This is complicated. At low current there is no photon generation because all you get is leakage. Then the generation increases with current, while a lot of pho

This is complicated. At low current there is no photon generation because all you get is leakage. Then the generation increases with current, while a lot of photons are lost to reflection and recombination and the leakage increases with thermal recombination. As the current increases, the carrier population densities become inverted and then most of the generation is through stimulated emission, which is much more efficient, and the reflections actually help drop the critical current to get to this point. The dream structure would be a very small junction region with high optical and electrical confinement, low defects, and high thermal conductivity. I would expect this to be a SiC wire with a thin strained CVD GaN crystal layer in a Bragg reflector outer shell. Just to be clear I know nothing of the last 15 years of progress.