Sputter yield measurements to evaluate the target state during reactive magnetron sputtering

The sputter yield and discharge voltage of fourteen target materials (Al, Cr, Cu, Mg, Mo, Nb, Pb, Ta, Ti, V, W, Y, Zn, and Zr) have been measured during reactive sputtering in argon/oxygen mixtures. The obtained oxide sputter yields strongly differ from the published data based on ion beam experiments. A second observation is that based on the discharge voltage behavior observed during target oxidation, the materials can be subdivided into two groups. For the first group, the discharge voltage increases on the target oxidation, while for the second group the opposite behavior is observed. Both observations are explained based on a model that accounts for oxygen implantation into the target, preferential oxygen sputtering, and additional oxygen loss mechanisms such as outdiffusion. The difference between both groups can be explained from the oxygen fraction in the gas discharge required to fully oxidize the target surface. This required fraction is lower for the first group, and higher for the second group, than the oxygen fraction when the reactive sputter process switches into poisoned mode. The required fraction is mainly defined by the oxide sputter yield. The lower sputter yield as compared to literature values can be attributed to implanted oxygen that dilutes the formed oxide and/or continuously replaces sputtered oxygen.The sputter yield and discharge voltage of fourteen target materials (Al, Cr, Cu, Mg, Mo, Nb, Pb, Ta, Ti, V, W, Y, Zn, and Zr) have been measured during reactive sputtering in argon/oxygen mixtures. The obtained oxide sputter yields strongly differ from the published data based on ion beam experiments. A second observation is that based on the discharge voltage behavior observed during target oxidation, the materials can be subdivided into two groups. For the first group, the discharge voltage increases on the target oxidation, while for the second group the opposite behavior is observed. Both observations are explained based on a model that accounts for oxygen implantation into the target, preferential oxygen sputtering, and additional oxygen loss mechanisms such as outdiffusion. The difference between both groups can be explained from the oxygen fraction in the gas discharge required to fully oxidize the target surface. This required fraction is lower for the first group, and higher for the second group, than the oxygen fraction when the reactive sputter process switches into poisoned mode. The required fraction is mainly defined by the oxide sputter yield. The lower sputter yield as compared to literature values can be attributed to implanted oxygen that dilutes the formed oxide and/or continuously replaces sputtered oxygen.A