GLOBALFOUNDRIES on Track to Deliver 7nm FinFET Technology

Megalith

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
GLOBALFOUNDRIES is announcing the availability of its 7nm Leading-Performance (7LP) FinFET semiconductor technology, which offers a 40-percent performance boost over 14nm FinFET. The first customer products based on 7LP are expected to launch in the first half of 2018, with volume production ramping in the second half of 2018. Initial production will utilize an optical lithography approach, but a migration to EUV lithography is planned.

...the first customer products based on 7LP are expected to launch in the first half of 2018, with volume production ramping in the second half of 2018. In September 2016, GF announced plans to develop its own 7nm FinFET technology leveraging the company’s unmatched heritage of manufacturing high-performance chips. Thanks to additional improvements at both the transistor and process levels, the 7LP technology is exceeding initial performance targets and expected to deliver greater than 40 percent more processing power and twice the area scaling than the previous 14nm FinFET technology. The technology is now ready for customer designs at the company’s leading-edge Fab 8 facility in Saratoga County, N.Y.
 
It sounds great but I still haven't seen Vega yet! But, glad to know GloFlo has some tricks up it's sleeve to hopefully keep the chip race competitive.
 
They lost me when they threw in the final statement at the end about migrating to EUV Lithography. EUV has been held out as the holy grail of chip fabrication for a so long now that it feels like the Duke Nukem Forever of fab techniques.

The commercial fab industry revolves around wafers per hour. EUV hasn't been about to get to the magic wafers per hour target the fabs want for it to start being viable for them. Available EUV equipment just isn't currently powerful enough to hit those production numbers. I think I recall reading the general consensus of fab companies has said they would require availability of a device in the 220w power source range, and the industry leader in EUV tech is only up to 150w. You can make things with EUV now but not very quickly and you still have a couple problems with physics that may never be resolved. The features you're making with EUV are so tiny, you actually get roughness along the edge of the cuts due to a phenomenon we don't fully understand. So if we can't control matter in a way to prevent the edge roughness, you then have to try to explain how you can continue to shrink the lithography node when those rough edges stay the same size regardless of the cut size. A rough edge starts looking for like a mountain range in comparison as the node shrinks further.

The only really viable way to keep going was X-ray, but it's development has been on ice because the patent holders couldn't come to terms on monetary funding of further development and ownership of the IP resulting from collaboration. EUV is quickly approaching the same cost and size of equipment concerns that prevented more involvement and development of x-ray in the 90s. Put your big girl panties on and forget EUV, our future way forward got left back in the 1990s.

If they had ended their announcement with a line about having taped out their Mr Fusion device and having recently acquired a gently used Delorean from Craigslist, it would have sounded like far less bullshit and shenanigans with node naming.
 
EUV is what they are using along with Samsung and IBM, obviously it has progressed to a point where they think it's viable and since 5nm is the limit we can possibly shrink to with known technology were kind of stuck anyway. As we get smaller in chip size the costs are going to go up because it gets massively harder to shrink things.
 
Gideon said:
EUV is what they are using along with Samsung and IBM, obviously it has progressed to a point where they think it's viable and since 5nm is the limit we can possibly shrink to with known technology were kind of stuck anyway. As we get smaller in chip size the costs are going to go up because it gets massively harder to shrink things.
Click to expand...

I've seen how some of them have announced their intent to proceed with EUV, but it just doesn't appear to be as imminent as they've been trying to make it sound. I understand there's been a lot of money spent on developing EUV and they'll want to recoup their investment so it enjoys more support. On a technical level it seems to be a half ass substitute for a better way forward though. The resolutions xray is now capable of (0.01nm) seem like the type of advance that causes tech to take a huge step forward. Imagine if we could stop talking about fab nodes in terms of nm entirely
 
Gman1979 said:
I've seen how some of them have announced their intent to proceed with EUV, but it just doesn't appear to be as imminent as they've been trying to make it sound. I understand there's been a lot of money spent on developing EUV and they'll want to recoup their investment so it enjoys more support. On a technical level it seems to be a half ass substitute for a better way forward though. The resolutions xray is now capable of (0.01nm) seem like the type of advance that causes tech to take a huge step forward. Imagine if we could stop talking about fab nodes in terms of nm entirely
Click to expand...

Well unless I remember wrong it's impossible for a gate to work still past 5nm. So while resolution xray may be better it seems the manufactures have chosen their way forward, patents are a nasty mess. If it was just Global Foundries going this path I might agree with you but IBM and Samsung also plan to use it so i think the kinks are worked out or are very nearly there. They might be putting the cart before the horse, sometimes engineers get ahead of themselves tho. I think Quantum processors will be the way forward once we hit 5nm, guess will find out.
 
You must log in or register to reply here.
Back
Top