Modelling thin film growth over realistic time scales

Energy security and supply is a key problem for the UK in the coming years. Photovoltaics have an important role to play in this. In order for demand to be met, continued research into new materials and methods of production is necessary. By modelling deposition techniques using classical molecular dynamics (MD), an atomistic scale understanding can be obtained. Combining this with long time scale dynamics (LTSD) techniques allows us to also model diffusion and surface growth over realistic time scales. The LTSD technique applied throughout this project is an on-the-fly Kinetic Monte Carlo (otf-KMC) method, which determines diffusion pathways and barriers, in parallel, with no prior knowledge of the involved transitions. These simulation techniques allow parameters such as deposition energy, substrate bias and plasma pressure to be easily changed to gain understanding of their effects. During this project, growth via industrial scale deposition techniques has been simulated, including evaporation (thermal and electron beam), ion-beam assisted evaporation and reactive magnetron sputtering. Metal thin films, of interest due to their uses in reflectors in concentrator photovoltaics, electrical conductors in the monolithic interconnect processes and back contacts, were investigated using otf-KMC. Ag and Al film growth was simulated for around 0.3 seconds of real time. It was found that Ag has the ability to grow smooth surfaces, using several mechanisms including multiple-atom concerted motion, exchange mechanisms, and damage and repair mechanisms. Ag (111) and (100) surfaces grew dense, complete and crystalline films when sputtering was simulated, however, evaporation deposition produced incomplete layers. The inclusion of Ar in the ion-beam assisted evaporation of Ag (111) aided growth by transferring extra energy to the surface allowing increased diffusion and atomic mixing. Al (111) and (100), however, show different patterns. Growth by evaporation deposition and magnetron sputtering actually produced very similar results. The inclusion of the ion-beam assist on the (111) surface actually damaged the film, producing subsurface Ar clusters where Al atoms were displaced, creating voids throughout the film. Otf-KMC methods enabled the investigation of specific mechanisms allowing film growth and a very important transition enabling the smooth and complete Al film growth was found to be the Ehrlich-Schwoebel (ES) barrier. The ES barrier involves an atom dropping off a step edge of an island and this barrier was found to be much smaller for the Al surfaces, therefore allowing the more complete growth from both evaporation and sputtering. Metal oxides are also of great interest in the photovoltaic industry. The rutile TiO$_2$ (110) surface was investigated using single point depositions, high temperature MD and otf-KMC. Otf-KMC enabled the simulation for up to 9 seconds of real time, totally inaccessible using traditional simulation methods. Results concluded that the evaporation deposition process produced a void filled, incomplete structure, even with the use of a low energy ion-beam assist, this material is of interest for dye-sensitised solar cells where a dye is injected into the voids. Sputtering, however, produced dense and crystalline film, which is much more applicable to anti-reflective coatings where a crystalline structure is required. Mechanisms which enabled crystalline rutile to form were also investigated, highlighting Ti interstitial annealing in the presence of an O rich surface as an important rutile growth mechanism. ZnO, an inorganic compound with many uses including transparent conductive oxides, is investigated in the most stable wurtzite phase. The O-terminated (000$\bar{1}$) polar surface was used as the substrate for otf-KMC growth simulations, where around 1 second of real time was simulated. Evaporation deposition of a stoichiometric distribution of deposition species was found to produce the best quality film, however, a phase boundary was observed where an area of zinc blende forms within the wurtzite. Sputtering resulted in a denser, more complete and crystalline structure due to the higher deposition energy of arriving species, similar to the TiO$_2$ results. Post-annealing at 770K did not allow complete recrystallisation, resulting in films with stacking faults where monolayers formed in the zinc blende phase. Annealing at 920K, however, in some cases enabled the complete recrystallisation of films back into the wurtzite structure. Although, the higher annealing temperature did not always enable recrystallisation and in some cases both wurtzite and zinc blende phases existed in the same layer, resulting in a phase boundary. An important mechanism for the nucleation of ZnO growth was found to be the formation and vibration of Zn$_x$O$_y$ strings on the surface, which after hundreds of milliseconds formed the desired hexagonal structure. Combining MD and otf-KMC enabled the simulation of systems over very large time scales which were previously totally inaccessible. Key mechanisms occurring during the growth of metals and metal oxides were investigated, providing a much more precise understanding of how growth occurs. It is clear from the work that the deposition technique used plays a significant role on the resulting film quality and surface morphology and we are now able to provide an insight into the optimum conditions under which complete, crystalline layers can form