You could have two layers, one for SiO2. But you might as well make the substrate SiO2. The BP method will not properly model the influence of an Si substrate anyway. On the other hand, your design should isolate the optical field from the Si substrate (and from the electrode), so this limitation doesn’t matter.
In the model, the electrode is simply a place where the applied potential takes the prescribed value. The electrode does not contribute anything to the model besides this, it doesn’t consider anything about the material properties of the metal except that it should be an equipotential at radio frequencies. In this sense, the electrode is already “any metal”. You can put those electrodes wherever you like, and they can have any shape.