W/Al₂O₃ sample catalysts analyzed in this work were prepared by the a single-pot method of pore volume impregnation in the presence or absence of citric acid (CA). First, the impregnation solutions were prepared with or without citric acid (CA, C₆H₈O₇·H₂O, PROLABO) and ammonium metatungstate ((NH₄)₆H₂W₁₂O₄₀·xH₂O Aldrich). Pretreated commercial γ-Al₂O₃ (Sasol, specific surface area of 252 m²g⁻¹ and pore volume of 0.84 cm³g⁻¹) was selected as catalyst support (sieved between 200 and 400 μm, precalcined in air at 723 K for 2 h). The W (CA/W = x) solutions (the pH was 0.01 for CA/W = 2) were added to the Al₂O₃, followed by strong stirring during 1 h and later matured for 2 h at room temperature. Finally, the catalysts were introduced in a rotatory furnace and dried at 383 K (3 K min⁻¹) for 16 h. It is emphasized that these catalysts were not calcined prior to the sulfidation in order to keep the species formed during the impregnation step. All catalysts samples were prepared with a constant amount of W (20 wt. %W) content corresponding to 3.3 W atoms nm⁻².

All the thiophene hydrodesulfurization (HDS) tests were performed in a differential glass reactor (diameter of 9 mm). It can be considered that the specific velocity corresponds to the initial velocity since the catalytic bed was very thin (mass of catalyst corresponds approximately to 50 mg without the use of a diluent agent). Catalysts were sulfided before the thiophene HDS test at different temperatures (573–773 K) with a heating ramp of 3 K min⁻¹ at 0.1 MPa for 2 h under a 30 ml min⁻¹ flow of 10% H₂S/H₂. Later, thiophene HDS test was carried through at 623 K and 0.1 MPa with an equivalent mass of 50 mg of sulfided catalyst. Thiophene (Alfa Aesar, 99%, extra pure) was introduced into the reactor by passing 70 ml min⁻¹ of H₂ flow through a saturator kept at 291 K and mixed with a flow of 20 ml min⁻¹ of 10% H₂S/H₂ in order to maintain the steady state of the catalyst. The partial pressures of thiophene, H₂, and H₂S in the mixture were 8, 90.2, and 1.8 kPa, respectively. The HDS reaction outlet gas was analyzed with a Varian 3,900 chromatograph provided with a Varian Factor 4 (VF-1MS) capillary column (15 m, 0.25 mm, 0.25 µm) and a flame ionization (FID) detector. To maintain a differential reactor regime, the thiophene conversion was controlled to be less than 10%. Using the following formula, the reaction rate (r_s) (mol h⁻¹ kg⁻¹) was calculated: r s = x × F m c a t (1) where x is the thiophene conversion (%), F is the molar flow rate of thiophene (mol h⁻¹), and m_cat is the mass of the catalyst after sulfidation, respectively. After the catalytic test, the spent catalysts were recovered to obtain their precise weight.

