That may depend on the environment you use and for how long you treat the sample. Kumar et al., Current Appl. Phys. 14:3 (2014) 491-5 showed dipping GaN samples in ammonium sulphide diluted in aqueous and alcoholic solvent produced increased Schottky barrier heights at room temp. For long, high temp. anneals S may diffuse into the GaN and eventually through the GaN into the Si substrate and cause reactions there.
It should be tested experimentally, of course, but I can estimate a propagation of S into GaN, especially for GaN/Si sample. In this case the red missmatch (between GaN and Si) is really high, this leads to bad crystalline structure of GaN epi-layer with a lot of disslocations.
Of cource, the thickness of GaN, the time of threatment and the used atmosphere define the final result.