Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

# Site-dependent reactivity of MoS2 nanoparticles in hydrodesulfurization of thiophene

## Abstract

The catalytically active site for the removal of S from organosulfur compounds in catalytic hydrodesulfurization has been attributed to a generic site at an S-vacancy on the edge of MoS2 particles. However, steric constraints in adsorption and variations in S-coordination means that not all S-vacancy sites should be considered equally active. Here, we use a combination of atom-resolved scanning probe microscopy and density functional theory to reveal how the generation of S-vacancies within MoS2 nanoparticles and the subsequent adsorption of thiophene (C4H4S) depends strongly on the location on the edge of MoS2. Thiophene adsorbs directly at open corner vacancy sites, however, we find that its adsorption at S-vacancy sites away from the MoS2 particle corners leads to an activated and concerted displacement of neighboring edge S. This mechanism allows the reactant to self-generate a double CUS site that reduces steric effects in more constrained sites along the edge.

## Introduction

Catalytic hydrodesulfurization (HDS) is an industrial process applied to remove sulfur heteroatoms from gas oil feeds to reduce sulfur content in fuels and petrochemicals to a <10 ppm level, compatible with legislative specifications. The primary catalysts used for removal of sulfur from crude oil are based on molybdenum disulfide (MoS2) nanoparticles, often promoted with Co or Ni, and supported on an alumina substrate1,2. Extensive studies involving reaction kinetics, atom-resolved microscopy, X-ray and IR spectroscopy, and density functional theory (DFT) have made it evident that the active sites are located on the edge of single-layer MoS2 catalyst3,4,5,6,7,8,9,10. The nature of these sites on the catalytically active edges was studied by selectively adsorbing probe molecules such as CO or NO11,12,13,14. These experiments allowed the titration of the number of these sites and its correlation with catalytic activity. Thiophene (C4H4S) has furthermore often been used as an S-containing reactant molecule to probe HDS activity of a catalyst15,16,17,18,19,20. This simple five-membered aromatic heterocyclic molecule is relatively inert as its aromatic ring makes its C–S bond quite resistant to rupture. Despite the many fundamental studies devoted to a theory- and experiment-based understanding of the HDS of thiophene a comprehensive mechanistic understanding of this process is still lacking. HDS may proceed by two independent routes: the dominant route for thiophene is the direct desulfurization (DDS) pathway, which involves the direct hydrogenolysis of the C–S bond. The generic type of active site considered for the DDS pathway is a coordinately undersaturated site (CUS) located at a sulfur vacancy (VS) on an edge site formed by reaction with H22,19,20,21,22,23,24,25,26,27. The concentration of the CUS sites has been shown to be dependent on the chemical potential of hydrogen and sulfur in the gas phase and the type of MoS2-edge5,28,29,30,31,32,33,34,35. In parallel to DDS, the hydrogenation pathway (HYD) can proceed, in which thiophene adsorption is followed by hydrogenation before C–S bond scission36,37,38,39,40. The HYD route involves a different type of active site4, proposed to be brim sites located at the top of the MoS2 particle edges6,8. The HYD route becomes more pronounced compared to DDS for large S-containing molecules such as dibenzothiophene and prevailing for alkyl-substituted derivatives, such as 2,5-dimethyldibenzothiophene41,42. However, whether the DDS route can still proceed efficiently for such large molecules, which are sterically hindered in their adsorption on a single VS site, remains an open question and a target of intense industrial HDS catalyst development due to the comparatively lower hydrogen consumption in DDS.

Herein, we use a combination of atom-resolved scanning tunneling microscopy (STM) and DFT on corresponding MoS2 nanoparticle structures to determine the precise adsorption configuration of thiophene on all types of CUS sites present on the MoS2-edges and corner sites present under HDS conditions. We first investigate the formation of VS sites by monitoring the distribution of sulfur vacancies and their associated formation energies. We then expose MoS2 to thiophene, which allows quantification of the accessibility of VS sites grouped by their position on the MoS2-edges and the corresponding adsorption configuration of thiophene. Thiophene adsorption is possible on all observed VS sites. However, whereas thiophene adsorbs directly in open VS sites at the corners between edges, the adsorption in VS sites located at the interior of the edge is observed to occur via a displacement of a neighboring S atom leading to simultaneous formation of an S2 dimer and thiophene adsorbed on a double-VS CUS site. The main finding is therefore that the adsorption of an S-containing molecule may generate its own, more accessible adsorption site, which facilitates the subsequent desulfurization of the molecule. This mechanism contrasts prevalent conclusions on the HDS active site as a static vacancy configuration, and can thus explain the apparent DDS reactivity observed for larger molecules than thiophene, such as dibenzothiophene; it further opens up the discussion for a more accurate description of catalytic inhibition in co-processing of e.g. aromatics and O- and N-bearing reactants in hydrotreating25,43,44,45,46.

