Bell Labs Plants Nanograss to Cool Mobile Chips

LoadWB writes "TechWeb has an article about Bell Labs' new liquid cooling technology for mobile processors. The tech, called 'nanograss' is described as 'tiny tubes that spray liquid on chip hot spots.' The use of this cooling technology reduces the power required to actively remove heat from mobile processors. Other applications are possible, but it seems it was primarily developed for use with mobile CPUs."

Comments

Re:But doesn't that mean ...

[prat393]

No, it's collected somewhere, cooled, and reused. It's a closed system; the article compares it to a car radiator.

Re:But doesn't that mean ...

[NeoThermic]

'the article compares it to a car radiator.'

And how many times have you re-filled (or your garage during an MOT) your radiator fluid?

I suspect that there might be a need to refill it after a long time...

NeoThermic

usually called MEMS fluidics

[Anonymous Coward]

This was called MEMS fluidics before, and is being researched at many places for at least 2-3 years now. The term nanograss is just a buzz word to take advantage of the term "nano". If the channels are 100s of microns, how can they be "nano"? Its obviously a marketing term. Lots of research is going on in MEMS fluidics, specially for molecular biology and diagnosis. Thinsg like lab-on-a-chip, etc.

fans power-hungry coolant systems? NOT

[Avian visitor]

Chip-generated heat, for example, can cause blade servers in densely packed racks to overheat, and can suck up notebook battery juice by requiring power-hungry coolant systems.

Your average CPU fan will consume approximately 0.5W of power. It is cheap, relatively reliable and it works.

This fan alone would run for about 70 hours on a 3000mAh battery. Compare that to the average battery life time of a modern laptop and you will see that fans are not exactly power-hungry coolant systems.

Why replace something that works well? I believe we need more efficient chips, not better cooling technology.

A word from technology inventor

[nanograss]

Guys, certainly a great pleasure to see so much interest in our technology.

Unfortunately, the TechWeb article is not that accurate. In particular, the statement that "nanograss" consists of tiny tubes that can spray liquid on chip hot spots is totally off mark.

What we call "nanograss" is a carpet of tiny nanocolumns (or nanoposts, but not tubes) each several hundred nanometers in diameter that cover the surface of say microchannel. The posts are treated with water repellant polymer coating and thus are not easy to wet. As the result the cooling liquid (such as water) can't penetrate inside this carpet and stays suspended on the tips of the nanoposts. Thus, flow of a liquid in a microchannel that has walls covered with the nanograss requires much less pressure head than in a regular channel. The liquid literally slides along the walls without touching them suspended by a tiny layer of air as in air hockey table.

Now, the trick is that we can intentionally design the nanograss such, that it can hold the liquid suspended on nanoposts only at the temperatures below a certain predetermined threshold. If the temperature exceeds this threshold the liquid sags through the nanograss and gets into direct contact with the wall. Needless to say that in this case thermal transfer from the wall to the liquid is greatly enhanced; the thin layer of air that isolates the wall from the cooling liquid is now gone. Thus the microchannels with the coolant that are located above the hottest areas on the chip (so-called hot spots) will have coolant penetrating through the nanograss and thus provide much better cooling exactly where the hot spots are. The system is self-adjusting and would automatically adapt to any arrangement of the hot spots. The obvious applications are in CPU and GPU cooling, as well as in telecom power electronics.

In addition to the application in cooling, there are multiple applications in other areas, ranging from electrical nano-batteries and biochem lab-on-a-chip devices to seagoing vessels. Indeed, wherever we have liquids we also have solid surfaces that contact them; thus you can think of a countless nanograss applications out there.

For those of you who are interested in further details the link to the Bell Labs press release is

http://www.newstream.com/cgi-bin/display_story.cgi ?12664

Also, the work will be published in May, 11 issue of Langmuir.

Best Regards, Tom Krupenkin

Re:A word from technology inventor

[nanograss]

There are two reasons. First of all, in order to achieve the most effective cooling at the minimal coolant flow rate what you really want is to keep the whole surface of your chip at a constant predetermined temperature (which is mostly defined by the nature of the chip and the temperature of the surrounding environment). The heat flux that you can transfer through the wall to the fluid is directly proportional to the temperature difference between the chip surface and the cooling liquid, with the coefficient of proportionality called heat transfer coefficient (this is sometimes referred as a Newton's Law of Cooling). The heat flux is distributed over the chip very unevenly, with the hot spots providing much higher flux than the rest of the chip. Thus, in order to keep the chip temperature constant over the whole area you need a way to adjust the heat transfer coefficient depending on the particular location on the chip surface. Nanograss gives you exactly this ability, as I described in my earlier message.

Secondly, nanograss provides you ability to strongly reduce the pressure drop required to push the liquid through the microchannels in your cooling system. Thus, you can use much smaller pump to push the coolant around.

In the existing systems no optimization like the one described above is attempted. As the result one ends up substantially overdesigning the cooling system in order to keep the chip (and especially the hot spots) cold enough.

Hope this helps.

Tom

Re:Where does the energy go?

[brokenbeaker]

There are already systems for moving miniscule amounts of liquids, for example for doing liquid chromatography (scroll down to "Environmental Sensors and Subsystems"). You could probably set up a system where the liquid circulates to a cooling area under the heating/cooling driving force. The advantage of this system would be to have this working fluid in much more intimate contact with the hot areas.

"I could imagine any microscopic particles in the fluid would eventually clog it up."

Given the standards for cleanliness in the IC industry, i'm sure this will not be a problem.

"For instance, the chemicals released as the processor ages"

Most metals and oxides are highly insoluble in most solvents.

"With a solid-state solution like this, you'd be dead in the water."

Since there is a liquid involved, this is not a solid-state solution - it's hard to get solids to circulate to draw off heat. One of the advantages of micromechanical systems is their reliability. Odds are the circulation system will not fail before the chip becomes obselete.