Originally posted by: RiverRicer
So confused! Everything I've read at various websites has indicated that 45nm will bring low-K interconnects. Now the AMD press release states it will be used in 'later 45nm' parts. This actually contradicts AMD's press kit which indicates low-K for 45nm, with high--K/metal gates coming later (probably sometime in 2009). So confused!
Well there is "low-k" and then there is "ultra low-k". That might be what is confusing folks.
AMD currently uses "low-k", blackdiamond-1 with k-value = 3.0, in their 65nm chips and initial 45nm production.
It's called "low-k" because the k-value is lower than the original dielectric used for decades (silicon dioxide, aka glass, aka sand, which has a k-value of 4.2).
AMD has stated they intend to eventually transition their 45nm to "ultra low-k", which will be blackdiamond-2 with a k-value = 2.4.
Shipping 45nm with low-k dielectric is expected, it's what Intel does (albeit they use ASM's Aurora dielectric material) as well as all the foundries (AMAT's BlackDiamond-1) and Texas Instruments (Novellus's Coral).
If AMD makes good on their stated intentions and does indeed ship 45nm parts with ultra low-k (called ULK for short) at some point in 2009 then they will have the technological advantage in terms of their copper interconnect wiring.
