Target poisoning during reactive magnetron sputtering: Part III: the prediction of the critical reactive gas mole fraction

Abstract

To predict the critical reactive gas mole fraction, needed to induce an abrupt change in the reactive magnetron sputtering process, a simplified scheme is proposed. The scheme is based on the equations proposed in part I and part II of this paper. A quite accurate prediction can be made using a small number of parameters. The pumping speed and the discharge current are the key experimental parameters, which define whether the gettering process or the target poisoning dominates the abrupt changes noticed during reactive magnetron sputtering. In this first case, i.e. when gettering dominates, the prediction is straightforward and the critical mole fraction can be calculated from the pumping speed, the current, the sputter yield of the target material and an average value for sticking coefficient of the reactive gas on the deposited target material. In the latter case, i.e. when the poisoning mechanism dominates, the prediction is a little more complicated, as one needs the reaction probability of the implanted reactive ions. It is shown that this can be estimated from the electronegativity difference between the reactive gas atoms and the target atoms.