IBM's next gen flash storage to feature spinning electrons

You don't usually think of IBM in the context of flash memory innovation, but all those processors Big Blue cranks out require some kind of cache , and the company's new joint venture with TDK is designed to create the next-gen of flash, using a technology called "spin torque transfer" that will allow scaling beyond 65nm. STT-RAM, as it's called, uses a current to "spin-polarize" electrons and align their magnetic fields to represent 1s and 0s. Intel and others have invested heavily in a rival next-gen tech called phase change, but IBM says STT is faster and may last longer. Obscure system-on-a-chip next-gen flash memory format war, here we come!

[alex @ Aug 21st 2007 12:15AM]

This is soooo 1960's when computers had magnetic memory cores.

[Tis---strange @ Aug 21st 2007 12:33AM]

No, this is so 2050, when all the processors will be quantumbased as well (maybe 2150...). For this is what it means: A quantum theorie based Flash...

[Jagannath A @ Aug 21st 2007 1:12AM]

You sir are an idiot.

[Bruce P. Henry @ Aug 21st 2007 11:14AM]

You sir are a genius! Indeed I have a desk lamp made of old magnetic core memory sitting on my desk right now. Spin as ones and zeros. I guess all old styles eventually DO come around again! :-)

[Ryhan @ Aug 21st 2007 12:32AM]

My 8-year old would say; "who's IBM"
My 13-year old would say; "You're old dad"
My wife would command me to call it Lenovo and chuck it out of the house out of fear of lead-contaminated nano-particles.

[Matt @ Aug 21st 2007 10:44AM]

And your wife would be incorrect. Lenovo only purchased the PC/notebook/server hardware business from IBM. Big Blue proper still conducts a great deal of research into new technologies that have nothing to do with Lenovo's business.

[Daniel @ Aug 21st 2007 12:38AM]

whichever one is cheaper will win, unless one is capable of storing a significant amount more.

[zorkon65 @ Aug 21st 2007 1:43PM]

so does this mean that flash chips goin bad after extended read/write usage will be over?

[tehpwnmstr @ Aug 21st 2007 1:32AM]

what causes problems when you go smaller than 65nm?

[Matt @ Aug 21st 2007 2:06AM]

>what causes problems when you go smaller than 65nm?

Ants seem like giant monsters, you can get stepped on, etc.

[Jack @ Aug 21st 2007 9:03AM]

"what causes problems when you go smaller than 65nm?"

Ever see "Inner Space?"

[nick @ Oct 19th 2007 3:22AM]

Ok, so this is pretty late, but I pulled this article up to have a laugh again and I saw that someone had a legitimate question which was answered with lame jokes.

First of all, Samsung has been on 45nm for a while, and Intel 45nm Penryn chips are at fab now.

But this is what happened when people tried to go below 65nm:
Dielectric breakdown -> Ionization of the Silicon -> Ruined Device

That probably didn't make any sense, so I'll do my best to explain.

Start by looking at this:
http://en.wikipedia.org/wiki/Image:Lateral_mosfet.svg

That is a MOSFET transistor, which is what most flash would be made from. The easiest way to describe this is to look at it as a faucet handle for electrons. Let's say you the left N+ area is the municipal water supply and the right hand N+ area is the spout. Electrons can't travel across a P region, so you turn the faucet handle by applying a voltage to the "Gate". This basically creates a path for the electrons to cross over, and they start coming out the other side (it would take a little while longer than this to actually explain what is happening, so I apologize if that explanation wasn't sufficient).

When you move past 65nm, one of the biggest problems is with the "oxide" between the gate and the P region. As you decrease the channel length (L in the diagram, this is what you're measuring when you say "65nm") you have to make the oxide thinner (and lower the gate voltage) to make the device operate. At 65nm, the oxide thickness was ~10s of atoms, so there wasn't really any where else to go to make it thinner. That was the problem.

What people ended up doing was changing the "oxide" material (see: high-k dielectrics) But since the old dielectric (SiO2) was one the reasons why we started using Silicon in the first place (it "grows" naturally on bare Si, so it's free and easy), it took people a long time to develop this and to retool the manufacturing to accommodate this new material.

[DarkAardvark @ Aug 21st 2007 2:22AM]

will it beat out their millipede project? i was really looking forward to that 1TB/in^2....................

[Wwhat @ Aug 21st 2007 4:00AM]

Hmm, magnetic spin? isn't that a bit sensitive to corruption by environmental influences like heat and (electro-)magnetic fields? guess not so suited for portable flash then?

[mattclarkie @ Aug 21st 2007 10:16AM]

I read about this in Scientific American back in 2003. They had an article about how all computers would use electron-spin, the direction of spin representing the binary value.

I am suprised it has taken this long to be picked up by a major company.

[stephen @ Aug 21st 2007 12:59PM]

In this case a format war is nice for the consumer. With Blu-ray vs. HD-DVD the consumer cannot do anything out of fear of supporting the losing camp.

With this type of format war, you can do either one and it doesn't effect the consumer experience much either way.

[DarkAardvark @ Aug 21st 2007 5:10PM]

LOL about the blu ray + dvd thing... heavenly swords has 10 gigs of music in it. thats a major slap in the face for HD DvD

[nick @ Aug 21st 2007 10:01PM]

Am I the only one who finds the title for this post hilarious? It'll feature spinning electrons? Really?

Here, I saved you the trouble of writing your next headline:

"Apple's newest iPod has mass, occupies finite space"