One‐pot Synthesis of Bulk NiMoS Catalysts: Influence of pH and Addition of Pluronic®P123. Hydrodesulfurization of Model Sulfur Molecules Representative of FCC Gasoline

The worldwide continual reinforcement of regulations for the production of cleaner fuels along with the larger demand for energy urge us to develop hydrotreatment sulfide catalysts exhibiting higher activity in hydrodesulfurization reactions and explore innovative synthesis strategies. In this work, a novel one‐pot synthesis method of unsupported NiMoS catalysts in water at ambient pressure and moderate temperature is described. The reaction conditions including pH variations and addition of a copolymer structuring agent were thoroughly investigated. Catalytic performances of the prepared bulk NiMoS catalysts were measured for the hydrodesulfurization of two model sulfur molecules, representative of FCC gasoline. Catalysts prepared at pH 7 without or with polymer exhibited roughly similar activity, resulting however from different characteristics favorable for catalyst efficiency probably compensating one another, i. e. a very high promotion rate for the first one and smaller particles along with larger specific surface area for the second.


Introduction
The worldwide continual reinforcement of regulations for the production of cleaner fuels combined with the heavier compositions of the new feedstock as well as the larger demand for energy urge us to develop hydrotreatment catalysts exhibiting higher activity in hydrodesulfurization reactions and explore innovative synthesis strategies. Conventional hydrotreating catalysts are based on nickel or cobalt promoted molybdenum or tungsten sulfides (MoS 2 , WS 2 ) supported on oxide carriers such as alumina. [1][2][3][4] These catalysts are usually prepared by incipient wetness impregnation of metal precursor solution on the support followed by an activation sulfidation step. [3,5] However, this method leads to poor dispersion and limited metal loading in contradiction with new activity overcome these limitations by maximizing the catalyst activity per volume of reactor. The efficiency of these catalysts then relies on optimized physicochemical characteristics of the active phase such as its size, shape and the stacking of the MoS 2 slabs.
Studies on the preparation of unsupported Ni-promoted MoS 2 catalysts at ambient pressure are found in the literature. Most of them reported a co-precipitation of mixed oxide phases, precursors of the catalyst and promoter, followed by a sulfidation step. [5][6][7][8] A metathesis-like procedure based upon the precipitation of ammonium thiomolybdate with a nickel salt (e. g. (NH 4 ) 2 MoS 4 + Ni(NO 3 ) 2 · 6H 2 0!NiMoS 4 + 2NH 4 NO 3 + 6H 2 O) was described as a fast and easy way of preparation of Nipromoted MoS 2 catalysts. [9][10][11][12][13] In order to improve the specific surface areas of the compounds and eventually, get a better control of the characteristics of the active phase, different organic compounds such as surfactants, polymers or ionic liquids have been added to the reaction medium during the synthesis of MoS 2 . [14][15][16][17] For instance, addition of polymers such as poly(vinyl pyrrolidone) (PVP) or poly(ethylene glycol) (PEG) led to a decrease of the MoS 2 slab stacking and to an increase of its specific surface area. [18][19][20] The metathesis-like procedure described above has also been carried out in the presence of organic structuring agents. [12,13] Genuit et al. prepared bulk Co(Ni)MoS catalysts in the presence of nonionic surface-active agent (poly(ethylene oxide) oligomers with aryl-alkyl tail), in a mixed water/ethylene glycol solution, leading to higher specific surface areas and to a gain in catalytic activity. [12] We have recently reported a similar synthesis of Ni-promoted MoS 2 bulk catalysts in water using a commercial polymer, i. e. the amphiphilic water-soluble triblock copolymer poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol), Pluronic® P123. [21] Once again, a larger specific surface area and a lower average slab stacking for the compound prepared in the presence of polymer was observed compared to the compound prepared in its absence.
In this work, a novel one-pot synthesis method of unsupported NiMoS catalysts, carried out in water at ambient pressure and moderate temperature, is explored. [22] The reaction conditions including pH variations and addition of Pluronic® P123, as structuring agent, were investigated. Catalytic performances of the prepared bulk NiMoS catalysts were measured for the hydrodesulfurization (HDS) of two model sulfur molecules -3-methylthiophene and benzothiophene -representative of FCC gasoline.

Pre-catalyst synthesis
The main driving idea when proposing a new procedure for the preparation of Ni-promoted MoS 2 bulk catalyst is to provide a one-pot synthesis where the Ni promotion would occur in an aqueous reaction medium at moderate temperature and ambient pressure, though avoiding the metathesis-like preparation, already explored. [9][10][11][12][13] To the best of our knowledge, such a procedure has never been investigated so far. As described in the introduction section, a two-step procedure with the second step of sulfurization being carried out on dried particles is usually carried out while the one-step processes are carried out under hydro or solvo-thermal conditions or at high temperature. [23][24][25] Ammonium thiomolybdate and nickel nitrate were considered as Mo and Ni precursors, respectively. A thiomolybdate reduction previous to the addition of nickel nitrate for promotion was chosen as the preparation procedure. Hydrazine was chosen as the reducing agent since it has already been reported as efficient for thiomolybdate reduction (Eq. 1). [26] The reducing power of hydrazine being stronger at high pH, a synthesis at high pH seemed to be appropriate. However, the addition of a nickel solution in a reaction medium at pH higher than 7.5 would lead to the precipitation of Ni(OH) 2 species. [27][28][29] Two synthesis procedures were then tested. In the first one, the thiomolybdate reduction proceeded at natural pH, i. e. 9.5, with adjustment at pH 7 before adding nickel nitrate; whereas in the second one, adjustment of [the reaction pH at 7 was done at the beginning of the reaction process.
The kinetics of the thiomolybdate reduction, controlled through the decrease of MoS 4 2À species by UV-vis spectroscopy, depended on the pH. Indeed, the reduction (in the absence of polymer) was completed within 15 min at pH 7, while it took 2 h at pH 9.5, which was not expected. However, if the main reaction at pH 9.5 is expected to be the reduction of MoS 4 2À by hydrazine, some other reactions involving the protons brought to the reaction medium by addition of hydrochloric acid, might have been overlooked at pH 7.
According to literature, other reactions can indeed occur in the reaction medium, in presence of protons.
One of these reactions is the fast formation of MoS 3 from MoS 4 2À and H + (Eq. 2). [15,30] Another one, proposed by Hadjikyriacou is the formation of Mo 2 S 7 2À by reaction of two MoS 4 2À : (Eq. 3). [31] 2 with the Mo 2 S 7 2À species eventually reacting further to form different complexes.
Thus, a competition between several reactions, which led to a rapid consumption of MoS 4 2À at pH 7 must have occurred. A point in favor of this assumption is the observation of a darkening of the solution that did not occur at pH 9.5.

Characterization of Ni-promoted MoS 2 catalysts
The two catalysts prepared at pH 9.5 in the absence of Pluronic® P123 or in its presence are hereafter called NiMoS-9.5 and NiMoS-9.5-P123, respectively, while the two catalysts prepared at pH 7 in the absence of Pluronic® P123 or in its presence are hereafter called NiMoS-7 and NiMoS-7-P123, respectively. Results of the chemical analyses of the compounds are reported in Table 1.
The ratios Ni/(Ni + Mo), comprised between 0.27 and 0.36, are slightly lower than the expected value, 0.4, corresponding to the amounts of Mo and Ni introduced in the reaction medium. When Pluronic® P123 was added during the synthesis, carbon was found with amount respectively equal to 6.1 wt.% for NiMoS-9.5-P123 and 12.7 wt.% for NiMoS-7-P123. The remaining of a carbonaceous phase can be explained by an incomplete decomposition of the polymer during the heat treatment in a reductive H 2 (5 %)/Ar atmosphere. The presence of similar or higher amounts of carbon after thermal treatment has already been reported in literature for NiMoS catalysts prepared in presence of organic compounds (solvent, surfactant or polymer). [11,12,21,32] The nitrogen adsorption-desorption curves of the catalysts are shown in Figure 1.
