Modeling of the Magnetron Discharge

Abstract

Because of the industrial importance of sputter deposition magnetrons, there is a strong drive to simulate the entire magnetron deposition process, to replace trial-and-error experiments. This can lead to serious cost reduction, both for the manufacturers and the users of magnetron sputter equipment. The reasons are straightforward. For a typical coating plant, one of the main costs is the installation cost. Hence, it is a major advantage to have equipment that can realize a large throughput which fits specifications like required deposition speed, uniformity, reproducibility, or target lifetime. This means that magnetron manufacturers strive to a minimum “setup time,” which requires from them the ability to predict whether the proposed design will work or not before the machine is actually built. This could be achieved by simulating the machines' characteristics. Of course, also the possibility of optimizing the deposition parameters without actually performing any real world experiment is attractive. The ideal magnetron sputter deposition simulation would use as input the desired coating characteristics (e.g., electrical resistance, adhesion, refractive index, etc.) and process requirements (e.g., deposition speed, price per square meter, etc.). It would yield as output the necessary process parameters, e.g., sputter mode (DC, RF, pulsed, etc.), gas pressure, magnetic field strength, electrical power input, etc. In reality, we are still far away from such a model. The model becomes more realistic when the input and output are switched (see Fig. 3.1). In this way, the parameters defining the deposition process are the input; the deposited film properties form the output. Such model is referred to as a “virtual sputter magnetron.” Efforts are made to develop such a simulation tool, e.g.,[1–4].