|Re: Continuous memory|
Message #4 Posted by Glynn on 10 May 2001, 10:59 p.m.,
in response to message #3 by Tom (UK)
I read an article recently touting IBM's newest semiconductor advance-- silicon on insulator. Lots less leakage, therefore much more frugal with power "to run tomorrow's sophisticated lineup of handheld internet appliances"...
Seems they figured out a way of using silicon oxide-- itself an insulator-- as a physical interface to a glass chip substrate, so they can arrange the electron pathways in a more contained way, less stray fields, and so you can get better circuit densities and smaller features that work as well as the standard cmos processes.
But something was tugging at me while I was reading that... and I remembered that HP had been experimenting with silicon-oxide barriers in a modified cmos process, many years before. And they were doing it for one reason: saving microwatts.
I'm pretty sure that was their rationale; to isolate each gate so it didn't have the "leakdown" of typical cmos. Believe it was a bit LESS dense and more prone to manufacturing glitches than typical cmos; certainly it involved many more steps in fabrication, which meant it was more expensive and yields would naturally be lower without extraordinary care taken to assure good results. But HP at that time was one of the few companies that would go out on a limb for a new technology and MAKE it work. I believe the results were incorporated into the Saturn cpus.
Now, of course, IBM hasn't just reinvented the wheel, by any means; their contribution-- to lay a consistent uniform film of oxide down (in a vapor-deposition process) so that silicon and glass "stick" and meld where they wouldn't before... that's really cool, and totally NEW. And it calls for a sort of "inverted" topology of cmos gate structures too. We can look forward to lots of battery-powered fun in the future, stuff that runs on a AA or two and outruns your current Pentium-class coffee-warmers.
But it IS kind of a reminder of the trailblazer HP has been over the years, that they were grappling-- and solving-- problems for themselves (like power consumption) A QUARTER OF A CENTURY AGO that still hold significant technical promise today.
May HP/Agilent continue that legacy always.