An epitaxial film has only one defined orientation and for good coalescence-promoting growth conditions is close to a single crystal.

Many properties starting with thermal and electrical conductivity depend on the degree of crystallinity. Anisotropic properties like piezoelectricity cannot be used at all if the material is polycrystalline.

Thank you Prof. Jürgen Weippert for answering.

Ok, if the epitaxial growth enhances the crystallinity that is good.

But is it possible to grow epitaxial thin film on polycrystalline substrate?

No, epitaxy literally means that the crystal structure of the substrate defines the structure of the layer. If the substrate is polycrystalline, you have no definition what epitaxial growth would be.

Thank you Prof. Jürgen Weippert for the clarification.

I have another query, in the thin film heterostructure as the substrate and the layer are have different crystal structure, then degree of lattice mismatching as well as on the smoothness of the substrate determines the formation of the thin film, so what fabrication should be good?

Actually I am working in the magnetoelectric composites and I want to prepare the thin film magnetoelectric heterostructure, so asking about the phenomena.

Well, a smaller lattice mismatch is usually favorable, but of course temperature stability is also an issue. For the iridium films which I grow, there are multiple available substrates. The yttria stabilized zirconia (YSZ) I use doesn't have the best room temperature mismatch, but since I need to grow a material which remains in its shape at temperatures >>1000K, the thermal expansion coefficients play into the whole thing and that favors YSZ.

Regarding smooth-/roughness: there may be correlations to your crystal quality, but there may be a quality minimum at some point, so it may happen that the epitaxy gets better if you manage to shrink the roughness to a certain value but when you shrink further, the growth gets worse again. Finding out the optimal point for your particular system is your job as a researcher and, trust me, it's a lot of lengthy work.

Thank you Prof. Jürgen Weippert for mentioning the possible points.