One pot synthesis of NiMo – Al 2 O 3 catalysts by solvent-free solid-state method for hydrodesulfurization †

A simple and solvent-free solid-state method was used to prepare NiMo – Al 2 O 3 hydrodesulfurization (HDS) catalysts using Ni(NO 3 ) 2 $ 6H 2 O, (NH 4 ) 6 Mo 7 O 24 $ 4H 2 O, and AlCl 3 $ 6H 2 O as the solid raw materials and polyethylene glycol (PEG) as an additive. The e ﬀ ects of PEG addition on the precursor thermal decomposition, catalyst properties and dibenzothiophene (DBT) HDS activity were investigated. The as-prepared catalysts were characterized by nitrogen adsorption – desorption measurements, powder X-ray di ﬀ raction (XRD), thermogravimetric analysis/di ﬀ erential scanning calorimetry (TGA/DSC), H 2 temperature-programmed reduction (H 2 -TPR), X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM) and high resolution transmission electron microscopy (HRTEM). The results showed that an increase in PEG addition dramatically increases speci ﬁ c surface area and pore volume of the catalyst, and improves Mo sul ﬁ dability and active MoS 2 dispersion by blocking the aggregation of metals, and consequently increases the number of HDS active sites. However, excess PEG leads to the decrease in speci ﬁ c surface area and pore volume attributed to the metal sintering caused by the strong heat release during thermal decomposition. As a result, dibenzothiophene HDS activity enhanced with increasing PEG addition and peaked at NiMoAl-15 (15% weight ratio of PEG to alumina), which exhibited a signi ﬁ cantly higher activity as compared to the NiMo/Al 2 O 3 catalyst prepared by wetness co-impregnation.


Introduction
Due to increasingly stringent environmental regulations for the sulfur content in fuel oils, 1 catalytic hydrodesulfurization (HDS) is widely used to remove suldes in petroleum rening. In typical commercial HDS catalysts, active components MoS 2 (WS 2 ) and Co(Ni) promoters are supported on g-alumina.
The HDS activity of the multicomponent catalysts depends on the synthetic procedure and their compositions. In recent years, different supporters such as SiO 2 , ZrO 2 , TiO 2 and mixed oxides have been widely studied for improving HDS performance. [2][3][4][5] Another source of activity enhancement can be an alternative active phase based on transition metals (e.g., Ru, Fe, Rh, Pt) or carbides, phosphides. [6][7][8] In addition, the addition of different organic compounds such as ethylene diamine tetraacetic acid (EDTA), citric acid, thioglycolic acid or polyethylene glycol also has crucial inuence. [9][10][11][12] Many approaches have been employed to improve metal dispersion of the catalysts for high HDS activity. The commonly used HDS catalyst preparation methods include simultaneous or sequential impregnation, hydrothermal deposition and solgel synthesis. [13][14][15] For these methods, a complicated and timeconsuming process is always required and active metal components on the support surface tend to agglomerate into large crystallites. Therefore, a novel and efficient solid-state reaction for preparing metal complexes, metal oxides or suldes with uniform sizes and shapes has been developed. Zhang et al. reported a low-temperature solid-state reaction for the synthesis of ammonium aluminum carbonate hydroxide (AACH), which used poly-glycol (PEG), NH 4 HCO 3 and Al(NO 3 ) 3 $9H 2 O as precursors. 16 The alumina obtained by the calcination of AACH showed unique bimodal porous structure and high content of surface acid. Without any solvents, Ren et al. synthesized zeolites with MFI structure by mechanically mixing of solid raw materials. 17 Because of the simple, rapid, solvent-free and energy-saving process, this method has attracted extensive research interest.
Recently, solid-state reaction has been applied to the synthesis of supported catalysts with a high dispersion of active metal, in which solid state reactions occur between hydrated transition metal salts and organics. With this method, Cu/ZnO/ ZrO 2 catalysts were prepared for CO 2 hydrogenation; the copper dispersion and catalytic activity of the catalysts are strongly inuenced by calcination temperature. 18 Our group systematically studied the properties of NiMo-Al 2 O 3 catalysts prepared by a similar process using starch and urea as additive. 19,20 The results showed that an appropriate concentration of starch improves the HDS activity, but an excess of starch addition decreases the activity because of agglomeration of Mo particles. Compared with conventional impregnation method, introducing active metal during the synthesis process of mesoporous alumina possesses especial advantages, including high dispersion of metallic composition and preferable anti-sintering.
As mentioned, such solid-state method has seldom been used to load active metal components and synthesis carrier in a one-step process, particularly in using this method to prepared supported HDS catalyst. Herein, we report the solventfree solid-state synthesis of NiMo-Al 2 O 3 catalysts using PEG as additive. The solid-state reaction is not only used for producing alumina but also used for dispersing Ni-Mo active species onto the support simultaneously. A detailed study on the inuence of PEG addition on the catalytic behavior of NiMo-Al 2 O 3 catalyst for dibenzothiophene (DBT) hydrodesulfurization has been performed.

