapsulated in the Y-zeolite nanocluster structural unit can be thermally decomposed into volatile gases, and the volatilization of these gases can play a role in pore formation. In addition, as shown in Figure 3, the introduction of Y-zeolite nanoclusters improved the pore size, and broadened the pore size range of unsupported Ni-Mo catalysts. However, the average pore size of the Y-NiMo-72h is only 3.65 nm less than that of the NiMo-1.5 (3.99 nm), the main reason is that the Y-zeolite nanocluster can gradually grow into larger particles with the prolongation of the crystallization time. A larger Y-zeolite nanocluster particle could destroy the macropores of the catalyst to produce more relatively small pores, so that the overall pore diameter is reduced. Additionally, the specific surface area of the Y-zeolite nanoclusters is 577 m 2 ·g −1 which is much larger than the modified catalyst. The reason is that the structure of the nanoclusters will be changed greatly than that of the pure nanoclusters after the zeolite nanoclusters are introduced to unsupported NiMo catalysts. On the other hand, when the structure of the Y-zeolite nanoclusters was measured by the N2 adsorption instrument, high-temperature treatment was required. In this process, the properties of the zeolite nanoclusters could be changed greatly, therefore, there is a great difference in the specific surface area between the Y-zeolite nanoclusters and the modified catalysts.    The SEM diagrams of the unsupported NiMo catalyst before and after modification were shown in Figure 4. The octahedral structures can be found in catalyst NiMo-1.5, which can be attributed to ammonium nickel molybdate crystal phase [18,41,42]. After modification, the octahedral ammonium nickel molybdate crystal phases disappear, the whole structure is in an amorphous state, indicating that the crystallization of ammonium nickel molybdate can be restricted by the Y-zeolite nanoclusters, and the same results are shown by the XRD (Figure 5). Additionally, the introduction of the Y-zeolite nanoclusters produces a more dispersed structure for the unsupported NiMo catalysts, which can improve the dispersion and utilization of active components. Figure 6A,B show the FT-IR spectra of different unsupported NiMo catalysts before and after calcination. Before calcination, the catalyst NiMo-1.5 shows five obvious absorption bands at about 486, 870, 1282, 1385, and 1629 cm −1 . The absorption peaks at 486 and 1282 cm −1 can be attributed to the Ni-O bond structures. The FT-IR peak at 1000-700 cm −1 is the Mo-O-Mo groups [10,30]. The absorption peaks at 1385 and 1629 cm −1 belong to the amino group of the ammonium-containing substance [10]. For modified unsupported NiMo catalysts, a new absorption peak was detected at about 1051 cm −1 , which is the characteristic absorption peak of the Si-O-Si structure [43][44][45], indicating that the Y-zeolite nanoclusters were successfully introduced into the unsupported NiMo catalysts. After calcination, the catalyst NiMo-1.5 retains the adsorption bands at 1282, 1385, and 1629 cm −1 , but the adsorption band at about 1385 cm −1 almost completely disappears, indicating that the unsupported NiMo catalysts modified by Y-zeolite nanoclusters are easier to be converted into an oxidized structure. In addition, the adsorption band strength at 1629 cm −1 decreases after calcination, which is the result of the thermal decomposition of the templates.
