Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Computational optimization of the thickness uniformity of magnetron-sputtered optical layers by means of particle in cell plasma simulations

Comparison of theory with experiments
: Vergöhl, M.

Society of Vacuum Coaters -SVC-, Albuquerque/NM:
Society of Vacuum Coaters. 55th Annual SVC Technical Conference 2012 : April 28-May 3, 2012; Santa Clara, Calif.; Proceedings
Albuquerque: SVC, 2012
ISBN: 978-1-878068-32-3
ISBN: 1-878068-32-6
Society of Vacuum Coaters (Annual Technical Conference) <55, 2012, Santa Clara/Calif.>
Fraunhofer IST ()

The optimization of the thickness uniformity of high precision optical filters is often a critical and time consuming procedure. In this paper it is shown that the model of particlein-cell Monte Carlo simulation can be effectively used to develop and optimize the magnetron sputter process, especially the distribution of film thickness over the wafer. For the experiments, a new sputter system "EOSS" was used to deposit Nb2O5 single films on a substrate area of 200 mm in diameter. The thickness distribution of the films on the substrates is investigated. The deposition is based on a dynamic deposition process using a rotating turntable with a speed of 250 rpm. A cylindrical double magnetron is used for the layer deposition. It is shown by simulation that the use of cylindrical magnetrons should yield a very constant and stable thickness distribution during the lifetime of the target. The thickness distribution on the substrates was measured by spectroscopic ellipsometry. The homogenization is carried out by shaping apertures. The distribution of the particle flow from the cylindrical magnetron was simulated using the particle in cell Monte Carlo code (PIC-MC) developed at Fraunhofer IST. The resulting simulated thickness profiles are compared with the experimental data. Is it shown that by a careful implementation of the coating setup into the simulation, a good agreement in theoretical thickness distribution with the experiment can be obtained. Thus, the method turns out not only to be very efficient for a quick and accurate process optimization but also it allows a better understanding of the deposition process itself.