Dopants used to be there just for their doping. But stress is now an important aspect as well, which means the dopant atoms must be sized appropriately as compared to their silicon hosts. This has worked for p-type, where compressive stress is desired. Germanium, which is larger than silicon, compresses the silicon, increasing hole mobility.
n-type should be the reverse: tensile stress is needed, meaning smaller dopant atoms. Phosphorus and carbon are both smaller and can work. Sounds simple, right?
Well, apparently not so. The n-type dopants have a tendency to migrate, and so far increased border security hasn’t worked. OK, kidding. About the security, that is. The migration has remained to be solved.
At Semicon West, Applied Materials announced that they had found a way to create a stable n-type epi layer. How do they manage it, you ask? Keep asking… they’re not telling. There was a mention of millisecond anneals helping to tweak any vagabonds before they get too far. And whatever they do sets up a strict thermal budget, although not so low that it affects the back-end interconnect processing.
Details aside, if this is all working as promised, then we have more control over how to optimize the performance of n- and p-type devices. You can read more in their release.