Synthesis samples
and alkapolpeg-6000 (PEG-6000, molecular weight: 6000) were purchased from Sinopharm Chemical Reagent (China). Commercial alumina (g-Al 2 O 3 , SSA 260 m 2 g À1 ) was obtained from Fushun Research Institute of Petroleum and Petrochemical, SINOPEC, Fushun, China. All of the above chemicals were used as purchased without any further purication.
A series of NiMo-Al 2 O 3 catalysts with different PEG addition were prepared by solid-state synthesis. In a typical preparation, the solid raw materials Ni(NO 3 ) 2 $6H 2 O, (NH 4 ) 6 Mo 7 O 24 $4H 2 O, AlCl 3 $6H 2 O and PEG-6000 (molecular weight: 6000) were mixed and grinded at ambient temperature for 10 min, then calcined at 500 C at a heating rate of 10 C min À1 for 4 h in muffle furnace. The samples were denoted as NiMoAl-x (x ¼ 0, 2, 6, 10, 15 and 20), where x represents the weight ratio of PEG to alumina. In the nal catalyst, the content of MoO 3 and NiO are 12 wt% and 4 wt%, respectively.
For comparison, commercial alumina (g-Al 2 O 3 , SSA 260 m 2 g À1 ) supported NiMo catalyst have been prepared by coimpregnation of aqueous solutions of ammonium heptamolybdate and nickel nitrate. The catalyst precursor was dried at 110 C for 12 h, followed by calcining at 500 C for 5 h, and the nal catalysts were denoted as NiMo/Al 2 O 3 . The content of MoO 3 and NiO in the nal catalyst was the same as the catalyst prepared by solid-state synthesis method.

Characterization
Thermogravimetric analysis/differential scanning calorimetry (TGA/DSC) of the precursor before calcination was conducted in owing air atmosphere up to 800 C at a heating rate of 10 C min À1 with TGA/DSC (SDT Q600) thermal analyzer.
Nitrogen adsorption-desorption measurements of the catalysts were performed on a Micromeritics Tristar 3000 adsorption automatic instrument at liquid nitrogen temperature. The samples were outgassed at 300 C for 3 h in vacuum. Surface areas were calculated by the Brunauer-Emmett-Teller (BET) method, and the pore size distribution and total pore volume were determined by the Barrett-Joyner-Halenda (BJH) method.
Powder X-ray diffraction (XRD) characterization was carried out on an X'Pert Pro automatic powder diffractometer operated at 40 kV and 30 mA using CuKa (l ¼ 0.15406 nm) monochromatized radiation in all cases.
Temperature-programmed reduction (TPR) experiments were performed with a GC-TPR apparatus. The samples of asmade NiMoAl-x (50 mg) were pretreated in an Ar stream at 300 C for 1 h and then cooled to room temperature. Next, the catalyst was heated to 900 C at a rate of 10 C min À1 under a 30 ml min À1 ow of an 5% H 2 /Ar mixture. H 2 consumption was determined by a thermal conductivity detector (TCD).
XPS spectra were obtained in MultiLab 2000 spectrometer using Mg Ka radiation under vacuum (<10 À9 mbar). The oxidic catalysts were pressed into thin sheets, sulfurized in a 15% H 2 S/ H 2 stream at 400 C for 2 h, and then cooled to room temperature. Before measurement, the sulded catalysts were kept under inert atmosphere without contact air.
Scanning electron microscopy (SEM) images were obtained on a eld emission scanning electron microscopy LEO-1530 device, which was manufactured by Leo Company in Germany. Emission voltage was 15 eV, and accelerating voltage was 20 kV.
High-resolution transmission electron microscopy (HRTEM) images of the samples were obtained using a Philips Analytical FEI Tecnai 30 electron microscope operating at 300 kV. Freshly sulded catalyst was ultra-sonically dispersed in ethanol. The suspension was then collected on a car bon-coated copper grid.

