We have previously developped a theoretical optical waveguide using a III-Nitrides substrate, but we nevertheless didn't find a lot of literature about this kind of waveguides.
Unless you are using an extremely lossy material, I think your efficiency comes down to coupling light into and out of the waveguide, not necessarily the waveguide itself. Also, if there are any tight bends or other discontinuities, they may effect your efficiency as well. Is this what you mean by "efficiency?"
Thank you dear colleague Raymond, but what I mean as eficiency is the light transport inside the waveguide.
Therefore our goal is to conceive a material continuum replacing electrical or air connections between an emitter and a receptor by an optical waveguide between photodevices.
We didn't try to investigate Silicon just because we hope to work with the III-Nitrides which are of direct bandgap, more suitable in the optoelectronics. Silicon has already been used as a waveguide in the SOI devices.