## Results

### Site-dependent sulfur vacancy formation

The reactivity of thiophene on MoS2 was first experimentally evaluated by atom-resolved STM imaging of single-layer MoS2 nanoparticles. We have specifically characterized S vacancies on hydrogen-activated MoS2-edge structures and then assessed the adsorption of thiophene on these sites by use of a model system composed of well-defined MoS2 nanoparticles synthesized on an Au(111) single crystal surface (see methods section). Figure 1 illustrates an atom-resolved STM image of a MoS2 nanoparticle (Fig. 1a) together with a top view of a structural model (Fig. 1b) based on previous extensive atom-resolved characterization and theoretical modeling27,47,48,49. The MoS2 particle in Fig. 1a represents the HDS active state (denoted by r-MoS2) induced in the experiment by dosing H2 gas at elevated pressure and temperature as presented in ref. 50 (see “Methods” section). The resulting r-MoS2 particle morphology is represented by a truncated triangular shape, which is bounded by two different types of edges referred to as Mo- and S-edges, respectively. The hydrogen activation leads to a reduced overall 50% S-coverage on the Mo-edge, represented by S monomers located at a bridge position between edge Mo atoms50 (Fig. 1b). The same 50% S-coverage on the Mo-edge structure was concluded to be present in situ at 1 bar H2 pressure in a recent reactor STM study51. Importantly, in Fig. 1a, a number of individual S vacancies can be clearly identified along the Mo-edge imaged as sites with an extinguished intensity in the STM image (see red arrows)50. Atomic defects are sometimes present within the basal plane too (e.g. in Fig. 1a), but these are not induced by the hydrogen treatment and their location suggest an impurity atom on the metal lattice of MoS2 rather than a basal plane VS. Based on a statistical analysis, the total number of vacancies at the Mo-edges is modest overall, corresponding to an average probability of 16% for a vacancy being formed at any edge site. This observation is, however, in full accordance with previous theoretical modeling of the S-vacancy formation, which results in a rather large energy cost required for VS formation by removal of S atoms from the 50% S covered Mo-edge34,52.

Importantly, our analysis shows that the specific VS probability varies significantly among the different sites. To see this, we labeled S atomic positions with a letter in the ball model representations (Fig. 1b, c) to identify S vacancies on the corner (C), adjacent corner (A) and in the middle (M) positions. These sites define the possible edge positions on the Mo-edge lengths most abundantly present in the synthesized MoS2 nanoparticle ensemble (Supplementary Fig. 1), corresponding to 4-6S monomers on the edge as shown in the side-view ball models in Fig. 1c. A breakdown of our statistical materials from the experimental images into each edge length (Supplementary Fig. 2) shows only a very slight variation of the vacancy numbers with the edge length, so for the experimental analysis we consider C, A and M sites on differently sized edges to be similar in terms their vacancy formation probability. The example in the STM image in Fig. 1a already points to the preferential formation of VS mainly at A and M locations (see top-view ball model). Figure 2a illustrates the frequency of observing a VS on these specific sites obtained by a direct counting of missing S atoms in all our STM images. Here, the most probable S vacancies are located at the A positions, with a VS probability at that site being 0.25 (i.e. 1 out of 4 of all A sites counted in the STM images were observed to have a VS). It is slightly less probable to find S vacancies at M positions, with 0.18 of all M sites containing a VS. The number of S vacancies observed directly at the corner site between the Mo-edge and S-edge is considerably lower at 0.06, indicating a higher initial stability of the terminal S monomer at the corner site.