The isotherms of NiMoS-9.5, NiMoS-9.5-P123 and NiMoS-7 are similar and correspond to inter-particle capillary condensation (type II according IUPAC classification). On the other hand, the isotherm of NiMoS-7-P123 corresponds to type IV isotherm characteristic of mesoporous materials. [33] The hysteresis loop can be attributed to pore-blocking/percolation in a narrow range of pore necks. [34] Such porosity could be due to the presence of remaining carbonaceous species (12.7 wt.% in this compound). Considering the t-plot from Harkins and Jura law, it is possible to assert that no microporosity is evidenced in this catalyst. [35] The specific surface areas of the catalysts were determined by the BET method and are reported in Table 1. They vary from 7 to 65 m 2 g À 1 with the catalysts prepared at pH 7 displaying specific surface areas higher than the catalysts prepared at pH 9.5. In both case, the addition of Pluronic®P123 led to an increase of the specific surface area.
These results are consistent with the morphology of the particles observed by SEM and shown in Figure 2.
Indeed, the SEM images of NiMoS-9.5 reveal very large agglomerates (> 500 nm) with low roughness, consistent with the low specific surface area of the catalyst (7 m 2 g À 1 ). The SEM images of NiMoS-7 show dispersed quasi-spherical particles with a size comprised between 35 and 60 nm, consistent with the larger specific surface area of 41 m 2 g À 1 . In both cases, addition of Pluronic® P123 during the synthesis led to a decrease of the size of the previously observed features, agglomerates and particles, consistent with the increase of the specific surface areas (44 m 2 g À 1 and 65 m 2 g À 1 , respectively) as usually reported in the literature. [9] However, addition of Pluronic® P123 also resulted in more heterogeneous compounds, particularly visible for NiMoS-7-P123 where large crystals of several microns in size are observed. TEM experiments coupled to EDS were then carried out to obtain additional information on the different observed structures. Results are given in Figures 3 and 7 and Table 2.
EDS analysis was used to identify the presence and relative amount of the elements, in particular Ni and Mo. The measurement of interplanar distances helped in identifying the observed crystallites (interplanar distances of~6 Å characteristic of hexagonal MoS 2 and interplanar distances of~2 Å and 2.6 Å characteristic of hexagonal NiS).
TEM images of NiMoS-7, shown in Figure 3, exhibit the presence of MoS 2 slabs tangled like balls of 20 to 50 nm in diameter, corresponding to a homogeneous 3D morphology, even though few rare NiO and NiS-rich crystallites were observed, in agreement with SEM observation.
TEM images of NiMoS-7-P123, shown in Figure 4, also confirmed SEM data, i. e. the heterogeneous nature of the compound. As a matter of fact, three different types of region were observed. TEM image, shown in Figure Table 2. The values are low, ranging from 2 to 4 slabs, and from 1.9 nm to 2.5 nm, respectively. The influence of polymer Pluronic® P123 can be i À 3ni þ1 with n i : number of Mo atoms along one side of a MoS 2 slab determined from its length and i being the total number of slabs measured by TEM. [36]     analyzed when comparing data for NiMoS-7 (Figure 6a, Figure 7a) and NiMoS-7-P123 (Figure 6c, Figure 7c).
Indeed, a shift in the distribution of both stacking and length towards smaller values for NiMoS-7-P123 compared to NiMoS-7 shows clearly the exfoliating effect of the polymer with 70 % of slab stacking being 1 or 2 in NiMoS-7-P123 compared to~12 % in NiMoS-7-P123. Considering a hexagonal model for MoS 2 , the dispersion defined as the ratio between the number of Mo atoms in edges and corners and the total number of Mo atoms, was calculated and resulted in 36 % for NiMoS-7, 43 % for NiMoS-7-P123 and 38 % for NiMoS-9.5-P123, thus slightly larger for NiMoS-7-P123 (Table 2).
X-ray diffraction patterns of the catalysts are reported in Figure 8.
