TiCp*(OMe)3 versus Ti(OMe)4 in atomic layer deposition of TiO2 with water--ab initio modelling of atomic layer deposition surface reactions.

It is a common finding that titanocene-derived precursors do not yield TiO2 films in ALD with water. For instance, ALD with Ti(OMe)4 and water gives 0.5 A/cycle, while TiCp*(OMe)3 does not show any growth (Me=CH3, Cp* = C5(CH3)5). This is apparently in contradiction with the computed reactivity of the ligands: the energetics of hydrolysis of the gas-phase precursor indicate that TiCp*(OMe)3 is more reactive to ligand elimination than Ti(OMe)4. However such a model of precursor reactivity neglects surface reactions such as adsorption, diffusion and desorption, all of which can have an important effect on ALD growth rate. A more accurate model of the surface reaction is needed to find the reason for the different behaviours of Ti(OMe)4 and TiCp*(OMe)3 in the ALD process. The more realistic surface model is a TiO2 slab that is periodic in three dimensions. These calculations reveal that TiCp*(OMe)3 does not chemisorb in the usual way because of extreme crowding of the Ti centre by Cp* and that this prevents ALD growth.