DFT calculations were performed to provide additional information on the probability of VS formation. The model chosen to represent experimentally observed MoS2 nanoparticles was a freely standing truncated MoS2 nanoparticle exposing a short S-edge with 100% S and a five S-monomers long Mo-edge (Mo-5S) comprising of sites of the C, A, and M type (Fig. 2b). The MoS2 nanoparticle models are important in our study since they possess higher site heterogeneity and allow a greater degree of S mobility on the edge, as opposed to semi-infinite stripe models used in previous work which only contain M sites5,18,19,24. We did not specifically include the Au substrate here, as recent calculations involving semi-infinite periodic calculations indicated that the substrate has a negligible effect on the bonding of S monomers to the Mo-edge50. As listed in Fig. 2b, the energy required to form S-vacancies (ES) (see “DFT calculations” in Methods) on the C, A, and M locations of the MoS2 nanoparticle is endothermic in all cases. However, the vacancy formation on the corner C site is significantly more unfavorable (ES = 1.97 eV) than the A or M position (ES = 1.19 eV and 1.05 eV, respectively). For reference, our corresponding calculation of ES on a semi-infinite stripe model gave a value of 0.99 eV, which is slightly lower than that of the M position. To quantitatively account for the effect of temperature and partial pressures of H2 and H2S in the experiment, we calculated the Gibbs free energy of VS formation from which the vacancy fraction at each of the three positions considered can be calculated (see “Theory methods”). We note that some H2S is present in the background gas even if pure H2 is used, due to residual gas and the fact that H2S is released during S vacancy formation, corresponding to a H2/H2S ratio of approximately 103–104 or higher (estimated from gas analysis using mass spectroscopy). Figure 2c shows the calculated vacancy fraction on the three locations at the experimental temperature of 673 K and a total pressure of 10−4 mbar as a function of the relative partial pressure of H2 and H2S from sulfur rich (H2/H2S > 1) to sulfur poor conditions (H2/H2S > 107). The model predicts a significant population of S vacancies at the M and A positions in the range of H2/H2S ratio between 103 and 104, in agreement with our experimental findings (Fig. 2a). Going to more sulfur poor conditions (H2/H2S > 105) from this point leads to almost unity vacancy fraction reflecting fully stripped M and A sites equivalent to a naked Mo-edge, which we never observed experimentally. On the other hand, the model predicts a very small fraction of corner (C) vacancies, below 10−7 as opposed to ~6% found in experiments. This difference may be due to the intrinsic errors in DFT (of 0.1–0.2 eV), which can introduce uncertainties in the calculated vacancy population. Furthermore, while the Au substrate is expected to not significantly influence ES on the M and A sites, which is captured well in semi-infinite models50, its influence on the corner site could be a source of error in our calculation of vacancy coverage for C positions in Fig. 2c. The DFT modeling also predicts that the VS formation energies on C sites are dependent on the edge length. Figure 2d shows that the sulfur vacancy formation energy (ES) at the corner site (see also Supplementary Fig. 4) decreases from ~2 eV to 1.8 eV (see Supplementary Fig. 4) upon going from Mo-edge length of four to six. Hence, this dependency will contribute to an increase of the overall VS fraction compared with the 5 S-atom Mo-edge, as shown in Fig. 2c (lightly shaded pink circles). The increasing trend for the VS probability for the corner sites is also noted in the breakdown of the experimental data for different edge lengths in Supplementary Fig. 2. For comparison, the calculated ES values at the M site in Fig. 2d (only present in 5S and 6S Mo-edges) shows that the vacancy coverage should be size independent, which is consistent with our STM observations (Supplementary Fig. 2). Furthermore, for the A sites, DFT predicts a variation where the S atom is hardest to remove on the 5S Mo-edge (1.19 eV for 5S compared to 1.12 eV and 0.99 eV for 4S and 6S respectively in Fig. 2d). We note that the DFT functional tends to trimerize Mo atoms particularly on the 5S Mo-edge which makes it generally easier to remove the S atom on the longer-bridge (e.g. M site of 5S) than others. We posit that this effect is the reason for variation of ES values with Mo-edge length, which is not reflecting the variation in the experiment.

### Thiophene adsorption modes on MoS2-edge and corner sites

To address differences in the bonding affinity of thiophene at C, A, and M-type VS sites, we performed plane wave DFT calculations for thiophene adsorption on pre-formed S-vacancies along the edge of the MoS2 nanoparticle model (see Fig. 2b) and identified the most energetically stable adsorbed states (see Fig. 4 for most stable structures and Supplementary Table 1 for alternative structures). We then compare these states with the corresponding edge structure observed on these sites in atom-resolved STM images, obtained after thiophene exposure at room temperature (Fig. 5). The most stable adsorbed thiophene state corresponds to adsorption directly at the corner S-vacancy site C (Fig. 4a). In this state, thiophene binds strongly (binding energy, BE = −2.50 eV) to the exposed Mo atom available at the C site. This is in line with our experiments (Fig. 3b), showing that the C vacancy sites react with thiophene readily at room temperature. In the STM images shown in Fig. 5a, we see a bright (see also linescan I in Fig. 5d), larger, and slightly displaced protrusion at the corner site (indicated with a C in the superimposed model structure), consistent with the presence of a thiophene molecule adsorbed in configuration shown in Fig. 4a.

