In commenting over at FCIWYPSC, I got to thinking about the issue of sub-disciplines. Materials science is already a small discipline, in terms of enrollment numbers. However, in terms of sub-disciplines, it's an enormous field. Here are some of the general areas researchers in materials science may work on:
Energy materials (oragnic or inorganic)
Computation (quantum through mesoscale)
Metallurgy (casting, welding, failure, alloying, forging...)
Amorphous Inorganic Solids (Glass, metallic glasses, sol-gels)
Theory (thermodynamics and statistical mechanics)
Within each of these branches are many sub-specialties. The diversity in research is in some ways fantastic, but in other ways, it can become problematic. Because students often learn both processing and characterization techniques, if advisor conflicts arise, changing groups means starting over. Skills are largely transferable, but results general are not. Leaving a group typically means leaving your results behind, because Sub-Discipline B doesn't care about Sub-Discipline A, and the relationship with Advisor A is probably such that you won't be publishing them (since you left A for a reason...).
On the other hand, if you decide you find fatigue tiring (no apologies for the pun), that diversity can be a good thing, allowing you to take your technical skills to something you find more interesting. As an undergraduate, I was able to be involved in a wider variety of research without leaving the comfort zone of my department right away.
How specialized is too specialized?