Phase Change Memory, anyone?

[jonesthewine]

found this on SA earlier, apologies if a repost:

http://www.scientificamerican....g-memory-chips-in-mass

 

[taltamir]

very cool.

 

[BTRY B 529th FA BN]

heat and then melt the info into the cell? did i read that correctly?

 

[Idontcare]

Originally posted by: BTRY B 529th FA BN
heat and then melt the info into the cell? did i read that correctly?

Works well for rewritable CD's/DVD's, works just fine for PCM. 16ns is about 1/16 as fast as I'd like it to be though for an entry speed, we'll just have to wait and see if it gets much faster without compromising reliability and price though.

 

[BTRY B 529th FA BN]

Originally posted by: Idontcare

Originally posted by: BTRY B 529th FA BN
heat and then melt the info into the cell? did i read that correctly?

Works well for rewritable CD's/DVD's, works just fine for PCM. 16ns is about 1/16 as fast as I'd like it to be though for an entry speed, we'll just have to wait and see if it gets much faster without compromising reliability and price though.

I need to re-read that article... i have a curiosity about the re-writing method; for continual re-writing...

 

[taltamir]

supposedly it is more reliable than regular flash ram, because once frozen, it stays the same essentially forever, unlike flash which has electron tunneling. Their main problem is speed and capacity. with their best units being 16ns and 512MB chips. they claim it will get much faster as you continue to decrease size, and can be decreased beyond the point where current flash tech becomes unusable.

 

[Idontcare]

Originally posted by: taltamir
the point where current flash tech becomes unusable.

I don't know, that will be a bit of a challenge. I understand why they are compelled to make the claim, every alternative memory technology has to claim it is more scalable than the existing technology it is targeted to compete with.

Fundamentally, flash could be scaled to single-electron charge wells, the only question being how small can the charge well itself be? The answer to that questions is of course already provided by nature, whereas the smallest fundamental unit of a phase-change system is going to be that of a unit cell comprising no fewer than 8 atoms.

The one that will win will be whichever is cheapest to manufacture in volume, naturally.

 

[jimhsu]

Time for spintronics then? Encode data on a per-atom level ... perhaps even more than 1 bit per atom. The question is ... how to do it?

Some totally new stuff in research: http://en.wikipedia.org/wiki/Spinplasmonics

 

[Idontcare]

Originally posted by: jimhsu
Time for spintronics then? Encode data on a per-atom level ... perhaps even more than 1 bit per atom. The question is ... how to do it?

Some totally new stuff in research: http://en.wikipedia.org/wiki/Spinplasmonics

Actually, not too get overly technical on physics bounded scaling limitations, but any component whose properties of interest (charge storage, spin manipulation, magnetic orientation, crystal structure, etc) are actually properties created by the wave function resulting from the geometric arrangement of multiple atoms is going to have fundamental scaling limits well short of reaching atomistic levels.

For example, both phase change memory as well as spinplasmonics rely on attributes of the system that owe their existence to the energy-levels created by the Brillouin zone...itself a product of the existence of enough atoms in the right geometry necessary to effect a Fundamental domain.

Now I'm not stating anything these guys don't already know, nor is there anything about this statement that negates the value of the devices they are developing, but it speaks to the fundamental scaling limits imposed by physics on these types of devices versus those devices which can be scaled to single-atom systems (e.g. spin-flipping in hydrogen or of more practical use as atom-centered spin-transition in small molecules).

Naturally we work towards whatever is the easiest to work on and to work with, even if we know beforehand that the end result will eventually be replaced by something better, something we know is plausible already.

 

[tommo123]

samsung are apparently rolling these out later in 2010 - according to the bbc anyway

http://news.bbc.co.uk/1/hi/technology/10093528.stm