The most stable adsorption mode found for thiophene on the A vacancy site is shown in Fig. 4b. In this state, thiophene adsorption is still exothermic by ~ −0.85 eV, but much less favorable than for corner S-vacancies, indicating that steric hindrance is significant on the A site. Importantly, the DFT calculation shows that the adsorption introduces a significant structural modification to “make space” for thiophene, thus displacing the neighboring S atom in the r-MoS2. Specifically, as the thiophene is adsorbed onto a S-vacancy on the A site, it either pushes away a neighboring S atom on the corner from a bridge to a top position (see the green arrow in Fig. 4b) or displaces a neighboring S atom on the interior (Fig. 4c, d) to its neighboring site leading to a relatively less exothermic adsorption (BE = −0.60/−0.65 eV for Mo-5S and Mo-6S edges, respectively). Indeed, this latter S displacement leads to the formation of a stable S2 dimer (formally $$S_2^{2 - }$$) (see red arrow and dashed circle in Fig. 4c) and places the thiophene in coordination with a double VS vacancy located over both the A and M sites. Note in Fig. 4c that the S in thiophene coordinates to one of the undercoordinated Mo atoms, whereas that carbon ring system is placed over the Mo which has become accessible due to the displacement of the S. Our calculations identified additional favorable thiophene adsorption states with the concurrent formation of S dimers and corner S atop a corner Mo atom (with BE ~ −0.5 to −0.6 eV, see structures (i)–(iii) of Supplementary Table 1). Our calculations also showed that thiophene adsorption on a S-vacancy on the M site can generate a favorable adsorption site by displacing all S atoms on the edge to a top position without S2 formation (see structure (iv) of Supplementary Table 1); we expect, however, that such a concerted displacement may involve a large kinetic barrier due to an additive contribution of each S displacement. As shown in Fig. 4e, for adsorption at the M vacancy site, the S2 dimer can also be formed on the other side of the edge closer to the C site, in a configuration with an exothermic BE of −0.39 eV (note here we used a six S-atom-long Mo-edge (Mo-6S)).

The STM images in Fig. 5b, c provide the experimental evidence that adsorption on A and M sites occurs via S displacement. To observe this, we point to the contrast asymmetry around the distinct dark edge region denoted with an A and S2 or M and S2, respectively, in the lower part of Fig. 5b, c (see also linescans II, III and IV in Fig. 5d). Here, the S2 dimer is located at the high contrast protrusion on only one side and the dark spot reflects the rather open region between thiophene and the S2 dimer seen in all cases in the structures in Fig. 4c–e. The edges reflecting linescans II and IV correspond to the adsorption of thiophene where the S2 dimer is located either away from the nearest corner site or towards the corner position. We also observe the modeled displacement of the corner S monomer from a bridge to on top in configuration 4b, as indicated by the position of the protrusion at the green arrow in the superimposed image below Fig. 5b.

Our calculations show that S2 dimerization, per se, is typically endothermic with energies ranging from 0.7–1.9 eV depending on the location of the dimers and the concomitant vacancy (see Supplementary Table 2). Since the adsorption with dimerization is, in all cases, exothermic, we checked if thiophene adsorption and S displacement occurs with a concerted transition state or sequentially. As shown in Supplementary Table 3, the barrier for S-dimerization starting from an edge vacancy and creating two adjacent vacancies without thiophene is very high (~2.1 eV). The barrier for the concerted step is 0.8 eV lower than the sequential pathway (at ~1.3 eV), which is fully consistent with the STM observation in Fig. 3b that thiophene adsorption on A and M sites is favorable, but with a surmountable reaction barrier.

## Methods

### Experimental details

The experimental approach is based on the synthesis method of well-defined nanocrystals of MoS2 on Au(111)27,30,53. The synthesis of MoS2 nanoparticles on Au(111) was carried out on a clean Au(111) single crystal in a standard ultra-high vacuum (UHV) equipped with a homebuilt Aarhus-type variable temperature STM. The sample temperature was measured with a K-type thermocouple in contact with the backside of the Au crystal. MoS2 nanoparticles were synthesized by physical vapor deposition of Mo evaporation onto the gold surface using an e-beam evaporator (Oxford Applied Research EGCO-4). H2S gas (AGA, purity 99.8%) was dosed for sulfidation corresponding to a background pressure of 1.0 × 10−6 mbar for 15 min, while the sample temperature was 400 K. In order to obtain full crystallinity of the MoS2 nanoparticles, the samples were post-annealed in H2S atmosphere to 673 K for 10 min. The MoS2 nanoparticles were subsequently activated under reducing conditions by back-filling the UHV chamber with H2 (99.999%, Praxair)50. The reduced (active) MoS2 nanoparticles (r-MoS2) used as the starting point of the experiments reported here were obtained by annealing the fully sulfided samples to 673 K in of 10−4 mbar of H2. For thiophene (Sigma Aldrich, 99% purity) dosage, the liquid was kept in a glass container and admitted to the UHV chamber through a leak valve connected to a stainless steel tube directed to the sample surface. Prior to dosing, the thiophene liquid was purified by several freeze-pump-thaw cycles to remove dissolved gas. The gas purity was checked and H2S and H2 partial pressures were measured by a quadrupole mass spectrometer (Hiden Analytical). The samples were exposed to thiophene at three different substrate temperatures: 300, 400, and 650 K; a background pressure of 1.0 × 10−7 mbar was used for 5 min during the thiophene dosage. After each thiophene exposure, the sample was cooled down to room temperature and transferred to the STM for a local inspection. STM images were recorded at room temperature using etched W tips in the constant current mode. The tunneling parameters are noted in the image text as Vt (tunneling bias) and It (tunneling current).

