If IBM can make a 150 GHz graphene transistor, why does Dr Phaedon Avouris, of IBM, say that it cannot replace silicon? Isn't clock frequency a measure of the transistor's ability to turn on and off quickly?
It's the raw manufacturability that silicon has. It's by far the easiest material to work with. It's abundant, cheap, and easy to obtain. It's got properties that lend it to easy junction doping and easy annealing. It's easy to deposit new layers on top of etched silicon. It's got a lattice that matches up nicely with the materials you'd want to use for contacts.
In short, the advantages of silicon aren't in it's ability to make transistors. It's in it's ability to make transistors while on a high volume assembly line.
It's going to be a while if we see whether graphene can match silicon for that.
Only time will show if dense packed transistors on a chip can be build cheap and reliably.
I've worked in the field of nanoscale electronic stuff, and I don't believe anything will replace silicon for a long time; at least not until we've completely run out of ways to do things better with silicon.
It's silicon that's the true miracle material -- it's versatile and cheap and amazingly easy to build stuff out of. We've been talking about replacing silicon for a while, but we keep coming back to it. It's all about silicon for the next couple of decades at least.