The SEM diagrams of the unsupported NiMo catalyst before and after modification were shown in Figure 4. The octahedral structures can be found in catalyst NiMo-1.5, which can be attributed to ammonium nickel molybdate crystal phase [18,41,42]. After modification, the octahedral ammonium nickel molybdate crystal phases disappear, the whole structure is in an amorphous state, indicating that the crystallization of ammonium nickel molybdate can be restricted by the Y-zeolite nanoclusters, and the same results are shown by the XRD (Figure 5). Additionally, the introduction of the Y-zeolite nanoclusters produces a more dispersed structure for the unsupported NiMo catalysts, which can improve the dispersion and utilization of active components. Figure 6A,B show the FT-IR spectra of different unsupported NiMo catalysts before and after calcination. Before calcination, the catalyst NiMo-1.5 shows five obvious absorption bands at about 486, 870, 1282, 1385, and 1629 cm −1 . The absorption peaks at 486 and 1282 cm −1 can be attributed to the Ni-O bond structures. The FT-IR peak at 1000-700 cm −1 is the Mo-O-Mo groups [10,30]. The absorption peaks at 1385 and 1629 cm −1 belong to the amino group of the ammonium-containing substance [10]. For modified unsupported NiMo catalysts, a new absorption peak was detected at about 1051 cm −1 , which is the characteristic absorption peak of the Si-O-Si structure [43][44][45], indicating that the Y-zeolite nanoclusters were successfully introduced into the unsupported NiMo catalysts. After calcination, the catalyst NiMo-1.5 retains the adsorption bands at 1282, 1385, and 1629 cm −1 , but the adsorption band at about 1385 cm −1 almost completely disappears, indicating that the unsupported NiMo catalysts modified by Y-zeolite nanoclusters are easier to be converted into an oxidized structure. In addition, the adsorption band strength at 1629 cm −1 decreases after calcination, which is the result of the thermal decomposition of the templates.     Table 2 show the Py-IR spectra and acid amount distributions of unsupported NiMo catalysts modified by Y-zeolite nanoclusters. As shown in Figure 7 and Table 2, the Y-zeolite nanoclusters Y-36h and Y-72h have all Lewis and Brønsted acid [46], but the unsupported NiMo catalysts modified by Y-zeolite nanoclusters have only Lewis acid, Brønsted acid was not detected, and the Lewis acid is obviously stronger than the unmodified catalyst. It may be that the Brønsted acid site could not be detected due to the weak Brønsted acid and low addition of the Y-zeolite nanoclusters. On the other hand, the Ni 2+ remaining in the unsupported NiMo catalyst could replace the H + of the Si-OH-Al to produce more Lewis acid sites.  Figure 8 and Table 3 show the NH 3 -TPD diagram and relative acid content of the unsupported NiMo catalysts modified by the Y-zeolite nanoclusters. All catalysts show the stronger peak at 100-300 • C and a weaker peak at 400-550 • C. The peak between 100 and 300 • C is a weak acid peak, and the peak between 400 and 550 • C can be classified as a strong acid peak, indicating that the unsupported NiMo catalysts before and after modification are mainly weak acid. Additionally, the peak area of the unsupported NiMo catalysts modified by Y-zeolite nanoclusters is obviously larger than that of the unmodified catalyst, and Table 3 also show higher relative acid content for modified unsupported NiMo catalysts, which indicates that the total acid content of the unsupported NiMo catalyst is obviously improved after the introduction of the Y-zeolite nanoclusters. Additionally, compared with the unmodified unsupported NiMo catalyst, the modified catalysts also show a weak desorption Crystals 2019, 9, 344 9 of 18 peak at 550-750 • C, which should be the deoxidization peak produced by thermal decomposition of template (TMAOH) in Y-zeolite nanoclusters and unsupported NiMo catalysts.    nanoclusters Y-36h and Y-72h have all Lewis and Brønsted acid [46], but the unsupported NiMo catalysts modified by Y-zeolite nanoclusters have only Lewis acid, Brønsted acid was not detected, and the Lewis acid is obviously stronger than the unmodified catalyst. It may be that the Brønsted acid site could not be detected due to the weak Brønsted acid and low addition of the Y-zeolite nanoclusters. On the other hand, the Ni 2+ remaining in the unsupported NiMo catalyst could replace the H + of the Si-OH-Al to produce more Lewis acid sites.     Table 3 show the NH3-TPD diagram and relative acid content of the unsupported NiMo catalysts modified by the Y-zeolite nanoclusters. All catalysts show the stronger peak at 100-300 °C and a weaker peak at 400-550 °C. The peak between 100 and 300 °C is a weak acid peak, and the peak between 400 and 550 °C can be classified as a strong acid peak, indicating that the unsupported NiMo catalysts before and after modification are mainly weak acid. Additionally, the peak area of the unsupported NiMo catalysts modified by Y-zeolite nanoclusters is obviously larger than that of the unmodified catalyst, and Table 3 also show higher relative acid content for modified unsupported NiMo catalysts, which indicates that the total acid content of the unsupported NiMo catalyst is obviously improved after the introduction of the Y-zeolite nanoclusters. Additionally, compared with the unmodified unsupported NiMo catalyst, the modified catalysts also show a weak desorption peak at 550-750 °C, which should be the Intensity/a.u.