### DFT calculations

All calculations (unless specified otherwise) were carried out for a single-layer hexagonal nanoparticle model of MoS2 with Mo-edge containing two to six molybdenum atoms while the S-edge contains four atoms, reflecting the typical truncated triangular shape of a single-layer particle observed in STM experiments. Two layers of sulfur atoms sandwich the molybdenum layer such that they are in the trigonal prismatic positions characteristic of the 2H phase of MoS2. Further, most of the calculations involve a Mo-edge of six molybdenum atoms (and 5S monomers, denoted Mo-5S), also consistent with our observations that metal edges are longer than the sulfur edges. All figures in this document with MoS2 structures and adsorption configurations show Mo atoms in blue, S atoms in yellow, C atoms in brown, and H atoms in gray. The Au(111) support was not included in these calculations; while the support can influence the stability of the edges, the single-layer stripe model used in periodic calculations have successfully captured the correct edge morphologies under sulfidation and reducing conditions, as inferred from STM studies We, therefore, suggest that the given model captures the local electronic structures adequately to provide a comparative analysis between different locations along the periphery of the MoS2 nanoparticles used in experiments. We further assume that the sulfur edge is 100% S-decorated while the metal edge is 50% S-decorated, consistent with ab initio phase diagrams and STM observations50. Importantly, we introduce S-vacancies on the Mo-edge on the center (C), adjacent (A) and middle (M) positions, as needed, to elucidate the energetics of S-vacancy formation and of thiophene adsorption on these vacancies.

The calculations were carried out with VASP54,55, a plane wave periodic DFT code. Generalized gradient approximation and projected augmented wave (PAW) potentials56 were used with PBE exchange correlation functional57 and the D3 Grimme dispersion correction58. All calculations were carried out in a box that had at least 10 Å of vacuum between two images in all directions. Spin polarization is included in all calculations: the effect of spin was found to be negligible (~0.01 eV on the vacancy formation energy). Plane wave and density wave cutoffs of 400 and 645 eV were used, respectively. A Gaussian smearing of 0.05 eV was used, and the energies were extrapolated to 0 K. Only gamma-point sampling was used in view of the large dimensions of the supercell. The convergence criterion for geometric relaxation was set to 0.02 eV/A.

The energy of vacancy formation, ES, is given by

$$E_{\rm{S}} = E_{{\rm{NP}},{\rm{with}}\,{\rm{CUS}}} \, + \, E_{{\rm{H}}_2{\rm{S}}}-E_{{\rm{NP}}}-E_{{\rm{H}}_2},$$
(1)

where ENP,with CUS is the energy of the nanoparticle with coordinative unsaturation (CUS or vacancy), $${E_{{\rm{H}}_2 {\rm{S}}}}$$ is the energy of H2S in the gas phase, ENP is the energy of the nanoparticle at the equilibrium termination (50% metal edge, 100% sulfur edge), and $${E_{{\rm{H}}_2 }}$$ is the energy of hydrogen gas. The free energy of vacancy formation, required to estimate vacancy fraction, was computed by calculating temperature-dependent enthalpy and entropy59. The entropy of surface S atoms was assumed to be of vibrational origin while rotational and translational components from statistical mechanics for an ideal polyatomic species were additionally included for gaseous molecules. Vibrational entropy was computed using the Harmonic approximation, which is sufficient for all systems considered here. The binding energy of thiophene, BE, is given by

$${\rm{BE}} = E_{{\rm{NP}},\,{\rm{with}}\,{\rm{CUS}} + {\rm{Thiophene}}}-E_{{\rm{NP}},\,{\rm{with}}\,{\rm{CUS}}}-E_{{\rm{Thiophene}}},$$
(2)

where ENP, with CUS + Thiophene is the energy of the nanoparticle with thiophene adsorbed on the vacancy (with and without sulfur rearrangements), ENP, with CUS is the energy of the nanoparticle with the vacancy (without rearrangements), and EThiophene is the energy of gas-phase thiophene.

For transition state calculations in Supplementary Table 1, a periodic single-layer stripe model with four rows of six Mo atoms each and commensurate S atoms was created based on our previous work10,25,59. We start with a 50% S covered Mo-edge and create a single vacancy in the middle of the edge to represent an internal vacancy (corresponding to an M vacancy site in the main text). The supercell was periodic in one direction (along the length of the edge) where an Mo-Mo distance of 3.19 Å was set based on our optimized lattice constant; the cell had more than 15 Å of vacuum in the other directions. The k-point sampling was 1 × 2 (along the edge length) x1 based on the Monkhorst Pack method60. All other settings for these DFT calculations were the same as above. The transition states were calculated using the climbing image nudged elastic band method61 such that the forces on the images were less than 0.1 eV/Å.

## Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

## References

1. 1.

Toulhoat, H., Raybaud, P. (eds). Catalysis by Transition Metal Sulphides (Editions Technip, Paris, 2013).

2. 2.

Topsøe, H., Clausen, B. S., Massoth, F. E. Hydrotreating Catalysis Vol. 11 (Springer Verlag, Berlin, Heidelberg, 1996).

