IBM Dipping Chips In 'Ionic Liquid' To Save Power

Nerval's Lobster writes "IBM announced this week that it has developed a way to manufacture both logic and memory that relies on a small drop of 'ionic liquid' to flip oxides back and forth between an insulating and conductive state without the need to constantly draw power. In theory, that means both memory and logic built using those techniques could dramatically save power. IBM described the advance in the journal Science, and also published a summary of its results to its Website. The central idea is to eliminate as much power as possible as it moves through a semiconductor. IBM's solution is to use a bit of 'ionic liquid' to flip the state. IBM researchers applied a positively charged ionic liquid electrolyte to an insulating oxide material — vanadium dioxide — and successfully converted the material to a metallic state. The material held its metallic state until a negatively charged ionic liquid electrolyte was applied in order to convert it back to its original, insulating state. A loose analogy would be to compare IBM's technology to the sort of electronic ink used in the black-and-white versions of the Kindle and other e-readers. There, an electrical charge can be applied to the tiny microcapsules that contain the 'ink,' hiding or displaying them to render a page of text. Like IBM's solution, the e-ink doesn't require a constant charge; power only needs to be applied to re-render or 'flip' the page. In any event, IBM's technique could conceivably be applied to both mobile devices as well as power-hungry data centers."

Comments

The Cost of the Liquid?

[ExploHD]

For an ionic liquid there should be no charge.

Re:The Cost of the Liquid?

[vuo]

Joking aside, ILs are expensive, because they're yet nothing but custom-manufactured small-batch chemicals for research. We're talking about 500-1200 €/kg for low grade (2-5% impurities). For high purity, you need very deep pockets, since producing pure ILs is not routine and may need expensive custom synthesis and research. If production is scaled up, though, then we're in the normal custom manufacturing range, order of magnitude being 10-100 €/kg. In this case, though, I think the price of the IL is not going to be a problem, simply because the amount needed is so small.

Personally, I think that what kills this eventually is the inability to control the degradation of the IL or the memory itself, and accumulation of harmful degradation products. Since this is a chip that you'll package and seal in, you wouldn't want to do an "oil change" now and then.

Re:The Cost of the Liquid?

[vuo]

And just to clarify, their IL was 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, or better known as [hmim][NTf2], fairly nonexotic as far as ILs go. Although I didn't find the price for this, the butyl version (whose synthesis is very similar) goes for 1150 €/kg at SigmaAldrich.

But...

[Anonymous Coward]

... does it have electrolytes?

Re:But...

[Anonymous Coward]

Yes, It's got what plants crave...

Good news...

[hcs_$reboot]

but there have been so many "IBM new revolutionary technologies" during the (recent) past years nobody has even been able to see in actual life, let's hope this one makes it up to the shops in a reasonably near future...

3 questions

[Alain Williams]

The really important things that were not mentioned are:

• * How fast is it -- to store/retrieve a bit
• * What is its life -- how many times can I do it before it breaks
• * How much heat does it generate ?

Re:3 questions

[rriegs]

While neither of the articles talk about the speed of the process, comparing it to e-ink makes it sound like it may take around 1 second before the conductor/insulator becomes reliable. If so, this may be hundreds of times slower than even an HDD and many millions of times slower than conventional RAM. That being said, after the conductor/insulator is formed, I imagine that read speeds would be similar to conventional RAM, as "reading" is still basically the same: testing whether current passes through a given bit of circuitry. This assumes, however, that the formed conductor/insulator is as reliable as a conventional transistor; if not, it may take a few cycles of "does current pass through?" before a definitive 1 or 0 is read.

Even if this is the case, there would still be huge application for this kind of technology in certain areas, such as massive, infrequently changing data centers. Of course not all of the memory/storage on the device would be replaced with this kind of non-volatile memory. Just as we currently have L1, L2, and L3 caches, main memory, and HDDs (and beyond that tapes!), I expect we'll see this find its niche.

Re:A good HOST file reduces power usage by ~37%

[cpicon92]

I like how this post seems to just sum up every Slashdot comment ever without actually saying anything.

Re:Garbled to hell

[Zontar The Mindless]

That's a pretty garbled summary.

Oh, absolutely, that it's actually marginally intelligible is rather worrisome.

I suspect that one of our Slashdot editors has been secretly replaced with Folger's Crystals.

Re:A good HOST file reduces power usage by ~37%

[Zontar The Mindless]

I like how this post seems to just sum up every Slashdot comment, ever.

Another phase change device

[Required Snark]

This is a phase change device http://en.wikipedia.org/wiki/Phase-change_memory

The track record of this type of technology has not been good. Ovonic device have never seen any significant deployment. They change state via an electric pulse that heats a cell and causes it to change between a crystalline and amorphous form. The cell holds the state without power consumption, and reading the value requires very little power.

The HP memristor is similar. The energy pulse moves oxygen ions in titanium dioxide which changes the conduction properties of the TiO2, which is a semiconductor. This has not hit the market so far either.

The IBM ionic liquid is even farther away from deployment. All they've shown is a phase change. They haven't even figured out how to do logic or non-volatile memory. It's interesting research, but nothing more so far.