Iron in the kinks, say boffins: Wrinkly graphene could one day make computer chips 'smaller and faster' Wrinkles and bubbles in wonder substance graphene are showing electronic properties researchers say could one day help solve limits on current microprocessor designs. A team at the UK's University of Sussex, along with international collaborators, have shown that creating kinks in the structure of graphene can make …
I think that might be a bit a bold claim.
Technology has now reached 7nm when it comes to engraving electronic chips. This technique is not doing much better.
Of course, this is like arguing that next year's Olympic champion is only going to be 0.1% faster/better than last year's. It's the final 0.1% that is the hardest.
From the perspective of the Olympic sprinter, those fractions of a second matter. But most applications aren't a competition - if these sprinters were put to work delivering letters, the difference in speed between a gold medal champion and a sprinter who (merely!) qualified for the race would most likely go unnoticed.
The process names like "7nm" have absolutely no relationship to any physical dimensions of the transistor, and haven't for a long time. Bleeding edge transistors are something like 40nm across these days, so something only 2nm in width would be a massive shrink.
Of course, figuring out how to wrinkle exactly where you want, connect that wrinkle to all the other wrinkles via buried metal vias, and punch out tens of thousands of wafers full of them every week is another matter entirely.
"Of course, figuring out how to wrinkle exactly where you want, connect that wrinkle to all the other wrinkles via buried metal vias, and punch out tens of thousands of wafers full of them every week is another matter entirely."
True, of course. But then the first silicon wafers were a bit pricey with very low yields too. What's interesting with this technique is they are talking about making a certain fold or wrinkle act like a transistor. What other components might they be able to create with different shaped folds, wrinkles or combinations of folds and wrinkles. It might take 20 years or it may never happen. But at this stage of the game, it's something new to play with and that's how progress happens :-)
At the nanometre scale, currents must by definition be tiny to avoid destruction of circuit elements by overheating. Thus the quantum nature of electric current becomes increasingly important - a current being a stream of discrete electrons. Some years back a team at Cambridge made a transistor the could be switched by a single electron, but the question of course is "which electron, the intended one or a stray one?".
In the real world there is so much random electrical noise that a good 6.5 digit DVM typically picks it up at levels up to 1μV (i.e. the least significant digit is noise) under ideal conditions. That roughly represents a stray current pickup of about 10-13 amps into its 10 megohm input impedance - roughly one stray electron every couple of microseconds - each one a potential disruptor of circuit function at this level of miniaturisation. Depending on circuit impedance, things can be much worse. Consequently, there are intrinsic practical lower limits to circuit element size in the real world even if theory allows them to be smaller.
"Consequently, there are intrinsic practical lower limits to circuit element size in the real world even if theory allows them to be smaller."
Maybe at such small scales but vastly higher speeds, you must repeat operations many times and take a statistical sample to get the right result? Some sort of complete paradigm shift in how operation are carried out? Or something more akin to analogue computing? Or it's a complete dead end, superseded by quantum computing before it's even born? From the little I understand of quantum computing, that's also a paradigm shift in how calculations are done.
I wonder we will actually be using in 10-20 years time.
This seems like an interesting development but it's years away from commercial availability. BTW, Swiss boffins just announced that they have developed kagome graphene which also may have semiconductor applications.
Right now, to keep Moore's Law going, 3-d chips seem closer to reality.
Still, a pint for the kinky boffins.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