Catalytic tests
Dibenzothiophene (DBT) HDS reaction was measured in a highpressure xed-bed continuous-ow stainless steel reactor with internal diameter 6.0 mm. Before the reaction, the catalyst was sulfurized in situ at 400 C for 2 h in a stream of 15 vol% H 2 S/H 2 under atmospheric pressure. The reaction protocol is as follows: 300 C with a total pressure of 2.0 MPa, a liquid hourly space velocity (LHSV) of 12 h À1 , and a H 2 /hydrocarbon volumetric ratio of 600, using 2.0 wt% DBT in decalin as a model compound. The liquid product was collected every hour aer a steady state had been reached and analyzed by a gas chromatograph equipped with a FID detector and a HP-5 capillary column. The GC spectra and product distribution of DBT HDS were shown in S- Fig. 1. †

Thermal analysis
Thermal oxidative decomposition of the various precursors, which determines the structure and property of the resultant catalyst, is signicantly affected by the PEG addition. The thermal analysis results for the NiMoAl-x precursors with different PEG contents are reported in Fig. 1. As seen from TG curves in Fig. 1A, the weight loss for NiMoAl-x at about 100 C corresponds to the removal of moisture. The major weight loss for NiMoAl-x between 150 C and 350 C is attributed to a series of overlapping oxidative decomposition reactions, including the process of dehydrationdecomposition and dehydroxylation of PEG and aluminum chloride and the subsequent total transformation of nickel nitrate and ammonium heptamolybdate. 21,22 The presence of PEG exhibits a new weight loss at 300 C, while it becomes greater with the increase of PEG content. The total weight losses coincided with the expected values, indicating that both the PEG and the precursors are completely decomposed.
As shown in Fig. 1B, the DSC curves indicate several endothermic processes at temperature lower than 250 C and an exothermic reaction at higher temperature. A weak endothermic band occurs at 60 C for PEG-containing samples corresponds to the melting of the extended chains, the other endothermic peak locates in 100-250 C is caused by the partial thermal decomposition of the precursors. Whereas, a strong exothermic peak at about 300 C is assigned to the combustion decomposition of PEG, in this process, most of aluminium chloride is decomposed into oxide. With the increasing PEG addition, the exothermic peak area and peak height at 300 C increase proportionally. Otherwise, most of the chloride ion is removed as gaseous HCl or NH 4 Cl during the calcining process, but there is still a trace residual on the oxide catalysts. Interestingly, we have not detected the presence of Cl on the sulde catalysts, which might generate HCl during the suldation process (Table S1 †).