All catalysts series were characterized in situ by low-temperature (∼100 K) CO adsorption followed by IR spectroscopy (IR/CO). The IRCO experimental method followed in this work has been the same as that has been described in previous work by our group. CO adsorption was carried out at low temperature to avoid as much as possible any reaction of the CO with the catalyst surface. The oxidic catalyst was pressed into a supported wafer (∼10 mg precisely weighted, for a disc of 2.01 cm²) and inserted into a quartz IR cell with CaF₂ windows. Prior to adsorption experiments, the sample was in situ sulfided in the IR cell at environmental pressure (0.1 MPa). For this purpose, the catalyst was sulfided with a mixture of gases consisting of 10% H₂S/H₂ (30 ml min⁻¹) at 673 K (the H₂S/H₂ flow introduced from room temperature) during 2 h, with a heating rate of 3 K min⁻¹. After the sulfidation step, the cell was flushed with Ar in order to displace the remaining H₂S and then proceed to evacuated down to p < 1 × 10^–4 Pa until the catalyst reached room temperature. IR/CO at low temperature (100 K) was performed introducing calibrated CO pressures. These small amounts of CO were successively introduced into the IR cell and lastly an equilibrium pressure of 133 Pa of CO was maintained. After each CO introduction, FTIR spectra were recorded with 256 scans and a resolution of 4 cm⁻¹ using a Nicolet Nexus FTIR spectrometer equipped with an MCT detector (FTIR = Fourier transform infrared; MCT = mercury−cadmium−tellurium). All spectra were normalized to a disk of 10 mg. The experiments were duplicated and the calculated experimental uncertainty is estimated to 5%. The bands of the adsorbed CO species were obtained by subtracting the reference spectrum (spectrum recorded after sulfidation and before CO introduction). For quantitative analysis, spectra corresponding to catalyst at full coverage in CO were decomposed using OMNIC program and pseudo-Voigt function. The center and full width at half height (FWHH) of each peak were allowed to vary in a small range (±3 cm⁻¹). The molar attenuation coefficient of CO adsorbed on M-edge (ε_M-edge) and S-edge (ε_S-edge) were determined using the same IR cell by introducing small doses of CO onto the sulfided W/Al₂O₃ catalyst at liquid nitrogen temperature (100 K). These catalysts were prepared with different metal loadings and the full details can be found in previous work of this group. (Zavala-Sanchez et al., 2020a). The reported values for (ε_M-edge) and (ε_S-edge) are 11.9 ± 2.0 μmol⁻¹ cm and 23.1 ± 6.0 μmol⁻¹ cm, respectively.

For the microscopy study, the catalysts in oxidic state were first sulfided at 673 K (heating rate of 3 K min⁻¹) at a pressure value of 0.1 MPa for 2 h under a 30 ml min⁻¹ flow of 10% H₂S/H₂. later, to limit the detrimental exposure to air, the sulfided catalysts were unloaded in a glovebox under argon flow from the sulfidation reactor and poured into absolute ethanol. Few drops of the catalyst suspension were deposited on a copper grid (300-mesh) with holey carbon film. The images of sulfided catalysts shown in this work have been taken using a double corrected JEOL ARM 200F cold FEG microscope operated at 200 kV. As a first step, parameters as length and stacking degree of the sulfide slabs were determined by measuring around 265 slabs per sample from the HRTEM images. The image treatment was carried out with help of the commercial software GMS3 from GATAN (DIGITALMICROGRAPH). The average length (L̅) and the average stacking degree (N̅) were calculated according to the references (Zavala-Sanchez et al., 2019b; Zavala-Sanchez et al., 2020b). The nanostructure and morphology of the WS₂ slabs were explored by high resolution scanning transmission electron microscopy (HR STEM) using a high angle annular dark field (HAADF) detector. All of the STEM images presented in this manuscript were acquired in HAADF mode. The images acquisition time was approximately 30 s with 1,024 pixels × 1,024 pixels resolution. During this image recording the stability of the particles was not affected, which means that the particles are sufficiently stable to the electron beam.

The values of the catalytic activity in thiophene HDS reaction of sulfided W (CA/W = x)/Al₂O₃ catalysts are presented in Figure 1. The HDS rate is the lowest on sulfided W (CA/W = 0)/Al₂O₃ catalysts and increases with the addition of citric acid was observed. The highest HDS rate was obtained when the (CA/W = 2) ratio was used. This result is comparable to what was earlier reported for Mo system. (Chen et al., 2014c).

When observing in detail the catalytic evolution over time (Figure 2), it can be appreciated that the sample prepared without CA showed a loss of catalytic activity over time. Later, when the ratio of citric acid (CA/W = 0.5) was increased, it seems to affect positively the catalytic stability and activity, but still appears to be insufficient to preserve a constant thiophene HDS rate. Nevertheless, when the CA ratio was increased to the values of (CA/W = 1) and (CA/W = 2), the catalysts exhibit a better stability in terms of catalytic activity.

The W (CA/W = x)/Al₂O₃ catalysts, sulfided at different temperatures were tested in thiophene HDS reaction for (Figure 3). Overall, the lowest catalytic performance was observed after sulfidation at 573 K while the highest after sulfidation at 673 K. It is noticeable that an increase in the CA/W ratio raises the rate of reaction whatever the sulfidation temperatures. Once again, it is observed that there was some improvement when adding the (CA/W = 0.5), however, it is observed that the effect of increasing the catalytic activity is more noticeable in those catalysts prepared with citric acid in higher concentrations (CA/W = 1.0 and 2.0). Beyond increasing the sulfidation temperature, the effect of improved catalytic activity is triggered by the action of the CA as chelating agent. A low concentration of citric acid has an improving effect in the thiophene HDS rate, but since it does not provide adequate catalytic stability, this amount is insufficient to provide desired performance.