The patterns exhibit similar features with broad diffraction peaks, characteristic of poorly crystallized MoS 2 and narrow peaks characteristic of NiS, the latter feature being in agreement with TEM observation and having already been reported several times. [11,21,22,37] A second Ni-phase, Ni 7 S 6 , is observed in the X-ray diffraction patterns of NiMoS-9.5-P123. The absence of any additional peak for NiMoS-7-P123 supports the TEM interpretation of NiS microcrystals surrounded by MoS 2 slabs. The structure of bulk MoS 2 is built up by stacking of hexagonal MoS 2 layers. The peak corresponding to MoS 2 (002) plans, distant of 6.2 Å, is expected at 2θ � 14°. Several information can be drawn from the characteristics of the peak. The size of the crystallites can be estimated from the full width at half maximum using the Scherrer formula. [38] The average diameter of the crystallites, given in Table 2, was estimated for the catalysts prepared without polymer and was found to be 2.3 nm for NiMoS-9.5 and 2.6 nm for NiMoS-7. In the case of catalysts prepared with Pluronic® P123, the peaks were too small to allow a reasonable estimation. Anyway, they indicate the presence of tiny crystallites, especially in the case of NiMoS-7-P123 that shows hardly any peak at all. The height of the peak is linked to the number of stacked slabs. Its decrease is then attributed to a decrease in the stacking, which occurs for the catalysts prepared in the presence of polymer and agrees with literature reports. [39] Finally, a shift in the position of the peak (002), is observed, which indicates an expansion of the layers with an interspace value of 6.5 Å for the two catalysts prepared without polymer and around 7 Å for NiMoS-9.5-P123. No shift is observed for NiMoS-7-P123, which might be correlated with the large exfoliation of the compound. From previous information, the average number of stacked slabs could be estimated to 3.6 for NiMoS-9.5 and 4.0 for NiMoS-7. This is consistent with TEM results that indicate an average number of slab stacking of 4.1 for NiMoS-7 and lower values (~2) for catalysts prepared in the presence of polymer.
X-ray photoelectron spectroscopy (XPS) was used to characterize NiMoS-7, NiMoS-7-P123 and NiMoS-9. The quantitative analysis of XPS data is reported in Table 3  and Table 4. Owing to the large uncertainty of XPS data (around 25-26 %) and to the heterogeneity of the compounds (especially those prepared with Pluronic® P123), the quantity of Mo and Ni can be considered similar for all compounds. The MoS 2 phase (S/Mo ratio between 2.1 and 2.5) was formed in a large amount (80 to 85 relative %), similar for all the catalysts. The Ni/ (Ni + Mo) atomic ratio values can be considered similar to those of elemental analyses and close to those of conventional NiMo/ Al 2 O 3 catalysts used in hydrotreament processes. [40,41] The most striking result is the change in the relative amount of NiS and NiMoS phases depending upon the synthesis conditions (Table 4). A large amount of Ni is involved in NiS phase when the polymer is present during the synthesis, i. e. 63 and 67  (relative %) for NiMoS-9.5-P123 and NiMoS-7-P123, respectively, whereas it drops down to 9 (relative %) for NiMoS-7. In the same time, the relative amount of the mixed NiMoS phase changes from 36 and 27 (relative %) in presence of polymer at pH 9.5 and 7 respectively, to 80 (relative %) in its absence. The promotion rate (%atNi × %NiMoS) and the ratio Ni/Mo slab (% atNi × %NiMoS)/(%atMo × %MoS 2 )) have been calculated and reported in Table 3. The promotion is clearly in favor of NiMoS-7 with values equal to 6.8 and 0.4, respectively, compared to 1.9 and 0.2 for NiMoS-7-P123 and 1.9 and 0.2 for NiMoS-9.5-P123. It can be seen that, whatever the way to appreciate the effectiveness of promotion, it is in favor of a much larger promotion for NiMoS-7. Therefore, the presence of polymer Pluronic® P123 when Ni(NO 3 ) 2 is added to the reaction medium is detrimental to the promotion, probably due to a more difficult access to Mo phases for Nickel.
The results above show clearly the importance of the synthesis parameters on the characteristics of the prepared catalysts. The synthesis at pH 9.5 with the thiomolybdate reduction as the main reaction mechanism, led to big agglomerates with hardly any specific surface area (7 m 2 /g), similar to that obtained for unpromoted MoS 2 compounds prepared in a similar way (10 m 2 /g). [26] The situation was somewhat improved by addition of polymer which led to a decrease of the size of the agglomerates and to reasonable specific surface area.
