I don't know what it is about the run up to Christmas this year, but it seems like every researcher in the world has a new scheme
for ultra-fast processors that they're just itching to tell us. The latest plan comes courtesy of Princeton University
boffins who have developed a system of putting transistors on a 'graphene' substrate – carbon formed into a honeycomb lattice.
Graphene-based processors promise – you guessed it – massive advantages in performance and power usage over traditional silicon-based chips. The reason we haven't been using it in the past is due to the difficulties in arranging graphene into a layer which is the right size to make chips out of. A standard silicon wafer used in the production of processors is between eight and twelve inches in diameter, and currently they can't get a slab of graphene anywhere near that.
Stephen Chou, professor of electrical engineering, reckons his team has the solution to the problem: put the graphene “only in the active areas of the chip
”, which neatly side-steps the whole too-small problem.
The actual process used is described by Chou as being “like printing
”, where a special stamp is pressed against a graphite block and then used to transfer layers just a few atoms thick to the wafer on which the transistors will be built. Like potato-printing
, but on a teeny-tiny scale and with extremely expensive potatoes.
The researchers think the new technology could rapidly find its way into devices which “require high-power output
” such as mobile 'phones. As usual, the sticking point is in scaling the technology to a level at which commercialisation becomes possible.
Do you think the technology holds promise, or will you believe it when you see it? Let us know via the forums