Résumé:

GaN nanowires (or nanocolumns) constitute a fascinating case of unstrained, nearly defect-free semiconductor systems. Their diameters (20-100 nm) are not small enough to induce 1D quantum confinement. Nevertheless some of their properties are clearly modified by the nanometric sizes. I will present the current state of understanding of (maybe) the most unimportant but certainly the most elusive -therefore interesting- of those properties : The vast majority of photoluminescence spectra of GaN nanowires reported in the literature present a doublet near 3.45 eV which is not observed (at least not with such an intensity) in bulk or epitaxial GaN. Since its first detailed observations in 2000, most discussions on this doublet tend to discard a number of hypotheses, but nearly none has ever risked any specific explanation, except that it is "most probably related to some surface defect". Yet a quite simple mechanism, involving the physics of donor-bound excitons has been suggested in 2009 by myself and co-workers at EPFL and I have gathered in the past years a number of quite supporting experimental and theoretical elements, if not definitive proof. I will propose these elements for an open discussion, along with the results of attempts by other groups to refute our hypothesis.