<t>Dominic is onto something here - as a research area formal methods in software engineering is not the most fertile and does not intersect much with more interesting topics. Can't comment on the second one, but if you must get your PhD from a CS department, something like machine learning or relat...