On the other hand, the preparation at pH 7 where the reduction reaction was in competition with reactions promoted by the presence of protons led to smaller particles, which produces larger specific surface area (41 m 2 /g), somewhat lower to that obtained for unpromoted MoS 2 compounds prepared at pH 7.5 in a similar way (108 m 2 /g). [26] The compounds comprised MoS 2 crystallites with an average slab stacking of~4, rather low for unsupported bulk catalysts. [23,24] Addition of polymer again led to a decrease of the particle size and led to a large exfoliation of MoS 2 slabs with slab stacking ranging between 1 and 2 for most of them (70 %). However, this choice turned out to be detrimental to the promotion step with nickel being involved in a majority NiS phase rather than in a majority NiMoS phase as it was the case for catalysts prepared in absence of Pluronic® P123.
The consequence of these different reaction mechanisms on the characteristics of the Ni-promoted MoS 2 catalysts obtained after thermal treatment of the dried pre-catalysts was then investigated.

Catalytic performances
The transformation of two model sulfur molecules, 3MT and BT, over NiMoS-7, NiMoS-7-P123 and NiMoS-9.5-P123, was investigated. Because of its low specific surface area (< 10 m 2 /g), NiMoS-9.5 was discarded from this evaluation. Results shown in Table 5 indicate a higher reactivity of both BT and 3MT over NiMoS-7 and NiMoS-7-P123 compared to NiMoS-9.5-P123, up to two times, whatever the way to calculate the catalyst activity (either reported to the catalyst mass, or per surface unit, or per Mo atom). It shows that the origin of the difference is not unique but involves multiple factors. For example, synthesis at pH 9.5 led to less performant active sites as shown by the poorer activity per molybdenum atom of NiMoS-9.5-P123 compared to the other catalysts prepared at pH 7. Moreover, NiMoS-9.5-P123 exhibits a lower specific surface area than that of NiMoS-7-P123 detrimental to the accessibility of the sites and a poorer amount of mixed NiMoS phase compared to that of NiMoS-7, detrimental to the quantity of active sites. NiMoS-7 and NiMoS-7-P123 exhibited roughly similar activity per gram for the transformation of 3MT and BT. Yet, these two catalysts possess very different characteristics, a higher promotion rate for NiMoS-7 (Table 3) and smaller particles along with the presence of some porosity as shown by SEM and BET experiments for NiMoS-7-P123. These different advantages for catalyst efficiency probably compensated one another. Indeed, the higher promotion rate led to a larger amount of active sites for NiMoS-7 compared to NiMoS-7-P123 as evidenced by a similar activity per molybdenum atom for both catalysts (~6 × 10 À 21 mmol/at Mo h) and a higher activity per m 2 for NiMoS-7 compared to NiMoS-7-P123 over both 3MT (11.4 × 10 À 2 mmol/ m 2 h versus 7.4 × 10 À 2 mmol/m 2 h) and BT (11.0 × 10 À 2 mmol/m 2 h versus 8.5 × 10 À 2 mmol/m 2 h). On the other hand, the accessibility to the sites for NiMoS-7-P123 is favored by its higher specific surface area.
Another interesting result is the similar reactivity of BT and 3MT over these catalysts, which differs from the usual observation of a higher reactivity of BT compared to 3MT over classical supported catalysts such as Ni(Co)Mo/Al 2 O 3 . [41] It could be due to different adsorption modes and adsorption forces of the molecules over these catalysts compared to classical ones. [22] It could be used advantageously for optimizing gasoline HDS processes and countering the important inhibiting effect of benzothiophenic compounds in the transformation of alkylthiophenic compounds.
The nature of the catalyst had hardly any impact on the selectivity. Indeed, whatever the catalyst, products resulting from 3MT transformation are essentially hydrodesulfurization (HDS) products, i. e. mainly pentanes and pentenes, which agrees with Scheme 1.
In the same way, whatever the catalyst, the transformation of BT led mainly to the HDS product, i. e. ethylbenzene (84 to 92 %), via the major hydrogenation (HYD) route, with some tetrahydrobenzothiophene, corresponding to the partial hydrogenation of BT, being still present, which agrees with Scheme 2.
These reaction schemes are consistent with those already reported for other catalysts. [41] Conclusions A one-pot synthesis method of unsupported NiMoS catalysts, carried out in water at ambient pressure and moderate temper-  ature with the promotion step being carried out in situ in the liquid reaction medium has been explored. Influence of addition of a polymer during the synthesis has also been investigated.
