Mechanism for the alpha -> epsilon phase transition in iron

The mechanism of the $\alpha$-$\epsilon$ transition in iron is reconsidered. A path in the Burgers description of the bcc/hcp transition, different from those previously considered, is proposed. It relies on the assumption that shear and shuffle are decoupled and require some peculiar magnetic order, different from that of $\alpha$ and $\epsilon$ phases as found in density-functional theory. Finally, we put forward an original mechanism for this transition, based on the successive shuffle motion of layers, which is akin to a nucleation-propagation process rather than to some uniform motion.