Atomistic Simulations from the TCM Group

The CASTEP code developed by Mike Payne and others in the TCM Group is able to model the behaviour of atoms over short timescales. The examples linked to below show it being used in four very different situations: a catalytised reaction in a zeolite, the melting of a metal surface, and the oxidation of a metal and the oxidation of a non-metal.

Production of Petrol in Zeolite

The conversion of methanol to dimethyl ether is the first stage in a series of reactions used industrially to convert methanol to petrol (gasoline) so the entire process is often referred to as the MTG or "Methanol to Gasoline" process. This initial reaction is catalysed by zeolites. Zeolites are materials made from silicon, aluminium and oxygen which have open framework structures. They are used extensively as molecular sieves and as catalysts. In the present case, the zeolite acts as a Bronsted acid - a substance which donates protons (H+) in this case to the molecules inside the framework. A catalyst speeds up a reaction by reducing the energy barrier, or activation barrier, that must be overcome before the reaction takes place.

It should be noted that the reaction of the methanols to form dimethyl ether is energetically unfavourable outside the zeolite. Remarkably the reaction is even more energetically unfavourable inside the zeolite. The only reason that the reaction proceeds is because of the increase in entropy due to the reaction means that, at the reaction temperature, the free energy of the products is lower than that of the reactants.

The reaction is too slow to take place within the timescales we can simulate quantum mechanically so, in the computer, we force the reaction to take place by pushing one of the methanol molecules towards the other.

Aluminium Surface Diffusion and Melting

This simulation shows an aluminium surface (the (110) surface) with an extra aluminium atom on top, this extra atom is called an adatom. The aluminium atoms are coloured according to their height so, at the beginning of the simulation, the adatom is gold, the atoms in the surface layer are light blue and the atoms in the second layer are dark blue. We use periodic boundary conditions for our calculations so the two adatoms that can be seen are repeated copies of the single adatom in our simulation cell. We gradually increase the temperature to see how diffusion occurs at the surface. Interestingly, the diffusion does not occur by the adatom jumping from one surface site to another but instead occurs by the adatom displacing an atom from the surface layer so that the displaced atom now becomes an adatom but at a different position - this is usually referred to an "exchange process". As the temperature increases further the adatom is absorbed into the surface layer which now becomes disordered - like a liquid. The phenomenon of a surface being a liquid while the bulk material is still a solid is called "pre-melting". Finally, we remove two atoms from the surface layer and the surface recrystallises leaving a surface vacancy. If you look carefully at the very end of the simulation you will see the surface vacancy also moves via an exchange process.

Aluminium Oxidation

This is a series of five simulations showing a sequence of oxygen molecules hitting an aluminium surface and reacting to form the oxide. In the process of reacting with the surface the bond between the two oxygen atoms is broken and the oxygen molecule dissociates. By the end of the fifth simulation a layer of oxide has formed at the surface of the aluminium (it is the oxide film that makes aluminium foil non-reactive - the pure metal is highly reactive). Perhaps the most interesting finding from these simulations is in the third simulation when we see that one of the oxygen atoms ends up below the surface aluminium atoms. Thus, a subsurface oxide is formed long before the entire surface is covered with oxygen. Note how the oxygen molecules are moved into positions which make the reaction with the surface more likely to occur - this effect is called "steering". In these simulations (and the following simulations showing the oxidation of silicon) the diffuse shading shows the spin density in the electronic system - orange represents regions of positive spin density and blue regions of negative spin density.

Silicon Oxidation

This is a series of simulations showing a sequence of oxygen molecules hitting a silicon surface and reacting to form the oxide. In the final simulation (for the 6th and 7th oxygen molecules) only one of the oxygen molecules reacts with the surface. The other molecule is positioned over an area of the surface which is already oxidised. This shows that the oxide is much less reactive than the original silicon. In the third simulation we see that one of the silicon atoms ends up fully (fourfold) bonded to oxygen atoms even though there are other silicon atoms which have no oxygens bounded to them. This shows that oxidation reaction is more likely if the silicon already has some oxygen atoms bonded to it - this is called a "cooperative" reaction. Steering effects are clearly observable in these simulations. Look carefully at the two simulations which show the first oxygen molecule reacting with the clean silicon surface. In one case the molecule dissociates but when the molecule is in a different position, although the molecule binds to the surface, the oxygen atoms remain bound to each other as a molecule. In this latter case, if we heat up the system the oxygen molecule eventually dissociates.

The silicon-silicon oxide interface is of crucial importance in most computer chips and before these simulations there was no convincing model for its structure. This is why we decided to grow one in the computer!

Si oxidation, first molecule, site 1

Si oxidation, first molecule, site 2

Si oxidation, second molecule

Si oxidation, third molecule

Si oxidation, fourth and fifth molecules

Si oxidation, sixth and seventh molecules

The People who Performed the Calculations

The zeolite simulation was carried out by Marek Hytha, who is now Chief Scientist at MEARS Technologies, and Ivan Stich, who is now Director of the Institute of Physics of the Slovak Academy of Sciences. The aluminium surface simulation was performed by Nicola Marzari, who now holds the Chair of Materials Modelling in Oxford University. The oxidation simulations were performed by Lucio Colombi Ciacchi, who now holds the Conrad Naber Endowed Chair of Hybrid Materials Interfaces at the University of Bremen.