Repository logo

Predicting Transitions in Fischer-Tropsch Reactors



Change log



The Fischer-Tropsch process has the potential to be fundamental to a future without dependence on fossil fuels. It converts syngas, a readily available resource, into high quality hydrocarbons, with water being the primary byproduct. Like many gas-to-liquid processes, it is catalysed on a transition metal surface, and the lifetime of the catalyst bed largely dictates the process’s economic viability. Predictive computational models can shed light on the mechanisms driving catalyst deactivation.

This work focuses particularly on reactors with a titania-supported cobalt catalyst. One part of this project is an investigation using VASP studying the adsorption and mobility of several cobalt species that might form on the TiO2 support surface. It is found that reactor species generally have a strong effect on the binding properties of cobalt, and that this effect could either strengthen or weaken its bond to the surface depending on how reactive the functional group is with the support surface. In particular, the carbon monoxide feedstock was found to favourably bind to surface- adsorbed cobalt and create highly mobile species. In practice the support surface is rarely dry, and this effect is also found on a model hydrated surface. This painted a clear picture that the carbon monoxide feed may have an effect on the sintering process by inducing surface and gas-phase transport of highly dispersed cobalt.

Plane-wave DFT suffers from unfavourable cubic scaling and its extended basis makes vacuum space costly, limiting its applicability to large clusters and surface structures. The linear-scaling DFT code ONETEP is a good candidate for investigating these classes of system. While its metals treatment for systems smaller than the thousands of atoms scale using ensemble DFT (EDFT) is also cubic-scaling, it is able to offload a lot of the cost of the calculation onto linear-scaling parts and also maintains a linear-scaling memory cost. Another part of this project was to develop ONETEP to be able to perform studies of chemistry on adsorbed and free catalyst nanoparticles. Two main pieces of crucial functionality have been added to ONETEP - free-spin EDFT and nudged elastic band transition-state searching. The former allows ensemble DFT to be performed at non-integer net spin and non-integer charge, and also adds the ability to relax the spin state during a calculation - previously ONETEP was constrained to fixed integer net spin. This is crucial in cases where the net spin of a magnetic system is not necessarily known or may be altered by, for example, surface adsorbates. The latter is a popular and robust transition state searching method that reliably minimizes a path connecting a product and reactant. The dimer method is also being developed for ONETEP primarily as a transition state refinement tool.

Early applications of the new functionality in ONETEP is demonstrated in an investigation of carbon monoxide binding on cobalt HCP and FCC nanoparticles of around 50 atoms. These adsorption energies are compared, where available, to surface adsorption energies from literature. Generally, the abundance of edges on these small particles make surface adsorption energies site dependent and different to clean surface values, and adsorption sites near edges are generally stronger than the surface values. Additionally, two CO dissociation pathways on the FCC cluster are examined, starting from the surface making up the majority on the FCC Wulff particle. A modest decrease in activation energy is identified, and the presence of new pathways involving particle edges is highlighted for future study.





Hine, Nicholas


ONETEP, Linear-scaling DFT, NEB, Transition State Search, Fischer-Tropsch, Gas-to-Liquid, DFT, Density Functional Theory, VASP, Plane-Wave DFT, Titania, Anatase, Cobalt


Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
This work was carried out with funding and advisor support from Shell Technology Center Bangalore. Funding was provided through the EPSRC CDT in Computational Methods for Materials Science. This work was performed in part using resources provided by the Cambridge Service for Data Driven Discovery (CSD3) operated by the University of Cambridge Research Computing Service (, provided by Dell EMC and Intel using Tier-2 funding from the Engineering and Physical Sciences Research Council (capital grant EP/P020259/1), and DiRAC funding from the Science and Technology Facilities Council ( I am grateful to the UK Materials and Molecular Modelling Hub for computational resources, which is partially funded by EPSRC (EP/P020194/1).