In order to ensure that the deposited thickness would be sufficient for the characterization process, we started by assessing the deposition rate for each set of parameters. To do so, we ran the deposition for 10 minutes, before measuring the resulting thickness of the film using a mechanical profilometer (Bruker DektakXT)
 

Film characterization

Design of experiment

Assuming that the responses of the experiment can be approximated well with a linear model, we opted for a fractional factorial design. For the first round of measurements, we chose to design the experiment such as to minimise the costs. As we have five factors, we used a 25-3 design (Table \ref{tab:experiment}) with the generator (1, 2, 3, 13, 23), yielding 8 measurement runs to perform. We added a null run in order to be able to test the lack-of-fit.
The system is of resolution III, meaning that the main effects are aliased with first order interactions (Table \ref{tab:aliasing}). The chuck temperature it is the only factor not defining the plasma, we thus expect it to have little interaction with the other factors. In order to limit aliasing on the other factors, we thus chose the temperature to be factor 3. If required, after analysis of the first round of experiments it is possible to extend the experiment matrix with 8 additional runs and obtain a 25-1 fractional factorial design, allowing de-aliasing of the main effects and the first order interactions.