A first procedure involving a thiomolybdate reduction by hydrazine at pH 9.5 followed by a Ni promotion from nickel nitrate at pH 7 led to compounds comprising large particle agglomerates, even in the presence of P123.
In a second procedure, pH was maintained at 7 all through the process, with several chemical reactions in competition owing to the presence of additional protons. Nevertheless, the obtained compounds comprised small particles (few tenths of nanometers), even smaller in presence of polymer and specific surface area of 41 m 2 /g for NiMoS-7 and 65 m 2 /g for NiMoS-7-P123. TEM characterization indicated a low slab stacking for both compounds, with an average value of~4 for NiMoS-7 and an even larger exfoliation in presence of polymer, 70 % of slab stacking being 1 or 2. XPS experiments showed that the in situ Ni-promotion in the liquid reaction medium was very effective with 80 % of the nickel present in NiMoS-7 being involved in the mixed NiMoS phase. On the other hand, it also showed that the presence of polymer was detrimental to Ni promotion with a larger amount of Ni involved in NiS phase rather than in the mixed NiMoS phase.
The catalytic efficiency of the compounds was tested for the transformation of 3MT and BT. NiMoS-7 and NiMoS-7-P123, while having very different characteristics, exhibited similar activity. It can be understood by a compensating effect of the different properties of the catalysts, a larger amount of active sites owing to a higher promotion for NiMoS-7 and larger accessibility to the sites for NiMoS-7-P123. The similar reactivity of 3MT and BT over the prepared catalysts needs to be highlighted. As a matter of fact, it could be an advantage to counter the inhibiting effect of benzothiophenic compounds in the transformation of alkylthiophenic compounds during the treatment of real charges containing both molecules.

Catalyst preparation
NiMoS catalysts were prepared by a reduction of ammonium thiomolybdate at 90°C followed by an addition of nickel nitrate in presence or in absence of a polymer as structuring agent. Ammonium tetrathiomolybdate, (NH 4 ) 2 MoS 4 , was first prepared according to the McDonald procedure. [42] For synthesis in absence of structuring agent, freshly prepared (NH 4 ) 2 MoS 4 (4 mmol) was dissolved in 40 mL of ultrapure water and introduced into a three-necked flask.
For synthesis in presence of structuring agent, two solutions were prepared, the first one by dissolving 4 mmol of (NH 4 ) 2 MoS 4 in 20 mL of ultrapure water and the second one by dissolving 2.4 g of Pluronic® P123 in 20 mL of ultrapure water. The two solutions were introduced into a three-necked flask.
In either case, the solution was then degassed by bubbling argon for 30 min to remove oxygen and hydrazine (40 mmol, 50-60 % in water) as a reducing agent, was added.
At this stage, two procedures with the pH of the solution being either 9.5 (natural) or 7 (adjusted by addition of~3 mL of hydrochloric acid) were investigated. The three-necked flask with H 2 S flowing through it was placed in an oil bath previously heated to 90°C. The reduction was let to proceed until the complete disappearance of MoS 4 2À (controlled by UV-visible spectroscopy) or after 4 h (in presence of Pluronic® P123). At this stage, whatever the previously chosen procedure, the pH was adjusted to 7 by addition of hydrochloric acid. A solution prepared by dissolving 1.2 mmol of Ni(NO 3 ) 2 in 8 mL of ultrapure water was then introduced dropwise in the reaction medium that was further stirred during 2 hours at 90°C to allow promotion. The amount of nickel, equivalent to a molar ratio Ni/Mo of 0.4, corresponded to the theoretical maximum of Ni that can be introduced as a promoter. [41] The products were extracted by centrifugation (20,000 rpm) using a Beckman Coulter Optima XPN ultracentrifuge with a 45 TI rotor and 94 mL PP tubes. They were then washed three times with 80 mL of water and one more time with 80 mL of ethanol. At each step the particles were dispersed in the solvent by 10 minutes of sonication and extracted by centrifugation. They were then dried under primary vacuum for 12 h at T = 25°C, then grinded and sieved below 350 μm. The obtained particles were heated at 350°C during 2 h in an Ar/5 %H 2 atmosphere. The catalysts were stored under vacuum before use.

