Finfet's successor is here, going to 7nm!

[Bsmith]

This size is of transistors is amazingly small if you ask me, unbelievable that they can do this, while they used to think that 14nm was the limit a few years back, with even more technologies showing up i wonder how small we actually are able to go!

translated from dutch.

The Belgium Research institute Imec showed a technology of kyoto transistors that could be produced at the size of 7nm.
These transistors are a further developed version of the finfet-technology introduced at 22nm.
.At the VLSI 2015 symposium in the Japanese city of Kyoto showed two types of transistors that could be used at a 7nm-node. The researchers showed transistors with channels of silicongermanium, where the gates get formed by a process called "RMG-HK". They also produced transistors which didn't use FinFet technology to produce the gates, but instead they used channels of nano fibers working with the gate wrapped around those fibers. To do this a thin layer of silicon get's sacrificed to make space for the gate that then lays around the channels, this should realize just like FinFet s bigger area of gates on smaller chips to make them(the transistors) more energy efficient, faster and smaller.

The GAA-technique has been demonstrated on both a cmos-process and so-process. besides that the researchers also showed FinFets a way where the gate-fins get produced through a combination of self-alignment and "double patterning" technologies, which where coated with quantum wells of strained germanium, which should lead to better performance, in the lab those transistors gained a 40% increase compared to older transistors making use of the strained germanium.

source:
Tweakers(dutch)
Imec's own article (tnx @Minibios )
http://www2.imec.be/be_en/press/imec-news/vlsi2015-germanium-multigate-gaanw-finfet.html

Despite still being in the lab and probably far away from CPU production nodes this seems very promising, before people start ranting/wondering about words in the quote, I admit there are words in there that came directly from the source and are terms or a abbreviation which i couldn't translate, i did my best on it but it's not so easy to translate a partly technical dutch into english which everybody can understand.

but basically, smaller chips, more performance! yeah!

[Aytex]

Japanese city of Kyoto 

skrullex pls no

[minibois]

This is an article from Imec themselves (in English)

http://www2.imec.be/be_en/press/imec-news/vlsi2015-germanium-multigate-gaanw-finfet.html

[minibois]

Japanese city of Kyoto 

skrullex pls no

... wut?

[Kamina]

The end of silicon is nigh.

[Aytex]

... wut?

skrillex-kyoto

[Bsmith]

This is an article from Imec themselves (in English)

http://www2.imec.be/be_en/press/imec-news/vlsi2015-germanium-multigate-gaanw-finfet.html

Thanks! Let me put it in OP, will do a nice edit once i can get my laptop again(just left home)

[minibois]

skrillex-kyoto

I knew the song, but that has nothing to do with the article

That's like me screaming "Nintendo!" because their HQ is there.. Or if I scream "the Kyoto protocol!"

[slightlyjaded]

Curious about overclocking something built on such a small process. When you start raising voltage, you start increasing the range a spark can jump. It's nice tech, but like all such I'll wait till it hits production before I jump on the bandwagon. I can see this being incredibly useful in phones, wearables, and various applications that don't exist yet as size is still an issue.

[WereCat]

The end of silicon is nigh.

Yea.. but it lasted quite long. I mean.. if you look in the past how fast the technology was changing and then we went on to silicon and we basicaly got stuck on it for ages. Time to move on further.

I wonder if the stacking will get more popular before graphene or if it will arrive all at once. Stacking will be maybe problem on silicon though.. temperatures (probably not very suited for GPUs and CPUs cores).

[Bsmith]

The end of silicon is nigh.

I wouldnt say that too fast, since this technology is only in it's childshoes at the given moment it might take a while, unless a big company like Intel suddenly jumps aboard, then it might go much faster.

I'm pretty sure this might be here earlier then graphene though, since graphene is pretty wildly spread at the moment and is the wonder material everyone is experimentating with, spreading the attention and possible fundings quite alot.

[Bouzoo]

Maybe we'll see this in consumer products... In 2020?

[MrDynamicMan]

Maybe we'll see this in consumer products... In 2020?

I'd day next 3 years... By the time this goes into the consumer space I'm sure silicone will be an outdated medium. Grapher or carbon nanotupes or some such like that.

[Bouzoo]

I'd day next 3 years... By the time this goes into the consumer space I'm sure silicone will be an outdated medium. Grapher or carbon nanotupes or some such like that.

Hm, we're on 22nm for 4 years now and just got to 14nm. Skylake is gonna be also 14nm, while Cannonlake should be 10nm in 2017. I'm not sure we might see it in 3 years.

[Guest Strangerbob]

Maybe we'll see this in consumer products... In 2020?

 

maybe well start seeing 28nm finfets by then for gpu`s.... by the looks of how things are going

[cesrai]

Yes, very promising.

[LukaP]

O-gate transistors were in plan for <14nm since that was in plans. the only way to reduce leakage and quantum tunelling is through adding more gate around the n-doped channel. 

[Commander Llama]

Wow, 7nm?  How is quantum tunnelling not happening all the time?  Are the new materials somehow resistant to it or something? I'm a bit layman on this, but thought it was already becoming an issue and by <10nm it would be a huge issue.

[Thony]

Future is so exciting ecerything is getting smaller (except smartphone screens lol).

I still wonder how cannonlake (or is it one after?) will deal with no silicon. Yay.

[azurezeed]

let's see how intel deal witn 7nm

[patrickjp93]

I wouldnt say that too fast, since this technology is only in it's childshoes at the given moment it might take a while, unless a big company like Intel suddenly jumps aboard, then it might go much faster.

I'm pretty sure this might be here earlier then graphene though, since graphene is pretty wildly spread at the moment and is the wonder material everyone is experimentating with, spreading the attention and possible fundings quite alot.

Intel has confirmed silicon is gone for 7nm, though many suspect it will use a III-V material for the new process. https://en.wikipedia.org/wiki/List_of_semiconductor_materials