The coating process, consisting of the generation of a bonding layer and the functional layer, was carried out in four consecutive phases. First, the magnetron was ramped up for three minutes at a constant argon flow of 120 sccm in front of a shading plate. During start-up, the sputter power was ramped up in two minutes from 0.5 kW to 7.0 kW followed by a ramp-down from 7.0 kW to 2.0 kW in one minute. During this process, the target surface was cleaned of any oxide layers. In the second phase, the shading plate was moved away from the magnetron and a pure titanium layer was deposited on the substrate for three minutes. The shading plate was subsequently moved back in front of the magnetron. In the third phase, the sputtering power was again increased to 7.0 kW. As soon as the sputtering power of 7.0 kW was reached, the oxygen control circuit was activated with the desired magnetron voltage. For a period of three minutes, the shading plate continued to be coated. During the fourth phase, the shading plate was moved away from the magnetron and the reactive coating process with titanium dioxide was carried out for five hours. The coating phase was followed by a one-hour cooling phase. All coating experiments were conducted with the same etching process, and with no additional heating and no BIAS during the coating process. The process gas flow was changed in three stages starting at 120 sccm, subsequently moving to 160 sccm, and finally up to 200 sccm. For each process gas flow, three operating points out of the transit region of target poisoning were chosen (Region 2 in Figure 2). The hysteresis behavior was generally divided into two stable sputtering areas, namely metallic (Region 1 in Figure 2) and reactive (Region 3 in Figure 2). An unstable transition area (Region 2 in Figure 2) was observed between the two stable areas. Although the metallic mode of operation has the advantage of high deposition rates, reactively deposited films from this region are substoichiometric. In the reactive sputtering region, the deposited layers are ideally stoichiometric with the disadvantage that the deposition rates are a factor of 5 to 10 times smaller than in the metallic sputtering region. The hysteresis behavior is prominent in the sputtering of oxide layers, e.g. Al2O3 or TiO2, and the process window for an economical and stoichiometric deposition is located in the transition area between the reactive and metallic sputtering areas. The sputtering behavior in the transition region poses unique challenges for process control since a relatively small change in the gas regulation of oxygen can lead to strong change target poisoning [15].