Characterizations
Nitrogen adsorption-desorption isotherm measurements after degassing overnight at 120°C were performed using a Micromeritics tristar apparatus (Micromeritics, Norcross, GA, USA). BET method was used for calculation of specific surface area. Elemental analysis (sulfur and carbon amounts) was carried out using an Elementar Vario Micro Cube (Elementar, Langenselbold, Germany) while the molybdenum and nickel amounts were measured by inductively coupled plasma optical emission spectrometer (ICP-OES) using a 5110 Agilent VDV analyzer (Agilent, Santa Clara, USA). X-ray powder diffraction (XRD) patterns were recorded with a Malvern Panalytical X'pert PRO X-ray diffractometer using Cu radiation (λ = 1.5406 Å) (Malvern Panalytical, Royston, U.K.). Micrographs of the catalysts were obtained with a Hitachi S-4800 scanning electron microscope (SEM) and a JEOL JEM2100LAB6 (JEOL, Tokyo, Japan) transmission electron microscope (TEM) equipped with an EDX JED JEOL. XPS spectra were recorded using a KRATOS AXIS supra-spectrometer (Kratos Analytical, Manchester, U.K.) equipped with an aluminium monochromatic source (hν = 1486.6 eV). Before analysis, catalysts were treated at 400°C during 10 h by a mixing of H 2 S (10 mol. %)/H 2 at atmospheric pressure and stored in Schlenk under Ar to avoid oxidation resulting in the formation of sulfates at the surface. The recorded spectra were analyzed using a CasaXPS software. The deconvolution of S 2p, Ni 2p and Mo 3d signals was carried out with respect to the appropriate standard samples (supported oxide and sulfated monometallic catalysts). The calibration was made with the peak of contamination carbon at 284.6 eV. For each catalyst, the metal and sulfur peaks were identified according to their binding energies. [43,44] The elemental surface composition of the catalysts, and therefore, the sulfur/metal atomic ratio (S/Me) and the active phase formation were determined from the area of the metal and sulfur peaks (the uncertainty of the values is around 25 %).

Catalytic activity measurements
As previously reported, catalytic activity measurements were carried out in a fixed bed reactor at 250°C under a total pressure of 2 MPa with a ratio H 2 /feed of 360 NL/L. [21,22] The catalyst was first sulfided in situ under H 2 S/H 2 flow (10 mol % H 2 S) for 10 h at 400°C at atmospheric pressure. These operating conditions correspond to those of the industrial process. [22,41] Catalytic performances of the NiMoS bulk catalysts were measured for the transformation of a single component feed, i. e. a sulfur model molecule (corresponding to 1000 ppm S), either 0.3 wt.% of 3-methylthiophene (3MT) or 0.42 wt.% of benzothiophene (BT) in nheptane. The sulfur model feeds were injected in the reactor by a HPLC Gilson pump (307 series, pump's head: 5 cm 3 ).
The different partial pressures of the reactants, H 2 S and H 2 , are reported in Table 6. n-heptane was not converted under these experimental conditions. No catalyst deactivation was observed for all the experiments.
The reaction products were injected on-line by means of an automatic sampling valve into a Varian gas chromatograph equipped with a PONA capillary column and a flame ionization detector as in previous works. [22,41] The catalytic activity (a) was calculated at isoconversion (30 %) in a differential regime according to the following equation: where F is the molar flow of the reactant in mmol/h, m cat is the mass of catalyst in g and X i is the reactant conversion (i = 3MT and BT). The catalysts were rapidly stable after two-three hours of reaction time corresponding to the setting of the system. Conversions of 3MT and BT were collected after the stabilization. The catalytic activity (a) was calculated at isoconversion (30 %) in a differential regime according to the following equation: where F is the molar flow of the reactant in mmol/h, m cat is the mass of catalyst in g and X i is the reactant conversion (i = 3MT and BT).The activity per active molybdenum atom (those located in the edges and corners of the slabs) is defined as followed: with n i : number of Mo atoms along one side of a MoS 2 slab determined from its length and i being the total number of slabs. [22,38] The activity per square meter (a: mmol/m 2 ) corresponds to the activity in mmol per gram divided by the specific surface area.