Not quite sure why you applied PCA after developing your clusters? Would it make more sense to reduce dimensionality via PCA and then performing clustering? Asking a fellow IO shrink making a transition to Data Science :)