Oxide fibers are attractive candidates to reinforce metal, intermetallic and ceramic matrices. Sapphire fibers produced by the edge-defined, film-fed growth (EFG) method [1] seem to be the only singlecrystalline fibers commercially available. Y3Al5O12 (YAG)[2] and some alumina-based eutectic fibers are reported to be obtained [3, 4] by using either EFG or laser heated pedestal growth (LHPG) method. Both growth methods yield fibers of high optical quality; however, the production cost does not allow the fibers to be used in structural composites. The same is certainly true with regard to so called-pulling-down method [5]. Hence, the only commercially available oxide fibers are fine-crystalline alumina or mullite based fibers such as the Nextel series [6] that certainly have a rather low temperature limit of microstructural stability.

Under these circumstances, a method to produce single-crystalline and eutectic oxide fibers based on the internal crystallization, that is crystallization of an oxide melt infiltrated into continuous channels made in an auxiliary matrix, for example of molybdenum [7, 8], and then extracting the fiber from the auxiliary matrix ie by chemical dissolution of the latter [9, 10] looks a promising starting point for the development of a fiber technology. The method is relative simple and requires a relatively small energy input into a real process. The method allows obtaining a variety of oxide fibers, such as sapphire of a homogeneous crystallographic orientation [10], YAG [11], or alumina-YAG eutectic [9]. All the fiber materials mentioned melt congruently. Hence, broadening a family of the oxide fibers, which can be obtained by using the internal crystallization method (ICM), by including oxides that melt incongruently seems an interesting and practically important problem as there is a large number of the materials with interesting properties among such oxides.