LiNbO3 is one of the most common perovskite-type crystals for optical applications. This a typical non stoichiometric material melts congruently at 48.6 mol % Li2O content and can be grown easily by Czochralski method from its melt. The congruent composition is a serious drawback causing a defect structure with about 1% anti-site Nb5+ ions in the lattice. Therefore stoichiometric crystals (50 mol % Li2O) are expected to show improved performance for a number of applications. In this structure the amount of anti-site Nb5+ is below 0.001%which results in the blue shift of the UV absorption edge, the narrowing of several spectral lines (EPR, NMR, Raman, IR OH- vibration lines , Cr3+luminescent line of the doped crystal.) An important reduction of the electric field (200 V instead of 2kV) required for ferroelectric domain inversion in the stoichiometric crystal makes easier the production of periodically polarized structures for quasi-phase matching.
Several methods exist for the preparation of stoichiometric LiNbO3 single crystals. The one which yields the composition closest to 50% Li2O is the high temperature top seeded solution growth method from the K2O- Li2O-Nb2O5 ternary mixture. Part of this system has been investigated by several authors who concentrated on the formation of the tetragonal tungsten bronzes. So far no systematic research was done on the formation of LiNbO3. Our aim is to explore the limiting chemical and thermal conditions of the growth of stoichiometric LiNbO3 single crystals from this three component solution.
For the construction of the phase diagram, besides the conventional thermal analytical and X-ray phase analysis, we used experimental crystal growth as well, for the identification of the consecutively crystallizing phases. With the combination of these methods we are able to determine the lay-out of the isotherms and to localize the singular points of the thermal surface representing the phase relations.