New Thermal Material Provides 72% Better Cooling Than Conventional Paste (techspot.com) 43
"Researchers at the University of Texas have unveiled a new thermal interface material that could revolutionize cooling, outperforming top liquid metal solutions by up to 72% in heat dissipation," writes Slashdot reader jjslash. "This breakthrough not only improves energy efficiency but also enables higher-density data center setups, cutting cooling costs and energy usage significantly." TechSpot reports: Thanks to a mechanochemically engineered combination of the liquid metal alloy Galinstan and ceramic aluminum nitride, this thermal interface material, or TIM, outperformed the best commercial liquid metal cooling products by a staggering 56-72% in lab tests. It allowed dissipation of up to 2,760 watts of heat from just a 16 square centimeter area. The material pulls this off by bridging the gap between the theoretical heat transfer limits of these materials and what's achieved in real products. Through mechanochemistry, the liquid metal and ceramic ingredients are mixed in an extremely controlled way, creating gradient interfaces that heat can flow across much more easily.
Beyond just being better at cooling, the researchers claim that the higher performance reduces the energy needed to run cooling pumps and fans by up to 65%. It also unlocks the ability to cram more heat-generating processors into the same space without overheating issues. [...] As for how you can get your hands on the material: it's yet to make it out of the labs. The UT team has so far only tested it successfully at small scales but is now working on producing larger batches to put through real-world trials with data center partners. The material has been detailed in a paper published in the journal Nature Nanotechnology.
Beyond just being better at cooling, the researchers claim that the higher performance reduces the energy needed to run cooling pumps and fans by up to 65%. It also unlocks the ability to cram more heat-generating processors into the same space without overheating issues. [...] As for how you can get your hands on the material: it's yet to make it out of the labs. The UT team has so far only tested it successfully at small scales but is now working on producing larger batches to put through real-world trials with data center partners. The material has been detailed in a paper published in the journal Nature Nanotechnology.
Saw this elsewhere (Score:3)
Re: (Score:2)
I have a hard time understanding how a better TIM is going to reduce pump energy usage by over half.
I guess you could say that if the thermal resistance is lower then for the same core temperature you would have a higher heat sink temperature. Which would mean that delta T between the heatsink and the cooling medium could be higher so to remove the same amount of heat the flow rate could be lower assuming everything else (e.g. heatsink size) is constant. But 65% savings still seems like a lot.
Re:Saw this elsewhere (Score:5, Informative)
And if you've ever seen the ratings stamps on modern server fans, you'll see why reducing their power consumption would matter. Example: Supermicro AS-4145 2U box has a bank of eight 2U height fans pushing air in... each one of them is rated to draw over 50W of power at maximum speed. So when the server's gas pedal is to the floor (which for a quad MI-300a system is around 2500W) nearly 10% of system power dissipation is the fans.
It's even worse for 1U systems because those tiny ass little fans are grossly inefficient.
Re: (Score:2)
Re: (Score:3)
The amount of work you need to do to move heat depends on the temperature difference. In an ideal world the entire heatsink would be exactly the same temperature as the transistor junction, but it's not because of the restrictions of conduction and the thermal interfaces. This means the headsink is cooler than the transistor junction, which means we are less efficient of using air (or liquid) to cool it, necessitating more power in terms of cooling fans.
In cooling everything can be thought of as some kind o
Yes, but... (Score:1)
Can kindergartners eat it?
Re: (Score:2)
Anything is edible, once.
Re: (Score:2)
Can kindergartners eat it?
Yes.
Galinstan is an alloy of gallium, indium, and tin.
Gallium and tin are non-toxic. Indium is an irritant but not toxic.
The amount a kindergartener can lick off a CPU isn't harmful.
Galinstan is liquid at room temperature.
Pure gallium is solid at room temperature, but your hands are warm enough to melt it. I bought 100 grams on Amazon for my kids to play with. We made some molds out of wax, poured in warm gallium, and popped it into the refrigerator to solidify. Lots of fun.
Does not matter in most cases (Score:3)
Thermal interface material on semiconductors drawing a lot of power is generally thin enough that this does not matter. It matters even less when there is a heat-spreader. Well, Intel has screwed thermal paste up in the past on some CPUs, but that is an anomaly.
Cool (Score:3)
$750 per kg (Score:2)
https://unitednuclear.com/chem... [unitednuclear.com]
6.44 grams/cc.
So 155 ml per kg. Or 6.44 kg/liter.
Re: (Score:2)
Most LM compounds used in exotic PC cooling setups cost about the same. The Thermal Grizzly stuff is quite expensive (Conductonaut).
Re: (Score:2)
Still a bargain compared to printer ink!
Bargain compared to printer ink from HP (Score:2)
Also $700 per liter which is about a kg.
Re: (Score:2)
You don't need a kg for a processor, you need a gram or even less... so like $0,75 each gram for such performance is pretty good.
Where are the "Laws of Thermodynamics" guys? (Score:1)
You shout always "Laws of Thermodynamics" and are usually wrong.
Now you could shout but you do not ...
