Jon Stokes at Ars writes yesterday about the latest announcement from IBM, the use of air as an insulator in its 32nm node. He also comments on the IBM, Intel "tit-for-tat press release exchanges" on process technology. link
Today he weighs in on AMD's Barcelona and upcoming 45nm node partnered with IBM. link
Its to give the impression of open space, I dont think he meant literally.
I didnt see direct quotes around the entire sentence either so it seems to be the authors interpretation of what the IBM guy said and not a direct translation, which is apparent from the first three words of the sentence.
stimulating crystallization the air pressure or lack of such may change the orientation, density or even rate of crystallization.
so buy controlling temperature and pressure you can change crystal shapes and sizes - basic technology to grow fake rocks - cubic zirconia or a diamond or a Bucky ball
the key here is to make crow around the wires and to produce the low k-factor needed to stop leakage.
the air gap may actually pull the crystals toward the direction needed such as the wires are up and down the air gap sucks air up and that causes crstaylation in the z axis - while limiting the x and y axis. just like coating a wire with plastic!
For the purpose at hand, a dielectric material, it doesn't matter if the gaps contain air or a vacuum. The difference in K is negligible. Both are pretty much 1, which is the lowest that K can get in a normal environment.
While everyone is going ga-ga over this wondrous new technology of IBM's, I would like to mention a structure called an air bridge is used in the more exotic technologies like GaAs. I would expect that there are a lot of companies that have something similar to what IBM is publicizing, as usual.