For circular magnetrons the uniformity on a substrate is determined by the coating flux distribution from the target surface. Geometry is a major factor as the target diameter relative to the substrate size, and target to substrate separation will have major influences. Generally for static coating of substrates the target needs to be much larger than the substrate. Typically a 10" target is required to coat a 6" substrate with good uniformity.
The magnetic field however can be tuned to 'fine-tune' the flux distribution. Below is an example where a different magnetic design can improve the coating uniformity from 4% to 1% for the same cathode and substrate geometry.

Gencoa have developed a program that can predict the coating uniformity on a substrate. The program calculates the sputtering probability across the substrate and hence a model of the coating flux from the target material can be generated. Correction factors can be added to the model to account for different target materials, since the angular sputtering dependance varies for each material. The program uses data from the finite element magnetic field modelling and can consider many factors including whether substrates are on or off axis, rotating or not and the required height of the substrate from the source. Also, for a given uniformity requirement, the optimum position of the substrate can be advised. Once the optimum position is found, the deposition rate in this position may be low, especially if offset distances are large. One method of increasing the rate is to tilt the magnetron in that position.
If the uniformity cannot be achieved then some manipulation of the magnetic field may be required. By using the simulation software it is possible to predict the corresponding change in uniformity. In this way the uniformity can be tuned to a specific application requirement.
12" High Uniformity circular
The figures below show a partially eroded target, 18mm thick, for a 12" high uniformity source and the corresponding actual erosion plot. The result of the erosion simulation using in-house software can also be seen, once again demonstrating good correlation between theoretical and experimental results.
|