1. Magnetron sputter would allow you to sputter at lower sputtering pressures.

     and this can improve the density in the film. The deposition rate would be low,

     and the total stress build-up in the film can be high.

     Diode sputtering would require higher pressures to strike and sustain the plasma

    The sputtering rate can be lower, but then higher pressures can allow thermalisation     of the ejected specie, and decrease he stress in the films.

2. You mention compound targets. I don't know what is the composition.

     Definitely oxide targets are costly, and need carefully bonding.

    RF-diode process will uniformly sputter the target area, whereas magnetron will create those circular tracks (ineffective use of target area). This can have implications on the composition in the deposited films especially with magnetron sputtering.

    Magntron sputtering would therefore lead to non stoichiometric variations on the compound target surface much faster, than the rf diode sputtering.

3.  Although magnetron is claimed for higher sputtering rates at higher powers.

     But there is always a problem with compound targets (especially oxide).

     They tend to crack if you use higher powers (either diode/magnetron).

      You need well bonded targets, and a good target cooling system.

4.  The technique (diode/magnetron) basically govern the deposition rate.

      Deposition conditions such as sputtering pressure (low/high) would

      determine the total stress in the film.

      High/low deposition rate actually depends on the power applied.  For the same power magnetron will give higher rates. But in the case of compound targets which may involve a reactive sputtering gas, one has to really check how much is the different in the deposition rates.

5. To really imagine the situation, and decide whether diode/magnetron will be good, it depend a lot on the composition of the compound-target material.

     Dielectric optical coatings - I dont' have  much idea what is the compositon.

     whether nitrides, or oxides, or oxy-nitrides.

     Probably to make a stack of different material coatings, you need very thin layers.

     which should be uniform in thickness.  For very thin layers - even rf diode sputtering would do.

     Magnetron sputtering basicaly caters for very high deposition rates, and build thicker films. 

6.  Thickness uniformity may be good with rf diode.

     thickness non-uniformity prolems are encountered more in the case of rf magnetron sputtering.

     In my case I have seen I need a good 6inch target for achieving good uniformity on a 3 inch target.

      whereas it depends a lot on magnetron electrode design, and the substrate-target distance must be carefully worked out. 

K. Sreenivas

 The previous answer seems very wholesome. I would just add that substrate rotation can help with the thickness non uniformity even if you choose to go with RF Magnetron. And you must make sure to stick with slow ramping rates to prevent cracking of brittle dielectric targets (of course proper bonding is also very critical in this case).

Thank you Drs. Sreenivas and Sharma for your helpful advice.  I am using an MRC 8667 sputtering system with 8-inch cathodes set for RF diode operation.  I have sputtered an AlN target with Ar:N2 and would like to try a ZnS target with Ar.  It seems worth trying.  Changing to magnetron operation would require a new cathode and a new power supply.