Textural properties and crystalline phase
Nitrogen adsorption-desorption isotherms of the NiMoAl-x catalysts are shown in Fig. 2A. All the samples exhibit type IV isotherms with H1 hysteresis loops, which indicate the feature of mesoporous structure. The area of the hysteresis loops and quantity of N 2 adsorbed for the NiMoAl-x samples increase signicantly with increasing PEG addition, which indicates that the porous system develops as PEG content increases. Texture parameters calculated by BET and BJH methods (Table 1) show that the BET surface area and pore volume increase obviously with increasing PEG addition. However, when the PEG addition is sufficiently high, comparing to the NiMoAl-15 sample, the BET surface area and pore volume of NiMoAl-20 decrease. On the one hand, swelling is not completely free, but is constrained by steric effects; on the other hand, the considerable heat released via roasting excessive amount of PEG results to the agglomeration of nickel-molybdenum particles. Pore size distribution in Fig. 2B shows that the most probable pore size improves with increasing PEG content, yielding a wide pore size distribution. Meanwhile, a series of Al 2 O 3 -x (where x represents the weight ratio of PEG to alumina, and x ¼ 0, 10,15,20) were synthesized via solvent-free solid-state route with different PEG addition (Table S2 and S- Fig. 2 †). All these results indicate that PEG plays a key role in directing framework of the as-prepared catalyst.
X-ray diffraction patterns of the oxide (A) and sulde (B) NiMoAl-x catalysts are shown in S- Fig. 3. † For the oxide catalysts, a weak peak corresponding to g-Al 2 O 3 can be observed, which indicates the formation of poorly crystalline. Typical features of molybdenum and nickel oxides are absent, which  can be due to the high dispersion of nickel and molybdenum species. For the sulde catalysts, the weak and broad XRD peaks at 14.4 , 33.5 , and 58.6 (JCPDS card 37-1492) can be identied, which corresponding to (002), (101) and (110) crystal planes of MoS 2 , respectively. The broad diffraction peaks indicate that no aggregated particles were formed aer sulfurization and that there was a high dispersion of active components over the Al 2 O 3 support surface.

Reducibility and surface composition
H 2 -TPR characterization of the as-prepared oxide NiMoAl-x catalysts is shown in Fig. 3. For all the catalysts, three main H 2consumption peaks could be clearly identied. On the basis of the previous reports, the rst peak between 400 C and 500 C, which slightly shied to lower temperature with increasing PEG addition, could be associated to the reduction (Mo 6+ / Mo 4+ ) of highly dispersed octahedral Mo species. 23 The second signal, identied in the temperature range of 500-700 C, could be ascribed to oxidic Ni 2+ species in weak interaction with the alumina. The high-temperature H 2 -consuming peak between 750 C and 900 C is related to the second step of reduction of octahedral Mo species and the rst step reduction of tetrahedral coordinated monomeric Mo species, which strongly bonded to the alumina support or enwrapped by alumina. It can be seen that a fraction of Mo oxide species possibly incorporated into alumina bulk phase, which are difficult to be reduced.
XPS spectra of the sulded NiMoAl-x catalysts with different PEG addition were collected and decomposed. Mo 3d spectra of the catalysts are shown in S- Fig. 4, † and the related parameters are listed in Table 2. For Mo 3d spectra, 3 different contributions corresponding to MoS 2 , MoO x S y , and Mo oxide species were determined according to binding energy values reported in previous works. 24 The three Mo 3d doublets are as follows: (1) the binding energy at approximately 228.7 eV (Mo 4+ 3d 5/2 ) associated with MoS 2 species, (2) the binding energy at 230.6 AE 0.2 eV (Mo 5+ 3d 5/2 ) correlated with a Mo oxy-sulde species, and (3) the binding energy at 232.6 AE 0.2 eV (Mo 6+ 3d 5/2 ) related to Mo oxide species strongly bonded to alumina or enwrapped by alumina that were not completely sulded. Furthermore, a broad peak at approximately 226.0 eV was assigned to the S 2s line and subtracted from the total spectrum of Mo 3d. For the NiMoAl-0 catalyst, a high percentage of MoO x (Mo 6+ ) was measured, and the MoS 2 species represented only about 60.7%. However, the atomic percentage of Mo 4+ in the sulde phase increases signicantly (from 60.7% to 89.5%) with the increase in the PEG content, demonstrating that the addition of PEG considerably improves the suldation and reducibility of Mo species.
Ni 2p XPS spectra of the sulded NiMoAl-x catalysts are shown in S- Fig. 5. † Ni 2p 3/2 envelope was decomposed by considering three major contributions. The binding energy at 853.5-854 eV corresponds to NiMoS phase. The peak at approximately 856.1 eV corresponds to Ni oxide species, including NiAl 2 O 4 and NiO x that   were strongly bonded to alumina support or enwrapped by alumina. Ni may also exist as NiS x (Ni 3 S 2 , Ni 9 S 8 , and NiS), which is located at 853.0 AE 0.1 eV. 25 The relative quantities of Ni species (NiMoS, NiS x , and Ni oxide) are reported in Table 3. The proportion of NiMoS phase, which is assumed to be the active phase for catalytic HDS, increases with PEG addition. A higher proportion of the NiMoS phase is obtained for the NiMoAl-15 sample, in which 76.3% of Ni atoms are engaged as compared to 65.1% in NiMoAl-0. The proportion of Ni oxide phase decreases with the increase of PEG amount, which can be attributed to the reduction of difficult-to-reduce species due to the blocking of PEG. These results suggested the presence of PEG suppresses the incorporation of Mo species into alumina bulk phase. The integrated areas for Al 2p, Mo 3d, Ni 2p and S 2p peaks of the sulded catalysts were used to calculate the atomic concentration ratios of Mo, Ni and S over Al. As shown in Table 1, the surface atomic ratio of Mo/Al increases with increasing PEG addition, which should be attributed to the transfer of Mo atoms from the bulk phase to the surface phase. However, Ni/(Ni + Mo) and Ni/Al ratios both reached a maximum value at a PEG/ alumina weight ratio of about 15, in line with surface atomic ratio of S/Al. The Mo suldation degree (Mo sul. ) of NiMoAl-x catalysts was correlated with the fraction of Mo 4+ determined by the decomposition of Mo 3d XPS spectra. The Mo suldation degree increases signicantly with PEG addition, indicating that the addition of PEG considerably improve the suldability of Mo.