Based on the evidence of the significant change in catalytic activity due to the addition of citric acid, the morphology of WS₂ slabs was investigated in detail. For this, the series of catalysts W(CA/W = x)/Al₂O₃ were sulfurized in-situ at 673 K and analyzed by IR/CO. The IR spectra of CO adsorption (100 K, 133 Pa CO at equilibrium) on these catalysts are compared in Figure 4. Regardless of the amount of citric acid used in the impregnation stage, four υ(CO) bands located at 2,183, 2,160, 2,141, 2,121 and 2065 cm⁻¹ were detected on these catalysts. These bands are attributed to CO adsorption on Lewis acid sites, OH groups of Al₂O₃ support, physisorbed CO, and M-edge and S-edge of WS₂ phase, respectively. (Zavala-Sanchez et al., 2020a). Comparing the spectra of the catalysts prepared without CA and W(CA/W = 0.5)/Al₂O₃ catalyst, a shift to lower wavenumber of ∼3 cm⁻¹ of the band attributed to CO adsorption on Lewis acid sites was recorded, meanwhile the rest of the bands remained in the same position with negligible variation. Conversely, some clear variations in the intensity of υ(CO) bands on WS₂ phase were detected. In the W(CA/W = 0.5)/Al₂O₃ catalyst the 2,121 cm⁻¹ band suffered a slight decrease in intensity meanwhile the one at 2065 cm⁻¹ increased.

With the progressive addition of CA, the υ(CO) band intensity of S-edge (band ∼2065 cm⁻¹) stepwise increased while that of M-edge (band ∼2,110 cm⁻¹) is maintained practically the same with minimal variation. When using a CA ratio (CA/W = 2), the υ(CO) band intensity of the S-edge reached a maximum, with a comparable value to the one of the M-edge. These results are evidence that the addition of citric acid favors the growth of the S-edge on WS₂ slabs in accordance with what was reported for MoS₂. However, contrary to what was detected for MoS₂, addition of citric acid does not decrease the M-edge contribution. The concentrations of M-edge and S-edge on this series of sulfide W(CA/W = x)/Al₂O₃ catalysts were calculated via spectral decomposition and are grouped in Figure 5. On sulfided W(CA/M = 0)/Al₂O₃ catalyst, about 75% of the total probed edges corresponded to M-edge. Similar behavior has been observed in MoS₂ under typical HDS conditions, where predominantly M-edge (∼80%) formation is observed. (Schweiger et al., 2002; Chen et al., 2014c).

Upon the addition of citric acid, the S-edge concentration stepwise increased reaching almost double value when the CA/W ratio is equal to 2. Meanwhile, the M-edge concentration remained in a steady value with fluctuations of less than 6%. With the increase of S-edge, the total edge site (M-edge + S-edge) concentration raised, reaching a maximum with the W(CA/A = 2)/Al₂O₃ catalyst. The fact that the concentration of M-edge remains with little change and that the one the S-edge increases with addition of citric acid, suggests an improved formation of active sites and a better dispersion degree of the WS₂ slabs, as previously reported on Mo catalysts.

Considering the global M-edge/S-edge ratio and detected by IR/CO, the schematic representation of the mean WS₂ morphology in this series of catalyst was depicted in the lower part of Figure 5. The degree of truncation of WS₂ slabs is greater with the addition of citric acid and therefore, the morphology of the WS₂ slabs is progressively changed from a slightly truncated triangle with mostly M-edge to an almost perfect hexagon with both M-edge and S-edge. Both the change in dispersion degree and in morphology can be explained by a decreased interaction between the support and the active phase. First, considering what is known for MoS_2, the interaction can be modeled by the linkages between M atoms and surface OH groups (M–O–Al). The amount of surface OH groups can be reduced through reaction with CA limiting the MS₂–Al₂O₃ interaction during the MS₂ phase formation. Moreover, formation of citrate complex with W could also modify the W-support interaction in similar way as was reported for Mo catalysts. (Chen et al., 2014a). The reduction of WS₂–Al₂O₃ interaction leads to a modification of the relative free energy of M-edge and S-edge and as a consequence, modifies the WS₂ morphology. In the specific case of W catalyst, the formation of W-CA complex could also have limited the formation of WS_xO_y oxides and thus increase the sulfidation degree of W.

