The project consisted of several scenarios. The first scenario revolved around a uniformal temperature across the entire model, in order to understand how the MD simulation behaves in response to temperature. Simulations done previously suggests it is possible to reform the diamond as a graphite crystal, however this demands a proper annealing schedule, as well as a scenario allowing expansion of the model itself, without periodic boundaries at least on the top and bottom facets of the simulation box. It might also be required to control the pressure within the crystal, as a small model tends to have a high pressure and rigid constraints under periodic boundary conditions. Taking these into an account was outside the scope of the project, thus we concentrated mostly on damage within a confined region.
The damage was done in two different manners. The first scenario induced the damage by a constant-temperature thermostat, keeping a constant high temperature across the entire region. This type of damage can be analogous to damage induced by a laser, and is currently under research. Below are the results of one of the simulations running this scenario. The model includes 13824 atoms, with a constant temperature of approximately 7000 K in a confined region.
The blue dots signify locations of SP3 hybridized atoms. Red lines signify bonds between SP2 hybridized atoms. One can see most of the lattice remains intact, as we applied the damage in a small region, keeping the rest of the model under a constant temperature of 300 K. The damaged area has turned into an amorphous SP2 material. Below are plots of the RDF of this model:
Below are plots of the Radial Distribution Function, RDF in short, of the previous model. The RDF, also known as the pair correlation function, describes the density of particles as a function of distance from a reference point. It is a valuable tool in the analysis of results, as peaks in the RDF, when calculated for a crystalline material, give information on the distances of the nearest neighbours for the different particles. Perfect crystals would take the shape of a sum of delta functions, while amorphous crystals would have wider peaks. The code used to calculate the RDF can be found in the Useful Scripts page. More on RDFs can be found in this Wikipedia page.
The left RDF provides additional proof to the claim that most of the model is still a diamond, as the first-neighbour peak of the RDF is on 1.51 Å, which is the bond length of the closes neighbour in a diamond lattice. The second RDF is computed on the damaged region, and the main peak is on 1.44 Å. The bond length of graphite under the Tersoff potential is 1.46 Å, and the discrepancy most likely originates from the lack of annealing after the damage in this particular example.
The second scenario is closer to the actual experiments, involving ion implantation. The model is once again damaged using controlled temperature. However, in this scenario, the damaged region has a piecewise constant temperature. We defined a set of concentric shells, the innermost with a temperature of 7500 K and the outermost with a temperature of 400 K. The rest of the model is kept under a temperature of 300 K.
The results of this scenerio are much better. Not only were we able to break the diamond bonds and receive SP2 hybridized atoms, the atoms also started arranging themselves in honeycomb planes, forming actual graphite. Below is a snapshot of a graphitized area in the diamond:
This result clearly shows a formation of a honeycomb lattice. The RDF for this model:
The marked peak is distinct from the adjacent peak, with the RDF dropping to 0 between the two peaks.
Bellow is yet another plot, the result of the simulation featured in About LAMMPS. It shows the bonds of SP2 sites. One can easily discern various planes as well as the honeycomb geometry of graphite.
The corresponding RDF plot, with a first-neighbour peak at 1.46 Å can be seen below: