They tried something vaguely like this with the big dome they tried to lower over it, didn't they? As I recall, a bunch of slushy hydrate ice started forming on the inside, clogging up the pipe and making the dome buoyant.
The dome was pretty small and made of metal. It seems like if the dome was very large and heavy it would work.
My point in all of this is to argue that it's possible to contain the spill using a brute force combination of the cap's size, weight and shape. And that the reason that the spill has not yet been fixed in not due to it being un-fixable, but likely other factors.