Temperature,℃
NiMo-1.5   Table 4 and Figure 9 show the DBT hydrodesulfurization products, conversion, and DDS/HYD ratios with different unsupported NiMo catalysts at 300 °C. As shown in Figure 9, the catalysts NiMo-1.5, Y-NiMo-10h, and Y-NiMo-24h show high DBT conversion (100%), indicates that the desulfurization activity of the unsupported NiMo catalyst was not affected by the Y-zeolite  Table 4 and Figure 9 show the DBT hydrodesulfurization products, conversion, and DDS/HYD ratios with different unsupported NiMo catalysts at 300 • C. As shown in Figure 9, the catalysts NiMo-1.5, Y-NiMo-10h, and Y-NiMo-24h show high DBT conversion (100%), indicates that the desulfurization activity of the unsupported NiMo catalyst was not affected by the Y-zeolite nanoclusters. In addition, the desulfurization activity of the modified unsupported NiMo catalysts decreases with the prolongation of the crystallization time, indicates that the crystallization time of the Y-zeolite nanoclusters is a key factor affecting the catalytic activity of the unsupported NiMo catalyst, and the desulfurization activity of the unsupported NiMo catalyst could be decreased by the Y-zeolite nanoclusters with too long a crystallization time. Additionally, compared with the catalysts NiMo-1.5, more biphenyls (BP) were produced for the catalysts Y-NiMo-10h and Y-NiMo-24h (Table 4), and the DDS/HYD of the unsupported NiMo catalysts was higher than NiMo-1.5 (Figure 9), indicated that the Y-zeolite nanoclusters modified catalysts were more likely to desulfurize through the DDS pathway. The TEM results of the sulfided unsupported NiMo catalysts were given in Figure 10. As shown in Figure 10, the concentration of the active components decreased, the reason was that some of the active components were replaced by nanoclusters, which would reduce the content of the active sites. In theory, it is not conducive to the improvement of desulfurization activity of the catalyst. However, the Lewis acid strength of the unsupported NiMo catalyst was improved by the Y-zeolite nanoclusters (Figure 7), and appropriate Lewis acid strength was beneficial to the increase of desulfurization activity of the catalyst [47]. It is the reason why the unsupported NiMo catalysts modified by the Y-zeolite nanoclusters still have high desulfurization activity. However, more nanoclusters structural units will be produced with the prolongation of crystallization time, which will cover more active sites, thus reducing the activity of unsupported NiMo catalysts. The Lewis acid could not make up for the activity loss of the unsupported NiMo catalyst. The desulfurization activity of the modified unsupported NiMo catalysts was finally reduced.   The TEM results of the sulfided unsupported NiMo catalysts were given in Figure 10. As shown in Figure 10, the concentration of the active components decreased, the reason was that some of the active components were replaced by nanoclusters, which would reduce the content of the active sites. In theory, it is not conducive to the improvement of desulfurization activity of the catalyst. However, the Lewis acid strength of the unsupported NiMo catalyst was improved by the Y-zeolite nanoclusters (Figure 7), and appropriate Lewis acid strength was beneficial to the increase of desulfurization activity of the catalyst [47]. It is the reason why the unsupported NiMo catalysts modified by the Y-zeolite nanoclusters still have high desulfurization activity. However, more The desulfurization pathways of the unsupported NiMo catalyst are related to the lamellar length and stacking layers of active component (MoS 2 ) [48]. The shorter lamellar length and more stacking layers favor direct desulfurization (DDS) and hydrogenation desulfurization (HYD) activity, respectively [49,50]. The lamellar length and stacking layers distribution of the MoS 2 particles in different unsupported NiMo catalysts was shown in Figures 11 and 12, and the average lamellar length and stacking layers of the MoS 2 particles were also calculated ( Table 5).

Catalytic Performance
NiMo-1.5 Figure 11. Distributions of slab length of MoS2 particles on different unsupported NiMo catalysts. The catalysts Y-NiMo-10h and Y-NiMo-24h had shorter MoS 2 lamellar (9.9 and 9.1) and fewer MoS 2 stacking layers (4.5 and 5.2) than the catalyst NiMo-1.5, so it produced the higher proportion of the DDS desulfurization route than the NiMo-1.5 catalyst, which occurred because shorter lamellar length can produce more sulfur vacancies which favor the DDS desulfurization route, and more MoS 2 stacking layers could result in the aggregation of the active components (MoS 2 ) with higher deep hydrogenation activity. After the introduction of the Y-zeolite nanoclusters, the active components were dispersed and the stacking layers were reduced, which is more favorable to the DDS route. Furthermore, the dispersion frequency of the active phase at edge angle site was also calculated (Table 5), and the catalyst Y-NiMo-24h has higher f Mo (0.13), indicating that the catalyst Y-NiMo-24h has more dispersed edge active sites, which is active for hydrogenolysis (C-S bond breaking). In addition, as the crystallization time of the Y-zeolite nanoclusters was prolonged, the modified catalysts had a longer lamellar length and less MoS 2 stacking layers, which could produce low hydrogenation activity.   The catalysts Y-NiMo-10h and Y-NiMo-24h had shorter MoS2 lamellar (9.9 and 9.1) and fewer MoS2 stacking layers (4.5 and 5.2) than the catalyst NiMo-1.5, so it produced the higher proportion of the DDS desulfurization route than the NiMo-1.5 catalyst, which occurred because shorter lamellar length can produce more sulfur vacancies which favor the DDS desulfurization route, and more MoS2 stacking layers could result in the aggregation of the active components (MoS2) with  On the other hand, the unmodified unsupported NiMo catalyst NiMo-1.5 has higher deep desulfurization activity which can produce more DHP (59.8%), so the higher hydrogen consumption is required, However, high hydrogen consumption is an unfavorable factor for industrial production due to the higher operating costs. The introduction of the Y-zeolite nanoclusters can reduce the deep hydrogenation activity of the unsupported NiMo catalyst which improves the proportion of biphenyls (BP) and decreases the DHP. For the Y-NiMo-10h, the proportion of DHP decreases from 59.8% to 28.9% than NiMo-1.5. It is conducive to reducing hydrogen consumption and saving costs than the unmodified unsupported NiMo catalyst (NiMo-1.5).

