GLOBALFOUNDRIES Details Advanced Technology Aimed at 22 nm and Beyond

[btarunr]

GLOBALFOUNDRIES today described an innovative technology that could overcome one of the key hurdles to advancing high-k metal gate (HKMG) transistors, bringing the industry one step closer to the next generation of mobile devices with more computing power and vastly improved battery life.

The semiconductor industry is celebrated for overcoming seemingly insurmountable odds to continue the trend toward smaller, faster, and more energy-efficient products. Performed in partnership with IBM through GLOBALFOUNDRIES' participation in the IBM Technology Alliance, the new research is designed to enable the continued scaling of semiconductor components to the 22 nanometer node and beyond.

At the 2009 Symposium on VLSI Technology in Kyoto, Japan, GLOBALFOUNDRIES reported the first demonstration of a technique that allows the equivalent oxide thickness (EOT) in a high-k metal gate (HKMG) transistor to scale down to well beyond the level required for the 22nm node, while maintaining a combination of low leakage, low threshold voltages, and superior carrier mobility.
"HKMG is a critical component of GLOBALFOUNDRIES' technology roadmap," said Gregg Bartlett, senior vice president of technology and research and development. "This development could eventually provide customers with another tool to enhance the performance of their products, particularly in the fast-growing market for ultra-portable notebooks and smartphones with extended battery life. In conjunction with IBM and the alliance partners, we are tapping our global knowledge base to develop advanced technologies that will allow our customers to stay at the leading edge of semiconductor manufacturing."

To maintain the switching precision of a HKMG transistor, the EOT of the high-k oxide layer must be reduced. However, reducing the EOT increases the leakage current, which can contribute to an increase in the power consumption of a microchip. GLOBALFOUNDRIES and IBM have developed a new technique that overcomes this barrier, demonstrating for the first time that EOT scaling to well beyond the 22nm node can be achieved while maintaining the necessary combination of leakage, threshold voltages, and carrier mobility. The results were successfully demonstrated through fabrication of an n-MOSFET device with EOT of 0.55nm and a p-MOSFET with EOT of 0.7nm.

View at TechPowerUp Main Site

 

[PP Mguire]

I think IBM is underrated compared to Intel.

 

[erocker]

I can't wait to see what a HKMG AMD processor can do.

 

[lemonadesoda]

I can't wait to see the market flooded with 40nm and even 32nm. Let's get that show on the road before we start braggin 22nm processors. But if IBM can goosehop an 11nm ARM7/8 processor within 18 months, they will have a knockout offering. Even better if they can CISC something for AMD. It might give them a fighting chance against the mighty Intel.

 

[happita]

The jump from 45nm isn't to 40, pretty sure its 32nm, then 28. But hearing about 22nm this early is ridiculous. Maybe this will get AMD in the state of mind Intel was in when they intro'd their Core Duos. All in all, this just makes for better competition which I am all for.

 

[cray86]

After 10nm quantum tunneling becomes a huge issue (ie voltage leak).

Now, this probably won't be an issue, but there will be no more traditional die shrinks after 2016-2020 because we'll get to the point where gates are insulated by less than a dozen molecules. You can't control electrons in that kind of an enviornment without magnetism, and that doesn't make for controlled voltage either.

The age of nanomachines and quantum computers (if possible this soon) is only 10-15 years away! I'm excited for that

 

[TheLaughingMan]

Not sure.

currently Intel is going for and will hit the 32nm processor segment. AMD released information a while back stating that they were going to try and skip straight to 28 nm, but said nothing else about that plan since then.

These 22 nm chips are not desktop processors and have no where near the power or complexity. This post is about mobile processors like the ARM chips in everyone's smart phones. So basically IBM is trying to enter the mobile chip market with this move and whil Global Foundaries is with AMD now, they have a none exclusive contract that allows them to work for others as well. I doubt that this will affect AMD's current plans in the slightest because this would be tech that only IBM should have access to at the moment.

 

[happita]

How does this not affect AMD in any way? Last I checked, AMD's processors are not exclusively used in only desktops/laptops. Their technology is used across many different platforms including computers, mobile phones, etc. And IBM has an interest in AMD, otherwise they wouldn't have helped them stay competitive against Intel when AMD shares were plummeting to say the least. This probably wouldn't affect them right now, but later on, I'm sure they will reap the benefits when their processors are manufactured using this 22nm tech.

