A monkey was once tried and convicted for smoking a cigarette in Indiana
SILICON CMOS is in for a treat: a plethora of new tech is in the works to improve on current CMOS logic and memory performance according to Sematech vice president, Raj Jammy.
Speaking at Semicon's TechXPOT (which was sadly not as sexy as it sounds) Jammy gushed about exciting innovations in high-mobility graphene channels, double-gate transistors built on SOI substrates - known as finFETs - complete with III-V materials and gates built around nanowires. All thrilling stuff and apparently the best way to make silicon sing with power and performance beyond 22nm.
Singing the blues, Jammy wailed that too many people were "entrenched in the silicon world" and needed redemption from the oppressive thinking chaining them to current CMOS transistor scaling methods.
"We need disruptive materials and technologies," said Jammy, calling for a semi revolution as R&D progresses to 22nms and beyond.
Germanium, which believe it or not isn't a flower your grandmother keeps on her porch, is proving exceptionally useful in positive channel field effect transistors (pFETs), whilst indium gallium arsenide (InGaAs) purportedly works wonders on the nFET.
Jammy told the audience how exciting memory R&D was becoming even in its old age, moving away from charge storage - which becomes virtually impossible the smaller devices become - towards phase-change memories (PCRAMs) and metallic resistive RAMs (ReRAMs). The only problem is that it's still unclear how well ReRAM will work below 20nm.
Adding his notes to the symphony of silicon wonders, Schubert Chu of Applied Materials discussed new breakthroughs in embedded silicon carbon (eSiC) and embedded silicon germanium (eSiGe), which hasn't proved easy because the pesky carbon atoms fidget about. Still, "silicon carbon is on track to be adopted at the 22nm generation," said Chu.
Another technology attracting a fair bit of attention is zero-leakage nanoelectrical-mechanical system (NEMS) devices which can exhibit instant on and off and are so safe they can even be used inside nuclear power plants.
But it wasn't all sweetness and light at silicon central, with Jammy offering stern advice to industry players about keeping product pricing in check. "When we hear that it may cost $80 million for an EUV scanner, we are not going in the right direction on costs," he tutted.
Well, no wafer no cry. µ
I'm puzzled. You claim the story is about advances in silicon processing, yet all the technologies quoted are not based on silicon at all, except maybe SOI, but the salient feature of SOI is not the silicon but the insulator base material.
It's time to move away from Si, it has had its day in the sun, there are much better materials.
BTW, I've figured out the trick behind the black on black on black control panel in Disaster Area's sunship. It's graphene on carbon, natch, with the displays using black OLEDs. I plan on incorporating the same technology into the next nuclear plant control panel I design. Should be quite a hit with the NRC, not to mention the operators.
Back in the day I cracked more then one OC transistor :)
http://www.donberg.ie/pics/o/oc_75.jpg
Good to see that old materials comes back in a new way.
At 90 nm, barton core has from 12,000,000 to 30,000,000 transitors. Now at 45 nm or half, trannie count is near 1 billion or more, up to 100 times more.
From 45 nm to 22 nm is half & from billion to ?100 billion transistors. it policy to make it so complex, little guys fall out.
Large Transistor Plantations of Old South, Come to Mind.Impossible?
Yet, if semicon right, theres still another doubling possible to 12 nm. 3 terraflops of trannie beds. Is That Practical Number?
5 nm is size of filter that oil comapnies use to convert heavy hydrocarbons into worthless gasoline. So You might end up with liquid trannies. thats about 13 years of developement, if engineers could keep pace. Then BOTS Will Rule.
Kill dave, Kill dave. an old adage with large scale twist.
13 More YEARS of DRASHEK, Extreme Luxury. Then What? 13 nm cell run on Body Heat. Hummm, BOTS ?Lose.
'Nuff said. If he really is a bot, I want to know anything and everything about it. If he's a real person, he's still awesome.
Wow.
ROFL, Drashek RULZ!
On the epic side of un-uniformity and aforementioned acronyms.