Daniel S. Main¹, Thomas G. Jenkins¹, Joseph G. Theis², Gregory R. Werner², Scott E. Kruger¹, John R Cary^1,2
1Tech-X Corporation, Boulder, CO
²University of Colorado, Boulder, CO
Magnetron sputtering devices have a wide range of uses in the coatings industry, where they are used for optical coatings, metallization in integrated circuits, and coatings for wear resistance. In most applications, the desired figure of merit is the deposition rate and uniformity onto the substrate. However, computing the deposition rate and uniformity correctly requires knowledge of the sputtered neutrals from the cathode, which in turn requires a plasma model that can correctly compute the sheath and ion distribution function impacting the cathode. For this reason, a kinetic approach is required. In this talk we present results from particle-in-Cell/Monte-Carlo collision (PIC-MCC) models in a 2D cylindrically symmetric system. The PIC-MCC simulation is expensive, but we have implemented an energy conserving force interpolation algorithm that considerably reduces the required computing resources by ~ 40 relative to more commonly used methods. In this talk we present results of a benchmark study that demonstrates agreement within ~ 15% between an under-resolved and well-resolved simulation study. We then present a workflow that allows for fast calculation of the deposition profile. In step 1, we rapidly compute the neutral flux emitted from the cathode in a PIC-MCC model using the energy conserving method. In step 2, we then fit this flux to an analytic function and ballistically transport the sputtered neutrals to the anode. In step 2, we model collisions between the sputtered neutrals and background gas but do not compute the self-consistent fields. Therefore, the ballistic transport can be computed rapidly (typically in 10’s of minutes). By performing coarse resolution energy-conserving PIC-MCC simulations coupled with the ballistic transport model, we show that a fully kinetic calculation of the deposition rate and profile is possible using widely available computing resources.