Experimental and theoretical investigations of Mg2+-doped single-crystal fibres of lithium niobate

Pure and Mg-doped single-crystal LiNbO3 fibres (molar concentrations up to 5 mol.% MgO) have been investigated via X-ray diffraction and micro-Raman scattering experiments in order to compare their structural (cell volume) and dynamical (modes) properties with those of bulk material. Novel first-principles calculations have been performed to ascertain definitely the frequency assignments of TO and LO zone-centre phonons involved in Raman spectra. Three main differences between fibres and bulk are (i) the fibre is not spatially homogeneous at low doping rates (macroscopic homogeneity is achieved when the doping rate reaches 5 mol.%); (ii) the doping dependence of Raman frequencies and linewidths is not the same for certain modes; (iii) the cell volume in the fibres decreases with increasing doping rate in the fibre in contrast with previously reported data on the bulk. Experimental results indicate the presence of two successive regimes when the Mg doping rate is increased up to 5%. This behaviour, already noticed though apparently to a lesser extent in the bulk, is discussed in the framework of a consistent double-vacancy model that encompasses previously proposed models as particular cases. This confirms that Mg2+ ions replace antisite Nb ions in a first stage (below 2% in the fibre) while, in a second stage (from 2% to 5% doping rates), Mg2+ ions are accommodated in remaining Li vacancies until these are completely filled. The reordering that ensues is undoubtedly linked with the recovered homogeneity of the fibre, the disappearance of photo-refractive effects and the considerable improvement of the resistance to optical damage observed for the 5 mol.% Mg doped crystal.

► Pure and Mg-doped LiNbO3 fibres: structural (X-ray) and dynamical (Raman) properties.
► Comparison between fibre and bulk properties.
► First-principles calculations of LiNbO3 zone-centre phonons.
► Discussion of results in the framework of a double-vacancy model.