3. 3.

Salmeron, M., Somorjai, G. A., Wold, A., Chianelli, R. & Liang, K. S. The adsorption and binding of thiophene, butene and H2S on the basal plane of MoS2 single crystals. Chem. Phys. Lett. 90, 105–107 (1982).

4. 4.

Daage, M. & Chianelli, R. R. Structure-function relations in molybdenum sulfide catalysts: the “rim-edge” model. J. Catal. 149, 414–427 (1994).

5. 5.

Raybaud, P., Hafner, J., Kresse, G., Kasztelan, S. & Toulhoat, H. Ab initio study of the H2-H2S/MoS2 gas-solid interface: the nature of the catalytically active sites. J. Catal. 189, 129–146 (2000).

6. 6.

Lauritsen, J. V. et al. Hydrodesulfurization reaction pathways on MoS2 nanoclusters revealed by scanning tunneling microscopy. J. Catal. 224, 94–106 (2004).

7. 7.

Hansen, L. P. et al. Atomic-scale edge structures on industrial-style MoS2 nanocatalysts. Angew. Chem. Int. Ed. 50, 10153–10156 (2011).

8. 8.

Lauritsen, J. V. & Besenbacher, F. Atom-resolved scanning tunneling microscopy investigations of molecular adsorption on MoS2 and CoMoS hydrodesulfurization catalysts. J. Catal. 328, 49–58 (2015).

9. 9.

Baubet, B. et al. Quantitative two-dimensional (2D) morphology-selectivity relationship of CoMoS nanolayers: a combined high-resolution high-angle annular dark field scanning transmission electron microscopy (HR HAADF-STEM) and density functional theory (DFT) study. ACS Catal. 6, 1081–1092 (2016).

10. 10.

Rangarajan, S. & Mavrikakis, M. On the preferred active sites of promoted MoS2 for hydrodesulfurization with minimal organonitrogen inhibition. ACS Catal. 7, 501–509 (2017).

11. 11.

Topsøe, N. Y. et al. Spectroscopy, microscopy and theoretical study of NO adsorption on MoS2 and Co-Mo-S hydrotreating catalysts. J. Catal. 279, 337–351 (2011).

12. 12.

Topsøe, N.-Y. & Topsøe, H. Characterization of the structures and active sites in sulfided Co-Mo/Al2O3 and Ni-Mo/Al2O3 Catalysts by NO Chemisorption. J. Catal. 84, 386–401 (1983).

13. 13.

Travert, A. et al. CO adsorption on CoMo and NiMo sulfide catalysts: a combined IR and DFT study. J. Phys. Chem. B 110, 1261–1270 (2006).

14. 14.

Chen, J., Maugé, F., El Fallah, J. & Oliviero, L. IR spectroscopy evidence of MoS2 morphology change by citric acid addition on MoS2/Al2O3 catalysts—a step forward to differentiate the reactivity of M-edge and S-edge. J. Catal. 320, 170–179 (2014).

15. 15.

McCarty, K. F. & Schrader, G. L. Deuterodesulfurization of thiophene: an investigation of the reaction mechanism. J. Catal. 103, 261–269 (1987).

16. 16.

Raybaud, P., Hafner, J., Kresse, G. & Toulhoat, H. Adsorption of thiophene on the catalytically active surface of MoS2: an local-density-functional study. Phys. Rev. Lett. 80, 1481–1484 (1998).

17. 17.

van der Meer, Y., Hensen, E. J. M., van Veen, J. A. R. & van der Kraan, A. M. Characterization and thiophene hydrodesulfurization activity of amorphous-silica-alumina-supported NiW catalysts. J. Catal. 228, 433–446 (2004).

18. 18.

Cristol, S., Paul, J.-F., Schovsbo, C., Veilly, E. & Payen, E. DFT study of thiophene adsorption on molybdenum sulfide. J. Catal. 239, 145–153 (2006).

19. 19.

Moses, P. G., Hinnemann, B., Topsøe, H. & Nørskov, J. K. The hydrogenation and direct desulfurization reaction pathway in thiophene hydrodesulfurization over MoS2 catalysts at realistic conditions: a density functional study. J. Catal. 248, 188–203 (2007).

20. 20.

Jin, Q. et al. A theoretical study on reaction mechanisms and kinetics of thiophene hydrodesulfurization over MoS2 catalysts. Catal. Today 312, 158–167 (2018).

21. 21.

Bataille, F. et al. Alkyldibenzothiophenes hydrodesulfurization-promoter effect, reactivity, and reaction mechanism. J. Catal. 191, 409–422 (2000).

22. 22.

Grange, P. Catalytic hydrodesulfurization. Catal. Rev. 21, 135–181 (1980).

23. 23.

Li, S., Liu, Y., Feng, X., Chen, X. & Yang, C. Insights into the reaction pathway of thiophene hydrodesulfurization over corner site of MoS2 catalyst: a density functional theory study. Mol. Catal. 463, 45–53 (2019).

