Lithium Air Batteries Get Boost From IBM and DOE

coondoggie writes "The Department of Energy and IBM are serious about developing controversial lithium air batteries capable of powering a car for 500 miles on a single charge – a huge increase over current plug-in batteries that have a range of about 40 to 100 miles, the DOE said. The agency said 24 million hours of supercomputing time out of a total of 1.6 billion available hours at Argonne and Oak Ridge National Laboratories will be used by IBM and a team of researchers from those labs and Vanderbilt University to design new materials required for a lithium air battery."

Comments

Re:Hopefully not vaporware.

[Rei]

Lithium-air is, IMHO, one of the least promising upcoming battery techs. It's really more like a fuel cell, and to be blunt, fuel cells suck. By that, I mean:

  * Expensive per watt
  * Short lifespans
  * Inefficient

There are many, many promising next-gen battery techs other than li-air. Here's just a couple of my favorites.

Lithium-sulfur: This has long been worked on, but only just recently one of its big problems has been worked around. It offers great energy density, but some of the intermediary reaction products -- various lithium polysulfides -- are rather soluble. They'd migrate across the membrane and precipitate out on the other side, being rendered permanently useless to the reaction and thus aging the cells very quickly. Older solutions to try to prevent this caused dramatically lower energy density. The latest technique involves wicking the sulfur into the pores of mesoporous carbon and then functionalizing the outside of the carbon with polyethylene glycol to keep the hydrophobic polysulfides inside when they form. The longevity improvements were amazing, without sacrificing energy density. We're talking that when they deliberately chose a worst-case solvent, one that's really good at dissolving polysulfides, the traditional Li-S cell lost 96% of its sulfur in 30 cycles while theirs only lost 25%.

Nickel-lithium: It is, quite literally, a hybrid NiMH/li-ion battery -- a traditional NiMH cathode that can hold a tremendous amount of lithium, and a lithium metal anode (almost obscene anode energy density). That's normally impossible, since you want to run a NiMH battery with an aqueous electrolyte and your various lithium-based cells with an organic electrolyte. They do both -- they use a new tech called a LISICON membrane to keep the two different electrolytes apart but allow lithium ions across. An additional problem with li metal anodes is that dendrites tend to form that rupture the membrane -- but LISICON membranes are a rigid ceramic that resists dendrite damage.

Digital quantum battery: This is my favorite, because it comes straight out of left field. It's really a type of capacitor. Now, capacitors normally hold a lot less energy than batteries; if the voltage gets too high, you get dielectric breakdown, it arcs across, and your energy is lost. But at very tiny scales, current must move as quanta. So if instead of a single big capacitor, you lithographically print an array of nanoscale capacitors, all of the sudden you can make it so that you essentially can't get dielectric breakdown. In fact, you can store so much energy that the stresses become so great that it's best to use a carbon nanotube for one of the electrodes in each nano-capacitor. :)

And even ignoring next-gen battery techs, there is still *huge* range for improvement in li-ion. In particular, for the cathodes, my favorites are layered manganese cathodes which alternate long-life forms and high energy density forms of magnanese oxides to get both properties; and fluorinated metal cathodes. For the anodes, there's many kinds of tin and particularly silicon anodes out there that store nearly an order of magnitude more lithium than conventional graphite anodes. Silicon anode li-ion cells are just this month starting to hit the market. The tech has finally matured to the point where their longevity is sufficient.

Re:DOE is serious?

[iamhassi]

"And after the 1.6 billion hours, does the computer self destruct? Just curious"

Sorry I'm back and I have answers.

The Oak Ridge "Jaguar" Supercomputer is the World's Fastest, with 37,376 six-core AMD processors. That puts it at 224,256 processors, so those 24 million hours should be done in 107 hours, or a little more than 4 days.

The 1.6 billion hours comes from the here: "....computing facilities at Oak Ridge and Argonne national laboratories will employ a competitive peer review process to allocate researchers 1.6 billion processor hours in 2010." That works out to be about 297 days.