The W(CA/W = 0)/Al₂O₃ and W(CA/W = 2)/Al₂O₃ catalysts were characterized by electron microscopy in HRTEM and in HR-STEM HAADF modes. The study aims to characterize the modification upon CA addition of sulfide phase morphology in terms of size, stacking and shape of the sulfide slabs. Figures 6A,B, shows HRTEM images of W(CA/W = 0)/Al₂O₃ and W(CA/W = 2)/Al₂O₃ catalysts. Analysis of ∼265 slabs was performed in order to get the distributions in length and stacking of the sulfide slabs. It should be mentioned that the local contrast of each zone of the HRTEM images was adjusted in order to observe slabs of about 1 nm. It can be concluded that the addition of citric acid leads to an appreciable decrease of the average slab length (2.7 ± 0.1 nm to 1.9 ± 0.1 nm). On the other hand, it was also detected a relevant variation in the average stacking number, passing from 2.60 ± 0.04 for W(CA/W = 0)/Al₂O₃ to 2.10 ± 0.05 for W(CA/W = 2)/Al₂O₃. This modification in stacking with CA addition was not observed in the case of Mo catalysts.

To analyze the morphology of the sulfided particles, STEM-HAADF analysis was carried out on the W catalysts prepared in absence (Figures 6C,E) or in presence of CA (Figures 6D,F)., The W(CA/W = 0)/Al₂O₃ sample consists of WS₂ plates distributed all over the alumina support. In some regions of the images, higher contrast is observed, attributed to a metal density increase due to agglomerated or stacked slabs. Conversely, the particles observed in the W(CA/W = 2)/Al₂O₃ sample are characterized by their considerable small size in comparison with the case of W(CA/M = 0)/Al₂O₃. Furthermore, clusters of a few atoms and even single atoms are observed in all the studied domain. We have detected a similar phenomenon in MoS₂/Al₂O₃ and CoMoS/Al₂O₃ prepared with CA as chelating agent. (Zavala-Sanchez et al., 2019b; Zavala-Sanchez et al., 2021). Such clusters and single atoms were also observed in the images of sulfided W(CA/W = 0)/Al₂O₃ sample but in much lesser extent.

Thus, both the HRTEM and HR STEM-HAADF modes demonstrated that addition of CA in the impregnation step significantly decreases the WS₂ slab size. This is in accordance with the increase in edge site concentration upon CA addition determined by IR/CO.

The W(CA/W = 0)/Al₂O₃ and W(CA/W = 1)/Al₂O₃ catalysts were in situ sulfided with 10% H₂S/H₂ at different temperatures (573, 623, and 673 K) and then characterized by IR/CO. The obtained IR spectra of CO adsorption (133 Pa at equilibrium, 100 K) are shown in Figure 7, for a catalyst prepared without CA and, in Figure 8 for a sample prepared with CA. When sulfiding the W(CA/W = 0)/Al₂O₃ and W(CA/W = 1)/Al₂O₃ catalysts at different temperatures, the υ(CO) bands described in the previous section are also observed with variations in intensity and some variations in position. Once again in both catalysts, a fairly strong υ(CO) band around 2,110 cm⁻¹ (ascribed to CO at the M edge of the WS₂ phase), an emergent shoulder located at ∼2066 cm⁻¹ (ascribed to CO at the S edge of the WS₂ phase), and two other bands located at ∼2,185 and ∼2,160 cm ⁻¹ (ascribed to CO on Al₂O₃ support). (Mueller et al., 1993; Travert et al., 2001; Travert et al., 2006; Chen et al., 2015a; Zavala-Sanchez et al., 2020a).

