The states of the target poisoning were assigned to the respective areas in the upper graph in Figure 2 which represents the progression of magnetron voltage over the linear increase in oxygen flow from 0 sccm to 96 sccm which was followed by a direct linear decrease in oxygen flow to 0 sccm. Recording a hysteresis plot during magnetron sputtering is an important step in detecting sputtering regions and the target poisoning state. In a hysteresis experiment, all settings are kept constant except for the variation of the reactive gas flow. At the beginning of the hysteresis experiment, the metallic sputtering region was present (approximately up to 30 sccm), since no target poisoning had occurred (a in Figure 2). As the oxygen flow increased, the target surface poisoning commenced outside the sputtering erosion (b in Figure 2). The target poisoning increased from 30 sccm to approximately 60 sccm oxygen flow to the point where the target surface was completely poisoned except for the sputtering erosion (c in Figure 2). The high ion density in this region provides enhanced sputtering in these areas of the target and is also responsible for keeping the sputtering erosion clear until shortly before the transition region (Region 2 in Figure 2). The magnetron voltage is very constant in the metallic region and only shows a small increase of approximately 20 V. Just past a slight increase in oxygen flow, in the metallic region at about 60 sccm oxygen flow, the process shifted to the unstable transition region (Region 2 in Figure 2). The transition region was characterized by an increase in target poisoning until partial saturation of the sputtering erosion (d in Figure 2). A relatively small increase of 15 sccm from 65 sccm oxygen flux to 80 sccm caused a rapid decrease in the magnetron voltage from 590 V to 450 V. In the transition area, the magnetron voltage displayed a rapid change over to the completely poisoned area. A small increase in oxygen flow of 10 sccm caused a sharp decline in magnetron voltage of 140 V and positioned the sputtering process to shift into the fully poisoned sputtering area. A further increase of the oxygen flow above 90 sccm only caused a slight decrease in the magnetron voltage but led to the complete poisoning of the target surface (e in Figure 2). Once the state of full poisoning was reached at 96 sccm, the flow was decreased to the starting value for the same duration. Based on the magnetron voltage plot, as the oxygen flow decreased, it became evident that the target poisoning was not behaving similarly to the increase of oxygen flow. The steep decrease in the transition region was not reversed to the same extent during the decrease in the oxygen flow. The total target poisoning only weakly decreased over a relatively large decrease from 70 sccm to approximately 25 sccm (f in Figure 2). If complete target poisoning occurs, the target is only atomized slowly due to the complete occupancy of titanium dioxide with the same decrease in oxygen flow, whereby the relatively slow atomization compared to the initially metallic region is the main influence. Since the target was completely covered with oxide and oxygen was constantly supplied, the sputtering process only reached the metallic sputtering region directly after a large decrease in oxygen flow. Since the magnetron voltage is a value that can directly be measured and the flux control is a manipulated value, both values can be used for a control loop to control the sputtering region of the reactive deposition. Usually, a magnetron voltage from the transition range, e.g. 500 V, is selected as desired value, and in a regulating concept of this kind, the oxygen flow is then adjusted to the desired magnetron voltage.
Such a feedback system is used on the PVD coating system Metaplas Domino.Mini, at the Niederrhein University of Applied Sciences, for the sputtering of oxides. Since hysteresis in reactive magnetron sputtering is always system dependent and influenced by the respective systems and plant conditions, the measurement of the hysteresis behavior is important for the investigation of an efficient and stoichiometric operating set point for the deposition of oxides.