24. 24.

Paul, J. F. & Payen, E. Vacancy formation on MoS2 hydrodesulfurization catalyst: DFT study of the mechanism. J. Phys. Chem. B 107, 4057–4064 (2003).

25. 25.

Rangarajan, S. & Mavrikakis, M. DFT insights into the competitive adsorption of sulfur- and nitrogen-containing compounds and hydrocarbons on Co-promoted molybdenum sulfide catalysts. ACS Catal. 6, 2904–2917 (2016).

26. 26.

Byskov, L. S., Nørskov, J. K., Clausen, B. S. & Topsøe, H. DFT calculations of unpromoted and promoted MoS2-based hydrodesulfurization catalysts. J. Catal. 187, 109–122 (1999).

27. 27.

Helveg, S. et al. Atomic-scale structure of single-layer MoS2 nanoclusters. Phys. Rev. Lett. 84, 951–954 (2000).

28. 28.

Schweiger, H., Raybaud, P., Kresse, G. & Toulhoat, H. Shape and edge sites modifications of MoS2 catalytic nanoparticles induced by working conditions: a theoretical study. J. Catal. 207, 76–87 (2002).

29. 29.

Bollinger, M. V., Jacobsen, K. W. & Nørskov, J. K. Atomic and electronic structure of MoS2 nanoparticles. Phys. Rev. B 67, 085410 (2003).

30. 30.

Lauritsen, J. V. et al. Atomic-scale insight into structure and morphology changes of MoS2 nanoclusters in hydrotreating catalysts. J. Catal. 221, 510–522 (2004).

31. 31.

Dinter, N. et al. Temperature-programmed reduction of unpromoted MoS2-based hydrodesulfurization catalysts: first-principles kinetic Monte Carlo simulations and comparison with experiments. J. Catal. 275, 117–128 (2010).

32. 32.

Sharma, L., Upadhyay, R., Rangarajan, S. & Baltrusaitis, J. Inhibitor, Co-catalyst, or Co-reactant? Probing the different roles of H2S during CO2 hydrogenation on the MoS2 catalyst. ACS Catal. 9, 10044–10059 (2019).

33. 33.

Rosen, A. S., Notestein, J. M. & Snurr, R. Q. Comprehensive phase diagrams of MoS2 edge sites using dispersion-corrected DFT free energy calculations. J. Phys. Chem. C. 122, 15318–15329 (2018).

34. 34.

Bruix, A., Lauritsen, J. V. & Hammer, B. Effects of particle size and edge structure on the electronic structure, spectroscopic features, and chemical properties of Au(111)-supported MoS2 nanoparticles. Faraday Discuss. 188, 323–343 (2016).

35. 35.

Saric, M., Rossmeisl, J. & Moses, P. G. Modeling the active sites of Co-promoted MoS2 particles by DFT. Phys. Chem. Chem. Phys. 19, 2017–2024 (2017).

36. 36.

Topsøe, H. et al. The role of reaction pathways and support interactions in the development of high activity hydrotreating catalysts. Catal. Today 107-08, 12–22 (2005).

37. 37.

Hermann, N., Brorson, M. & Topsøe, H. Activities of unsupported secand transition series sulfides for hydrodesulfurization of sterically hinderede 4,6-dimethylbenzothiophene and of unsubtituted dibenzothiophene. Catal. Lett. 65, 169–174 (2000).

38. 38.

Morales-Valencia, E. M., Castillo-Araiza, C. O., Giraldo, S. A. & Baldovino-Medrano, V. G. Kinetic assessment of the simultaneous hydrodesulfurization of dibenzothiophene and the hydrogenation of diverse polyaromatic structures. ACS Catal. 8, 3926–3942 (2018).

39. 39.

Li, X., Wang, A. J., Egorova, M. & Prins, R. Kinetics of the HDS of 4,6-dimethyldibenzothiophene and its hydrogenated intermediates over sulfided Mo and NiMo on γ-Al2O3. J. Catal. 250, 283–293 (2007).

40. 40.

Schachtl, E., Yoo, J. S., Gutiérrez, O. Y., Studt, F. & Lercher, J. A. Impact of Ni promotion on the hydrogenation pathways of phenanthrene on MoS2/γ-Al2O3. J. Catal. 352, 171–181 (2017).

41. 41.

Knudsen, K. G., Cooper, B. H. & Topsøe, H. Catalyst and process technologies for ultra low sulfur diesel. Appl. Catal. A 189, 205–215 (1999).

42. 42.

Choudhary, T. V., Parrott, S. & Johnson, B. Unraveling heavy oil desulfurization chemistry: targeting clean fuels. Environ. Sci. Technol. 42, 1944–1947 (2008).

43. 43.

Rota, F. & Prins, R. Role of hydrogenolysis and nucleophilic substitution in hydrodenitrogenation over sulfided NiMo/γ-Al2O3. J. Catal. 202, 195–199 (2001).