 

[Frogger]

mabe not the size of the chip But the cost will lead to sizes leveling off???



"By 2014, however, the high cost of semiconductor manufacturing equipment will threaten Moore's Law, "altering the fundamental economics of the industry," according to a report released on Tuesday by iSuppli.

"The usable limit for semiconductor process technology will be reached when chip process geometries shrink to be smaller than 20 nanometers (nm), to 18nm nodes," said Len Jelinek, director and chief analyst, semiconductor manufacturing, for iSuppli. "At those nodes (levels), the industry will start getting to the point where semiconductor manufacturing tools are too expensive to depreciate with volume production, i.e., their costs will be so high, that the value of their lifetime productivity can never justify it."

here

 

[PCpraiser100]

Ooooo, a Linux-powered wristwatch.

Seriously, the technologists were right on the Discovery Channel. Maybe Technology will be so small that we will need holograms to display it.

 

[btarunr]

I can't wait to see what a HKMG AMD processor can do.

Low TDP at best, bad overclocker at worst.

 

[Hayder_Master]

Low TDP at best, bad overclocker at worst.

did you think about overclock if you have CPU like this

 

[lemonadesoda]

The jump from 45nm isn't to 40, pretty sure its 32nm, then 28. But hearing about 22nm this early is ridiculous. Maybe this will get AMD in the state of mind Intel was in when they intro'd their Core Duos. All in all, this just makes for better competition which I am all for.

90nm to 65nm to 45nm to 32nm to 22nm to 16nm. Note the "halving" of lengths every two jumps. Actually, the area is halving in each jump. These are the new-process technologies, referred to as "technology nodes".

55nm, 40nm and 28nm are called "half-nodes" and reflect the ability to reduce scale through lithographic and process improvements without changing the underlying gate technology.

Have you noted that there are GPUs made on 55nm and 40nm "half-node" points? CPU's could also be done, in theory, on these half-nodes too. Although the lifecycle of a CPU is longer than a GPU so usually there is no need to reduce CPUs to half-nodes on existing CPU architecture. It could be done however esp. in low power critical situations. The Atom CPU and ARM7/8/9 are prime candidates for half-node fabrication.

 

[LaidLawJones]

This all sounds great, but aren't they having a hard time with the current 40nm process? I thought the lack of 4770's was due to low production and/or high failure rate. It would be great to hear that they have fixed all the problems and that production is expected to increase significantly, or are we going to have the equivalent of a M.S. hardware manufacturer? Inside G.F. boardroom, "Yeah, the 40nm has a lot of problems but we can skip it going right to 32nm and if that doesn't work we'll add a few patches and call it 22nm".

 

[lemonadesoda]

What you say is quite true... but the truth lies in that 40nm is a scale of 45nm and is causing all kinds of problems, whereas the 32nm technology is a *new* technology, involving different production methods and silicon substrates; the so called "high-k dielectric metal gate process". I'm no expert, so I'll leave any further interest you have in the topic to this article: http://en.wikipedia.org/wiki/45_nanometer

 

[1c3d0g]

About damn time someone took over the lead from TSMC, which, dare I say has been doing a shoddy job lately with its fab process... :shadedshu

 

[TheLaughingMan]

You are right.

How does this not affect AMD in any way? Last I checked, AMD's processors are not exclusively used in only desktops/laptops.

I had almost forgot that most LCD TV use some for of ATI GPU's and other random stuff.

While you are correct, I don't think this is the a new tech. I think this is the same die shrinking techniques used to bring down the core size on desktops being moved over to mobile processors. While the size seems unrealistically small, think about this. The chips being built at this size have like 1/4 the number of transistors as a AMD or Intel Quad core processor at 45 nm.

 

[devguy]

This all sounds great, but aren't they having a hard time with the current 40nm process? I thought the lack of 4770's was due to low production and/or high failure rate. It would be great to hear that they have fixed all the problems and that production is expected to increase significantly, or are we going to have the equivalent of a M.S. hardware manufacturer? Inside G.F. boardroom, "Yeah, the 40nm has a lot of problems but we can skip it going right to 32nm and if that doesn't work we'll add a few patches and call it 22nm".

It's TSMC that is having issues with the 40nm process, not GlobalFoundries.

 

[LaidLawJones]

Sorry, my mistake.