Morphology and dispersion of MoS 2 clusters
The external morphologies of the NiMoAl-x and NiMo/Al 2 O 3 are presented in S- Fig. 6. † NiMoAl-0 with large particle size is mainly formed by the accumulation of akes, and has a smooth surface, whereas NiMoAl-x (x ¼ 6, 10, 15, 20) catalysts seem to be formed by the agglomeration of the uniform spherical particles with the size of 10 nm.
TEM characterization of the sulded NiMoAl-x and NiMo/ Al 2 O 3 catalysts was performed for the determination of the MoS 2 dispersion and crystallites. Representative HRTEM micrographs of the catalysts are shown in S- Fig. 7. † Typical MoS 2 fringes with 6.2Å interplanar distance are observed in all the micrographs of the sulded catalysts. The slab lengths and stacking layer numbers of MoS 2 slabs on the sulded NiMoAl-x catalysts were obtained through statistical analyses based on at least 20 images of 500-600 slabs taken from the different parts of each catalyst. The average layer number (N) and the average slab length (L) have been calculated according to the rst moment of the distribution: 26 where l i is the length of slab i, N i is the layer number of slab i, x i is the number of slabs with length l i , and y i is the number of slabs with N i layers.
The statistical results of the length and stacking layer number distributions of the MoS 2 slabs over the sulded NiMoAl-x catalysts are showed in Fig. 4. Over 45% of the observed MoS 2 crystals are longer than 4 nm on the NiMoAl-0 catalyst, while the percentage decreases with increasing PEG content, as a result, the average slab length of MoS 2 gradually reduces. However, when the amount of PEG increases to 20 (NiMoAl-20), more MoS 2 crystals longer than 4 nm are detected.
The NiMoAl-15 catalyst shows an average MoS 2 slabs length of 2.9 nm, which is much shorter than that of NiMo/Al 2 O 3 catalyst prepared by impregnation method (L, Table 1). Particularly, the NiMoAl-15 catalyst exhibits the most centralized length distribution, more than 70% of the observed MoS 2 crystals length is between 2 nm and 4 nm. The stacking layer number distributions show that the proportion of multilayered structures (>4) increases with increasing PEG addition, which results an increase in the average stacking layer number (N, Table 1).  (3), by assuming that the MoS 2 crystals are perfect hexagons. 29 In eqn (3), the numerator is the number of atoms in the active surface (edges) and the denominator is the total number of Mo atoms in the crystal; t is the total number of slabs shown in the HRTEM images, and n i is the number of Mo atoms in one edge, determined from the slab length L as shown in eqn (4) The f Mo values in Table 1 show a noticeably higher MoS 2 dispersion on the sulded NiMoAl-x catalysts prepared using PEG, as compared to the NiMo/Al 2 O 3 catalyst. The MoS 2 dispersion over the NiMoAl-x catalysts increases from 0.23 to 0.34 with increasing PEG content.

