The Co-Mo-S model

Other surface techniques that have been applied to this system (e.g. XANES, XPS, and high-resolution transmission electron microscopy) also provided further support for the overall picture. As for the precise geometry of the active sites, the debate continues. 

From EXAFS measurements it appears that each Co is pentacoordinated in an approximately square-base pyramidal arrangement of sulfur ligands (as in structure (a) in Fig 1.3) with two neighboring Mo atoms, while each Mo is in turn hexacoordinated by a trigonal prismatic array of sulfur atoms with three neighboring Mo atoms and one Co atom (partially shown in (b) in Fig 1.3). 

Although very few complexes capable of closely mimicking the structure of HDS active sites are known, in the following Chapters a wealth of results obtained from coordination and organometallic chemistry will be presented, some of which are very clearly in line with a number of the possibilities discussed above, thus adding to this very lively discussion. 

---

Fig. I.S. A representation of the Co-Mo-S model site from STM measurements Adapted from ref, 47).

The overall degradation of (103) assisted by the cluster [(Cp )2 M o2Co2S3(CO)4] (Cp = CH3C5H4) is the model reaction that best resembles the heterogeneous counterparts, particularly those classified as Co/Mo/S phase,158 in terms of both structural motif and HDS activity.229 Morever, the Co/Mo/S cluster has successfully been employed to show that the C—S bond scission in the desulfurization of aromatic and aliphatic thiols occurs in homolytic fashion at 35 °C and that thiolate and sulfido groups can move over the face of the cluster as they are supposed to do over the surface of heterogeneous catalysts.230... 

What is the structure of this Co-Mo-S phase A model system, prepared by impregnating a MoS2 crystal with a dilute solution of cobalt ions, such that the model contains ppms of cobalt only, appears to have the same Mossbauer spectrum as the Co-Mo-S phase. It has the same isomer shift (characteristic of the oxidation state), recoilfree fraction (characteristic of lattice vibrations) and almost the same quadrupole splitting (characteristic of symmetry) at all temperatures between 4 and 600 K [71]. Thus, the cobalt species in the ppm Co/MoS2 system provides a convenient model for the active site in a Co-Mo hydrodesulfurization catalyst. 

The data analysis in Table 9.3 summarizes the crystallographic information of the Co-Mo-S phase active for hydrodesulfurization. The Co-S distance in Co-Mo-S is 0.22 nm, with a high sulfur coordination of 6.2 1.3. Each cobalt has on average 1.7 0.35 molybdenum neighbors at a distance of 0.28 nm. Based on these distances and coordination numbers one can test structure models for the CoMoS phase. The data are in full agreement with a structure in which cobalt is on the edge of a MoS2 particle, in the same plane as molybdenum. 

A third model therefore attributed the promotion effect to cobalt present in the Co-Mo-S phase, with cobalt atoms located at the M0S2 surface a significant contribution of separate Co9S8 was excluded (29). This so-called Co-Mo-S model (or Ni-Mo-S model for Ni-Mo catalysts) is currently the one most widely accepted. 

Fig. 1.3. Schematic representation of TopsOe s Co-Mo-S model, showing perspective views of (a) the coordination around the Co site and a MOjCoS, unit at the edge Co-Mo-S site,...

It has previously been found (3., 11, 18, 31-3 ) that unsupported catalysts exhibit a HDS activity behavior quite similar to that of supported catalysts. This suggests that although the support is of importance, it does not have an essential role for creation of the active phase. Thus, it is very relevant to study unsupported catalysts, both in their own right and also as models for the more e-lusive supported catalysts. Many different explanations have been proposed to explain the similarity in behavior of unsupported and supported catalysts ( 3, 31-3b). Recently, we have observed that for both types of catalysts the HDS activity behavior can be related to the fraction of cobalt atoms present as Co-Mo-S (9-11 35). 

Thus, theoretical calculations can be a very useful tool in mechanistic understanding. As it is now known that the most active Co-Mo-S or SBMS catalysts consist of very small clusters of Mo atoms (as few as seven), it is within the realm of practical computational procedures to completely model the catalyst/S-molecule interactions. Assumptions made about steric crowding around the catalytic site may be quite different for such small clusters as the catalytic site is not an extended planar surface, as discussed in the previous section. Future work in this area should provide new insight into the true limitations in HDS of dialkyldibenzothiophenes. 

---