So, this "cooling method" makes it easier to cool a motherboard.
And? Where is the heat going?
Obviously into the room/hall of the data center.
It does not affect cooling costs at the slightest. For that you would need fancy set ups, like blowing a lot of "normal air" through the data center to blow the hot air out.
If you just cool the data center as usual: this tech makes no difference.
But at home in a Pc it
Re: (Score:2)
It does not affect cooling costs at the slightest.
Not actually true. With this material you might get away with a slightly higher temperature in the DC, which means better heat pumping efficiency. Obviously that is only for the CPU and hence will not matter at all.
Re: (Score:2)
Re: (Score:2)
I was answering a DC (!) statement about thermodynamics (!). Seriously, get some context.
Re: (Score:2)
Re: (Score:1)
The CPU and GPU produce the same waste heat as before.
It just gets moved faster out of the computer chassis.
Someone above pointed out, that about 10% of the power consumption of a computer is cooling. So there is a point of saving power.
My point was: you still need the same energy to cool the server room. As the same heat energy goes into that room.
Re:Where are the "Laws of Thermodynamics" guys? (Score:4, Informative)
The CPU and GPU produce the same waste heat as before.
Not necessarily true.
Higher temperatures decrease the band gap, increase leakage current, and increase heat.
Heat also increases the resistance of copper traces.
Heat begets heat.
you still need the same energy to cool the server room.
Not true. With better cooling, you can run the DC at a higher temperature. As you get closer to ambient, the energy needed to remove the heat decreases. Once you're above ambient, you don't need any cooling at all, just ventilation.
Re: (Score:2)
I finally understood why there is such a huge gap in performance per watt in the low vs the high end processors of the same model line. The top models always have a higher performance, but the difference is always much much smaller than their increased TDP would suggest. I was always curious why double the heat doesn't even come close to double the performance.
Re: (Score:1)
Higher temperatures decrease the band gap, increase leakage current, and increase heat.
That is pretty nitpicking.
Heat also increases the resistance of copper traces.
Also pretty nitpicking.
you still need the same energy to cool the server room.
Not true. With better cooling, you can run the DC at a higher temperature. As you get closer to ambient, the energy needed to remove the heat decreases. Once you're above ambient, you don't need any cooling at all, just ventilation.
It is true.
The energy converted into heat: is exactly the same. Except for your minour nit
Re: (Score:2)
The energy converted into heat: is exactly the same. Except for your minour nitpicking points.
For every 10C increase in temperature, there's about a 4% efficiency reduction (and thus 4% more heat).
Re: (Score:2)
Yes, and?
This 4% is nitpicking.
I talk about the data center cooling. So, you get with thermal paste 4% more efficient.
Good. But the claim it reduces costs for cooling "significantly" is simply wrong.
If it goes down 4%, then that is hardly relevant.
Re: (Score:2)
Wrong. On top of what Shanghai bill said cooling depends not just on thermal resistivity but also on temperature difference. By reducing or eliminating the resistance between the junction and the heatsink, the heatsink will have a higher temperature meaning that it will naturally transfer more heat between heatsink-air. I.e. you don't need to run your cooling fan as hard.
You are German, if you want to understand this better go look up how the radiator in the corner of your room works, and why its relative s
Re: (Score:2)
Are you daft?
The amount of heat moved into the data center is the same.
Does not matter how FAST you transfer it.
So cooling the data center cost the same.
Re: (Score:2)
Oh wait.
Heat flows down temperature gradients at a rate proportional to the temperature gradient and inversely proportional to the thermal conductivity. If you improve thermal conductivity in one place, you reduce the end-to-end temperature gradient required to maintain the same heat flux. Therefore you can choose a combination of increased operating power at the s
Re: (Score:2)
The energy moved to the data center is the same.
FACEPALM
Actual numbers? (Score:1)
All the claims are in terms of how much better the new material is compared to existing materials, without giving actual delta-T across a given area at a given power. No specific reference to the comparison material. No statement of the actual bulk thermal conductivity, nor a comparison of bulk thermal conductivity to copper or diamond.
Last time I looked, the thermal paste was not the most significant part of the silicon-to-air heat path.
Re: (Score:2)
Yea, I read the summary and knew off the bat that a basic graphite TIM pad can do pretty much the same thing this stuff claims, because the aluminum to air interface is the real limiter in actual heat transfer.
Re: (Score:2)
Ho hum (Score:2)
With the prevelance of SSDs and the death of physical media in most data centres, you're virtually always going to be better off with total emersion cooling.
Wait, what? (Score:2)
Did Techspot really say "The material pulls this off by bridging the gap between the theoretical heat transfer limits of these materials and what's achieved in real products"?
That literally translates to "The material achieves this breakthrough performance by being better than existing products".
Wow, such explanation.
Probably for the new HPE 224 Nvidia (Score:2)
Corrosion (Score:2)
Doesn't do anything to address one of the most important limitations of liquid metal based TIMs: corrosion due to the gallium (and to a lesser extent, the indium) content.
If it's so good ... (Score:2)