HDS performance of the NiMoAl-x catalysts
The HDS reaction results of dibenzothiophene (DBT) for the NiMoAl-x catalysts are reported in Table 4. The conversion of DBT occurs through two parallel routes: the direct desulfurization pathway (DDS) yields mainly biphenyl (BP), whereas the hydrogenation route (HYD) generates cyclohexylbenzene (CHB) and dicyclohexyl (DCH). The specic rate of DBT HDS increased with increasing PEG addition and peaked at NiMoAl-15, decreasing thereaer. Compared to NiMo/Al 2 O 3 catalyst, the DBT HDS conversion was 10 percent higher for NiMoAl-15 catalyst under the same conditions. It is noted that the CHB selectivity over the NiMoAl-15 catalyst is higher than that over NiMoAl-0 and NiMoAl-20 catalysts. As a result, the NiMoAl-15 catalyst shows a large HYD/DDS ratio, and the total HDS activities of DBT over the NiMoAl-x catalysts have the same changing tendency as HYD/DDS ratio, which can be attributed to more accessible NiMoS edge sites. However, the selectivity of BP produced via the DDS pathway is much higher than that of CHB and DCH formed through the HYD route, which was similar to the ndings of previous studies. 30 This feature should be due to the low hydrogen pressure and high liquid hourly space velocity (LHSV).

Effect of PEG on catalyst structure and HDS activity
As shown in Fig. 5, the solid-state method using PEG as additive permit to synthesize highly active Al 2 O 3 supported NiMo HDS catalysts. The addition of PEG results in large surface area and pore volume, which allow the catalyst to expose more active sites. Indeed, the addition of PEG improves the reducibility and suldability of Mo, as proved by TPR and XPS results, which suggests a weak metal-support interaction. However, a larger Mo dispersion can be observed from the HRTEM images with PEG addition, which can be attributed to suppression of active metal aggregation by PEG blocking. 27,28 As previous studies described, this positive effect may be caused by the fact that there is a coordination/interaction between numerous hydroxyl (OH) in PEG and active metals on the alumina support. During the drying step, higher dispersion of Ni and Mo precursors was maintained on the alumina surface even aer water removal, since PEG has higher boiling point than water. These results are the major factors for the signicant improvement in the HDS activity of dibenzothiophene.

Conclusions
In the present work, we develop a simple and solvent-free solidstate route for synthesizing Al 2 O 3 supported NiMo catalysts with different PEG addition, which also can be used to synthesis Al 2 O 3 supported other metal catalysts. An appropriate amount of addition of polyethylene glycol (PEG) can dramatically increases specic surface area and pore volume of the catalyst, and improves Mo suldability and active MoS 2 dispersion by  blocking the aggregation among metals. As a result, dibenzothiophene HDS activity enhanced with increasing PEG addition and peaked at NiMoAl-15, decreasing thereaer. Compared with the NiMo/Al 2 O 3 catalyst prepared by incipient wetness coimpregnation, NiMoAl-15 exhibited a signicantly higher HDS activity. Concerning the selectivity, the HYD/DDS ratio is parallel to the activities of the NiMoAl-x catalysts for the HDS of DBT, which is in good agreement with the total population of Ni-Mo-S active sites.

Conflicts of interest
There are no conicts to declare.