The gain falls off as the RMS surface roughness becomes a fraction of a wavelength; see figure 1 here: https://arxiv.org/pdf/1007.4600.pdf https://arxiv.org/p

The gain falls off as the RMS surface roughness becomes a fraction of a wavelength; see figure 1 here: https://arxiv.org/pdf/1007.4600.pdf https://arxiv.org/pdf/1007.4600.pdf So it would be interesting to see how that effects it. I suppose a crystal fractured along the lattice for an atomically smooth surface.