In both analyzed catalysts, two important aspects were detected with the increase in sulfidation temperature: 1) the υ(CO) bands corresponding to the OH and to M-edge suffered a progressive shift to higher wavenumbers, being this more noticeable for the υ(CO) of the M-edge (∼7 cm⁻¹). The υ(CO) band position results from the effect of sigma donation from CO to TMS and Pi back-donation from TMS to CO. The sigma donation shifts the vibration frequency to higher wavenumbers, while the Pi back-donation shifts it to lower wavenumbers. (Hadjiivanov and Vayssilov, 2002). The shift to higher wavenumber of the υ(CO/M-edge) is thus indicative of a decrease of electron density on M-edge sites. 2) with increasing sulfidation temperature, the υ(CO/M edge) band stepwise decreases, while the υ(CO/S edge) band gains intensity.

This result is a strong indication of WS₂ morphology change with sulfidation temperature on these two catalysts. Reviewing in more detail the spectra of W(CA = 1)/Al₂O₃ catalyst at different sulfidation T (Figure 8), first, it can be seen that the υ(CO) band on M-edge (∼2,110 cm⁻¹) was once again detected with high intensity and similar width than the υ(CO) band observed for W(CA = 0)/Al₂O₃ catalyst. Meanwhile, the υ(CO) band on the S-edge (band at 2066 cm⁻¹) appeared slightly more defined than the ill-defined shoulder of the W(CA = 0)/Al₂O₃ catalyst.

The obtained IR spectra at different sulfidation temperatures were further decomposed, and the calculated concentration of M-edge and S-edge sites as well as the M-edge/S-edge ratio are plotted versus the sulfidation temperature in Figure 9 for W(CA = 0)/Al₂O₃ and W(CA = 1)/Al₂O₃ catalysts. On both catalysts, 1) the concentration of M-edge detected by IR/CO catalyst decreases with increasing sulfidation temperature 2) the concentration of the M-edge is greater than that of the S-edge. 3) the concentration ratio of S-edge/M-edge is steadily increased with sulfidation temperature. Such effects are more pronounced for the W(CA = 1)/Al₂O₃ catalyst than on W(CA = 0)/Al₂O₃ catalyst. Therefore, the S-edge/M-edge ratio on the W(CA = 1)/Al₂O₃ is about twice than that on the W(CA = 0)/Al₂O₃ catalyst.

Knowing that increased sulfidation temperature leads to the increase of slab length and stacking, (Chen et al., 2014b), the decrease in M-edge site concentration with increasing sulfidation temperature could be linked to the decrease in WS₂ dispersion.

For the catalyst series prepared with increasing amount of CA, the increase in total edge site concentration (due to the increase in S-edge sites) corresponds well with the increase in thiophene HDS activity. However, for the series prepared varying the sulfidation temperature, the increase in HDS activity does not correspond to the decrease of the total edge site concentration. Thus, the global edge site concentration does not allow to explain the catalytic activity. One can note that with addition of CA or increase in sulfidation temperature, the S-edge site concentration is increased as well as the thiophen HDS activity. This would mean that the S-edge sites are more active than the M-edge sites and that they contribute more significantly to the HDS activity. Accordingly, for the series prepared with CA, intrinsic activity expressed as Turn Over Frequency (TOF) was calculated for each edge site considering that the global activity could be mathematically expressed by a simple relation such as: r HDS = n M − edge × TOF M − edge + n S − edge × TOF S − edge Where r HDS   is the overall thiophene HDS rate; n M − edge  and n S − edge are the concentration of M-edge and S-edge, respectively; TOF_M-edge and TOF_s-edge are the intrinsic activity of M-edge and S-edge, respectively (Chen et al., 2014a). The solution of the set of equations was obtained using the multifactor line regression tool of Microsoft Excel 2019. It was found that the TOF_M-edge is 19 ± 4 h⁻¹ and TOF_S-edge is 54 ± 8 h⁻¹, with R ² = 0.998. This result shows that the S-edge in WS₂ is ∼70% more active than the M-edge in thiophene HDS reaction. These results are in agreement with what was obtained on Mo catalysts, (Chen et al., 2014a) although the difference in TOF values for the two edges is greater in the case of W catalysts.