44. 44.

Prins, R. Catalytic hydrodenitrogenation. Adv. Catal. 46, 399–464 (2002).

45. 45.

Gutiérrez, O. Y. et al. Effects of the support on the performance and promotion of (Ni)MoS2 catalysts for simultaneous hydrodenitrogenation and hydrodesulfurization. ACS Catal. 4, 1487–1499 (2014).

46. 46.

Albersberger, S. et al. Simultaneous hydrodenitrogenation and hydrodesulfurization on unsupported Ni-Mo-W sulfides. Catal. Today 297, 344–355 (2017).

47. 47.

Lauritsen, J. V. et al. Size-dependent structure of MoS2 nanocrystals. Nat. Nanotechnol. 2, 53–58 (2007).

48. 48.

Bruix, A. et al. In situ detection of active edge sites in single-layer MoS2 Catalysts. ACS Nano 9, 9322–9330 (2015).

49. 49.

Lauritsen, J. V. et al. Size-dependent structure of MoS2 nanocrystals. Nat. Nanotechnol. 2, 53–58 (2007).

50. 50.

Grønborg, S. S. et al. Visualizing hydrogen-induced reshaping and edge activation in MoS2 and Co-promoted MoS2 catalyst clusters. Nat. Commun. 9, 2211 (2018).

51. 51.

Mom, R. V., Louwen, J. N., Frenken, J. W. M. & Groot, I. M. N. In situ observations of an active MoS2 model hydrodesulfurization catalyst. Nat. Commun. 10, 2546 (2019).

52. 52.

Schweiger, H., Raybaud, P., Kresse, G. & Toulhoat, H. Shape and edge sites modifications of MoS2 catalytic nanoparitcles induced by working conditions: a theoretical study. J. Catal. 207, 76–87 (2002).

53. 53.

Lauritsen, J. V. & Besenbacher, F. Model catalyst surfaces investigated by scanning tunneling microscopy. Adv. Catal. 50, 97–143 (2006).

54. 54.

Kresse, G. & Furthmuller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci., 6, 15–50 (1996).

55. 55.

Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).

56. 56.

Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59, 1758–1775 (1999).

57. 57.

Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).

58. 58.

Grimme, S., Antony, J., Ehrlich, S. & Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 132, 154104 (2010).

59. 59.

Rangarajan, S. & Mavrikakis, M. Adsorption of nitrogen- and sulfur-containing compounds on NiMoS for hydrotreating reactions: a DFT and vdW-corrected study. AIChE J. 61, 4036–4050 (2015).

60. 60.

Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 13, 5188–5192 (1976).

61. 61.

Henkelman, G., Uberuaga, B. P. & Jónsson, H. A climbing image nudged elastic band method for finding saddle points and minimum energy paths. J. Chem. Phys. 113, 9901–9904 (2000).

## Acknowledgements

The computational work was supported by the U.S. Department of Energy (DOE)-Basic Energy Sciences (BES), Office of Chemical Sciences, Catalysis Science Program, under Grant No. DE‐FG02‐05ER15731. Part of the calculations were conducted using supercomputing resources from the National Energy Research Scientific Computing Center (NERSC) and the Center for Nanoscale Materials (CNM) at Argonne National Laboratory (ANL). CNM and NERSC are supported by the U.S. Department of Energy, Office of Science, under contracts DE‐AC02‐06CH11357 and DE‐AC02‐05CH11231, respectively. S.R. acknowledges the use of Extreme Science and Engineering Discovery Environment (XSEDE) resources, which is supported by National Science Foundation grant number ACI-1548562. J.V.L., J.R.F., and N.S. acknowledge financial support from the Danish Research Council—Technology and Production (HYDECAT), Villumfonden and Haldor Topsøe A/S. J.V.L. acknowledges the Aarhus University Centre for Integrated Materials Research (iMAT).

## Author information

Authors

### Contributions

N.S. and J.R.F. performed the experiments. N.S. analyzed the experimental data. S.R. performed the theory. J.V.L. and M.M. planned and organized the studies. N.S. wrote the first draft. J.V.L. wrote the final version. All authors contributed to the final version of the manuscript.

### Corresponding authors

Correspondence to Manos Mavrikakis or Jeppe V. Lauritsen.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Rights and permissions

Reprints and Permissions

Salazar, N., Rangarajan, S., Rodríguez-Fernández, J. et al. Site-dependent reactivity of MoS2 nanoparticles in hydrodesulfurization of thiophene. Nat Commun 11, 4369 (2020). https://doi.org/10.1038/s41467-020-18183-4

• Accepted:

• Published:

• DOI: https://doi.org/10.1038/s41467-020-18183-4

• ### Adsorption performance of CeO2@NS-HNbMoO6 for ethyl mercaptan in methane gas

• Lifang Hu
• Xianyun Zheng
• Jie He

Chemical Papers (2022)