The Action Mechanism of Y-Zeolite Nanoclusters
In order to study the action mechanism of Y-zeolite nanoclusters, the particle size distribution was measured by a laser particle size analyzer ( Figure 13). The schematic diagram of Y-zeolite nanoclusters with different crystallization times is shown in Figure 14.

The Action Mechanism of Y-Zeolite Nanoclusters
In order to study the action mechanism of Y-zeolite nanoclusters, the particle size distribution was measured by a laser particle size analyzer ( Figure 13). The schematic diagram of Y-zeolite nanoclusters with different crystallization times is shown in Figure 14. Intensity(%)  As shown in Figures 13 and 14, when the crystallization time is 10 h, most of the samples are in gel state, and the particle size distribution of nanoclusters is uneven. When the crystallization time is 24 h, the DLS shows a narrower range of particle size distribution, indicating that the nanocluster particles become uniform by rearrangement, and then the uniform nanocluster particles gather to

The Action Mechanism of Y-Zeolite Nanoclusters
In order to study the action mechanism of Y-zeolite nanoclusters, the particle size distribution was measured by a laser particle size analyzer ( Figure 13). The schematic diagram of Y-zeolite nanoclusters with different crystallization times is shown in Figure 14. Intensity(%)  As shown in Figures 13 and 14, when the crystallization time is 10 h, most of the samples are in gel state, and the particle size distribution of nanoclusters is uneven. When the crystallization time is 24 h, the DLS shows a narrower range of particle size distribution, indicating that the nanocluster particles become uniform by rearrangement, and then the uniform nanocluster particles gather to form clusters with uneven particle size. Finally, these clusters will form zeolite nanocrystals by As shown in Figures 13 and 14, when the crystallization time is 10 h, most of the samples are in gel state, and the particle size distribution of nanoclusters is uneven. When the crystallization time is 24 h, the DLS shows a narrower range of particle size distribution, indicating that the nanocluster particles become uniform by rearrangement, and then the uniform nanocluster particles gather to form clusters with uneven particle size. Finally, these clusters will form zeolite nanocrystals by rearrangement and crystallization. The Y-24 h has a smaller and more uniform particle size distribution, so it is easier for nanoclusters to enter the interlamellar space of the unsupported NiMo catalysts, and fewer nanoclusters are exposed to the outer surface of unsupported NiMo catalysts which results in less Lewis acid and total acid content than other catalysts (Tables 2 and 3). On the other hand, for Y-NiMo-24h, less active components in the unsupported NiMo catalysts are covered due to less nanocluster distribution on the outer surface, which is beneficial to higher hydrogenation activity, so the desulfurization was more inclined to the HYD pathway ( Figure 9).

Conclusions
The Y-zeolite nanoclusters were synthesized successfully to modify the unsupported NiMo catalysts. The specific surface area, pore volume, and average pore size of the unsupported NiMo catalysts modified by the Y-zeolite nanoclusters were increased obviously, and the dispersion of the unsupported NiMo catalysts was improved. The Lewis acid and total acid of the unsupported NiMo catalysts were improved by the Y-zeolite nanoclusters. The DBT HDS results showed that the crystallization time of the Y-zeolite nanoclusters was an important factor affecting the desulfurization activity of the catalysts, the desulfurization activity of the modified catalysts decreased with the increased crystallization time of the Y-zeolite nanoclusters, and when the crystallization time was 10-24 h, the modified catalysts showed higher DBT desulfurization rate.