The presence of heteroatoms (e.g. S, N) in crude oil poses formidable challenges in petroleum refining processes as a result of their irreversible binding on catalytically active sites at industrially relevant conditions. Such poisons render these sites inaccessible to reactants, negatively impacting overall process performance, and require replacement and/or regeneration of the catalysts, leading to high material and time costs. With increasing pressures from legislation that continues to lower the permissible levels of sulfur content in fuels, hydrodesulfurization (HDS), the aptly named reaction for removing heteroatoms from organosulfur compounds, has become an essential feedstock pretreatment step to remove deleterious species from affecting downstream processing. Extensive research in the area has identified the paradigm catalysts for desulfurization; MoSx or WSx, promoted with Co or Ni metal; however, despite the vast library of both empirical and fundamental studies, a clear understanding of site requirements, the elementary steps of C-S hydrogenolysis, and the properties that govern HDS reactivity and selectivity have been elusive. While such a lack of rigorous assessments has not prevented technological advancements in the field of HDS catalysis, fundamental interpretations can inform rational catalyst and process design, particularly in light of new requirements for “deep” desulfurization and in the absence of significant hydrotreatment catalyst developments in recent decades.

We report HDS rates of thiophene, which belongs to a class of compounds that are most resistant to sulfur removal (i.e. substituted alkyldibenzothiophenes), over a range of industrially relevant temperatures and pressures, measured at differential conditions and therefore revealing their true kinetic origins. These rates, normalized by the number of exposed metal atoms, on various SiO2-supported, monometallic transition metals (Re, Ru, Pt), range several orders of magnitude. Under relevant HDS conditions, Pt and Ru catalysts form a layer of chemisorbed sulfur on surfaces of a metallic bulk, challenging reports that assume the latter exists as its pyrite sulfide phase during reaction. While convergence to a single phase is expected and predictable from thermodynamics at a given temperature and sulfur chemical potential, metastability of two phases can exist. We demonstrate, through extensive characterization and kinetic evidence, such behaviors exist in Re, where structural disparities between its phases lead to kinetic hurdles that prevent interconversions between layered ReSx nanostructures and sulfur-covered Re metal clusters. Such features allowed, for the first time, direct comparisons of reaction rates at identical conditions on two disparate phases of the same transition metal identity. Rigorous assessments of kinetic and selectivity data indicated that more universal mechanistic features persist across all catalysts studied, suggesting that differences in their catalytic activity were the result of different densities of HDS sites, which appeared to correlate with their respective metal-sulfur bond energies.

Kinetic responses and product distributions indicated that the consumption of thiophene proceeds by the formation of a partially-hydrogenated surface intermediate, which subsequently produces tetrahydrothiophene (THT) and butene/butane (C4) via primary routes on similar types of sites. These sites are formed from desorption of weakly-bound sulfur adatoms on sulfur-covered metal surfaces, which can occur when the heat of sulfur adsorption is sufficiently low at high sulfur coverage as a result of increased sulfur-sulfur repulsive interactions. Relative stabilities and differences in the molecularity of the respective transition states that form THT and C4 dictate product distributions. THT desulfurization to form C4 occurs via readsorption and subsequent dehydrogenation, evidenced by secondary rates that exhibited negative H2 dependences. These behaviors suggest that C-S bond activation occurs on a partially (un)saturated intermediate, analogous to behaviors observed in C-C bond scission reactions of linear and cycloalkanes on hydrogen-covered metal surfaces. Our interpretations place HDS in a specific class of more general C-X hydrogenolysis reactions, including hydrodeoxygenation (HDO) that has gained popular appeal in recent biomass conversion processes.

These hydrodearomatization routes, hydrogenolysis and hydrogenation, act as probes for studying hydrogen spillover, a frequently observed phenomenon in bifunctional systems. Indeed, we observe enhancements solely in the rates of thiophene hydrogenation when monofunctional catalysts, which generate equilibrated concentrations of surface H-species, are mixed with materials (e.g. Al2O3) that cannot dissociate H2. Conventional mechanisms that suggest gas phase or surface diffusion of atomic H-species (or H+-e- pairs) are implausible across distances along insulating surfaces (i.e. SiO2, Al2O3). We propose, with kinetic-transport models that are consistent with all observed behaviors, that mobility of active H-species occurs through gas phase diffusion of thiophene-derived molecular H-carriers, whose rate of formation on HDS sites can control maximum spillover enhancements. This synergy is disrupted when the ability of thiophene to form these H-carriers is suppressed, leading to an absence of spillover-mediated rates and further challenging any diffusive roles of atomic H-species. Such implications help guide optimal designs of bifunctional cascades to permit the uninhibited access and egress of larger molecules within both catalytic functions.