That's how I believe to understand the abstract you linked
http://www.ncbi.nlm.nih.gov/pubmed/150806670.5% Cr
2O
3 in Al
2O
3 is what makes the final product, ruby. Cr makes the colour centers that give the colour or possibly lase.
Instead of melting (or dissolving if possible) the ruby constituents, then solidifying it, the authors added a flux that helps melting at a lower temperature, and chose MoO
3 because it evaporates easily. Upon evaporation of the flux, ruby is no more a liquid at the chosen temperature but a solid - it crystallizes. Well done.
The authors did not control the crystallization at all, so many small (1.7mm) crystals formed spontaneously. A slower evaporation of the flux (or more usually, cooling of the pure ruby melt) would let pass the liquid-to-solid transition more gently, hence have fewer seeds from which crystals grow, and this would produce bigger crystals.
Some uses desire single crystals of high quality (lasers, particle detectors, semiconductors) and these don't rely on random seeds. The process provides a seed and lets the crystal grow in one direction under controlled conditions, generally a temperature gradient. These are the Bridgeman and Czochralski methods.
Alternately, the crystal can begin spontaneously at the coolest end of the target part, pass through a narrow point that selects one crystal from the many spontaneous seeds, the proceed with the desired shape. This is done for gas turbine blades of nickel alloy, presently single-crystal, which reduces creeping at heat.
Among these possibilities, I see Czochralski that could be adapted to the vaporizing flux to make big crystals of ruby and more.