SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Yousef who wrote (35121)7/26/1998 10:09:00 PM
From: kash johal  Read Replies (2) | Respond to of 1572630
 
Yousef,

I think you are spot on with this.

In Very deep sub-micron the loading capacitance is by far the limiting factor on overall chip performance. The use of low K dielectrics will be much more relevant at 0.18, 0.13 and even 0.1 than Cu.

One can solve the electromigration problems with thicker power/gnd rails etc.

So in 0.18 micron or even 0.13 one would rather have low K dielectrics than Cu if that was the tradeoff.

Ideally you want both, but Intels approach makes a lot more sense to me than the rush to Cu. It seems to be logical and methodical.

Having said that, I suspect IBM at least has the low K dielectric program under way just like Intel.



To: Yousef who wrote (35121)7/26/1998 10:09:00 PM
From: Bill Jackson  Read Replies (1) | Respond to of 1572630
 
Yousef, Why not optimise all aspects at the same time? Does it get out of hand with too many process variables being fiddled with? I would expect them to play with each separately at the same time and then attempt to find an optimal blend

Bill



To: Yousef who wrote (35121)7/26/1998 10:28:00 PM
From: Maxwell  Read Replies (6) | Respond to of 1572630
 
Yousef:

Do you know why I like to reply your posts so much? The reason is that you are supposed to know more process technology than me. You have a PhD and go to many conferences and symposiums. I on the other hand never go to those geek fests. Yet I understand and know more than you. Your posts have demonstrated a lack of understand in the technology. I realize that many people you have to talk to them many times before it sinks it. Let me repeat.

<< I see you are still HYPE'ing the Copper damascene process. >>

1) Copper process reduces backend process by 40%.
...This improves cycle time and hence reduces manufacturing time.
...Reduce manufacturing time reduces cost.
...Reduce cost improves profit margin.
...Companies are in the business of making money.
...Improve margin means more money in the pocket.
...Understand?

2) Copper conductivity is about 50% better than Al.
...Ohm Law:...Power=R*I*I
...Lower power means lower power dissipation
...Power=C*f*V^2 for a CPU. Lower power means you can go to higher frequency without burning the CPU and can be put in notebooks.
...Lower resistance means you can scale the die size down further since the metal trace resistance is now 50% less.

3) Low K is a breeze to integrate to copper to lower capacitance.
...Fluorinated HDP (K~2.5-3) is cakewalk with Cu damascene.
...AMD will have both Low R and Low K
...Low cost and superior process is a COMPETITIVE ADVANTAGE.
...Two is better than one.

4) Many companies opted for low K because it is EASIER than Cu.
...To go for Cu, companies have to resolve issues which you discussed.
...That means big R&D budget.
...AMD got that problem SQUARED away.

5) Cu has no electromigration problem.
...I will let you figure this one out.

Conclusion:

Intel will go for Cu sooner or later. They will SPEND BIG BUCKS on R&D to get that process MANUFACTURABLE. AMD will ENJOY YEARS of LOW COST MANUFACTURING CPUs while Intel engineers battle it out when to attack the COPPER MONSTER.

Maxwell


HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2025 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News